EPA
United States
Environmental
Protection
Agency
Olfice of Air Quality
Planning and Standards
Research Triangle Park
North Carolina 27711
EPA-452 \R-96-013
July 1996
Review of the National Ambient Air Quality
       Standards for Participate Matter:

     Policy Assessment of Scientific and
              Technical Information
                OAQPS Staff Paper
                                                    Botton/
                                                    Watertown,
                                                     A
                                                    •w Haven, CT
                                                  NYC, NY

                                                  lladelphla, PA
        Seattle. WA

     Taeome. WA
          Spokane, WA
     Trl-eltlea, WA
                                oronto,
                                Canada/i Alb
                                   NY
                                  Buffalo
                                  NY
                                leveland, O
                             SteubeAVlHa. OH
                                O mortality only
                                • mortality and morbidity
                                  morbidity only
                                                 Additional Locations
                                                 Athens, Greece
                                                 Barcelona. Spain
                                                 Erfurt, East Germany
                                                 London. England
                                                 Sao Paulo, Brazil
                                                 Santiago, Chile
                                                 Switzerland
                                                 Wagenmgen, Netherlands
             Office of Air Quality Planning and Standards
                U.S. Environmental Protection Agency
                 Research Triangle Park, NC 2771L ,. .
                                        u. .->. r".

-------
Cover illustration: Locations of recently published community epidemiology studies finding
statistically significant associations between shortrterm concentrations of paniculate matter and
health effects (CD, Tables 12-2 through 12-5). Studies conducted on three continents have found
both increased morbidity and mortality to be associated with a variety of particle measurement
devices, including mass measurements of TSP, PM10, PM2.5, sulfates, and acids, and optical
based approaches including BS, KM, and COH. Although the highest PM-10 concentrations in
the U.S. are in the West, most of the results in North America are from eastern communities, at
PM-10 concentrations that are generally below those permitted by the current standards.

-------
                                   DISCLAIMER
      This report has been reviewed by the Office of Air Quality Planning and Standards
(OAQPS), U. S. Environmental Protection Agency (EPA), and approved for publication.
This OAQPS Staff Paper contains the findings and conclusions of the staff of the OAQPS
and does not necessarily represent those of the EPA.  Mention of trade names or commercial
products is not intended to constitute endorsement or recommendation for use.

-------
                             ACKNOWLEDGEMENTS

      This Staff Paper is the product of the Office of Air Quality Planning and Standards
(OAQPS). For the primary standards, the principal authors include John Bachmann,
Dr. Jane C. Caldwell, Patricia D. Koman, Harvey M. Richmond, Eric G.  Smith of OAQPS,
Dr. Tracey J. Woodruff of the Office of Policy, Planning and Evaluation (OPPE), and
Dr. Karen M. Martin of OAQPS who also managed the project. For the secondary
standards,  principal authors include Rich Damberg, Chebryll Edwards,  and Bruce Polkowsky
of OAQPS.  The authors would like to acknowledge John Haines for his guidance and
expertise, and Terence Fitz-Simons, David Mintz, and Miki Wayland for providing expertise
and analysis.  Finally, the authors wish to acknowledge Tricia Crabtree for providing
substantial wordprocessing  and general support for the Staff Paper.  The Staff Paper includes
comments  from OAQPS, the Office of Research and Development, OPPE, Regional Offices,
and the Office of General Counsel within EPA.
      On  two different occasions,  draft chapters of this document were formally reviewed
by the Clean Air Scientific  Advisory Committee (CASAC).  Helpful comments and
suggestions were also submitted by a number of independent scientists,  by  officials from
several State and local air pollution organizations, by  environmental groups, by industrial
groups, and by individual companies.

-------
                                         1

                              TABLE OF CONTENTS


List of Tables  	vii

List of Figures	xiii

I.     PURPOSE	1-1

H.    BACKGROUND	II-1
      A.     Legislative Requirements	II-1
      B.     History of PM NAAQS  Reviews	II-3
             1.     Establishment of the NAAQS for Particulate Matter	  II-3
             2.     First Review of NAAQS for Particulate Matter	  II-3
             3.     Recent Litigation	II-4
             4.     Current Review of the Particulate Matter NAAQS	  II-4

DI.   APPROACH  	Ill I
      A.     Bases for Initial  Analytical Assessments	III-l
             1.     Primary Standards  	III-l
             2.     Secondary Standards	III-2
      B.     Organization of Document	III-2

IV.   AIR QUALITY: CHARACTERIZATION AND IMPLICATIONS	IV-1
      A.     Characterization  of U.S. Ambient Particulate Matter	IV-1
             1.     Multi-modal Size Distributions	IV-2
             2.     Properties of Fine and Coarse Fraction Particles	IV-4
                   a.      Sources and Formation Processes	IV-4
                   b.      Chemical  Composition, Solubility, and  Acidity  	IV-5
                   c.      Atmospheric Behavior  	IV-6
                   d.      Correlations Between PM, 5 and Coarse Fraction
                          Mass	".	IV-7
                   e.      Summary   	IV-8
      B.     PM Air Quality  Patterns	IV-8
             1.     PM Concentrations and Trends       	IV-8
                   a.      PM10 Concentrations and Trends  	IV-8
                   b.      Fine Particle Concentrations and Trends  	IV-9
                   c.      Trends in Emissions of Fine Particle Precursor
                          Gases	IV-11
             2.     Background Levels	IV-12

-------
      C.     Air Quality Implications for Interpreting Epidemiological Studies  .... IV-14
             1.     Representativeness of Central Monitor Measurements
                   of PM Exposures	IV-15
             2.     PM25 and PM10 Comparisons in Areas Relevant to the
                   Health Studies	   >'-16
      D.     Air Quality Implications for Risk Management Strategies	iV-17

V.    CRITICAL ELEMENTS IN THE REVIEW OF THE PRIMARY
      STANDARDS   	V-l
      A.     Introduction	V-l
      B.     Mechanisms	V-2
      C.     Nature of Effects  	V-8
             1.     Mortality	V-ll
                   a.     Mortality from Short-term Exposures to PM	V-ll
                          i.     Historical Findings from Community
                                Epidemiology  	V-ll
                          ii.     Recent Findings	V-12
                          iii.    Specific Causes of Mortality Associated with PM   . V-l3
                          iv.    Experimental Animal Studies	V-l4
                   b.     Mortality from Long-term Exposures to PM	V-l4
                   c.     Extent of Life Shortening   	V-18
             2.     Indices of Morbidity Associated with Respiratory  and
                   Cardiovascular Disease   	V-20
                   a.     Hospital Admissions and Emergency Department Visits  . . V-20
                   b.     School Absences, Work Loss Days, and Restricted
                          Activity Days	V-21
             3.     Altered Lung Function and Symptoms  	V-22
                   a.     Effects Related to Short-term Exposures to PM	V-22
                          i.     Community Air Pollution Studies   	V-22
                          ii.     Controlled Exposures to Laboratory Aerosols .... V-23
                   b.     Effects Related to Long-Term Exposures	V-26
             4.     Morphological Damage   	V-27
                   a.     Acid Aerosols  	V-27
                   b.     Silica, Crustal Dusts, and Other PM Components	V.-28
             5.     Effects on Host Defense Mechanisms	V-29
      D.     Sensitive Subpopulations	V-31
             1.     Individuals with Respiratory and Cardiovascular Disease	V-33
             2.     Individuals with Infections  	V-34
             3.     The Elderly   	V-34
             4.     Children  	V-35
             5.     Asthmatic Individuals  	V-35

-------
                                         Ill

      E.     Evaluation of the Epidemiological Evidence  	V-36
             1.     Interpretation of Individual PM Study Results	V-38
                   a.     Model Selection and Specification  	V-38
                   b.     Measurement Error  	V-39
                   c.     Potential Influence of other Covariates in Short-
                          Term Studies	V-43
                          i.     Weather	V-44
                          ii.     Confounding by Other Pollutants  	V-45
             2.     Consistency and Coherence of the Epidemiological Studies  .... V-54
                   a.     Consistency	V-54
                   b.     Coherence	V-56
      F.     Health Effects Associated with Fine and Coarse Fraction Particles  .... V-58
             1.     Epidemiological Studies Using Fine Particle Indicators	V-60
                   a.     Short-Term Studies  	V-60
                   b.     Long-Term Studies  	V-61
             2.     Community Studies Comparing Effects of Fine and Coarse
                   Fraction PM	V-63
                   a.     Short-Term Comparisons	V-63
                   b.     Long-Term Comparisons	V-65
             3.     Epidemiological Studies of Areas Dominated by Coarse Particles . V-67
             4.     Relevant Physicochemical Differences Between  Fine and Coarse
                   Fraction Particles	V-69
                   a.     Comparisons of Fine and Coarse Component Toxicity in
                          Laboratory Studies	V-69
                   b.     Toxicity of Fine and Coarse Mode Chemical Components . V-71
                   c.     Physical Aspects of Fine and Coarse Particles	V-73
                   d.     Deposition in Sensitive Individuals	V-75
             5.     Summary and Conclusions  	V-76

VI.   RISK ASSESSMENT	VI-1
      A.     General  Scope  	VI-2
      B.     Components of the Risk Model	VI-4
             1.     Air Quality Information	VI-4
             2.     Concentration-Response Functions	VIrll
             3.     Baseline Health Effects Incidence Rates  	VI-20
             4.     Limitations and Uncertainties	Vl-24
      C.     Risk Estimates for Philadelphia and Los Angeles  Counties  	VI-25
             1.     Base  Case Risk Estimates Associated with "As Is" PM Levels   . VI-25
                   a.     Philadelphia County	VI-26
                   b.     Los Angeles County	VI-27
                   c.     Key Uncertainties   	VI-31
             2.     Base  Case Risk Estimates Upon Attainment of Current Standards VI-31

-------
                                         IV

             3.     Uncertainty Analyses of Estimated Risks Associated with  "As Is"
                   PM Levels in Philadelphia County and Attaining Current  PM10
                   Standards in Los Angeles County	VI-35
                   a.     Sensitivity Analyses of Individual Key uncertainties  . . .  VI-35
                   b.     Integrate :  " vertainty Analysis	Vi  3
             4.     Risk Estimates Asi..^ied with Alternative PM2 5 Standards  ...  VT--+4
                   a.     Base Case Ris^ Estimates  	\ i-44
                   b.     Individual  Sensi       nalysis Concerning Air Quality
                          Rollbacks	VI-53
                   c.     Integrated  Uncertainty Analysis	VI-54
             5.     Key Observations from the Risk Analyses	VI-58

VH.  STAFF CONCLUSIONS AND RECOMMENDATIONS ON PRIMARY
      NAAQS  	VIM
      A.     Adequacy of the Current Primary Standards for Particulate Matter  . . .  VII-2
      B.     Alternative PM  Indicators and Risk Management Implications	  VII-4
             1.     PM10  as Surrogate Indicator for  Fine and Coarse Fraction
                   Particles  	        	VII-4
             2.     Alternative Surrogate i..u.^tors  for Fine and Coarse Fraction
                   Particles  	VII-11
                   a.     Surrogate Indicators for the Fine Fraction of PM10  .... VII-12
                   b.     Surrogate Indicators for the Coarse Fraction of PM10 . . . VII-16
             3.     Staff Conclusions and Recommendations for Particle Indicators . VII-17
      C.     Alternative PM2.5 Standards for Control of Fine Fraction Particles  . . . VII-18
             1.     Averaging Time	VII-18
                   a.     Short-term PM25 Standard	VII-19
                   b.     Long-term PM25 Standard	VII-20
             2.     Form — General Approaches	VII-21
                   a.     24-Hour PM2 5 Standard  	VII-21
                   b.     Annual PM2 5 Standard	VII-22
             3.     Level and Specific Forms	VII-22
                   a.     24-Hour PM2 5 Standard  	VII-24
                   b.     Annual PM2 5 Standard	VII-32
      D.     Alternative PM10 Standards for Control of Coarse Fraction Particles  . . VIIr37
             1.     Averaging Time	VII-37
             2.     Level and Form for Alternative  Averaging Times  	VII-38
                   a.     Annual PM10 Standard  	VII-38
                   b.     24-Hour PM10 Standard	VII-39
             3.     Summary of Coarse Fraction (PM1Q) Standard Conclusions and
                   Recommendations	VII-40
      E.     Summary of Key Uncertainties and Research Recommendations	VII-4.
      F.     Summary of Staff Recommendations on Primary PM NAAQS

-------
VHI. CRITICAL ELEMENTS IN THE REVIEW OF THE SECONDARY STANDARD
      FOR PARTICIPATE MATTER	VIII 1
      A.     Introduction	VIII-1
      B.     Effects of PM on Visibility 	VIII-2
             1.      Definition of Visibility and Characterization
                    of Visibility Impairment	VIII-2
             2.      Significance of Visibility to Public Welfare	VIII-3
             3.      Mechanisms of and Contributors to Visibility Impairment  ....  VIII-3
             4.      Background Levels of Light Extinction	VIII-5
                    a.     Rayleigh Scattering   	VIII-6
                    b.     Light Extinction Due to Background Paniculate Matter  .  VIII-6
             5.      Overview of Current Visibility Conditions	VIII-7
                    a.     Role of Humidity in Light Extinction  	VIII-9
                    b.     Significance of Anthropogenic Sources of Fine
                          Particles	VIII-10
                    c.     Regional Differences in Specific Pollutant
                          Concentrations	VIII-11
                    d.     Regional Variation  in Urban Visibility	VIII-11
             6.      Policy Considerations Pertaining to the Effects of PM
                    on Visibility	VIII-12
      C.     Effects of PM on Materials Damage and Soiling  	VIII-15
             1.      Materials Damage  	VIII-15
                    a.     Effects on Metals  	VIII-16
                    b.     Effects on Paint  	VIII-17
                    c.     Effects on Stone	VIII-18
                    d.     Effects on Electronics	VIII-19
             2.      Staff Considerations Pertaining to the Effects of PM on
                    Materials Damage  	VIII-19
             3.      Soiling  	VIII-19
             4.      Societal Costs	VIII-22
                    a.     Soiling/Property Value	VIII-22
                    b.     Soiling/Materials	VIII-23
             5.      Staff Considerations Pertaining to the Effects of PM on
                    Soiling  	VIIIT24
      D.     Summary of Staff Conclusions and Recommendations on Secondary
             NAAQS	VIII-24
APPENDDC A       Considerations in Selecting Particle Size Cut Point for Fine
                    Particles   	A-l
APPENDIX B       Measurement  Methods from Epidemiology Studies   	  B-1
APPENDK C       PM)0 National Concentration Maps and Definitions of Regions  . .  C-l

-------
                                         VI

APPENDIX D      I.  Hypothetical Mechanisms of Action for PM	  D-l
                   II. Extrapolation of Results From Laboratory Studies to Those
                   Of Epidemiologic Studies: Strength and Limitations Of
                   Controlled Human and Animal Stuci^     	   D
APPENDIX E      Concentration-Response Relationships for Model Sensitivity
                   Analysis in Risk Assessment	  E-
APPENDIX F      Sensitivity Analyses of Key Uncertainties in the Risk Assessment    F-l
APPENDIX G      Measures of Visibility Impairment and Light Extinction  	  G-1
APPENDIX H      Clean Air Scientific Advisory Committee Closure Letters   	  H-l

-------
                                         Vll

                                 LIST OF TABLES


Number                                                                        Page

IV-1         Particle Size Fraction Terminology Used in this Staff Paper  	IV-3a

IV-2         Comparison of Ambient Fine and Coarse Mode Particles	IV-4a

IV-3         PM,0 and PM2.5 Regional Background Levels  	IV-13

IV-4         PM25 Concentrations in Selected Cities	  IV-17a

IV-5         Summary of PM10 Non-Attainment Areas by Source Type   	IV-19

IV-6         PM10 NAAQS Implementation  Case Studies Summary	  IV-19a

V-l          Modeled 24-Hr Regional Deposition for Measured Ambient Particle
             Size Distributions  	V-4

V-2          Hypothesized Mechanisms of PM Toxicity	V-6

V-3          Estimated Mortality Increase Per 50 /zg/m3 Increase in 24-h
             PM10 Concentrations from U.S. Studies	V-13a

V-4          Comparison of Total Mortality With Age- and Cause-Specific Mortality
             for Short-Term Exposure Studies	V-13b

V-5          Relative Risk Between the Most Polluted and Least Polluted Cities
             for Total Population and Former and Current Smokers in the
             Prospective Cohort Studies  	V-14a

V-6          Estimated Increased Hospital Admissions for the Elderly Per 50 /ug/m3
             Increase in 24-h PM,0 Concentrations from U.S. and
             Canadian Studies	V-20a

V-7          Estimated Lung Function Changes and Respiratory Symptoms per
             50 /zg/m3 Increase in 24-h PM10 Concentrations from U.S.  and
             Canadian Studies	V-22a

V-8          Morbidity Effects Estimates Per Increments in Annual Mean Levels
             of Fine/Thoracic Particle Indicators from U.S. and Canadian Studies . .  V-26a

-------
                                         Vlll

                              LIST OF TABLES (Cont'd.)


Number                                                                          Page

V-9          Qualitative Summary of Recent PM  Community Epidemiologic Results
             for Short and Long-Term Exposure   	V-32

V-10        Quantitative Coherence of Acute  Mortality and Hospitalization Studies  .  V-57a

V-ll        Short-Term Exposure Epidemiological Studies Of Mortality Using
             Optical Fine Particle Indicators	V-60

V-12        Fine Particle Indicator (PM25, SO4,  H+) Effects Studies from the U.S.
             and Canada	V-60b

V-13        Effect Estimates Per Increments in Annual Mean Levels of Fine/Thoracic
             Particle Indicators from  U.S. and Canadian Studies  	  V-61a

V-14        Estimated Increase in Daily Mortality, 95% Cl, and t Statistic by City
             and Combined Estimate Associated with a 10 /ig/m3 Increase in
             Paniculate Mass Concentrations.   Effect of Each Particle Mass
             Measure Associations Estimated Separately, Controlled for Long-Term
             Trends and Weather	V-63a

VI-1         Cities Examined in  PM  Risk Analysis	VI-6

VI-2        Selected Epidemiological Studies and Associated Relative Risk
             Estimates Used in Risk  Analyses	VI-12

VI-3        Concentration-Response "Cutpoints" Examined in Uncertainty Analyses  VI-18

VI-4        Relevant Population Sizes for Philadelphia County and Southeast Los
             Angeles County  	Vl-22

VI-5        Baseline Health Effects  Incidence Rates	VI-23

VI-6        Estimated Annual Health Risks Associated with "As Is" PM
             Concentrations in Philadelphia County, September 1992-August 1993
             (for base case assumptions)  	VI-28

-------
                                          IX

                              LIST OF TABLES (Cont'd.)

Number                                                                          Page

VI-7         Estimated Annual Health Risks Associated with "As Is" PM
             Concentrations in Southeast Los Angeles County, 1995 (for base case
             assumptions)  	VI-30

VI-8         Estimated Annual Health Risks Associated with Attainment of Current
             Standards in Southeast Los Angeles County, 1995 (for base case
             assumptions)  	VI-33

VI-9         Summary of Selected  Sensitivity Analyses on Estimates of Risk
             Associated with PM in Philadelphia County  	VI-37

VI-10        Summary of Uncertainties Incorporated Into Integrated Uncertainty
             Analysis	VI-40

Vl-lla       Controlling Monitors  for Rollbacks to Attain Alternative PM-2.5
             Standards  	VI-45

Vl-llb       Controlling Standards and Percent Rollbacks Necessary to Attain
             Alternative PM2 5 Standards	VI-45

VI-12a       Estimated Changes in Health Risks Associated with Meeting Alternative
             PM-2.5 Standards in Philadelphia County,  September 1992 - August
             1993 (for base case assumptions)	VI-46

VI-12b       Estimated Changes in Health Risks Associated with Meeting Alternative
             PM-2.5 Standards in Philadelphia County,  September 1992 - August
             1993 (for base case assumptions)	VI-47

VI-13a       Estimated Changes in Health Risks Associated with Meeting Alternative
             PM-2.5 Standards in Southeast Los Angeles County, 1995 (for base
             case assumptions)  	VI-48

VI-13b       Estimated Changes in Health Risks Associated with Meeting Alternative
             PM-2.5 Standards in Southeast Los Angeles County, 1995 (for base
             case assumptions)  	VI-49

VI-14        Sensitivity Analysis: Effect of Alternative Rollback  Methods on
             Mortality Estimates Short-term Exposure (Pooled Function) and Long-
             term Exposure PM-2.5 Mortality Functions. Philadelphia County,
             September 1992-August 1993	VI-50

-------
                              LIST OF TABLES (Cont'd.)
Number

VII-1

VII-2


VII-3


VII-4

VIII-1

VIII-2


VIII-3


VIII-4


VIII-5


VIII-6



VIII-7



E-l



E-2
Percentage of Counties Not Meeting Alternative PM10 Standards   ....  Vn-8

Predicted Percentage of Counties Not Meeting Alternative PM2 5
Standards	VII-26

Predicted Comparison of Alternative Forms for a 24-Hour PM2 5
Standard	VII-30

Comparison of Alternative Forms for a 24-Hour PMK> Standard	VII-41

Comparison of Residential Visibility Valuation Study Results	  VIII-3a

Average Natural Background Levels of Aerosols and Light
Extinction	VIII-6a

Dry Particle Light Extinction Efficiency Values Used in 1996
Analysis of IMPROVE Data	VIII-9

Comparison of Total Light Extinction to Estimated Background
Light Extinction for Several Eastern and Western Locations  	  VIII-lOb

Visibility Model Results:  Anthropogenic Light Extinction
Budgets 	VIII-1 la

Percentage Contribution by Source Category to Fine Particle
(and Precursor) Emissions in the East, Southwest, and
Northwest	VIII-1 Ib
Percentage Contributions of Aerosol Constituents to Annual
Average Total Light Extinction in the Washington, DC and
Southern California Areas	
Potential Concentration Cutpoints of Interest for Assessing the
Sensitivity of Risk Estimates Derived from Short-Term Exposure
Studies 	

Relationship Between Relative Risk of Death and PM-10 in Utah
Valley	
VIII-12
   E-6a
                                                                                  E-7a

-------
                                          XI

                              LIST OF TABLES (Cont'd.)

Number                                                                          Page

E-3          Potential Concentration Outpoints of Interest for Assessing the
             Sensitivity of Risk Estimates Derived from Long-Term Exposure Studies E-lOa

F-la         Sensitivity Analysis: The Effect of Alternative Background Levels on
             Predicted Health Effects Associated with  "As-Is" PM-10 Philadelphia
             County, September 1992-August 1993	F-la

F-lb         Sensitivity Analysis: The Effect of Alternative Background Levels on
             Predicted Health Effects Associated with  "As-Is" PM-2.5 Philadelphia
             County, September 1992-August 1993	F-lb

F-2          Sensitivity Analysis: Effect of Alternative Rollback Methods on
             Predicted Health Effects of PM-2.5. Philadelphia County,
             September 1992-August 1993	F-2a

F-3a         Sensitivity Analysis: The Effect of Alternative Cutpoint Models on
             Predicted Health Effects Associated with  "As-Is" PM-10.  Slope Adjustment
             Method 1. Philadelphia County, September  1992-August 1993  	F-8a

F-3b         Sensitivity Analysis: The Effect of Alternative Cutpoint Models on
             Predicted Health Effects Associated with  "As-Is" PM-10.  Slope Adjustment
             Method 2. Philadelphia County, September  1992-August 1993  	F-8b

F-3c         Sensitivity Analysis: The Effect of Alternative Cutpoint Models on
             Predicted Health Effects Associated with  "As-Is" PM-2.5. Slope Adjustment
             Method 1. Philadelphia County, September 1992-August  1993	F-8c

F-3d         Sensitivity Analysis: The Effect of Alternative Cutpoint Models on
             Predicted Health Effects Associated with  "As-Is" PM-2.5. Slope Adjustment
             Method 2. Philadelphia County, September 1992-August  1993	Fr8d

F-3e         Sensitivity Analysis: The Effect of Differing Cutpoints on
             Estimated Mortality Associated with Long-term  Exposure to PM-2.5.
             Philadelphia County, September  1992-August 1993	F-8e

F-4          Sensitivity Analysis: Effect of Combining Different Averaging  Times
             in Pooled Short-Term Exposure Mortality Functions on Predicted Health
             Effects Associated with "As-Is" PM-10.  Philadelphia County,  September
             1992-August 1993	F-1 la

-------
                                         Xll

                             LIST OF TABLES (Cont'd.)

Number                                                                         Page

F-5a         Sensitivity Analysis: Effect of Copollutants.  Relative Risks for Change
             of 50 /xg/m3 PM-10 or 25 ^g/m3 PM-2.5 ...................  F-12a
F-5b         Sensitivity Analysis: Effect of Copollutants on Predicted Health
             Effects Associated with "As-Is" PM.  Philadelphia County, September 1992-
             August 1993  .................................... F-12b

F-6          Sensitivity Analysis: The Effect of Concentration-Response Function
             Slope on Estimated Mortality Associated with Long-term Exposure to
             PM-2.5. Philadelphia County, September 1992-August 1993  .......  F-14a

-------
                                        Xlll

                                 LIST OF FIGURES

Number                                                                      Page

IV-1         Idealized Distribution of Ambient Particulate Matter	IV-2a

IV-2         Distribution of Particles by Number, Surface Area, and Volume  .... IV-2b

IV-3         Major Constituents of PM25, Coarse Fraction, and PMIO (CD, Figures
             6-85a-c)	IV-5a

IV-4         Areas Designated Nonattainment for Particulates (PM-10)	IV-8a

IV-5a        PM-10 Trend, 1988-1993 (Annual Arithmetic Mean)	IV-9a

IV-5b        PM-10 Trend, 1988-1993 (90th Percentile)	IV-9a

IV-6         Fine Mass Concentration Derived from Nonurban IMPROVE/
             NESCAUM Networks  	IV-9b

IV-7         United States Trend Maps for the 75th Percentile Extinction
             Coefficient	IV-lOa

IV-8         Trends in Visibility and Sulfur Emissions in the Eastern U.S	  IV-lla

IV-9         Locations Where Community Epidemiology Studies Associating
             Short-Term  PM Exposure with Mortality were Conducted
             in North America  	IV-16a

V-l          Human Respiratory Tract PM Deposition  Fraction Versus Mass Median
             Aerodynamic Diameter (MMAD) with Two Different Geometric Standard
             Distributions  	V-3a

V-2          Relationship Between Relative Risk Per 50 /xg/m3 PM10 and Specific
             Causes of Mortality and Morbidity in Adults and  Children	V-55a

V-3a         Relationship Between Relative Risk of Mortality Associated with
             PM-10 and Maximum Levels of SO2, CO, NO2, and  Ozone  	V-56a

V-3b         Relationship Between Relative Risk of Mortality Associated with PM10
             and Mean Values of S02 and NO2	V-56b

-------
                                         XIV

                             LIST OF FIGURES (Cont'd.)

Number

V-4          Relative Risks of Acute Mortality in the Six City Study, for
             Thoracic Particles (PM,0, PM15), Fine Particles (PM25) and
             Coarse Fraction Particles (PM15-PM2.5)	 V-63a

V-5          Adjusted Relative Risks for Mortality are Plotted Against
             Each of Seven Long-Term Average Particle Indices in the Six
             City Study, from Largest Range (Total Suspended Particles,
             Upper Right) Through Sulfate and Nonsulfate Fine Particle
             Concentrations (Lower Left)	V-65a

V-6          Age-Sex-Race Adjusted Mortality Rates from ACS Study Plotted
             Against Mean Sulfate and TSP Levels for U.S. Metropolitan
             Areas (Pope, 1995)	V-66a

V-7          % of Children with <85% Normal FVC vs. Annual Fine
             and  Coarse Fraction  Mass in 24 City Study  	V-67a

VI-1         Major Components of Paniculate Matter Health Risk Analysis	VI-5

VI-2         Daily Average PM Concentration Frequencies Philadelphia County,
             September 1992 - August 1993	VI-8

VI-3         Daily Average PM-10 Concentrations for Southeast Los Angeles
             County,  1995	VI-9

VI-4         Daily Average PM-2.5 Concentration Frequencies  for Southeast Los
             Angeles County, 1995   	VI-10

VI-5         Schematic Representation of Alternative Interpretations of Reported
             Epidemiologic Relative Risk (RR) Findings with Regard to Possible
             Underlying PM Mortality Concentration-Response  Functions (CD,
             Figure 13-5)  	VI-17

VI-6         Slope Adjustment Methods Used in Sensitivity and Uncertainty
             Analyses	VI-19

VI-7         Effect of Several Uncertainties on Mortality Risk Associated With
             Short-Term Exposure to  PM-2.5 in Philadelphia County   	VI-41

-------
                                         XV

                             LIST OF FIGURES (Cont'd.)

Number                                                                        Page
VI-8         Effect of Several Uncertainties on Mortality Risk Associated With
             Short-Term Exposure to PM-2.5 After Meeting Current PM-10
             Standards in Los Angeles County  	VI-43

VI-9         Effect of Several Uncertainties on Reductions in Mortality Risk
             Associated With Short-Term Exposure to PM-2.5 Upon Attaining PM-
             2.5 Standards of 15 /*g/m3 Annual and 50 /*g/m3 Daily in Los Angeles
             County	VI-55

VI-10        Effect of Several Uncertainties on Reductions in Mortality Risk
             Associated With Short-Term Exposure to PM-2.5 Upon Meeting
             Alternative PM-2.5 Standards in Los Angeles County	VI-57

VIII-1        Average Light Extinction Coefficient (in Mnr1) for Each of the
             Reported Sites in the IMPROVE Network, 1992-1995	VIII-7a

VIII-2        Annual Average Visibility Impairment in Deciviews Calculated
             from Total Light Extinction (Rayleigh included), IMPROVE Network,
             1992-1995	VIII-7a

VIII-3        Average PM2 5 Mass Concentration (in /^g/m3) for Each Site in the
             IMPROVE Network,  1992-1995  	VIII-8a

VIII-4        Average PM10 Mass Concentration (in ^g/m3) for Each Site in the
             IMPROVE Network,  1992-1995  	VIII-8a

VIII-5        Fine Mass as a Percent of PM10 for Each Site in the IMPROVE
             Network,  1992-1995   	VIII-8b

VIII-6a      Average Winter Visibility Impairment in Deciviews Calculated
             from Total Light Extinction (Rayleigh included), IMPROVE Network,
             1992-1995	VIII-9a

VIII-6b      Average Spring  Visibility Impairment in Deciviews Calculated from
             Total Light Extinction (Rayleigh included), IMPROVE Network,
             1992-1995	VIII-9a

VIII-7a      Average Summer Visibility Impairment in Deciviews Calculated from
             Total Light Extinction (Rayleigh included), IMPROVE Network,
             1992-1995	VIII-9b

-------
                                         XVI

                             LIST OF FIGURES (Cont'd.)

Number                                                                         Page

VIII-7b      Average Autumn Visibility Impairment in Deciviews Calculated from
             Total Light Extinction (Rayleigh included),  IMPROVE Network,
             1992-1995	 . VIII-9b

VIII-8       Spatial Variation in Average Relative Humidity and the Sulfate
             Humidity Correction Factor	 VIII-lOa

VIII-9       Perceptible Change in Visibility as a Function of Fine Mass
             Concentration	VIII-lOc

C-l          PM-10 Air Quality Concentrations, 1992-94	C-2

C-2          PM-10 Air Quality Concentrations, 1992-94	C-3

C-3          Regions Used in Air Quality Analyses in this Staff Paper	  C-4

E-l          Relationship Between Relative Risk of Death and PM-10 in Birmingham   E-6b

E-2          Relationship Between Relative Risk of Death and PM-10 in Utah Valley . E-6b

E-3          Relationship Between Relative Risk of Death and PM-10 in the Utah
             Valley	E-7a

E-4          Relationship Between Relative Risk of Pneumonia Admissions Among the
             Elderly and PM-10 in Birmingham	E-7b

E-5          Relationship Between Ischemic Heart Disease Admissions Among the
             Elderly and PM-10  	E-7b

E-6          Relationship Between Relative Risk of Pneumonia Admission Among the
             Elderly and PM-10 in Detroit	E-8a

E-7          Relationship Between the Odds of Cough Incidence Versus PM-10
             Concentration  from the Six City Study  	E-8b

E-8          Relationship of the Odds of Lower Respiratory Symptoms  Incidence Versus
             PM-10 Concentration from the Six City Study	E-8b

E-9          Relationship Between Mortality Risk Ratios and Inhalable Particles
             (PM15/IO) in the Six City Study  	E-lOb

-------
                                         XV11

                             LIST OF FIGURES (Cont'd.)

Number                                                                         Page

E-10         Relationship Between Mortality Risk Rate Ratios and PM-2.5 in the
             Six City Study	E-lOb

E-ll         Relationship Between Adjusted Mortality and PM-2.5 in the American
             Cancer Society Study	E-lOc

F-l          Distribution of PM2 5 Concentrations and of Estimated Mortality Risks
             from Short-Term Exposures in Philadelphia County   	  F-5

F-2          Comparison of Smoothed Nonlinear and Linear Mathematical Models for
             Relative Risk of Total Mortality  Associated with Short-Term TSP
             Exposure   	F-10

G-l          Visual Range and Extinction Coefficient as a Function of Haziness
             Expressed in Deciview	G-4a

-------
                                        1-1
       REVIEW OF THE NATIONAL AMBIENT AIR QUALITY STANDARDS
                          FOR PARTICIPATE MATTER:
    POLICY ASSESSMENT OF SCIENTIFIC AND TECHNICAL INFORMATION

I.      PURPOSE
       The purpose of this Office of Air Quality Planning and Standards (OAQPS) Staff
Paper is to evaluate the policy implications of the key studies and scientific information
contained in the EPA document, "Air Quality Criteria for Paniculate Matter" (U.S. EPA,
1996, henceforth referred to as the CD), and to identify the critical elements that EPA staff
believes should be considered in review of the national ambient  air quality standards
(NAAQS) for paniculate matter (PM).  This assessment is intended to help bridge the gap
between the scientific review  contained in the CD and the judgments required of the
Administrator in  setting ambient standards for PM.  Thus, emphasis is placed on identifying
those conclusions and uncertainties in the available scientific literature that the staff believes
should be considered in selecting particulate pollutant indicators, forms, averaging times, and
levels for the primary (health) and secondary (welfare) standards.  These specifications must
be considered collectively in evaluating the health and welfare protection afforded by PM
standards.
       While this Staff Paper should be of use to all parties interested in the standards
review, it is written for those decision makers, scientists, and staff who have some
familiarity with the technical discussions contained in the CD.  This Staff Paper presents
factors relevant to the evaluation of current primary and secondary NAAQS, as well as staff
conclusions and recommendations of suggested options for the Administrator to consider.

-------
                                           II-l
H.    BACKGROUND
A.    Legislative Requirements
    Two sections of the Clean Air Act govern the establishment and revision of NAAQS (42
U.S.C. 7401 to 7671q, as amended).  Section 108 (42 U.S.C. 7408) directs the
Administrator to identify pollutants which "may reasonably be anticipated to endanger public
health and welfare" and to issue air quality criteria for them.  These air quality criteria are
intended to "accurately reflect the latest scientific knowledge useful in indicating the kind and
extent of all identifiable effects on public health or welfare which may be expected from the
presence of [a] pollutant in the ambient air .  . ."
       Section 109 (42 U.S.C. 7409)  directs the Administrator to propose and promulgate
"primary"  and "secondary" NAAQS for pollutants identified under section 108.  Section
109(b)(l) defines a primary standard as one "the attainment and maintenance of which in the
judgment of the Administrator, based  on such criteria and allowing an adequate margin of
safety, are requisite to protect the public health."1  A secondary standard, as defined in
section 109(b)(2), must "specify a  level of air quality the attainment and maintenance of
which, in the judgment of the Administrator, based on such criteria, is  requisite to protect
the public  welfare from any known or anticipated adverse effects associated with the presence
of [the] pollutant in the ambient air."  Welfare effects as defined in section 302(h) [42
U.S.C. 7602(h)] include, but  are not limited  to, "effects on  soils,  water, crops, vegetation,
manmade [sic] materials, animals,  wildlife, weather, visibility and climate, damage to and
deterioration of property, and  hazards to transportation, as well  as effects on economic
values and on personal  comfort and well-being."
       The U.S. Court of Appeals for the District of Columbia  Circuit has held that the
requirement for an adequate margin of safety for primary standards was intended  to address
uncertainties associated with inconclusive scientific and technical information available at the
time of standard setting.  It was also intended to provide a reasonable degree of protection
   'The legislative history of section 109 indicates that a primary standard is to be set at "the maximum permissible
ambient air level  ... which will protect the health of any [sensitive] group of the population," and that for this
purpose "reference should be made to a representative sample of persons comprising the sensitive group rather than
to a single person in such a group"  (S. Rep. No. 91-1196, 91st Cong., 2d Sess. 10 (1970)).

-------
                                         II-2
against hazards that research  has not yet identified (Lead Industries Association v. EPA, 647
F.2d 1130, 1154 (D.C. Cir.  1980), cert, denied. 101 S. Ct. 621 (1980); American Petroleum
Institute v. Costle, 665 F.2d  1176,  1177 (D.C. Cir.  1981), cert, denied. 102 S. Ct.  1737
(1982)).  Both kinds of uncertainties are components of the risk associated with pollution at
levels below those at which human  health effects can be said to occur with reasonable
scientific certainty. Thus,  by selecting primary standards that provide an adequate margin  of
safety, the Administrator is seeking not only to prevent pollution levels  that have been
demonstrated to be harmful but also to prevent lower pollutant levels that she finds may pose
an unacceptable risk of harm, even  if the risk is not  precisely identified as to nature or
degree.
       In  selecting a margin  of safety, the EPA considers such factors as the nature and
severity of the health effects  involved, the size of the sensitive population(s) at risk, and the
kind and degree of the uncertainties that must be addressed.  Given that the "margin of
safety" requirement by definition only comes into play where no conclusive showing of
adverse effects exists, such factors which  involve unknown or only partially quantified risks
have their inherent limits as guides  to action.  The selection of any particular approach to
providing an adequate  margin of safety  is a policy choice left specifically to the
Administrator's judgment  (Lead Industries Association v. EPA, supra.  647 F.2d at  1161-62).
       Section  109(d)(l) of the Act requires that "not later than December 31,  1980, and at
5-year intervals thereafter, the Administrator shall complete a thorough  review  of the criteria
published under section 108 and the national  ambient air quality standards ... and shall make
such revisions in such  criteria and standards ... as may be appropriate ...."  Section
109(d)(2) requires that an independent scientific review committee be appointed and provides
that the committee "shall complete a review of the criteria ... and the national primary and
secondary ambient air  quality standards ... and shall recommend to the  Administrator  any  ...
revisions  of existing criteria  and standards as  may be appropriate ...."  Since the early
1980's, this independent review function has been performed by the Clean Air Scientific
Advisory  Committee (CASAC) of EPA's Science Advisory Board.

-------
                                         II-3
B.     History of PM NAAOS Reviews
1.     Establishment of the NAAQS for Paniculate Matter
       National ambient air quality standards for PM were first established in  1971, based on
the original criteria document (DREW, 1969).  Paniculate matter is the generic term for a
broad class of chemically and physically diverse substances that exist as discrete particles
(liquid droplets or solids) over a wide range of sizes.  Particles originate from a variety of
anthropogenic stationary and mobile sources as well as natural sources.  Particles may be
emitted directly or formed  in the atmosphere by transformations of gaseous emissions such as
sulfur oxides, nitrogen oxides, and volatile organic substances.  The chemical and physical
properties of PM vary greatly with  time, region, meteorology, and source category, thus
complicating the assessment of health and welfare effects.
       The reference method specified for determining attainment of the original standards
was the high-volume sampler, which collects PM up to a nominal size of 25 to 45
micrometers (/xm) (so-called total suspended paniculate or TSP).  The primary standards
(measured by the indicator TSP) were 260 micrograms per cubic meter  (/ig/m3), 24-hour
average, not to be exceeded more than once per year, and 75 /*g/m3, annual geometric mean.
The secondary standard (measured as TSP) was 150 jig/m3, 24-hour average, not to be
exceeded more than once per year.
2.     First Review of NAAQS for Paniculate Matter
       In October 1979 (44 FR 56731), EPA announced the first review of the criteria
document and NAAQS for PM and, after a lengthy and elaborate process, promulgated
significant revisions of the  original standards in 1987 (52 FR 24854,  July  1, 1987).2  In
that decision, EPA changed the indicator for particles from TSP to PM10, the latter
referring to particles with a mean aerodynamic diameter less than or  equal to
       The revised standards were based on a revised Criteria Document (U.S. EPA, 1982a), a corresponding
Staff Paper (U.S. EPA, 1982b), and subsequent addenda to those documents (U.S. EPA, 1986a; U.A. EPA.
1986b). A detailed description of the process followed in reviewing and revising the original Criteria Document
and NAAQS appears in the notice of final rulemakmg (52 FR at 24636-37).

-------
                                         11-4
10 fj.m.3  EPA also revised the level and form of the primary standards by  1) replacing the
24-hour TSP standard with a 24-hour PM10 standard of 150 /xg/m3 with no  more than
one expected exceedance per year and 2) replacing  the annual TSP standard with a PM10
standard  of 50 /tg/m3, expected annual arithmetic mean.  The secondary standard was
revised by replacing it with 24-hour and annual standards identical in all respects to the
primary standards.  The revisions also included a new reference method for the measurement
of PM10 in the ambient  air and rules for determining attainment of the new standards.  On
judicial review,  the revised standards were upheld in all respects  (Natural Resources Defense
Council v. Administrator. 902 F. 2d 962 (D.C. Cir. 1990),  cert,  denied. Ill  S. Ct. 952
(1991)).
3.     Recent Litigation
       The American Lung Association filed suit in February 1994 to compel EPA to
complete the present review of the PM  NAAQS by December 1995.   The U.S. District Court
for the District of Arizona subsequently ordered EPA to  complete its review and any revision
of the PM NAAQS by publishing a final decision in the Federal Register by January 31,
1997, with publication of a proposed decision required by June 30, 1996 (American Lung
Association v. Browner. CIV-93-643-TUC-ACM (D. Ariz., October 6, 1994)).  As
subsequently modified,  the court-ordered schedule requires publication of the proposed and
final decisions by November 29, 1996, and June 28, 1997, respectively.
4.     Current Review of the Paniculate Matter NAAQS
       In December 1994, EPA presented its plans for completing review  of the criteria
document and NAAQS for PM under the court order to the CASAC.  In addition, EPA's
OAQPS  completed a PM NAAQS Development Project Plan in January  1995, which
incorporated CASAC comments, identifying key issues to be addressed in  this Staff Paper
    The more precise term is 50 percent cut point or 50 percent diameter (Dx).  This is the aerodynamic particle
 diameter for which the efficiency of particle collection is 50 percent.  Larger particles are not excluded altogether,
 but are collected with substantially decreasing efficiency and smaller particles are collected with increasing (up to
 100 percent) efficiency.  Ambient samplers with this cut point provide a reliable estimate of the total mass  of
 suspended particulate matter of aerodynamic size less than or equal to 10 jim.

-------
                                        II-5
as well as the basis for the additional scientific and technical assessments needed to address
the policy issues.
      EPA desires to incorporate as much peer review and public input into the review as is
possible under the court-ordered schedule. Accordingly, as part of the development of the
CD, EPA hosted a public PM-Mortality Workshop in November  1994, at which seminal new
studies on particles and health effects were presented and discussed.  In January 1995, the
EPA's National Center for Environmental Assessment (NCEA) hosted three public peer-
review workshops on drafts of key chapters of a revised CD.
      Successive external review drafts of the revised CD were reviewed by CASAC and
the public at public meetings held on August 3-4, 1995 and December 15-16, 1995.  The
first external review draft of this Staff Paper was also reviewed by CASAC and the public at
the December 16, 1995 meeting.  Based on  CASAC and public comment, NCEA  revised the
CD and submitted chapters the committee had requested for additional review (namely CD
chapters 1, 5, 6, and 13) to CASAC and the public for review at a public meeting held
February 29, 1996.  At this meeting, CASAC also discussed the plan and methodologies for
the risk assessment presented in this Staff Paper.  On March 15,  1996, CASAC sent a letter
to the EPA Administrator indicating the committee's satisfaction with the CD (Wolff,
1996b).  NCEA made additional revisions to the document to respond to comments from
CASAC and the public and completed the CD on April 12,  1996.  At a  public meeting held
on May  15-16, 1996, CASAC and the public reviewed the revised Staff Paper, provided
additional  comments, and came to closure on the document.  On June 13, 1996, CASAC sent
a closure letter on the Staff Paper to the EPA Administrator (Wolff,  1996c).  Both CASAC
closure letters are reproduced in Appendix G of this Staff Paper.

-------
                                         III-l
HI.    APPROACH
       This Staff Paper is based on the scientific evidence reviewed in the CD and takes into
consideration CASAC and public comments received on the previous drafts.  The staff has
also considered comparative air quality and quantitative risk analyses in evaluating the
appropriateness of retaining or revising the current primary NAAQS and  in assessing
potential alternative NAAQS.  Technical  and economic analyses examining visibility
impairment and soiling and materials damage have also been considered in evaluating the
appropriateness of retaining or revising the current secondary NAAQS and  in assessing
potential alternative NAAQS.
       The approach taken in this Staff Paper is to assess and integrate the above information
in the context of those critical elements that the staff believes should be considered in
reviewing the primary and  secondary standards.  Attention  is drawn to judgments that must
be made based on careful interpretation of incomplete or uncertain evidence.  In such
instances, the Staff Paper provides the staffs evaluation, sets forth alternatives the staff
believes should be considered, and recommends a course of action.
A.     Bases for Initial  Analytical Assessments
       The staff identified several possible policy alternatives to provide  a basis  for
commencing initial analytical assessments of air quality, human exposure, and health risks.
1.     Primary Standards
       As in the 1987 review of the NAAQS, selecting the most appropriate indicator for
PM is  a major issue for this review. Thus, the staff planned for initial analytic assessments
of the assumption  that this  PM  NAAQS review might result in  setting or retaining one or
more primary standards from the following possibilities:
       •      Short-term Standard:  A 24-hour standard using a fine particle  indicator, a
              PM10 indicator, or  both; and
       •      Long-term Standard:  An annual standard using a fine particle indicator, a
              PM]0 indicator, or  both.
The staff also recognized that other indicators of PM pollution (e.g., sulfates  and acids) may
be important in relating effects  to PM pollution.

-------
                                         III-2
2.     Secondary Standards
       In revising the secondary standards, the staff has focused primarily on two types of
effects:  1) visibility impairment and 2) soiling and materials damage.  In the case of
visibility, this Staff Paper briefly assesses available scientific information in order to
determine an appropriate regulatory approach  for addressing regional haze.  A key
consideration in this assessment is that a number of factors that influence visibility
impairment vary significantly between  the eastern and western parts of the U.S.   Thus, this
Staff Paper examines the advisability of a uniformly implemented and attained secondary
NAAQS as contrasted to the establishment of  a regional haze program  under section 169A of
the Clean Air Act.  This Staff Paper also examines the available literature on  material
damage and soiling to ascertain whether  such information provides a basis for establishing a
separate national secondary NAAQS to protect against such effects.
B.     Organization  of Document
       The remainder of this Staff Paper is organized as outlined below.  Chapter IV
summarizes differences among  the various fractions of PM10, air quality trends for both PMU,
and fine particles,  characterizations of  average "background" concentrations, information on
relationships between PM and population exposures, and the air quality implications of
ongoing PM control  programs designed to attain the current PM NAAQS.
       Chapter V discusses available information on PM dosimetry and hypotheses regarding
mechanisms of toxicity, the  nature of health effects associated with PM, sensitive
subpopulations,  and integrated evaluations of the scientific evidence. Chapter V also presents
alternative interpretations of the evidence and  uncertainties surrounding reported health
effects associations and specific agents  of concern which are important for the Administrator
to consider in selecting appropriate primary standards.
       Chapter VI summarizes health risk assessments conducted for two urban areas to
provide quantitative estimates of the risks to public health associated with  1) existing PM  air
quality levels, 2) projected air quality levels that would occur upon attainment of the current
PMjo standards, and  3) projected air quality levels associated with attainment of alternative
PM25 standards.

-------
                                         III-3
       Drawing on these factors and on information contained in the previous chapters,
Chapter VII presents staff conclusions and recommendations for the Administrator to
consider in reaching decisions on the retention and/or revision of the primary NAAQS. The
chapter addresses alternative pollutant indicators, averaging times, forms, and levels, with
summary sections highlighting both key uncertainties and related staff research
recommendations as well as staffs overall recommendations for a suite of primary standards.
       With respect to review of the secondary standards, Chapter VIII presents information
on visibility impairment and soiling and materials damage, discusses pertinent scientific,
technical,  and policy considerations, and offers staff conclusions and recommendations for
the Administrator to consider in reaching a decision  on retention and/or revision of the
secondary NAAQS.

-------
                                          IV-1
IV.    AIR QUALITY: CHARACTERIZATION AND IMPLICATIONS
       This chapter defines the various subclasses of paniculate matter (PM) and then briefly
discusses the chemical and physical properties of PM in the atmosphere, recent PM
concentrations and trends, the relationships between PM and population exposures, and the air
quality implications of PM10 controls. This information is important both in interpreting the
available health effects and welfare information and in making recommendations for
appropriate indicators for PM.
A.     Character! ration of US. Amhient Particulate Matter
       PM represents a broad class of chemically and physically diverse substances.  The
principal common feature of PM is existence as discrete particles in the condensed (liquid or
solid) phase spanning several  orders of magnitude in size, from molecular clusters of 0.005
/xm in diameter to coarse particles on the order of 100 /xm.1  In addition to characterizations
by size, particles can be described by their formation mechanism or origin, chemical
composition, physical properties, and in terms of what is measured by a particular sampling
technique.
       In most locations, a variety of diverse activities contribute significantly to PM
concentrations, including fuel combustion (from vehicles, power generation, and industrial
facilities), residential fireplaces, agricultural and silviculture! burning, and atmospheric
formation from gaseous precursors (largely produced from fuel combustion). Other sources
include construction and demolition activities, wind blown dust, and road dust.  From these
diverse sources come the mix of substances that comprise PM.  The major chemical
constituents of PM10 are sulfates, nitrates, carbonaceous compounds (both elemental and
organic carbon compounds), acids, ammonium ions, metal compounds, water, and crustal
materials.  The amounts of these components vary from place to place and over time.
        In this Staff Paper, particle size or diameter refers to aerodynamic diameter, which is defined as the
diameter of a spherical particle with equal settling velocity but a material density of 1 g/cm3, normalizing particles of
different shapes and densities (CD, page 3-8).

-------
                                           IV-2
1.      Multi-modal Size Distributions
       The health and environmental effects of PM are strongly related to the size of the
particles.  The aerodynamic size and associated composition of particles determines their
behavior in the respiratory system (i.e., how far the particles are able to penetrate, where
particles are deposited, and how effective the body's clearance mechanisms are in removing
them as discussed in Chapter V).  Furthermore, particle size is one of the most important
parameters in determining atmospheric lifetime of particles, which is a key consideration in
assessing health effects information because of its relationship to exposure. The total surface
area and number of particles, chemical composition, water solubility, formation process, and
emission sources all vary with particle size.  Particle size is also a determinant of visibility
impairment, a welfare consideration linked to fine particle concentrations. Thus, size is an
important parameter in characterizing PM, and particle diameter has been used to define the
present standards.
       The multi-modal distribution of particles based on diameter has long been recognized
(Whitby et al., 1972; Whitby et al., 1975; Willeke and Whitby,  1975; National Research
Council, 1979; U.S. EPA, 1982a; U.S. EPA,  1982b;  U.S. EPA, 1986b; CD Section 3.1.3.2).
Although particles display a consistent multi-modal distribution over several physical metrics
such as volume and mass, specific distributions may vary over place, conditions, and time
because of different sources, atmospheric conditions, and topography.  Based on particle size
and formation mechanism, particles can be classified into two fundamental modes:  fine and
coarse modes. Figure IV-1 illustrates an idealized mass distribution of the fine and coarse
modes. A depiction of typical number, surface area, and volume distribution of ambient
particles is shown in Figure IV-2.  This latter figure illustrates that fine particles can be further
subdivided into nuclei or ultrafme, and accumulation modes.2  As illustrated  in the figure,
       rj                                                         •
         Typically, the accumulation mode can be characterized by mass median aerodynamic diameter (MMAD)
of 0.3 to 0.7 /im and a geometric standard deviation (sigma-g) of 1.5 - 1.8 (CD, page 13-5).  The CD defines
ultrafine particles as _£_ 0.1 /un in diameter (CD, Sections 3.1.3 and 13.2.1). Nuclei or ultrafme particles tend to
exist as disaggregated particles for very short periods of time (minutes) and  rapidly coagulate into accumulation mode
particles (CD page 3-10).  Accumulation mode particles, however, do not grow further into the coarse particle mode.

-------
                                      IV-2a
    0)
    a.
    a
    o
   CO
   (0
        70
        60
50
        40
        30
   S   20
        10
                 Fine Mode Particles
                                         Coarse Mode Particles
  0.1    0.2      0.5   1.0     2        5     10     20
             Aerodynamic Particle Diameter (Da), urn

     4	Total Suspended Particles (TSP)	
                               PM
                                  10
                         PM
                            2.5
                                        (10-2.5)
 TSP
HiVol
                                                                     50    100
Figure IV-1. Idealized Distribution of Ambient Particulate Matter

Distribution shows fine and coarse mode particles and fractions collected by size-selective
samplers such as the wide range aerosol collector (WRAC) and the TSP high volume sampler.
(Adapted from Wilson and Suh (1996); CD Figure 3-3).

-------
                                  IV-2b
o

X
I

o
          15
          10
   .-  Q

   i  <
                    I  I Mll|
                                   Nn = 7.7 x 10'
                                DGNn = 0.013
                                               N. = 1.3 x 10'
                                           DGNa = 0.069
                                              oga= 2.03
   
-------
                                           IV-3
 even when the fine mode contains about 40 percent of the volume or mass of PM10, it accounts
 for most of the surface area and number of particles.
       The CD concludes that an appropriate cut point3 for distinguishing between the fine and
 coarse modes lies in the range of 1.0 pm to 3.0 jum where the minimum mass occurs between
 the two modes (CD, Section 3.1.2; Miller et al., 1979).  The CD states that the data do not
 provide a clear choice of cut point given the overlap that occurs between the modes.  Most
 ambient measurements of fine particle mass in the U.S.  have used instruments with cut points
 of 2.5 or 2.1 urn. Appendix  A outlines the policy  considerations involved in making the staff
 recommendation  for using 2.5 pm as the cut point for measuring fine particles.
       Table IV-1 introduces some of the size-related terminology used in this Staff Paper.
 For the purposes of this document, PMX (e.g., X = 1, 2.5,  10, 15, 10-2.5) is used to refer to
 gravimetric measurements with a 50 percent cut point of X /im diameter while the terms fine
 or coarse particles will be used more generally to refer to the fine and coarse modes of the
 particle distribution.  The distinction highlights the role  of formation mechanism and chemistry
 in addition to size in defining fine and coarse  mode particles. Any specific measurement (e.g.,
 PM2 5) is only an approximation  for fine particles.4
       In addition to gravimetric fine particle measurements, PM has been characterized in the
 U.S. and abroad  using a variety of filter-based optical techniques including British or black
 smoke (BS), coefficient of haze (COH), and carbonaceous material (KM), as well as estimates
 derived from visibility measurements (CD, Chapter 4 and 12; see Appendix B of Staff Paper
 for limitations in  determining mass).  In locations where they are calibrated to standard mass
 units (e.g. London), these measurements can be useful as surrogates for fine particle mass
 (CD, Chapter 4).
         When used in the context of sampling, outpoint is a term used to describe the separation efficiency curve
for samplers. The cut point is typically described by the aerodynamic diameter at which the sampler achieves 50
percent collection efficiency.
       4
         Monitor design, measurement temperature, and inlet efficiency can also affect which particles are included
in the definitions of the various size fractions (CD, Chapter 4). Sampling protocols may also affect the amount of
semivolatile organics and nitrates and particle-bound water included in a measurement.

-------
                                        IV-3a
TABLE rV-1. PARTICLE SIZE FRACTION TERMINOLOGY USED IN THIS STAFF
                                         PAPER
              Term
Description
Size Distribution Modes
  Fine particles
  Coarse Particles
Sampling Measurements

  Total Suspended
  Particles
  PM
      10
   PM
      2 5
   Coarse fraction of PM10
   PM(10.2.5)
Fine mode particles which are generally formed through
chemical reaction, nucleation, condensation of gases, and
coagulation of smaller particles; contains most numerous
particles and represents most surface area.

Coarse mode particles which are mostly generated from
mechanical processes through crushing or grinding.
Particles measured by a high volume sampler as described in 40
CFR Part 50, Appendix B.  This sampler has a cut point of
aerodynamic diameters1 that varies between 25 and 40 /*m
depending on wind speed and direction.

Particles measured by a sampler that contains a size fractionator
(classifier) designed to have an effective cut point  of 10 ^im
aerodynamic diameter.  This measurement includes the fine
mode and part of the general coarse mode and is an indicator
for thoracic particles (i.e., particles that penetrate to the tracheo-
bronchial and the gas-exchange regions of the lung).

Particles measured by a sampler that contains a size fractionator
(classifier) designed to have an effective cut point  of 2.5 ^m.
The collected particles include most of the fine mode.  Some
small portion of the coarse mode may be included  depending on
the sharpness of the sampler efficiency curve and the size of
coarse mode particles present.

Particles measured directly using a dichotomous sampler or
by subtraction of particles measured by  a PM2 5 sampler from
those measured by a PM,0 sampler.
 When discussing samplers, cut point is a term used to describe the separation effipiency curve for particle
collection devices. The cut point is typically described by the particular aerodynamic diameter at which the
sampler achieves 50% collection efficiency. Aerodynamic diameter is defined as the diameter of a spherical
particle with equal settling velocity but a material density of 1 g/cm3. This normalizes particles of difteient
shapes and densities.
2PMX indicates an 50 percent cut point of X /im diameter.  Because samplers have a collection efficiency that
varies around the cut point, not all particles less than X /xm diameter will be collected and some particles
greater than X /*m diameter will be collected.

-------
                                          IV-4
       The distinction between any specific measurement of fine particles and fine mode (or a
 measurement of coarse particles and coarse mode) is important because in the subsequent
 chapters of this Staff Paper, the staff draws public health conclusions regarding fine and coarse
 mode particles and in doing so the staff relies on the available measurements. Examples of fine
 particle measurements include PM2 5, BS,  COH, and concentrations of specific chemical
 classes predominantly in the fine fraction such as sulfates and acids all judged to be surrogates
 for fine mode particles.  Measurements of coarse particles include PM10_2 5, PM15_2 5, and
 TSP minus PM10.
 2.     Properties of Fine and Coarse Fraction Particles
       As  summarized in Table IV-2, fine and coarse particles can be differentiated by their
 sources and formation processes, chemical composition, solubility, acidity, atmospheric
 lifetime and behavior, and transport distances (CD Chapter 3). The key properties of fine and
 coarse particles are described below.
       a.     Sources and Formation Processes
       Fine and coarse particles generally have distinct sources and formation mechanisms
 although there may be some overlap.  Primary fine particles are formed from condensation of
 high temperature vapors during combustion (CD, page 3-2).  Fine particles are usually formed
 from gases in three ways:  (1) nucleation (i.e., gas molecules coming together to form a new
 particle), (2) condensation of gases onto existing particles, and (3) by reaction in the liquid
 phase (CD, page 13-7).  Particles formed from nucleation also coagulate to form relatively
 larger particles, although such particles normally do not grow into the coarse mode  (CD,
 Section  3.1.3.2). Particles formed as a result of chemical reaction of gases in the atmosphere
 are termed  secondary particles because the  direct emission from a source is a gas that is
 subsequently converted to a product that either has a low enough  vapor pressure to form a
 particle or reacts further to form a low vapor pressure substance.  Some examples include the
 conversion  of sulfur dioxide (SO2) to  sulfuric acid droplets that further react with ammonium
 to form  particulate sulfate, or the conversion of nitrogen dioxide  (NO2) to nitric acid which
reacts further with ammonia to form particulate ammonium nitrate (NH4NO3) (CD, Section
3.2.2).  Although directly emitted particles are found in the fine fraction (the most common

-------
                                        IV-4a
      TABLE IV-2.  COMPARISON OF AMBIENT FINE AND COARSE MODE
                                    PARTICLES
                                  Fine Mode
                                      Coarse Mode
 Formed from:
 Formed by:
 Composed of:
 Solubility:
Gases
Chemical reaction;
Nucleation;
Condensation;
Coagulation;
Evaporation of fog and cloud
droplets in which gases have
dissolved and reacted.

Sulfate, SO4=;
Nitrate, NOj;
Ammonium, NH4+-
Hydrogen ion,  FT;
Elemental carbon
Organic compounds
(e.g., PAHs, PNAs);
Metals (e.g., Pb, Cd, V,
Ni, Cu, Zn,  Mn, Fe);
Particle-bound  water.

Largely soluble,
hygroscopic and deliquescent.
Large solids/droplets
Mechanical disruption
(e.g., crushing, grinding,
abrasion of surfaces);
Evaporation of sprays;
Suspension of dusts.
Resuspended dusts (e.g., soil
dust, street dust);
Coal and oil fly ash;
Metal oxides of crustal
elements (Si, Al, Ti,  Fe);

CaCO3, NaCl, sea salt;
Pollen, mold spores;
Plant/animal fragments;
Tire wear debris.

Largely insoluble  and non-
hygroscopic.
 Sources:
 Lifetimes:

 Travel Distance:
Combustion of coal, oil,
gasoline, diesel, wood;
Atmospheric transformation
products of NOX, SO2, and
organic compounds including
biogenic species (e.g.,
terpenes);
High temperature processes,
smelters, steel mills, etc.


Days to weeks

100s to 1000s of kilometers
Resuspension of industrial
dust and soil tracked onto
roads;
Suspension from disturbed
soil (e.g., farming,  mining,
unpaved roads);
Biological sources;
Construction and demolition;
Coal and oil combustion;
Ocean spray.

Minutes to hours

< 1 to 10s of kilometers
Source:  Adapted from Wilson and Suh (1996); CD (Table 3-15, p. 3-145)

-------
                                          IV-5
being particles less than 1.0 /im in diameter from combustion sources), particles formed
secondarily from gases dominate the fine fraction.
       By contrast, most of the coarse fraction particles are emitted directly as particles and
result from mechanical disruption such as crushing, grinding, evaporation of sprays, or
suspensions of dust from construction and agricultural operations.  Simply put, most coarse
particles are formed by breaking up bigger particles into smaller ones.  Energy considerations
normally limit coarse particle sizes to greater than 1.0 /xm in diameter (CD, Chapter 3). Some
combustion-generated particles such as fly ash are also found in the coarse fraction.
       b.     Chemicql Composition, Solubility, and Acidity
       Fine and coarse mode particles generally have distinct chemical composition, solubility,
and acidity. Fine mode PM is mainly composed of varying proportions of several major
components:   sulfates, nitrates, acids, ammonium, elemental carbon, organic carbon
compounds, trace elements such as metals, and water.  By contrast, coarse fraction
constituents are primarily crustal, consisting of Si, Al,  Fe, and K (note that small amounts of
Fe and K are also found among the fine mode particles but stem from different sources).
Biological material such as bacteria, pollen, and spores may also be found in the coarse mode.
As a result of the fundamentally different chemical compositions and sources of fine and
coarse fraction particles, the chemical composition of the sum of these two fractions, PM10, is
more heterogenous than either mode alone.
       Figure IV-3 presents a synthesis of the available published data on the chemical
composition of PM25 and coarse fraction particles in U.S. cities by region described in
Chapter 6 of the CD.  The CD concludes that the fine and coarse fraction are composed of
different chemical constituents and that each fraction also has regional patterns resulting from
the differences in sources and atmospheric conditions (CD, Section 6.6).  Differences across
the country in sources and atmospheric conditions contribute to the variability.  In addition to
the larger relative shares of crustal materials in the West, total concentrations of coarse
fraction particles are generally higher in the arid areas of the Western and Southwestern U.S.
      In general, fine and coarse particles exhibit different degrees of solubility and acidity.
With the exception of carbon and some organic compounds, fine particle mass is largely

-------
         PM2J5 Mass^Apportlonment
     Unknown 22.
                             Minerals 4.3V.
            NO;  1.1%
             Nitrate based on 3 studies
                                    S04 34.1%
                                     )- 13.0%
        IV-5a

  Coarse Mass Apportionment
                                                   Unknown 41 5%
                                                                           Minerali 51 8%
 (NH;>-  1.8%—'  \	so; 4.9%
Insufficient Nitrate, OC, and EC data available
PM10 Mass Apportionment
                                                                                           Unknown 28 9%
                                                                                                                    Mineral! 19 6%
                                                                                                                       SO. 27.8%
                  -(NH, )* 10 7%
     Nitrate based on 2 studies
  to
  0
          EC 14 7%
                               Mineral! 14.6%
        OC x 1.4 38.9%-

             Reconstructed sum « 102.2%
                                                  Unknown 27.
                                                                                                   EC 5.1%
                       -Minerals 69.9%

Insufficient Nitrate. OC. and EC data available
                                                                                         OC x 1.4 30.0%
                                                                                                                  •Mineral* 36 3%
                        -SO, 4 6%
      240%	'        ^—(NH«)'67%

   Reconstructed sum = 111.4%
Figure IV-3.  Major constituents of PM2 5, Coarse Fraction, and PM10 (CD, Figures 6-85a-c).* Sulfate ions, ammonium ions, and
organic carbon account for most of the PM25 mass.  Eastern PM25 has more sulfate; whereas, many western sites have a larger
nitrate contribution and twice the proportion of organic carbon compounds of eastern sites.  In contrast, minerals dominate the
coarse fraction, ranging from over 50 percent in the Eastern U.S. to 70 percent in the Western U.S. of coarse fraction mass.  Total
concentrations of coarse fraction particles are generally higher in the arid areas of the Western and Southwestern U.S. than in the
Eastern U.S.

*The analysis focuses on data from the Harvard Six-City Study and the Inhalable Particle Network (IPN) as well as other published data shown in CD Tables A-
2a-c. (NH4*)* represents the concentration of NH4* that would be required if all the sulfate ion were present as ammonium sulfate and all the NO3' as NH4NO3.
Therefore, (NH/)* represents an upper limit to the true concentration of (NH/)*. The unknown fraction of fine mass is assumed to be mainly water.

-------
                                          IV-6
soluble in water and hygroscopic (i.e., fine particles readily take up and retain water).  The
fine particle mode also contains the acidic fraction (CD, Section 3.3.1).  By contrast, coarse
particles are mostly insoluble, non-hygroscopic, and generally basic.
       C.      Atmospheric Behavior
       Fine and coarse particles typically exhibit different behavior in the atmosphere.  These
differences affect several exposure considerations including the representativeness of central-
site monitored values and the behavior of particles formed outdoors once inside homes and
buildings where people spend most of their time (as discussed below in Section C).
       Fine accumulation mode particles typically have longer atmospheric lifetimes (i.e.,
days to weeks) than coarse particles and tend to be  more uniformly dispersed across an urban
area or large geographic region, especially in the Eastern  U.S. (CD Sections 3.7, 6.3, and 6.4;
Wilson et al.,  1995; Eldred and Cahill, 1994; Wolff et al., 1985; Shaw and Paur 1983;
Altshuller 1982; Leaderer et al., 1982).  As noted above,  secondary fine particles are formed
by atmospheric transformation of gases to particles. Such atmospheric transformation can take
place locally during atmospheric stagnation or during transport over long distances.   For
example, the formation of sulfates from  SO2 emitted by power plants  with tall stacks can  occur
over distances exceeding  300 kilometers and 12 hours of transport time; therefore, the
resulting particles are well mixed in the air shed (CD, Sections 3.4.2.1, and 6.4.1)  Once
formed, the very low dry deposition velocities of fine particles contribute to their persistence
and uniformity throughout an air mass (CD, Sections 6.4  and page 7.2; Suh et al., 1995;
Burton etal., 1996).
       Larger particles generally deposit more rapidly than small  particles; as a result, total
coarse particle mass will be less uniform in concentration  across an urban area than are fine
particles (CD, Sections 3.7,  and 13.2.4). Because  coarse particles may vary in size from
about 1 pm to over  100 jim, it is important to note  their wide range of atmospheric behavior
characteristics.  For example, the larger coarse particles (> 10 pm) tend to rapidly fall out of
the air and have atmospheric lifetimes of only minutes to hours depending on their size and
other factors (Wilson and Suh, 1995; Chow et al.,  1991; CD, Section 3.2.4). Their spatial
impact is typically limited by a tendency to fallout in  the proximate area downwind of their

-------
                                          IV-7
emission point.  Such large coarse particles are not readily transported across urban or broader
areas, because they are generally too large to follow air streams and they tend to be easily
removed by impaction on surfaces (DRI, 1995; CD, Sections 7.2.2 and 13.2.4). The
atmospheric behavior of smaller "coarse fraction" particles (PM10_2 5) is intermediate between
that of the larger coarse particles and smaller fine particles. Thus, coarse fraction particles
may have lifetimes on the order of days and travel distances of up to 100 km or more.5 While
it may be reasonable to expect that coarse fraction particles would be less homogeneously
distributed across an urban area than fine particles in areas with regionally high fine particle
concentrations (e.g. the eastern U.S.), this is not consistently true in a variety of locations
(DRI,  1995). In some locations, source distribution and meteorology affects the relative
homogeneity of fine and coarse particles, and in some cases, the greater measurement error in
estimating coarse fraction mass (Rodes and Evans,  1985) precludes clear conclusions  about
relative homogeneity.
       Nevertheless, because fine particles remain suspended for longer times (typically on the
order of days to weeks as opposed to days for coarse fraction particles) and travel  much farther
(i.e., hundreds  to thousands of kilometers) than coarse fraction particles (i.e., tens to hundreds
of kilometers),  all else being equal,  fine particles are theoretically likely to be more uniformly
dispersed across urban and regional  scales than coarse fraction particles. In contrast, coarse
particles tend to be less evenly dispersed around urban areas and exhibit more localized
elevated concentrations near sources (CD, Section 13.2.7;  DRI, 1995).
       d.     Correlations between PM2 5 and Coarse Fraction  Mass
       As might be expected from the differences in origin, composition, and behavior,
ambient daily fine and coarse fraction mass concentrations generally are not well correlated.
An analysis  (SAI,  1996) of several data sets conducted for this review reported the R-squared
statistic between daily PM2 5 and PM10.2 5 mass to be 0.13 for all  non-rural sites and  0.21
particles.
        In extreme cases, dust storms occasionally cause very long-range transport of the smaller size coarse

-------
                                           IV-8
when rural sites were included.6 The results indicate a poor correlation between daily
averages of the fine and the coarse fractions. In some specific instances, however, fine and
coarse fractions may be correlated. For example, a vehicle moving on a dusty road would emit
fine particles from the exhaust and produce coarse particle emissions from the road dust.  In
locations with poorly controlled industrial emissions of both fine and coarse particles, R2 as
high as 0.7 have been reported (Schwartz et al., 1996a).
       e.     Summary
       In summary, the fine and coarse mode particles are distinct entities with differing
sources and formation processes, chemical composition, atmospheric lifetimes and behaviors,
and transport distances.  The CD concludes that these profound differences alone justify
consideration of fine and coarse fraction particles as separate pollutants for measurement and
development of control strategies.  The fundamental differences between fine and coarse
particles are also important considerations in assessing the available health effects and exposure
information.
B.     PM  Air Quality Patterns
       This section outlines geographic distributions of PM  as well as ambient concentration
trends and background levels for PM10 and fine particles.
1.     PM Concentrations and Trends
       a.      PM.Q Concentrations and Trends
       State and  local air pollution control agencies have been collecting PM10 mass
concentration data using EPA-approved reference samplers and reporting these data  to EPA's
publicly available AIRS database since mid-1987.  Figure IV-4 shows geographic distribution
of the 83 areas that are listed as not attaining the current PM10 standards as of September
       6 SAI (1996) reported the following:
(1) R2 = 0.13 of daily PM2 5 with daily coarse fraction mass concentrations (n = 8,676) between 1988 and 1993
using the Aerometric Information Retrieval System (AIRS), Interagency Monitoring of Protected Visual Environments
(IMPROVE), California Air Resources Board (CARS) Dichotomous Network (1990-1993 data), with rural sites
removed.
(2) R2 = 0.21 of daily PM2 5 with daily coarse fraction mass concentrations (n = 31,510; 57% rural data) between
1985 and 1993 using AIRS, IMPROVE, CARB Dichot Network (1990-1993 data), and South Coast Air Basin
(SCAB) Intensive Monitoring Network (IMN) (1985-1986).

-------
AK
         [  FIGURE IV-4.
O Eagle River [
_"•_*-. _j Areas Designated  Nonattainment for Particulates (PM-10)
            KEY TO PRINCIPAL EMISSION TYPE
   O  ARI'.AS NONA11AINMENT DUE TO STATIONARY SOURCE EMISSIONS
   •  AREAS NONATTAINMENT DUE IN PART TO WOOD SMOKE EMISSIONS
   (. ^  ARPAS NONATFAINMENT DUF IN PART TO FUGITIVE DUST EMISSIONS
   O  AREAS NONA1TAINMENT DUE TO MUL1 IPI.E TYPES OF EMISSIONS
                                                          CIRCLE DIAMETER O
                                                          INDICATES RELATIVE SI7E
                                                          OF AFFECT ED POPU1. A TION
                                                                                                                 00
 Designated Nonattainment Areas as of September 1994
                                                                        Note: Unclassified areas are not shown.

-------
                                          IV-9
1994; the figure also summarizes the prevalent contributing sources and size of population
residing in nonattainment areas. Most of the non-attainment areas are in the Western U.S.
with fewer in heavily populated or industrialized eastern areas. Many of the highest values
occur in western areas with fugitive dust sources and in mountain valleys impacted by wood
smoke during winter inversions (CD, Section 6.5).
       National trends may readily be examined for the 6-year period from 1988 to 1993 as
illustrated in Figures IV-5a and IV-5b. The figures represent 799 trend sites, mostly from
urban and suburban locations as well as a few remote locations; monitoring sites with data in
at least five of the six years are included. The figures show the trend and site-tb-site
variability in the composite annual mean and the ninetieth percentile of 24-hour PM10
concentrations.7 The trend for the composite annual mean shows a steady decline totaling 20
percent over the six-year period from 1988 to 1993. The ninetieth percentile similarly
decreases 19 percent over the same period (U.S. EPA, 1994a).  Annual average PM10
concentrations ranged from 25  to 35 /xg/m3 for most U.S. regions by 1994. Additional
information about current PM10 concentrations are presented in Appendix C.
       b.     Fine Particle Concentrations and Trends
       The PM2 5 concentration data are considerably more limited than for PM10. From
1983 to 1993, fewer than 50 sites reported data to AIRS in any given year.8 Figure IV-6
displays a quarterly smoothed geographic distribution of the IMPROVE and Northeast States
Coordinated Air Use Management (NESCAUM) networks' PM2 5 data.  These data generally
do not include urban concentrations but represent the regional non-urban concentrations.  The
figure shows both the regional  character of elevated fine particle levels in the Eastern U.S. and
         The ninetieth percentile statistic is used because PM10 sampling frequency varies among sites and may
change from one year to the next at some sites. This statistic is less sensitive to changes in sampling frequency than
are the nuucinmm or second muTimmn peak values. Most PM10 sites sample on a once every six day schedule.

         Additional special studies have also monitored PM2.S, but these data are not reported in AIRS. For this
review, EPA assembled other available data sets for analysis (see CD, Section 6.10 and SAI, 1996). The databases
assembled to support this Staff Paper include AIRS, Inhalable Particle Network (IPN) (1982-1984), IMPROVE (1987-
1995), CARS Dichotomous Network (1990-1993), and SCAB MN (1985-1986).   Figure C-4 in Appendix C
provides a summary of the available data for fine particles.

-------
                           IV-9a
 FIGURE iv-sa.   PM-10 TREND, 1988-1993
                (ANNUAL ARITHMETIC MEAN)
   CONCENTRATION, JJG/M3
          1988
1989
1990
1991
1992
1993
                                           -95th PERCENT1LE
                                           -90* PERCENTILE
110
    FIGURE iv-5b. PM-10 TREND, 1988-1993
                      (90th PERCENTILE)
    CONCENTRATION, JJG/M3
                                                              -7Mi PERCENTILE
                                                              -COMPOSITE AVERAGE
                                            -25* PERCENTILE
                                            -IMiPERCEXTILE
                                            -MiPERCEXTILE
                                       t 2-1. BliMratan tf ftettmg cammtum of
100-
 90-
 80
 70
 60-
 50-
 40-
 30
 20 H
 10
  0
                                                 799 SITES
        I      i
                i     i
         ¥      V     ¥     ¥
           1988
1989
1990
1991
1992
1993

-------
                                                                                      Quarter 2
                                                                                 Source MPROVEandNESCAUM
                                                                                      Quarter 4
                                                                                                  i
                                                                                                  VD
                                                                 Fine Mass
                                                                                 Source IMPROVE and N6SCAUM
FIGURE IV-6.
FINE MASS CONCENTRATION DERIVED FROM NONURBAN IMPROVE/NESCAUM

NETWORKS.  (CD, Figure 6-8).

-------
                                          IV-10
California as well as a strong seasonality.  In the Eastern U.S. high fine particle levels
dominated by sulfates occur in the summer often in conjunction with elevated ozone levels.
       National PM2 5 trends are not available because of the limited number of sites
measuring PM2 5 and the sampling period at most sites is restricted to a few years. The
development of national trends is further hindered because PM2 5 is measured using a variety
of sampling frequencies and a variety of non-standard sampling equipment (because there is
currently no federal reference and equivalency program for PM2 5).
       However, visibility data can be used as a reasonable surrogate to estimate fine particle
trends because the extinction coefficient (Bext) is directly related to fine particle mass (CD,
page 6-216). Sufficient visibility data are available to produce national trends from 137 U.S.
sites (principally airports) since 1948 (CD, Section 6.10.2;  NAPAP, 1991). The location of
these sites reflects  suburban and urban  locations with airports. Figure IV-7 depicts trends
maps for the 75th percentile extinction  coefficient for summer and winter quarters. The
figures show significant regional and seasonal trends.  In the northeastern states, winter haze
shows a 25 percent decrease while in the southeastern states, there is a 40 percent increase in
winter haze (NAPAP, 1991).9  The summer haziness in the Northeast shows an increase up to
the mid-1970s followed by a decline.  In the Southeast, there was an 80 percent increase in
summer haziness, mainly occurring in the 1950s and 1960s (NAPAP, 1991).   During the
summer months, haziness (extinction coefficient) in the East can be dominated by sulfate (with
associated water and ammonium). In this situation, visibility trends may be a better surrogate
for sulfate than for non-sulfate related fine particle components (see subsection c below).
       Visibility and fine particles have been monitored with more precision by the IMPROVE
network from 1987 to present.  In eastern remote locations, air quality data from  1982 to 1992
showed roughly a 3 percent annual increase in sulfate  mass concentration during the summer
and a smaller negative (although not statistically significant) trend in the winter (Eldred and
         For the NAPAP analyses, the Northeast was defined as Indiana, Ohio, Pennsylvania, New York,
Kentucky, West Virginia and New England states, and the Southeast was defined as states south of the Ohio River
and east of the Mississippi (NAPAP, 1991).

-------
                                            IV-lOa
                                                                   Y
Figure IV-7. U.S. trend maps for the 75th percentile extinction coefficient, Bex,for winter (Ql)
and summer (Q3) (after CD, Figure 6-112). Bext (km'1) is derived from visual range (VR) data by
Bat=3.9/VR.  Data obtained under natural impairment conditions (i.e. rain, snow, fog) were
eliminated.  Because of the reationship between extinction and fine particle mass, these trends can
be used to make some inferences about regional fine particle trends.  As noted in the text and
Figure IV-8, summertime visibility trends in the eastern U.S. are greatly influenced by the sulfate
fraction of fine particles.

-------
                                       IV-11
Cahill 1994). Western visibility monitoring through the IMPROVE network has not shown
any trends for the period.
      C.     Trends in Emissions of Fine Particle Precursor Oases
      SO2, nitrogen oxides (NOX), which encompasses NO and NO2, and certain organic
compounds are major precursors of secondarily formed fine particles, as described above.  The
relationship between precursor emission reductions and ambient PM2 5 is nonlinear in many
aspects; thus, it is difficult to project the impact on PM2 5 arising from expected changes in
PM precursor emissions without air quality simulation models that incorporate treatment of
complex chemical transformation processes. In general terms, one would expect that emission
reductions of SO2 should lead to reductions in sulfate aerosol, but reductions will vary by
season, depending on both emission fluctuations and changes in prevailing meteorology and
photochemistry.
      Figure VI-8 shows comparisons of sulfur emissions for summer and winter with
extinction measurements derived from airport visibility data over the Northeast and Southeast
in the winter and summer seasons where sulfates are currently the major contributor to light
extinction (NAPAP, 1991). The correspondence between sulfur emissions and extinction
coefficient is fairly close, particularly in the summer, but not  an absolute match. For some
years there are increases or decreases in extinction coefficient without corresponding changes
in sulfur emissions, which likely reflect changes in non-sulfate particles as well as changes in
meteorology and errors in emissions and visibility data.  Overall, these data point to a strong
relationship between sulfur emissions and regionally occurring fine particle concentrations in
the Eastern U.S.  (NAPAP, 1991).
      It is noteworthy that major reductions in precursor emissions have occurred in the past,
such as the large S02 reductions that were achieved in the 1970s and 1980s in  some locations
because of other CAA programs such as the SO2 NAAQS implementation, prevention of
significant deterioration (PSD) program, and later from the new source performance standards
(NSPS)  program. Similarly, NOX emissions increases have been limited due to PSD, NSPS,
and mobile source control programs.  Future reductions in SO2 of slightly less than  1 percent
per year for the next 9 years are projected for the Eastern U.S., primarily from electric

-------
                                                IV-lla
    FIGURE IV-8.
TRENDS IN VISIBILITY AND SULFUR EMISSIONS IN THE
EASTERN U.S.
    6-
 * 3  «;-
 J*4  3-
  n  4
« c
3 0
    3-
    2-
    C
    1940
                        Sulfur Emissions
                         Haziness
                   NORTHEAST WINTER  (January)
                                              o <"
           1950
                   1960
                          1970
       1980
              1990
             Comparison of sulfur emission trends
             and extinction coefficient (+) for the
             northeastern region during the winter
             months.
                                 3
                                28-
                                26-
                                24-
                             z -22-
                             O *
                                                           08'
                                                        w-06
04-
02-
 0-
 1940
                                                                           Sulfur Emissions
                                                                           SOUTHEAST WINTER (January)
                                                                    1950
                                                                           1960
                                                                                  1970
                                                                1980
                                                                                                 1990
                                           Comparison of sulfur emission trends
                                           (Dj and extinction coefficient (+) for the
                                           southeastern region during the winter
                                           months.




to

V] IM
»H *3
X 10
w
to
ft C
££
D C
IB O
^ ^
< z
bl ^





6-


5-


4-

3-

2-


1-

0-

Haziness
JQ * J.
£0^*yy\./^\
/V "\Bn \
M // ^^R \
B \ /* Z S B \fi
7 J^§*yD^'! ^ Sulfur Emissions
* TWfl






NORTHEAST SUMMER (July)

> 500


r
3-
• 400 r>
Ss
n Z
-300 ?™
D

Z
• 200 3.


1940 1950 1960 1970 1980 1990
Comparison of sulfur emission trends (0)





«
z
o
m
z
U
ft
£
^
«
^
j
c
U




3-] 	 r
2 8-

26-

24-
tT22-
3
« 1 8-
^ 1 6-
° 1 4
g 1.2-

— < -J .
^08-
0.6-
04
02




A J
fCW _)A
/ l/\ O^]I
// ^
/p
Haziness J"
fr
/(/ Sulfur Emissions
A/a-
r-f^
j^j^*^^^
sf V«^
SOUTHEAST SUMMER (July)
500




400 P

:r
w

-300 5^
S z
rt W
Q W
3
-200 ?
_-


1940 1950 1960 1970 1980 1990
             and extinction coefficient (+)for the
             northeastern region during the summer
             months.
                                            Comparison of sulfur emission trends
                                            (D) and extinction coefficient (+) for the
                                            southeastern region during the summer
                                            months.
     Source:  NAPAP,  1991

-------
                                         IV-12
utilities (U.S. EPA 1995b). These projected reductions are due to the Acid Deposition
Program, as required under Title IV of the 1990 CAA Amendments.  Substantial NOX controls
are also required for motor vehicles and utilities under the CAA Amendments.
2.     Background Levels
       Natural sources contribute to both fine and coarse particles in the atmosphere. For the
purposes of this document, background PM is defined as  the distribution of PM concentrations
that would be observed in the U.S. in the absence of anthropogenic emissions of PM and
precursor emissions of VOC,  NOX, and SOX in  North America. Estimating background
concentrations is important for the health risk analyses presented in Chapter VI and the
assessment of fine particle concentrations and visibility effects in Chapter VIII.
       Background levels of PM vary by geographic location and season.  The natural
component of the background arises from physical processes of the atmosphere that entrain
small particles of crustal material (i.e., soil) as  well as emissions of organic particles and
nitrate precursors resulting from natural combustion sources such as wildfire. In addition,
certain vegetation can emit fine organic aerosols as well as vapor phase precursors or organic
particles.  Biogenic sources and volcanos also emit sulfate precursors.  The exact magnitude
of this natural portion of PM  for a given geographic location can not  be precisely determined
because it is difficult to distinguish from the long-range transport of anthropogenic particles
and precursors.   Based on published reports that attempt to construct a representation of total
PM mass from the sum of estimated natural contributions for the PM components noted above,
the criteria document provides broad estimates of background PM levels for longer averaging
times as shown in Table IV-3.

-------
                                           IV-13
         TABLE FV-3. PM10 AND PM2 5 REGIONAL BACKGROUND LEVELS

PMIO , annual average
PM2 5 , annual average
Western U.S. (/tg/m3)
4-8
1-4
Eastern U.S. (jig/m3)
5-11
2-5
Source: CD, page 6-44.  The lower bounds of the above ranges are based on compilations of natural versus human-
made emission levels, ambient measurements in remote areas, and regression studies using human-made and/or
natural tracers (NAPAP, 1991; Trijonis, 1982). The upper bounds are derived from the multi-year annual averages
of the "clean" remote monitoring sites in the IMPROVE network (Malm et al., 1994).  It is important to note,
however, that the IMPROVE data used here reflect the effects of background and anthropogenic emissions from
within North America and therefore provide conservative estimates of the upper bounds.
As noted in the estimates, there is a definite geographic trend to these levels with the lower
values applicable to the Western U.S. and the higher values applicable to the Eastern U.S.
The Eastern U.S. is estimated to have more natural organic fine particles and more water
associated  with hygroscopic fine particles than the West.
       The range of expected background concentrations on a short-term basis is much
broader. Specific natural events such as wildfires, volcanic eruptions, and dust storms can
lead to very high levels of PM comparable to or greater than those observed in polluted urban
atmospheres.   Because such excursions are essentially uncontrollable,  EPA has developed an
"natural events" policy that removes consideration of them from attainment decisions.10
Disregarding such large and unique events, some estimate of the range of "typical" background
on a daily  basis can be obtained from reviewing various multi-year data as  well as special field
studies.  On very clean days, IMPROVE daily measurements are less than  1 /zg/m3 of PM2 5.
On some days  atmospheric conditions are more conducive to accumulation  and formation of
PM from both  natural and anthropogenic emissions sources. Upper bound  estimates of daily
     Under the most recent statement (Nichols, 19%), EPA will exercise its discretion not to designate areas as
nonattainment and/or to discount data in circumstances where an area would attain but for exceedances that result
from uncontrollable natural events. Three categories of natural PM events are specified: volcanic or seismic activity,
wildland fires, and high wind dust events.

-------
                                         IV-14
background as high as 12 /tg/m3 PM10 have been made based on short-duration studies in
remote "clean" areas of the Eastern U.S. (Wolff et al., 1983).   Observed peak to mean ratios in
natural areas over much longer time periods can provide a rough guide to the highest 24 hour
levels arising from "routine" natural emissions and  meteorology conducive to maximum particle
accumulation.  Because such meteorology appears prevalent in the Southeastern US, staff
developed 24-hour peak to annual mean ratios for PM2 5 data taken from the four Southeastern
IMPROVE sites (Bachmann, 1996).  If one assumes that the broad  regional distribution of
anthropogenic and natural sources of PM are somewhat similar,  present day observed peak to
mean ratios of 2 to 4 can be assumed to apply to the background annual values in Table IV-3.
This estimation approach suggests that the highest  background 24 hour PM2 5 levels over the
course of a year could be on the order of 15 to 20 ftg/m3.
C.     Air Quality Implications for Interpreting F.pideminlngical  Studies
       Based on the examination of the substantial body of data, the CD concludes that the
differences in exposure relationships alone of fine and coarse fraction particles are sufficient to
justify the consideration of fine and coarse particles as separate classes of pollutants (CD page
13-94). The CD  notes that the likelihood of ambient fine mode particles being significant
contributors to PM-related health effects in sensitive populations (discussed in Chapter V of
this Staff Paper) is related to the linkages between fluctations in outdoor concentrations of PM
and personal exposure to outdoor PM, particularly in indoor environments where people spend
most of their time and where many chronically ill elderly can be expected to spend all their
time (U.S. EPA 1989a; Spengler et al.,  1981). In this regard, while both fine and coarse
fraction particles  can penetrate indoors with similar efficiency (CD, Sections  7.2, 7.7, and
13.2.7; Wallace,  1996; Koutrakis et al., 1992; Lioy et al., 1990), once inside, the longer
residence time of fine particles compared to coarse fraction particles enhances the probability
of a linkage between fluctuations in outdoor concentrations and day-to-day population
exposures for fine mode particles of outdoor origin,  as compared to coarse fraction particles of
outdoor origin (DRI, 1995; CD, Sections 7.6 and 13.2.7; Wallace,  1996; Anuszewski et al.,
1992).  In addition, the more uniform distribution of fine particles expected across many
urban areas with regionally elevated concentrations and their well-correlated variation from

-------
                                         IV-15
site to site within a given city mean that fine particle measurements at central monitors may
provide a better indicator of day-to-day variations in potential exposure to outdoor particles
(CD, Section 13.2.7; Burton et al., 1996; Wallace,  1996; Wilson and Suh, 1996).
1.     Representativeness of Central Monitor Measurements of PM Exposures
       The CD concludes that central monitoring can be a useful, if imprecise, index for
representing the average exposure of people in a community to PM of outdoor origin (CD,
Chapter 7; Tamura et al., 1996; Wallace, 1996;  Tamura and Ando, 1994; Suh et al., 1993).
Thus, for both the prospective cohort and time series epidemiological studies, it appears
reasonable to use a representative central monitor or spatially averaged group of monitors to
represent the mean community exposure to outdoor  PM.
       In addition, the CD concludes that fixed-station ambient PM measurements (e.g.,
PM10, TSP) generally approximate total ambient fine particle exposure more closely than
coarse fraction PM exposure (CD  Chapter 13.4.3).  Within the fine fraction, fixed-station
measurements of ambient sulfates  likely approximate total exposure to sulfates better than
similar measurements of H+ characterize total exposure to acidity because a higher proportion
of SO4=  persists indoors (whereas, H+ is neutralized by indoor ammonia). Thus, the CD
concludes that on balance, available health effects estimates from community studies, whatever
their magnitude and direction, are subject to more uncertainty for the coarse fraction than the
fine mode, and  for H+ than for SO4= (CD,  page 13-52).
       Individual personal exposures to PM can  vary considerably from the concentrations
measured at a monitoring station.  Typically, in the  U.S. PM personal exposure measurements
are higher than  the ambient  PM concentrations due to indoor sources of particles such as
cooking, smoking,  and cleaning.  Because of relative day-to-day consistency within any given
residence of indoor sources  and sinks of PM, the longitudinal (time series) correlation of
personal  exposure of a specific individual  to total indoor PM10 (from both outdoor and indoor
sources) and ambient PM10  can be very high. In homes with minimal indoor sources of PM10,
the R2 values can range above 0.9 when these sources are consistent from day-to-day (CD,
page 7-164).

-------
                                        IV-16
       The CD reports similar high correlations between personal and ambient values of
sulfate in a cross-sectional exposure study (R2 = 0.92 reported in Suh et al. (1993); CD, page
7-105). Similar high correlations for total sulfur were found by Ozkaynak et al. (1996) in the
PTEAM study. These results are noteworthy because unlike PM10 which has both indoor and
outdoor sources, sulfate is virtually all of outdoor origin.  Consequently, only the traits of the
indoor environment, such as air conditioning, modify personal exposures to sulfates while
indoors (CD, page 7-105).  By contrast, the strength of cross sectional comparisons between
total PM10 or PM2 5 personal exposures and ambient concentrations can vary greatly depending
upon the presence of smoking, cooking, or other strong indoor/personal sources (Wallace,
1996).
       The day-to-day relationship between PM concentrations monitored at a central station
and measurements of personal exposure is important to interpreting the time series community
health studies. The CD notes that longitudinal exposure studies are more relevant to
interpreting the time series epidemiologic studies than the cross-sectional exposure analyses
because the cross-sectional studies often are more influenced by the variations in indoor
sources (e.g., one household with a smoker and a smoke-free household)  and sinks between
subjects (CD, Section  7.4.2; Wallace, 1996). Cross-sectional regression analyses of indoor on
outdoor PM2 5 and PM10 concentrations generally explain less than half of the variance (R2 <
0.50); however, longitudinal regressions (for a single home measured over a series of days)
often have much better indoor-outdoor relationships (R2 ranging up to 0.9) (CD,  Section 7.8).
       Thus, the CD concludes that measurements of daily variations of ambient PM
concentrations, as used in the time series epidemiologic studies presented  in Chapter V, have a
plausible linkage to the daily variations of human exposures to PM from ambient sources for
the populations represented by the ambient monitoring stations (CD,  Chapter 7).  The CD
concludes that this linkage will be better for indicators of fine particles than for indicators of
fine plus coarse particles (i.e., PM10 or TSP).
2.     PM2 5 and PM10 Comparisons in Areas Relevant to the Health Studies
       Figure IV-9 shows the locations of selected community health studies which reported
positive, statistically significant associations between  short-term exposure to PM and excess

-------
FIGURE  IV-9  Locations where community epidemiology studies associating short-term PM exposure with
               mortality were conducted in North America.
                                                                                                          Boston/
                                                                                                          Watertown,
                                                                                                            A
                                                                                                    Philadelphia, PA
 Chicago, IL  j Steubenvllle,OH

        Incinnattl, OH
 Locations of PM studies using a variety of PM indicators (e.g. PM,0
 (See CD tables 12-2 through 12-5)
, SQ, , TSP) and reporting statistically significant results

-------
                                          IV-17
mortality, which are discussed in Chapter V.  Significantly, despite the fact that most of the
PM10 non-attainment areas are mainly in the Western U.S. (see Figure IV-4), the mortality
studies were conducted mainly in Eastern U.S. cities, many of which attain the current
standards.  The eastern sites where studies were conducted have a higher level of regional fine
particles (as shown in Figures IV-6 and IV-7). Table IV-4 presents available information about
fine particle concentrations in selected cities relevant to the health studies.
       By contrast, the coarse fraction in the eastern U.S. is lower, on both an absolute
concentration and relative fraction of PM10 basis than in the Western U.S. In the Eastern
U.S., less than half of the daily PM10 mass concentration is coarse fraction material. The
seasonal coarse fraction to PM10 ratios in the Northeast, for instance, range from 0.36 to 0.38,
with an average of all seasons of 0.37 (SAI, 1996).
       The Western U.S. has a more complicated pattern of fine and coarse particles because
of its more complex mix of sources, topography, and seasonal variability.  In some western
urban areas, fine particle levels can be equal to or greater than those observed in the Eastern
U.S. (see Table IV-4).  Urban areas such as Los Angeles, CA, Utah  Valley, UT, and Denver,
CO, have relatively high contributions of local precursor emissions that may contribute to the
formation of fine particles.
D.     Air Quality Implications for Risk Management Strategies
       Through the state implementation plan process, State and local agencies are responsible
for adopting strategies to control PM in areas  with violations of the PM NAAQS.11
Conversely, areas  that currently meet the PM10 NAAQS are not required to implement any
controls. In non-attainment areas, the implementing agency typically selects control strategies
based on its evaluation of which strategies are most effective at reducing PM10 concentrations
contributing to an exceedance, considering the ability of the area or source to implement the
controls and cost.  Accordingly, implementing agencies take into account financial costs,
         In moderate non-attainment areas, the CAA requires the application of reasonably available control
measures (RACM) and the attainment of the NAAQS as expeditiously as practicable.  The expeditiousness test
requires the application of reasonably available control technology (RACT).  EPA provides guidance on
RACM/RACT. Under the guidance, States have flexibility in choosing the mix of controls used to attain the NAAQS.

-------
                        TABLE  IV-4.   PM25 CONCENTRATIONS  IN SELECTED CITIES
LOCATION
Boston, MA
Detroit, MI
Harriman, TN
Los Angeles-Long
Beach, CA
Minneapolis-St .
Paul, MN
New York, NY
Philadelphia, PA
Portage, WI
Riverside-San
Bernardino, CA
Salt Lake City-
Ogden, UT
St. Louis', MO
Steubenville-
Weirton, OH-WV
Topeka, KS
MONITOR TYPES
(Number)
SPM ( 1 )
SPM (4)
SPM**
SPM (2)
Unknown ( 1 )
SPM (4)
SLAMS (1)
SPM**
SLAMS ( 1 )
SPM (1)
Unknown ( 1 )
SLAMS ( 2 )
SPM (5)
Unknown ( 1 )
SPM (4)
SPM**
YEARS OF
DATA
COLLECTION
1986-88
1988-92
1980-87
1988-89
1986-87
1986-93
1986-91
1979-87
1988-89
1986-88
1985-93
1990-91
1979-88
TOTAL NUMBER
OF
OBSERVATIONS
AT SELECTED
MONITOR*
193
149
1,481
90
98
309
249
1,436
111
121
44
51
1,432
AVERAGE PMjS
ANNUAL MEAN
(pg/m')
19.2
22.4
20.8
32.0
13.0
39.5
20.9
11.2
42.8
29.3
16.0
25.7
12.2
2nd HIGHEST
PM,5 VALUE
(A/g/m')
55
73
-
88
38
91
47
-
114
91
49
81
-
YEAR OF
2nd
HIGHEST
VALUE
1986
1989
-
1988
1986
1988
1987
-
1989
1988
1987
1990
-
                                                                                                              I
                                                                                                              I—•
                                                                                                              03
Key:  SPM   - Special Purpose Monitor
      SPM** - Data from dichotomous  virtual  impactors  reported  in Schwartz et al. (1996a)
      SLAMS - State + Local Air Monitoring System
      *With multiple monitors in an  area, monitor with highest  in annual mean selected.

-------
                                         IV-18
availability of technology, suitability of the measure to the specific problem, legal authority of
the implementing agency over the emission source (e.g., local sources within a jurisdiction are
normally controlled rather than sources of long range transport), and other factors.  Because
the current standards use a PM10 indicator, the extent to which any strategy controls fine or
coarse particles is not currently a consideration. As long as the strategies adopted can be
reliably demonstrated to provide for expeditious attainment of the standards, EPA does not
require one specific measure over another in moderate non-attainment areas. Coarse fraction
particles may be preferentially controlled because of their larger contribution to PM10 mass
concentration in some areas, their local impact, and the relatively  lower cost per ton removed.
       Of the 83 PM10 nonattainment areas shown in Figure IV-4, 37 are eligible for
redesignation to attainment, based on air quality data for 1992 to 1994, and an additional seven
have preliminary data which suggest they may also be meeting the current standards. The
implementation of the PM10 NAAQS encompasses diverse sources and solutions.  The major
sources contributing to PM non-attainment areas include fugitive dust, woodsmoke,  stationary
sources (e.g., including stacks and materials processing fugitive emissions from steel mills),
and mixed  areas (that may include the above sources plus additional sources such as regional
transport or motor vehicles).
       Table IV-5 presents additional information on the non-attainment areas and the progress
towards attainment based on air quality data.  Areas dominated by residential woodsmoke and
stationary sources have made the most improvement to meet the PM NAAQS, as measured by
the number of areas with improved air quality data. Areas with fugitive dust problems and
mixed sources (most of which have a fugitive dust problem  from activities such as
construction and road dust as well as primary and secondary motor vehicle contributions and
other sources) have made less progress because local areas with large mobile source
contributions have difficulty reducing these emissions and  areas with windblown fugitive dust
problems are often unable or have limited ability to control the major sources of their problems
from soil erosion.

-------
                                        IV-19
 TABLE IV-5.  SUMMARY OF PM10 NON-ATTAINMENT AREAS BY SOURCE TYPE
Dominant Source
Type
Fugitive Dust
Woodsmoke
Stationary Sources
Mixed Sources
Total
Number of PM10
Non-attainment
Areas
23
32
23
5
83
Areas eligible for
redesignation based
on air quality data*
5
20
12
0
37
Difference
18
12
11
5
46
  *    Areas with complete data shown only. Implementing agencies must complete other
      requirements to be redesignated.
      Although implementing agencies have no requirement to consider the relative
contributions of fine and coarse particles to the control strategies adopted, national emission
inventories and special studies provide some limited information about the relative
contributions of fine and coarse fraction particles.  Generally, fugitive dust sources tend to
produce predominantly coarse fraction particles; residential woodsmoke is predominantly
composed of fine particles; and stationary sources typically emit a mixture of fine and coarse
fraction particles from a facility (U.S. EPA, 1995b).
      Because of the heterogenous nature of the sources of PM10, several different types of
complex situations confront implementing agencies. Table IV-6 summarizes the relative
contributions of PM10 sources and solutions in five areas typical of how successful
implementing agencies have dealt with the PM10 NAAQS in each of the broader categories
described above (Blais, 1996).  The additional details in this table make apparent that even  in a
typical community affected mostly by fine particle residential woodsmoke such as Klamath
Falls, OR, as much as 17 percent of the PM10 can be attributed to coarse fraction geological
material prompting the implementing agency to take appropriate steps to curb these coarse
PM10 emissions. Some mixed source areas may be able to meet the NAAQS by preferentially
controlling the locally emitted coarse fraction particles without controlling fine particles.
      The PM  NAAQS program has not historically focused on the reduction of PM
precursors to reduce PM concentrations except in a few special situations (e.g., Los Angeles,

-------
                                       TABLE IV-6.   PM10 NAAQS IMPLEMENTATION CASE STUDIES SUMMARY
                 Location
                                              Percent of
                                            Annual PM10
                                            Concentration   Sources
                                                                               Control Strategies
                                                                                                           Predominant
                                                                                                              Fraction
Fugitive Dust    Coachcella Valley, CA
                   Highest daily average PM1Q
                    712ug/m3 (in 1989)

                      Annual Average PM10
                    90.2 ug/m3 (in 1989)
Woodsmoke     Klamath Falls, OR
                   Highest dally average PM10
                    792 ug/m3  (in 1988)
Stationary
Sources
Steubenville, OH
                   Highest dally average PM10
                    176 ug/m3  (in 1989)

                      Annual Average PM10
                    45.4 ug/m3  (in 1989)
Typical Eastern  Philadelphia, PA
U.S. Attainment
Area
                   Highest dally average PM1Q
                     97 ug/m3  (in 1989)

                     Annual Average PM10
                    40.3 ug/m3 (in 1989)
                 "Hot spot" PGW
                   Highest dally average PM10
                     567 ug/m3 (in 1993)

                      Annual Average PM1Q
                     110 ug/m3 (in 1994)
                               30     Windblown dust from erosion
                               20     Windblown dust from human activities such as
                                      resuspension by vehicle traffic and suspension by
                                      constniction, agricultural and recreational activities
                               13     Vegetative burning
                               8     Motor vehicle emissions
                               7     Ammonium nitrate (transported from LA Basin)
                               6     Ammonium sulfate (transported from LA Basin)
                                                                                                              •Miz* parking Ms wid unpmd ro»d» and ihoukton.
                                                  No controls
                                                  PlMordwmollyl
                                                  limit mhid* ipewts on unpwed read«. tracton of windbn«lu. itrad Mmping
                                                  wHenng. rangetifion, * restriction! on conttniction. dMnoMon.t agricultural «cttvHi.«
                                                  Transferring waste to energy-conversion plant
                                                  Conversion of county's diesel bus fleet to natural gas
                                                  No controls
                                                  No controls
                                                           C
                                                           C
                                                           F&C
                                                           F
                                                           F
                                                           F
                               73     Residential wood combustion
                               17     Geological material

                               2      Secondary aerosols
                               2      Vegetative burning
                               1      Industrial
                                                  Woodstove replacement and burning bans
                                                  Replace highway sanding with liquid deicing. street sweeping,
                                                  control of track out from unpaved roads and constniction sites
                                                  No controls
                                                  No controls
                                                  No controls
                                                           F
                                                           C

                                                           F&C
                                                           F
                                                           F
 56*     Steel mills (stack and fugitive process emissions,
         fugitive dust from paved and unpaved roads,
         storage piles, and parking lots)

 32*     Electric utilities
  6*      Mobile sources
  6*      Road dust
	* based on Emissions Inventory estimates
Increased chemical wet suppression of unpaved roadways; parking        F&C
areas; raw material, scrap and slag separation, processing, * storage
piles; enclosure of rail and truck unloading station; switch boiler fuel;
add control equipment to blast furnace; vent blast furnace bleeder to boilers
No additional controls beyond Acid Rain program            F
No additional controls                                       F
No controls                                                C
Western Denver, CO 35*
U.S. Mixed
Sources



Highest dally average PM10
189ug/m3 (in 1987)
Annual Average PM10
49 ug/m3 (in 1987)
33*
7*

* .
                                                       Utilities and industrial boilers
                                                        (Ammonium nitrate & sulfate)
                                                       Reentrained road dust
                                                       Residential wood combusion
                                                 * apportionment from high concentration day
                                                                               Restrictions on oil use, limits for NOx and SOx emissions     F

                                                                               Switching to alternative materials; enhanced street sweeping  C
                                                                               Restriction on burning, conversion to cleaner heating          F
                                                                               technologies
                                      Sources not characterized because area
                                      attains PM10 Standards
                                                  No additional controls
                                                                                                                                                            F&C
                                      Sorting scrap metal; processing slag from casting
                                      Melting, smelting, and refining; fugitive emissions
                                      from furnace

                                      Source: Blais, 1996
                                                  Control of emissions from slag piles
                                                  Enclose blast furnace
                                                  (enforced via consent order, not SIP)
                                                           C
                                                           F
                                                  Key.
                                                  F = Fine Mode
                                                                                                 C = Coarse Fraction

-------
                                         IV-20
CA, and Provo, UT).  Although the CAA requires consideration of secondary PM,
implementing agencies are not required to control sources which are not within their non-
attainment area or if source-receptor relationships are not established.  Many non-attainment
areas explicitly do not consider the control of secondary fine PM transported into their area
from other sources (e.g., regional background from Ohio River Valley affecting Steubenville,
OH, and secondary fine particles from LA Basin affecting Coachella Valley, CA).  Instead,
implementing agencies preferentially control locally generated coarse and fine fraction sources.

-------
                                          V-l
V.     CRITICAL ELEMENTS IN THE REVIEW OF THE PRIMARY STANDARDS
A.     Introduction
       This chapter summarizes key information relevant to assessing the known and
potential health effects associated with airborne PM, alone and in combination with other
pollutants that are routinely present in the ambient air.  A more comprehensive discussion of
this information can be found in Chapters 10 - 13 of the Criteria Document (EPA, 1996).
The presentation here organizes the key health effects information into those critical elements
essential for the evaluation of current and alternative standards for PM. Specifically, this
chapter summarizes: 1) key dosimetry information and hypotheses regarding mechanisms by
which particles that penetrate to and deposit in various regions of the respiratory tract may
potentially exert effects; 2) the nature of effects that have been reported to  be associated with
PM in  community air, largely drawn from the more recent epidemiologic information, 3) the
identification of sensitive populations and subgroups that appear to be at greater risk to the
effects of community  air containing PM; 4) issues raised in assessing community
epidemiologic evidence on PM, including alternative interpretations of the evidence; and 5)
evidence and alternative interpretations of the effects associated with the two major
components of ambient PM10, fine and coarse fraction particles.
       The discussions of hypothesized mechanisms, effects,  sensitive populations, and
epidemiology include  consideration of the full range of particle sizes and composition
commonly found in urban and regional air.  The PM epidemiological data base has greatly
expanded since the last review, and suggests a variety of health effects are  associated with
ambient PM at concentrations extending from those found in  the London episodes down to
levels currently experienced in a number of U.S. cities (CD,  p 13-1).  Although a number of
measures of PM have been used in such studies, based on an integrated assessment of the full
range of laboratory and observational data, the revised CD and this staff assessment conclude
that the ambient particles of greatest concern to health remain those smaller than 10 /*m
diameter. Accordingly, the discussion of effects, sensitive populations, and epidemiology
highlights quantitative information on PM10, but also includes some quantitative and
qualitative information derived from studies of physical  and chemical components of PMU).
Based on atmospheric considerations summarized here in Chapter IV and supporting health

-------
                                          V-2
evidence,  the CD recommends separate consideration of the fine and coarse fractions of
PM10.  The final section of this Chapter evaluates the extent to which the available
quantitative and qualitative evidence might be used to support  separate standards for the fine
and coarse fractions of PM10.
B.     Mechanisms
       This  section briefly summarizes available information concerning the penetration and
deposition of particles in the respiratory tract and outlines hypothesized physiological and
pathological responses to PM.  It is important to emphasize that, at present, available
toxicological and clinical information yields no demonstrated biological mechanism(s) that
can explain the associations between ambient PM exposure and mortality and morbidity
reported in community epidemiologic studies.  Thus,  any discussion of possible mechanisms
linking ambient PM exposures to mortality and morbidity effects is necessarily limited to
hypotheses derived from animal or human studies conducted at exposure levels of PM
constituents  far higher than  found  in ambient air. The major purposes of the discussion
presented  here is to identify available information of greatest relevance that helps identify
those fractions of PM that are most likely to be of concern to health, to examine possible
links between ambient particles deposited in various regions of the respiratory tract and
reported effects in humans,  and to focus attention on  the kinds of  mechanistic research
needed to provide a biological basis for elucidating mechanisms that may provide support for
a causal link between ambient PM exposures and reported  health effects.  An expanded
treatment  of key particle dosimetry considerations, potential mechanisms by which  PM
exposure  is  hypothesized to produce effects in humans at ambient  exposure levels, and  the
limitations of the current human clinical and toxicological database can be found in Appendix
D and in  Chapters 10, 11, and 13 of the CD.
        An  evaluation of the ways by which inhaled particles  might ultimately affect human
health must take account of patterns of deposition and clearance in the respiratory tract. The
human respiratory tract can be divided into three main regions: (1) extra-thoracic, (2)
tracheobronchial, and (3) alveolar regions (CD, Table 10-1, Figure 10-5).  The regions differ
markedly in structure, function, size, mechanisms of deposition, and sensitivity or  reactivity
to deposited particles (U.S. EPA,  1982b, CD,  Figure 10-6).  The junction of conducting and

-------
                                          V-3
respiratory airways appears to be a key anatomic  focus; many inhaled particles of critical
size are deposited in the respiratory bronchioles that lie just distal to this junction,  and many
of the changes characteristic of emphysema involve respiratory bronchioles and alveolar ducts
(Hogg et al., 1968). Retention of deposited particles depends on clearance and translocation
mechanisms that vary with each of the three regions (See Appendix D).  Coughing,
mucociliary transport, endocytosis by macrophages  or epithelial cells, and dissolution and
absorption into the blood or lymph are important  mechanisms of clearance in the
tracheobronchial region.  Endocytosis by macrophage or epithelial cells and dissolution and
absorption into the blood or lymph are the dominant mechanisms of clearance in the alveolar
region (CD, pp. 10-55, 56).
      Figure V-l illustrates the regional deposition of particle distributions of varying
aerodynamic diameter.  In essence,  regional deposition of ambient particles in the
respiratory tract does not occur at divisions clearly  corresponding to the atmospheric aerosol
distributions shown in Chapter IV.   The CD provides simulations of deposition of ambient
particle  distributions that indicate fine and coarse  particles are deposited in  both the
tracheobronchial and alveolar regions (CD, Chapter 10). Table V-l provides estimated
deposition patterns in the human lung for typical particle size distributions found in
Philadelphia and Phoenix; these simulations  are for adult males with normal breathing.  The
CD shows that as mouth-breathing or workload increases so does deposition in the bronchial
and alveolar regions.  For those individuals  considered to be  mouth breathers, deposition
increases  for coarse particles in the tracheobronchial region (CD, pp. 166-168).
      Evidence from epidemiological studies of occupational and  historical community
exposures and laboratory studies of animal and human responses to simulated ambient particle
components suggests that at exposures well above current standards, particles may produce
physiological and ultimately pathological effects by  a variety  of mechanisms. The previous
criteria  and standards review included an integrated extensive examination of available literature
on the potential mechanisms, consequences,  and observed  responses to particle deposition
organized according to major regions of the  respiratory  tract  (EPA, 1982b).  Based on this
assessment and the composition of typical urban PM, staff  concluded, with CASAC concurrence

-------
                                           V-3a
                                                                        100
                                MMAD (urn) with og = 1.8
         0.01
          1              10
MMAD (|jm) with og = 2.4
100
            0   Alveolar (Normal)   *  TB (Normal)    °  Total Thoracic (Normal)

            0   Alveolar (Mouth)    *  TB (Mouth)    D  Total Thoracic (Mouth)
FigureV-1.  Human respiratory tract PM deposition fraction versus mass median aerodynamic
diameter (MMAD) with two different geometric standard distributions (og = 1.8 or og = 2.4).
Alveolar, tracheobronchial, or total thoracic deposition fractions predicted for normal augmenter
versus mouth breather adult male using a general population (ICRP66) minute volume activity
pattern and the 1994 ICRP66 model.  After CD, Figure 13-3.

-------
                                             V-4
TABLE V-l.  MODELED 24-HR REGIONAL DEPOSITION FOR MEASURED
AMBIENT PARTICLE SIZE DISTRIBUTIONS  (After CD Tables 10-21, 23)*
City
Philadelphia
Phoenix
Particle
Fraction
Fine
Coarse
Fine
Coarse
Mode Size
(MMAD)
0.436 /xm
28.8 /xm
0.188 jum
16.4 /xm
Total Mass
Deposition
84 Mg
270 - 330 jxg**
42 Mg
440 - 530 Mg**
Tracheobronchial
Deposition
9Mg
3 - 7 Mg**
8/xg
10- 15 Mg**
Alveolar
Deposition
37 Mg
1-12 ^g**
26 ME
12 -29/xg**
'Results for normal breathing for adult males. Particle size distribution from impactor data.  Total mas.s assumed
50 /xg/m3.
**Separate estimated deposition of "intermodal" peak of 2.3 to 2.6 //m in the original table is excluded for clarity,
and because this peak may be an artifact of the sampling.  Because it is possible that much of this mass (intermode)
may be the "tail" of the coarse mode fraction, a range is given for coarse mode mass.  The lower bound is the
original estimate for the coarse mode.  The upper bound is the sum of the estimates for the coarse model plus the
intermode. This may tend to overstate coarse mode deposition relative to fine, which also contributes to the
intermode.
(Friedlander, 1982), that particles that deposit in the thoracic region (tracheobronchial and
alveolar regions),  i.e. particles smaller than 10 txm diameter, were of greatest concern  for
standard setting.  The staff identified a number of potential mechanisms and supporting
observations by which common components of ambient particles that deposit in  the thoracic
region, alone or in combination with pollutant gases, might produce health effects (Table 5-2.
EPA,  1982b).  While there has been little doubt in the scientific community that the
historical London  air pollution episodes had profound effects on daily mortality  and
morbidity, no combination of the mechanisms/observations advanced in the last  review has
been sufficiently tested  or generally accepted as explaining the historical  community results.
Moreover, as  noted above, the potential mechanisms cited in the last review were based on
insights developed from laboratory and occupational/community epidemiological studies that
involved concentrations that are substantially higher than those observed  in current U.S.
atmospheres, and in many cases  using laboratory  generated particles that may be of limited
relevance to community exposures.

-------
                                          V-5
       As discussed in the CD, the significant body of new epidemiologic evidence that has
accumulated since the last review of PM criteria and standards provides "evidence that
serious health effects (mortality, exacerbation of chronic disease, increased hospital
admissions, etc.) are associated with exposures to ambient levels of PM found in
contemporary U.S. urban airsheds even at concentrations below current U.S.  PM standards"
(CD, p. 13-1).  This increasing evidence has prompted renewed interest in generating
testable hypotheses regarding potential  mechanisms that might ultimately provide support for
a causal link  between health effects and particle exposure at these much  lower levels.  Table
V-2 provides a very general summary of recent thinking concerning how particles may affect
sensitive subpopulations as more fully discussed in the Criteria Document (CD,  pp. 13-67 to
72, CD, pp.  11-179 to 185)  and in Appendix D of this paper.
       Because Table V-2 condenses and groups a number of hypotheses that have appeared
in the literature and the CD  in a summary  fashion, several points should be noted.  A
complete definition of mechanisms of action for PM would involve description of the
pathogenesis or origin and development of any related diseases or processes resulting in
premature mortality; this is not currently possible.  Some of the entries in the Table,  on the
other hand, may be more accurately described  as intermediate responses potentially caused
by PM exposure rather than  complete mechanisms.  The descriptions provide some rationale
as to how such responses might conceivably contribute to the types of clinically relevant
health endpoints reported in  the literature,  although evidence for action at low concentrations
is presently lacking. It appears unlikely that the complex mixes of particles that are present
in community air pollution would act alone though any single pathway of response.
Accordingly,  it is plausible that several responses might occur in concert to produce reported
health endpoints. Some of the hypotheses  in the Table may be more  likely to be associated
with effects from short-term rather than long-term exposure to PM, while others may relate
to both.  It is also important to note that a number of recent investigations have begun to
examine promising new approaches involving new animal models, methods of concentrating
ambient particles, and examination of the possibly more toxic constituents of PM such as
ultra-fine particles and transition metals. This  work, as well as  future research, should
provide important insights on mechanisms  for the next standards review.

-------
                                                           V-6
Table V-2. Hypothesized Mechanisms of PM Toxicity*
  Response
Description
  Increased Airflow
  Obstruction
PM exposure may aggravate existing respiratory symptoms which feature airway obstruction.
PM-induced airway narrowing or airway obstruction from increased mucous secretion may
increase abnormal ventilation/perfusion ratios in the lung and create hypoxia.  Hypoxia may lead
to cardiac arrhythmias and other cardiac electrophysiologic responses that in turn  may lead to
ventricular fibrillation and ultimately cardiac arrest.  For those experiencing airflow obstruction,
increased airflow into non-obstructed areas of the lung may lead to increased particle deposition
and subsequent deleterious effects on remaining lung tissue, further exacerbating existing disease
processes.  More frequent and severe symptoms may be present or more rapid loss of function.
  Impaired Clearance
PM exposure may impair clearance by promoting hypersecretion of mucus which in turn results
in plugging of airways.   Alterations in clearance may also extend the time that particles or
potentially harmful biogenic aerosols reside in the tracheobronchial region of the lung.
Consequently alterations  in clearance from either disturbance of the mucociliary escalator or of
macrophage function may increase susceptibility to infection, produce an inflammatory response,
or amplify the response to increased burdens  of PM.  Acid aerosols impair mucociliary clearance.
  Altered Host Defense
Responses to an immunological challenge (e.g., infection), may enhance the subsequent response
to inhalation of nonspecific material (e.g., PM).  PM exposure may also act directly on
macrophage function which may not only affect clearance of particles but  also increase
susceptibility and severity of  infection by altering their immunological function.  Therefore,
depression or over-activation  of the immune system, caused by exposure to PM, may be involved
in the pathogenesis of lung disease.  Decreased respiratory defense may result in increased risk of
mortality from pneumonia and increased  morbidity (e.g., infection).

-------
                                                           V-7
 Cardiovascular
 Perturbation
Pulmonary responses to PM exposure may include hypoxia, bronchoconstriction, apnea, impaired
diffusion, and production of inflammatory  mediators that can contribute to cardiovascular
perturbation.  Inhaled particles could act at the level of the pulmonary vasculature by increasing
pulmonary vascular resistance and further increase ventilation/perfusion abnormalities and
hypoxia.  Generalized hypoxia could result in pulmonary hypertension and interstitial edema that
would impose further workload on the heart.  In addition, mediators released during an
inflammatory response could  cause release of factors in the clotting cascade  that may lead to
increased risk of thrombus formation in the  vascular system.  Finally, direct stimulation by PM
of respiratory  receptors found throughout the respiratory  tract may have direct cardiovascular
effects (e.g., bradycardia, hypertension, arrythmia,  apnea and cardiac arrest).
  Epithelial Lining
  Changes
PM or its pathophysiological reaction products may act at the alveolar capillary membrane by
increasing the diffusion distances across the respiratory membrane (by increasing its thickness)
and causing abnormal ventilation/perfusion ratios.  Inflammation caused by PM may increase
"leakiness" in pulmonary capillaries leading eventually to increased fluid transudation and
possibly to interstitial edema in susceptible individuals.  PM induced changes in the surfactant
layer leading to increased  surface tension would have the same effect.
  Inflammatory Response
Diseases which increase susceptibility to PM toxicity involve inflammatory response (e.g.,
asthma, COPD, and infection).  PM may induce or enhance inflammatory responses in the lung
which may lead to increased permeability, diffusion abnormality, or increased risk of thrombus
formation in vascular system.  Inflammation from  PM exposure may also decrease phagocytosis
by alveolar macrophages and therefore reduce particle clearance. (See discussions above  for
other inflammatory effects from  PM exposure.)
*Summarization from the CD (p. 13-67 to 72; p. 11-179 to  185) and Appendix D of this document.

-------
                                           V-8
       In conclusion, dosimetric information shows that both fine and coarse fraction
particles smaller than 10 fj.m can penetrate and deposit in the tracheobronchial and alveolar
regions of the lung. Particles also may carry other harmful substances with them to these
regions with the smaller particles having the greatest  surface area available for such transport
(see section IV). While a variety of responses to constituents of ambient PM have been
hypothesized to contribute to the reported health effects, there is no currently accepted
mechanism(s) as to how relatively low concentrations of ambient PM may cause the health
effects that have been reported in the epidemiologic literature.  Therefore, there is an urgent
need to expand ongoing research on the mechanisms by which PM, alone and in combination
with other air pollutants, may cause adverse health effects.
C.     Nature of Effects
       The evidence for the kinds of health effects associated with exposures to PM comes
from a large body of literature dating back more than 40 years.   This  section reviews and
discusses the findings and conclusions concerning the principal health effects associated with
PM exposure contained in the CD (CD, Chapters  11,12,13).  Evidence for such conclusions
and findings  as well as  for associations drawn from epidemiological studies, controlled
human exposures,  and animal toxicology is discussed  and evaluated in the  CD (CD, Chapters
11, 12,  and  13), Appendix D of this document, and below.  For reasons presented in the
previous section, it is more likely that such  effects  are primarily  related to particles smaller
than 10 /xm in diameter.  Evidence with respect to  the fine and coarse fractions of PMIO is
discussed in Section V.F.
       The scientific information discussed and evaluated in the  CD and in this staff paper
suggests that the key health effects categories associated with PM include:
       • Increased Mortality
       • Indices of Morbidity associated  with  Respiratory and Cardiovascular Disease
              • Hospital Admissions and Emergency Department Visits
              • School Absences
              • Work  Loss Days
              • Restricted Activity Days

-------
                                              V-9
       • Effects on Lung Function and Symptoms
       • Morphological Changes
       • Altered Host Defense Mechanisms
Most of the effects categories listed above have been consistently associated with PM
exposure from a number of community epidemiological studies,  with supporting insights
from animal toxicology and controlled human exposures of various constituents of PM
conducted at higher-than-ambient levels.  Primary  evidence of PM-related morbidity comes
from indicators of aggravation of existing disease.  In addition, while mechanisms of lung
injury by particles have not been elucidated, there  is agreement that the cardio-respiratory
system is the major target.
       Before discussing the effects, it is important to note some key characteristics and
limitations of the kinds of studies used to identify them.  The strengths and weakness of
epidemiological studies in general are discussed  in some detail in the CD throughout
Chapters 12 and 13.  While epidemiological studies alone cannot be used to demonstrate
mechanisms of action, they can provide evidence useful in making inferences with regard to
causal relationships, as in the case of cardiovascular disease and cigarette smoking (CD,
Chapter 12). The CD discusses criteria  for the use of epidemiological studies as an aid to
inferring cause-effect relationships rather than merely establishing associations (CD, Section
12.1.2). It  then reviews the criteria used to assess the scientific quality of epidemiological
studies  of community air pollution containing PM1.  Particularly important issues and
uncertainties for evaluation of the PM epidemiology studies  are related to model
specification, control for potential confounders, exposure  misclassification, and consistency
and  coherence.  These issues are discussed in detail in the CD and summarized here in
Section 5.E.
       Based on a comprehensive evaluation of the extensive published community data, the
CD  concludes that  "the weight of epidemiologic evidence indicates that ambient PM exposure
has affected the  public health of U.S. populations" (CD, p.  13-27).   As the CD points out,
    1 Community air pollution refers to the mix of outdoor ambient PM and other pollutants that occur in typical
urban/suburban atmospheres.

-------
                                          V-10
however, "little non-epidemiologic evidence is presently available to either support or refute
a causal  relationship (i.e., to construct an exposure-dose-response continuum) between low
ambient  concentrations of PM and increased morbidity and mortality risks" (CD, p.  13-27 to
28).
        Under ideal circumstances,  animal toxicology and controlled human exposure studies
can provide qualitative and quantitative support for environmental epidemiology.  In  the case
of PM, however, the lack of published experimental human and laboratory animal studies
involving relevant exposure levels and experimental subjects representative of sensitive
subpopulations identified in the epidemiological studies presents problems in providing an
integrated assessment (CD, p 13-2).  Epidemiological studies describe relationships between
regionally and temporally variable mixtures of particles and gases in community air pollution
and mortality and  morbidity in  sensitive populations — most notably the elderly and
individuals with cardiopulmonary disease, which includes adults and children with asthma.
In contrast,  experimental studies of PM effects  in humans tend to use healthy young  adult
humans (or  those with only mild disease) and examine mainly reversible physiologic  and
biochemical effects from exposure to laboratory-generated acidic aerosols, sulfates or
nitrates.  Similarly, experimental  studies on laboratory animals have tended to use genetically
homogenous healthy animals to examine a broader range of effects from individual
components of the PM mix.  In both animal and human studies, the limited number of
individuals exposed greatly limits the ability to detect effects at concentrations close to
ambient  levels.  In addition, extrapolation of quantitative and qualitative results from animal
studies to human is encumbered by  methodologic difficulties from differences in dosimetry.
The various species used in inhalation toxicological studies do not receive identical doses in
comparable  respiratory tract regions  when exposed to identical aerosols  (see Appendix D).
Consequently few  laboratory experiments have used appropriate models of susceptibility to
PM which limits evaluation of possible mechanisms and potential quantitative effects
comparisons.
       However, at least qualitative  support for some of the epidemiologic observations has
been reported for specific components of the ambient  particle mix in controlled clinical
studies of humans  as well as studies  in animals.  For  such studies,  the biological responses

-------
                                         V-ll
occurring in the respiratory tract following PM inhalation encompass a range of effects
including: respiratory symptoms such as wheeze and coughing, changes in pulmonary
function, altered mucocilary clearance, inflammation, changes in lung morphology and tumor
formation (CD, p. 13-70, p. 11-1).  In the vast majority of studies, however,  results were
observed only at concentrations of specific substances or simple mixtures that  are
significantly higher than those found in contemporary atmospheres.  Because the health
effects produced by PM exposure are dependent on  the chemical composition, size, and
concentration of particles, as well as species tested,  these aspects of experimental paradigms
used to characterize PM toxicity are noted in  the following discussion.  However, in this
discussion, the emphasis  is placed on reported effects of PM in general,  rather than a  specific
emphasis on particle size or composition.
       Key evidence illustrating each of the major effects categories listed above  is outlined
below, with an emphasis  on the more recent information.
1.      Mortality
       a.      Mortality From  Short-Term Exposures to PM
       i.      Historical Findings From Community Epidemiology
       The most notable  reports of the health effects from community air pollution
containing high PM have come from the dramatic pollution episodes of Belgium's industrial
Meuse Valley (Firket,  1931); Donora, Pennsylvania (Schrenk et al., 1949); and London,
England (Ministry of Health, 1954). In these cases, winter weather inversions led to very
high particle concentrations in ambient air, which were associated with large simultaneous
increases in mortality and morbidity (especially among  individuals with preexisting cardio-
pulmonary conditions).  In a ten year follow-up study,  survivors  of the Donora, Pennsylvania
pollution episode with either chronic disease prior to the episode, or those who became
acutely ill during the episode,  were  found to have higher subsequent rates of mortality and
illness (Ciocco and Thompson, 1961).
       Analyses of a series of episodes in London indicated an excess of mortality (mostly
from cardiopulmonary causes) occurred with abrupt  increases in particles (including sulfuric
acid) accompanied by simultaneously high levels of  SO2 (Martin, 1964; Martin and Bradley.
1960).  Although the London studies measured PM as  British Smoke (BS), gravimetric  mass

-------
                                          V-12
calibrations permitted development of quantitative mass-concentration relationships.  There
was general acceptance in the 1982 CD (EPA, 1982a) and in critical reviews of PM-
associated health effects (Ware et al,  1981; Holland et al,  1979) that London air pollution at
high levels (at or above 500 - 1000 /xg/m3 of both pollutants) was causally related to
increased mortality.
       During the previous review of the PM standards, the London mortality  studies were
augmented by several more extensive time-series analyses examining the PM
pollution/mortality relationship across 14  London winters (e.g, Mazumdar et al, 1982;
Schwartz and Marcus, 1986; Ostro, 1984). These  studies  used more sophisticated statistical
techniques  to examine relationships between routine variations in PM and sulfur dioxide
levels and mortality.  Such analyses showed a continuum of response across the full range of
PM levels in London and suggested that effects from exposure to  PM occurred at levels more
similar to those observed in the U.S..  Some of these studies suggested, although  not
conclusively, that  particles were more likely to be responsible for the associations of health
effects with air pollution than SO2 (e.g., Mazumdar et al 1982).  These studies and analysis
of associations of  health effects with the lower levels of PM measured in the 14 London
winters (150 jig/m3 as BS) was influential in the selection of the level of the current 24-hour
PM10 standard (EPA, 1982b; 1986).
       ii. Recent  Findings
       Beginning in  1987, two important  developments took place.  Investigators  began to
use more sophisticated statistical techniques, originally based on econometric techniques, to
further evaluate  the association between short-term  variations in PM and mortality (CD, p
12-32).  In addition the expansion of particle monitoring, related to  the revision of the
standard, increased the  information concerning size-specific PM levels in  cities throughout
the U.S.. From 1987 to present, numerous epidemiological studies  have  reported statistically
significant positive associations2 between short-term exposures to PM and mortality.  In
these studies, investigators have observed statistically significant associations between
   - Unless otherwise noted, statistically significant results are reported at a 95% confidence level.

-------
                                          V-13
increased daily or several-day average concentrations of PM (as measured by a variety of
indices: TSP, PM10, PM2.5> COH, KM, and BS) and excess mortality in communities across
the U.S. as well as in Europe and South America.  Of 38 studies published between 1988
and 1996, most found statistically significant associations between increases in ambient PM
concentration and excess mortality (CD, Table  12-2). These studies are consistent with the
earlier analyses of the London winters, but extend the association to lower concentrations for
a large number of areas with differing climate,  aerosol composition, and amounts of co-
occurring gaseous pollutants such as SO2 and O3.
       Table V-3 presents a comparison of relative risk estimates reported  for PM-related
mortality expressed in terms of a PM10 increment.  A generally consistent association is
found between changes in PM10 levels and mortality in most of these studies, with a range of
2 percent to 8 percent increase in daily mortality for a 50 Mg/m3 increase in PMK) for those
with statistically significant results.  In the studies with statistically  significant results, mean
PM]0 concentrations ranged from 18 to 58 ^g/m3 and maximum daily concentrations from 80
to 365  /ig/m3.  These studies were conducted in a number of different geographic locations in
North America. Each of these locations differ  significantly in pollution and weather patterns.
Yet most of these studies finds a  statistically significant association  between increased
mortality and PM10 that is relatively consistent across the studies. It is of note that a rough
estimate of the  relative risk for a 50 /xg/m3  increase in PM (as PM1U) for the 1952  episode in
London (1.06) is in the range of those reported for the recent studies (Schwartz et.  al..
1994).
       iii.    Specific Causes of Mortality  Associated with PM
       Table V-4 summarizes the relative risks  for total mortality, respiratory  and
cardiovascular causes of  death, and mortality among the elderly for the community studies
evaluating cause of death.  Reported cases of "respiratory related" deaths were assigned to
individuals who had been diagnosed with acute  respiratory illness (e.g., symptoms involving
the upper respiratory tract and pneumonia),  as well as COPD and paeumoconioses when  they
died.   In general, these  studies reported stronger significant relationships  between short-term
PM concentrations and deaths in those with  respiratory and cardiovascular disease  than for

-------
Cover illustration: Locations of recently published community epidemiology studies finding
statistically significant associations between short-term concentrations of paniculate matter and
health effects (CD, Tables 12-2 through 12-5). Studies conducted on three continents have found
both increased morbidity and mortality to be associated with a variety of particle measurement
devices, including mass measurements of TSP, PM10, PM2.5, sulfates, and acids, and optical
based approaches including BS, KM, and COH. Although the highest PM-10 concentrations in
the U.S. are in the West, most of the results in North America are from eastern communities, at
PM-10  concentrations that are generally below those permitted by the current standards.

-------
                                         V-13a
TABLE V-3.  ESTIMATED MORTALITY INCREASE PER 50 //g/m3 INCREASE IN
24-h PM10 CONCENTRATIONS FROM U.S. STUDIES (After CD, Table 13-3)
Study Location
Increased Total Acute
Six Cities"
Portage, WI
Boston, MA
Topeka, KS
St. Louis, MO
Kingston/Knoxville,
Steubenville, OH
St. Louis, MOC
Kingston, TNC
Chicago, ILh
Chicago, ILg
Utah Valley, UTb
Birmingham, ALd
Los Angeles, CAf
RR (± CI)
Only PM
in Model
Mortality

1.04(0.98, 1.09)
1.06(1.04, 1.09)
0.98(0.90, 1.05)
1.03(1.00, 1.05)
TN 1.05(1.00,1.09)
1.05 (1.00, 1.08)
1.08(1.01, 1.12)
1.09(0.94, 1.25)
1.04(1.00, 1.08)
1.03(1.02, 1.04)
1.08(1.05, 1.11)
1.05 (1.01, 1.10)
1.03(1.00, 1.055)
RR (± CI)
Other Pollutants
in Model

—
—
—
—
—
—
—
1.06(0.98, 1.15)
1.09(0.94, 1.26
—
1.02 (1.01, 1.04)
1.19(0.96, 1.47)
—
1.02(0.99, 1.036)
Reported
PM10 Levels
Mean (Min/Max)T


18 (±11. 7)
24 (±12.8)
27 (±16.1)
31 (±16.2)
32 (±14.5)
46 (±32. 3)
28 (1/97)
30 (4/67)
37 (4/365)
38 (NR/128)
47(11/297)
48 (21, 80)
58( 15/177)
References:

•Schwartz et al. (1996a).
'PopeetaJ. (1992. 1994)/O,.
Ttockery et al. (1992)/O,.
"Schwartz (1993).
*Ito and Thurston (1996)/O,.
•Kinney et al. (1995)/O,, CO.
'Slyer et al. (1995).
 'Min/Max 24-h PMW in parentheses unless noted
 otherwise as standard deviation (± S.D), 10 and
 90 percentile (10, 90). NR = not reported.
'Means of several cities.

-------
                                     V-13b
TABLE V-4. COMPARISON OF TOTAL MORTALITY WITH AGE- AND CAUSE-
        SPECIFIC MORTALITY FOR SHORT-TERM EXPOSURE STUDIES
Study
  Total Mortality,
Relative Risk per 50
   jtg/m3 PM10
Age- and Cause-specific
Mortality per 50
        PM10
Respiratory Related

  Utah Valley, Pope et al. (1992)


  Chicago, Styer et al. (1995)


  Chicago, Ito and Thurston (1996)


  Birmingham, Schwartz (1993)*


  Santiago, Chile,  Ostro et al. (1995a)

Elderly

  Chicago, Styer et al. (1995)


  Santiago, Chile,  Ostro et al. (1995a)

Cardiovascular Related

  Utah Valley, Pope et al.  (1992)


  Chicago, Styer et al. (1995)


  Chicago, Ito and Thurston (1996)


  Birmingham, Schwartz (1993)


  Santiago, Chile, Ostro et al. (1995a)
       1.08
    (1.05- 1.11)

       1.04
    (1.00- 1.08)

       1.03
    (1.01, 1.04)

       1.05
    (1.01 - 1.10)

       1.04
   (1.035- 1.06)
        1.04
    (1.00- 1.08)

        1.04
   (1.035 - 1.06)
        1.08
    (1.05 - 1.11)

        1.04
    (1.00- 1.08)

        1.03
    (1.01 - 1.04)

        1.05
    (1.01 - 1.10)

        1.04
    (1.035 - 1.06)
         1.20
     (1.11 - 1.29)

         1.12
     (0.99 - 1.26)

         1.07
      (1.02,  1.12)

         1.08
     (0.88- 1.32)

         1.06
     (1.03 - 1.10)
         1.08
      (1.03 - 1.13)

         1.05
      (1.03 - 1.06)
         1.09
      (1.02 - 1.17)

         1.03
      (0.98 - 1.09)

         1.02
      (1.00- 1.03)

         1.08
      (1.02 - 1.14)

         1.04
      (1.02 - 1.06)
 The Schwartz (1993) study was ot COPD.

-------
                                          V-14
other conditions, as well as a larger effect in the elderly (> 65) than in the general population
(CD, Chapter 12; Styer et al.,  1995; Ostro, 1995a; Schwartz, 1994a; Pope et al., 1992).
The CD notes that the relative risk for respiratory-related mortality was up to 4.3 times as
large as that for total mortality (CD, p. 12-77).   As noted in the CD, such results are
supportive of the biological plausibility of a PM/air pollution effect on mortality.
       iv.     Experimental Animal Studies
       The vast majority of studies examining short-term exposures to animals of
components of PM have found mortality only at concentrations well  above ambient levels of
PM, even in sensitive species (e.g., guinea pig).  Such  studies appear to be of little relevance
to the effects observed in humans  at ambient levels (CD, Table 11-18, p.  11-42,43).
       b.      Mortality From Long-Term Exposures  to PM
       Prior to 1990, cross sectional studies were generally used to evaluate the relationship
between mortality and long-term exposure to PM.  These, as well  as more recent cross-
sectional studies, are summarized  in Tables 12-14 and 12-15 in the CD.  These studies have
reported, for at least one of the experimental designs  used in each  study, statistically
significant positive associations linking higher long-term concentrations of various indices of
PM with higher mortality rates across communities.  However, absent other  supporting
evidence,  the unaddressed confounders and methodological problems inherent in these studies
have limited their usefulness.  The previous staff paper  concluded  that such studies provided
only suggestive evidence of long-term mortality  associated  with PM  exposure (EPA,  1982b).
In the recent literature, however, new prospective cohort studies have reported results that
may lend additional support to  the earlier results.  These studies use subject-specific
information  and appear to provide more reliable findings (CD, section  13.4.1.1), although
the uncertainties in controlling  for a number of factors such as smoking, lifestyle, and
exposure patterns are improved by the design of cohort studies, they remain  greater than for
short-term studies conducted in single communities. The results of three recent studies
(Abbey et al.,  1991; Dockery et al., 1993; Pope et al.,  1995) are summarized in Table V-5
and described briefly below.
       Dockery et al., (1993) analyzed survival  of 8,111 adults followed for 14 years in six
cities in the  eastern U.S. (Six City Study). Extensive information  was obtained regarding

-------
   TABLE V-5. RELATIVE RISK BETWEEN THE MOST POLLUTED AND LEAST POLLUTED CITIES FOR
   TOTAL POPULATION AND FORMER AND CURRENT SMOKERS IN THE PROSPECTIVE COHORT STUDIES

A)     Harvard Six City Study, Dockery et al. (1993)
Endpoint
Total Mortality
Cardiopulmonary
Disease
Lung Cancer
Total Population
RR*
(1
(1
(0.
1.26
.08 - 1.47)
1.37
.11 - 1.68)
1.37
.81 -2.31)
Non-Smokers
RR*
1.19
(0.90- 1.57)
—
—
Former Smokers
RR*
1.35
(1.02- 1.77)
—
—
Current Smokers
RR*
1.32
(1.04- 1.68)
—
—
No Occupational Exposure
RR*
1.17
(0.93 - 1.47)
—
—
                                                                                                                       I
                                                                                                                       ^_1
The results (and 95 percent confidence intervals) were reported in the paper between the city with the highest level of PM2, (Steubenville, OH, average     js
29.6 /xg/m3) and the lowest level of PM2J (Portage, WI, 11.0 /ig/m3).                                                                   *

B)     American Cancer Society Study, Pope et al. (1995)
Endpoint
Total Mortality
Cardiopulmonary
Lung Cancer
Total Population RR**
(1
(I
(0.
1.17
.09 - 1
1.31
.17 - 1
1.03
.80 - 1
.26)
.46)
.33)
Non-Smokers RR**
(1
(1
1
.07
I
.18
.22
- 1
.43
- 1
0.59
(0.23 - 1
.39)
.72)
.52)
Current and Former
(1
(1
(0
1
.05
1
.08
1
.82
.15
- 1
.24
- 1
.07
- 1
Smokers RR**
.26)
.42)
.39)
The results (and 95 percent confidence intervals) were reported in the paper between the city with the highest and the lowest level of PM23 of the 47 cities
examined.
* Per 18.6 /xg/m' increase in PM,5.
**Per 24.5 /ig/nv1 increase in PM:J.

-------
                                          V-15
potential confounders for each individual, including, smoking, education level, and
occupation.  After adjustment for these co-variates, the authors found elevations in several
measures of long-term PM concentration (PM15/10, PM2 5 and  sulfates) were significantly
associated with increases of total mortality.  The adjusted  increase in risk  (26 percent, CI of
8-47 percent) from PM exposure was nearly equal for PM15/10, PM2 5 and sulfates between the
cities with highest and lowest levels of air pollution.
       A second prospective cohort study was conducted by Pope et al. (1995) which used 7-
year survival data, between 1982 and 1989, for over half a million adults  in  151 U.S. cities
[American Cancer Society  (ACS) study].  This study was  designed to follow-up on the
suggestion made from the Six City study that long-term exposure to fine particles is
associated with increased mortality.  To test this hypothesis, the association between multi-
year concentrations of two fine particle indicators, sulfates and PM2 _s, and mortality  was
evaluated.  As in the Six City study, information for each individual was used to adjust for
important risk factors, such as age, sex, race, smoking, passive  smoking,  and occupation.
After adjustment for the other risk factors, PM2 s concentrations were found to be associated
with a 17 percent (CI of 9-26 percent) increase in total mortality, with sulfate concentrations
associated with a 15 percent  (CI of 5-26 percent) increase in  total mortality, between cities
with the least and  most polluted air.
       The Six City study  found somewhat higher RR estimates for mortality than the ACS
study. The  sensitivity of the RR estimates to important confounders can be assessed by
evaluating the effects estimates for different subgroups of the populations  (Table V-5).  Two
subgroups in this population  with high potential  for confounding are smokers and those with
occupational exposures to PM.   With regard to smokers, both the Six City and ACS studies
evaluated the association between fine particle levels and total and cause-specific  mortality by
smoking status.  The ACS  study compared the risk of mortality  associated with PM
separately for those who never  smoked and those who have at one time smoked.  The Six
City study compared risk of  mortality associated with exposure to fine particles for the total
population, former smokers,  current smokers, and nonsmokers.  All categories showed
elevated  risk; only the non-smoking category failed to achieve statistical significance. The
ACS  study,  which had a much  larger population and consequently greater statistical  power.

-------
                                          V-16
found a statistically significant association with total mortality and nonsmokers as well as for
the total population and current and former smokers.  It is possible that the RR estimates are
sensitive to specification of smoking and occupational  exposure, and as such adjusting for
these variables in the Six City study may have been inadequate to fully capture the potential
confounding from these variables.
       The Six City study also evaluated the RR of mortality for the population non-
occupationally exposed, defined as those who report no exposure to gases, fumes or dust.
The RR for non-occupationally exposed individuals similar to that for  non-smokers, but also
did not achieve statistical significance.  The ACS study did not evaluate the occupational
subgroup separately.  However, the authors note that the RR was not  sensitive to the
inclusion of occupational exposure variables after adjusting  for cigarette smoking.
       Some reviewers have raised concerns regarding the adequacy of the adjustment for
confounders in the prospective cohort studies, maintaining that other uncontrolled factors
may well be responsible for the observed mortality rates (Lipfert, 1995; Moolgavkar and
Luebeck, 1996; Moolgavkar,  1994).  In particular, these authors have suggested that  the Six
City Study did not control adequately for smoking and other factors.   However, both  the Six
City Study and the ACS study evaluated the association between  PM and mortality among
never smokers and found relative  risks that were similar in  magnitude, and for the much
larger population in the ACS study, statistically significant.  Lipfert (1995) evaluated  the Six
Cities using State average sedentary lifestyle data.  Based on this evaluation, he suggested
that much of the mortality associations in the  Six Cities might be explained by this additional
factor, if it had been included in the original study.  Aside from  the fact that such State
average data  suffers from the same problems  that have plagued past cross-sectional analyses,
both the Six City Study and the ACS study adjusted for body mass index as well as other
factors using individual specific data that should provide adjustments that are related to
sedentary lifestyle.  The CD notes that it is unlikely that these studies  overlooked plausible
confounders, although the addition of unaccounted factors might  well alter the  magnitude of
the association (CD, 12-180).
       Both the Six City and the ACS studies evaluated  specific causes of mortality
associated  with PM (Table V-5).  As with  the short-term studies, the increase in risk  of

-------
                                          V-17
 mortality associated with PM was mostly attributed to increases in mortality from
 cardiopulmonary causes.  The Six City study reported a 37 percent (CI of 11-68 percent)
 increase in mortality from cardiopulmonary causes associated with PM2 5 levels, after
 adjusting for covariates, between the most polluted and least polluted city.  Similarly, the
 ACS study reported a 31 percent (CI of 17-46 percent) increase in such mortality associated
 with PM2.5 levels, after adjusting for covariates, between the most polluted and least polluted
 city.  Taken together, the ACS study and the Six City study did not find any other
 statistically significant associations between PM levels and specific causes of mortality other
 than from cardiopulmonary causes.
       Neither study showed any statistically significant increase in risk for lung cancer
 associated with undifferentiated fine PM exposure, although the ACS study found a
 significant association with sulfates.   While earlier studies  provided some evidence suggestive
 of an association of increased cancer at high PM exposure levels, the 1982  CD could not
 draw any conclusions with regard to  such an association. Thus,  there continues to be little
 epidemiological evidence for an effect of ambient PM on cancer  rates.  Evidence of potential
 cancer risk from specific paniculate matter components comes from  laboratory studies.
 Polycyclic aromatic hydrocarbons (PAHs), commonly found as combustion  products, are
 perhaps the best studied class of potential carcinogens in PM.  Extracts of organic material
 from particle emissions  have been shown to induce tumors in a variety of studies (CD, p.  11-
 123).  Extrapolation to  human risk from such studies are difficult because of different species
 and age, route of exposure (e.g., not inhalation assays in animals), physico-chemical
 properties of the material, and exposure concentration.  In any event, no clear evidence of
 sulfates acting as a carcinogen have been reported in  the toxicological literature  in the CD.
       A third prospective cohort study of about 6,000 white, nonhispanic,  non-smoking
 long-term residents of California (Abbey et al.,  1991, California  Seventh Day Adventist
 Study), did not find a significant association between  total  mortality and TSP. However, this
 study has more limited  statistical  power than one of the other two studies because of the
 smaller number of deaths  (4 percent of deaths reported in the ACS study).  More
importantly, the PM indicator (days of high TSP) is of questionable usefulness as an
indicator of levels of exposure to PMID or PM2 _,, particularly for cohorts residing in  various

-------
                                          V-18
locations in California.  Cohorts classified with equivalent TSP exposure could experience
varying exposures to fine and coarse fraction particles.  For example, frequently high TSP
exposures to cohorts near the South Coast could have less days of exposure to fine particle
smog, while other cohorts could have similar high TSP exposures from dust storms.
       The CD concludes that the Six City study and the ACS study, taken together with the
earlier cross-sectional studies, suggest possible increases in mortality for specific disease
categories that are consistent with long-term exposure to airborne particles.  Moreover, as
discussed in Chapter 13 of the CD and below, at least some fraction of these deaths likely
reflect cumulative PM impacts above and beyond those seen from acute exposures  (CD, p.
13-34).  To the extent that this is true, additional caution must be used in interpreting these
studies because some of the effects may  be due to historical exposures that are significantly
higher than those used as an index of population exposures in these studies.
       c.     Extent of Life Shortening
       An  important consideration in evaluating mortality effects in  a public health context is
the potential shortening of lifespan ("mortality displacement" or "prematurity of death")
associated  with PM exposure in these studies.  Epidemiological findings suggest ambient PM
exposure affects mortality both in the short and long term, and promotes potentially life-
shortening chronic illness in the long term (CD, p. 13-44).  The relative risk estimates from
the PM mortality cohort studies are considerably larger (Dockery et al,  1993) to somewhat
larger (Pope et al, 1995)  than those from the daily mortality studies, suggesting that a
substantial portion of the deaths associated with long-term PM exposure may be independent
of the daily deaths associated with short-term exposure (CD, p. 13-44).
       Information concerning life shortening of only a few days comes from the daily time-
series studies.  These studies indicate greater incidence and severity of effects are  associated
with  PM exposure in vulnerable individuals, primarily the elderly (i.e., 65 years of age or
older) and individuals with  preexisting respiratory disease.  Thus, it is reasonable  to expect
that some  of the mortality associated with short-term pollution is occurring in the weakest
individuals who might have died within  days even without PM exposure ("harvesting
effect").  Such a pattern is  often seen for some other environmental insults,  such as high

-------
                                          V-19
temperature (Kalkstein, 1991). However, direct evidence from short-term PM exposure
studies concerning the degree of mortality displacement observed is limited (CD, p. 13-44).
       The CD cites only two studies, Spix et al. (1993) and Cifuentes and Lave (1996), that
have attempted to quantitatively test this hypothesis. Their analyses are based on the premise
that if short-term "harvesting" is occurring, an observed increase in  mortality on a  day with
high pollution should result in a corresponding decrease in mortality in subsequent  days.
The analysis by Spix et al. suggests a small portion of the PM-associated  mortality occurs in
individuals who would have died anyway.  The authors speculate, on the  other hand, that
exposure to PM may also lead to the extra stress that causes the death of  a seriously ill
person who may have  otherwise recovered.
       Cifuentes and Lave used two different methods to evaluate the potential for  a
"harvesting  effect" from exposure to PM.  In the first method, they  examined a series of
correlations to test the hypothesis that an increase in mortality in one day  leads to a decrease
in mortality in subsequent days (as evidenced by  negative correlations). They report a
negative correlation for a 2 day lag for all deaths, but it was not significant. While this
result indicates some portion of deaths may be from those who would have died anyway, it is
not an adequate test since it does not consider the effect of previous days  of pollution.  They
extended the analyses by considering "episodes" of pollution, which are defined as multi-day
periods  of relatively high air pollution  that are  preceded and followed by  periods of relatively
low air  pollution.  Their result suggests that there is some mortality  displacement of a few
days occurring in a portion of the population.  However, the Cifuentes  and Lave estimates
are for those deaths which  occur in addition to deaths estimated from the  regression model.
The authors conclude "more research is needed to estimate which fraction, if any of the total
deaths estimated ... is  due to mortality displacement of a few days only".
       An alternative explanation of the observed daily mortality results is that the sensitive
subpopulations for PM effects could be continually changing as people contract disease and
recover (Schwartz,  1994b;  Samet et al.,  1995;  and  Bates, 1992).  Thus, it is possible that
death might be substantially premature if a person becomes  seriously ill and without the extra
stress of PM would otherwise have recovered.  This hypothesis can  be  explored by
evaluating deaths that occur outside the hospital, based on the premise that patients with

-------
                                          V-20
current life-threatening symptoms of disease would be more likely to be in a hospital.
Schwartz (1994c) has reported an increase in sudden deaths for individuals who were not
hospitalized on days with high PM levels in Philadelphia.
       The CD suggests that a portion of deaths associated with long-term exposure to PM
are independent of the short-term exposures and could be on the order of years (CD, p. 13-
45). Quantification of the degree of life shortening observed in the long-term cohort
mortality studies (Dockery et al., 1993; Pope et al., 1995) is difficult and requires
assumptions about life expectancies given other risk factors besides PM exposure, the ages at
which PM-attributable deaths occur, and the general levels of medical care available  in an
area to sensitive subpopulations.  Because of the  uncertainties discussed above,  the CD
concludes that it is not possible to confidently estimate quantitatively  the number of years  lost
(CD, p. 13-45).
2.     Indices of Morbidity Associated with  Respiratory and Cardiovascular Disease
       Given the statistically significant positive associations between community PM
concentrations and mortality outlined above,  it is reasonable to anticipate that the same kinds
of community-based observational studies should  find increased morbidity with elevated
levels of PM.  This is indeed the case where morbidity effects are measured through
increased hospital admissions indicating aggravation of existing disease in the elderly (Table
V-6). There is coherence across these morbidity studies, the mortality studies discussed
above, and discussions of sensitive  subpopulations presented  in section C below.  The
majority of such studies  find effects associated with PM  exposure to be linked to
subpopulations with respiratory or cardiovascular disease (CD, section 13.4.3.5). Numerous
studies have observed positive associations between exposure to PM and responses ranging
from severe effects (e.g., increased hospitalization for respiratory and cardiovascular
conditions) to moderate exacerbation  of respiratory conditions. The key evidence for
associations of PM exposure with such effects is  summarized below.
       a.      Hospital Admissions  and Emergency Department Visits
       A number of epidemiological  studies report statistically significant positive
associations between short-term exposures to PM and hospital admissions for respiratory-
related and cardiac diseases.  Hospital admissions and emergency room visits for these

-------
                                      V-20a
    TABLE V-6.  ESTIMATED INCREASED HOSPITAL ADMISSIONS FOR THE
 ELDERLY PER 50 fig/m3 INCREASE ESI 24-h PM10 CONCENTRATIONS FROM U.S.
                               AND CANADIAN STUDIES
                                     (After CD, Table 13-3)
Study Location
Respiratory Disease
Toronto, CAN1
Tacoma, WAJ
New Haven, CTJ
Cleveland, OHK
Spokane, WAL
CQPD
Minneapolis, MNN
Birmingham, ALM
Spokane, WAL
Detroit, MI°
Pneumonia
Minneapolis, MNN
Birmingham, ALM
Spokane, WAL
Detroit, MI°
Ischemic HP
Detroit, MIP
RR (± CI)
Only PM
in Model

1.23 (1.02, 1.43)*
1.10(1.03, 1.17)
1.06(1.00, 1.13)
1.06(1.00, 1.11)
1.08 (1.04, 1.14)

1.25 (1.10, 1.44)
1.13 (1.04, 1.22)
1.17 (1.08, 1.27)
1.10(1.02, 1.17)

1.08 (1.01, 1.15)
1.09 (1.03, 1.15)
1.06(0.98, 1.13)
—

1.02 (1.01, 1.03)
RR (± CI) Reported
Other Pollutants PM10 Levels
in Model Mean (Min/Max)+

1.12 (0.88, 1.36)* 30-39*
1.11 (1.02, 1.20) 37(14, 67)
1.07(1.01, 1.14) 41 (19, 67)
— 43 (19, 72)
- 46 (16, 83)

- 36 (18, 58)
- 45 (19, 77)
- 46 (16, 83)
- 48 (22, 82)

- 36(18,58)
- 45 (19, 77)
- 46 (16, 83)
1.06(1.02, 1.10) 48 (22, 82)

1.02 (1.00, 1.03) 48 (22, 82)
References:

Thurston et al. (1994)/O,.
'Schwartz (1995)/SO:.
"Schwartz et al. (1996b).
"-Schwartz (1996).
"Schwartz (1994e)
"Schwartz (19940-
"Schwartz (1994d).
pSchwartz and Morris (1995)/O3, CO. SO,
TMin/Max 24-h PM10 in parentheses unless noted
otherwise as standard deviation (± S.D). 10 and
90 percentile (10, 90).  NR = not reported.
Means of several cities.
*RR refers to total population, not just > 65 years

-------
                                          V-21
diseases reflect prevalence, severity, and patterns of health care utilization. Table V-6
summarizes the results for admissions for all respiratory disease and specific respiratory or
cardiovascular diseases such as COPD (emphysema, chronic bronchitis,  bronchiectasis,
asthma, etc,), pneumonia, and heart disease (see also CD, Tables 12-8 to  12-11). Of the  13
studies included in the CD tables, 12 found statistically significant associations between
increases in PM level and increased risk of admission to the hospital,  including evaluation of
cause-specific admissions for respiratory diseases when only PM was  in the model.  As with
the mortality studies, associations between PM exposure and hospital admissions (Table V-6)
have been observed in communities throughout North America (Birmingham, Detroit,
Spokane, Tacoma, New Haven, Utah Valley, New York State, Ontario, Canada).  These
studies reported 6 to 25 perceni increases in hospital admissions  for respiratory disease
associated with a 50 /ig/m3 increase in PM10. Specifically, studies reported 6 to 9 percent
increases in admissions for pneumonia, and  10 to 25 percent increases for COPD for  the
elderly, associated with a 50 ng/m?  increase in PM10.  A recent study  of hospital admissions
for cardiovascular illness (Schwartz and Morris,  1995) reported that PMK, was positively and
significantly associated with daily admissions for ischemic heart disease, with SO2, CO, and
O3 making no independent contribution to the effect.  In the  same study PM10 and CO were
both independently associated with congestive heart failure admissions.
       When viewed together, these studies demonstrate an association between hospital
admissions for respiratory and cardiac  causes and PM exposure (CD,  Chapter  13).  These
results also suggest a  greater effect  on admissions for COPD that for other causes from
exposure to PM, and  are consistent  with those of the mortality studies which also found a
stronger association between respiratory-related mortality and PM exposure than for all
causes of mortality.
       b.      School  Absences. Work Loss Days and Restricted Activity Days
       School absences, restricted activity days,  and work loss days can also be used  as
indicators of acute respiratory conditions, though these are indirect measures compared to
actual diagnosis and measurement of respiratory  conditions.  However, it is not clear  whether
the effects reported in this way result from aggravation of chronic disease (e.g., COPD),
acute infection, or non-specific symptomatic effects. Nevertheless, the results of these

-------
                                          V-22
 studies show consistent statistically significant associations between such measures of
 morbidity and increased short-term levels of indicators of PM.  Ransom and Pope (1992)
 have reported a statistically significant association between PM levels and school  absences;
 this is consistent with an effect from PM exposure, since respiratory conditions are the most
 frequent cause of school absences (CD, Chapter 12).  In addition, three other  studies
 reported statistically significant associations between community air pollution,  as  indicated by
 PM, and work loss days and restricted activity days (Ostro,  1983; Ostro and Rothschild,
 1989;  Ostro, 1987). More specifically, a study by  Ostro and Rothschild (1989) reported
 significant associations between PM exposure  and respiratory-related restricted activity days.
 All of these studies used two- to  four- week lag times between elevations in PM levels and
 school absences, work loss days, and restricted activity days.  This suggests that  not only are
 there immediate effects after elevations of PM exposure (e.g., increased hospital admissions),
 but PM may elicit effects  which are exhibited  at a later time.  These results are consistent
 with a hypothesis of increased susceptibility to respiratory infection resulting from exposure
 to PM.
 3.     Altered Lung Function and Symptoms
       Community epidemiology studies  of ambient PM levels, and studies of exposure of
 humans (clinical studies) and laboratory animals to  PM components, show that PM exposure
 is also associated with  altered lung function and increased respiratory symptoms.   Effects on
 respiratory mechanics can range from mild transient changes with  little direct health
 consequence to incapacitating impairment of breathing.  Symptomatic effects also vary in
 severity, but at minimum  suggest a biological  response that is often more sensitive than lung
 function  measurements.
       a.     Effects Related to Short-Term Exposures To PM
       i.      Community Air Pollution Studies
       Table V-7 lists a number of community studies  highlighted in the CD from U.S.
communities that show associations between PM exposure and both respiratory symptoms
and immediate pulmonary function changes [e.g., forced expiratory capacity for one second
(FEV]) and peak expiratory flow  rate  (PEFR)]. Studies reporting  symptoms have found
associations  between short-term exposures of PM and upper  respiratory  symptoms (e.g.,

-------
                                     V-22a
   TABLE V-7.  ESTIMATED LUNG FUNCTION CHANGES AND RESPIRATORY
  SYMPTOMS PER 50 /ig/m3 INCREASE IN 24-h PM10 CONCENTRATIONS FROM
                  U.S. AND CANADIAN STUDIES (After CD, Table 13-3)
Study Location
Increased Respiratory
Lower Respiratory
Six Cities0
Utah Valley, UTR

Utah Valley, UTS
Cough
Denver, COX
Six Cities0
Utah Valley, UTS
RR (± CI)
Only PM
in Model
Symptoms

2.03 (1.36, 3.04)
1.28 (1.06, 1.56)T
1.01 (0.81, 1.27)*
1.27(1.08, 1.49)

1.09(0.57, 2.10)
1.51 (1.12, 2.05)
1.29 (1.12, 1.48)
RR (± CI) Reported
Other Pollutants PM10 Levels
in Model Mean (Min/Max)+


Similar RR 30 (13,53)
- 46(11/195)

- 76(7/251)

- 22 (0.5/73)
Similar RR 30 (13, 53)
- 76(7/251)
Decrease in Lung Function
Utah Valley, UTR
Utah Valley, UTS
Utah Valley, UTW
55 (24, 86)"
30 (10, 50)"*
29(7,51)"*
- 46(11/195)
- 76(7/251)
- 55 (1,181)
References:

QSchwartz et al. (1994).
"Pope et al. (1991).
'Pope and Dockery (1992).
TSchwartz (1994g)
wPope and Kanner (1993)
*Ostro et al. (1991)
 'Mm/Max 24-h PM,0 in parentheses unless noted
 otherwise as standard deviation (± S.D), 10 and
 90 percentile (10, 90). NR = not reported
 'Children.
 'Asthmatic children and adults.
 Means of several cities.
"PEFR decrease in ml/sec.
"~FEV, decrease.

-------
                                         V-23
hoarseness, sore throat), lower respiratory symptoms (chest pain, phlegm, and wheeze),
fever, cough, and acute respiratory illness.  Additional studies of European communities are
reported in Table 12-12 of the CD.  Four studies from Table 12-12 evaluated respiratory
symptoms in all children (Schwartz et al., 1994; Hoek and Brunekreef, 1993; Hoek and
Brunekreef,1995; Schwartz et al., 1991), ard all but one found positive statistically
significant associations with  exposure to PM with one or more symptoms.  Two studies
evaluated respiratory symptoms in asthmatic children (Pope et al., 1991, Ostro,  1995) and
found statically significant positive associations  with exposure to PM, although  in the Ostro
(1995) study, the effect could not be separated from O3. A study of non-asthmatic
symptomatic and asymptomatic children in Utah Valley found statistically significant positive
associations between increased PM levels and all symptoms in the symptomatic children.
For asymptomatic children,  statistically significant positive and consistent associations were
found between PM exposure and cough, although no statically significant associations were
found for lower respiratory symptoms and inconsistent results for upper respiratory
symptoms (Pope and Dockery, 1992).  The four studies in adults were inconsistent.  Taken
together, these studies  suggest that sensitive individuals, such as children (especially those
with asthma or pre-existing respiratory  symptoms) may have increased or aggravation of
symptoms associated with PM exposure, with or without reduced lung function.
       ii.    Controlled Exposures to  Laboratory Aerosols
       The 1982 CD (EPA,  1982a) and staff paper summarized  earlier literature on
controlled human and occupational exposures to a variety of paniculate substances.  This
summary (Table 5-2, EPA 1982) highlights studies which report that broncho-constriction
and associated symptoms may be induced by chemical or mechanical irritation by high
concentrations of inert dusts (e.g. Andersen et al., 1979; Constantine et al., 1959),  re-
suspended urban dust (Toyama, 1964),  coarse organic dusts (e.g. Dosman, 1980), fine acid
aerosols (e.g. Utell et al. 1981),  and fine particles in combination with pollutant gases
(Koenig et al,  1981; McJilton et al., 1976).
       Measurements of pulmonary  function and symptoms resulting from acid sulfate
aerosols have been a primary focus of PM research in short-term (< 24 hours) controlled
human clinical and animal studies (CD, Table 11.2).   Short exposures to fine H2SO4

-------
                                          V-24
aerosols in environmental chambers, with short periods of exercise, have been reported to
cause a slight concentration-related  increase in lower respiratory symptoms (cough, sputum,
dyspnea, wheeze, chest tightness, substernal irritation) (Avol et. al.,1988a,b).
       Asthmatic subjects appear to be more sensitive than healthy subjects to the effects of
acid aerosols on lung function (Utell et al.,  1982), but the reported effective concentration
differs widely among studies (CD, Table 11-2).  Adolescent asthmatics may be more
sensitive than adult asthmatics and may experience small decrements in lung function in
response to H2SO4 at exposure levels less than 100 jtg/m3 (Koenig et al., 1989; CD, p.  11-
24).  A more recent study of H2SO4 (< 1/zm diameter) on subjects with asthma and COPD
(emphysema or chronic bronchitis)  found pulmonary function decrements at acid  levels as
low as 90 ng/m* (Morrow et al., 1994).  Even in studies reporting an overall absence of
effects on lung function, some individual asthmatic  subjects appear to demonstrate clinically
important effects (CD, p. 11-31).
       Relevant to considerations of the characteristics of acid aerosols that may  elicit effects
in asthmatic subjects, lung function effects in asthmatic  subjects have been correlated with
hydrogen ion content of the sulfate  aerosol (CD, p. 11-17) and affected by neutralization by
oral ammonia (Utell et al., 1983; 1989) and buffering capacity  of the aerosol (Fine et al.,
1987b).  Recent studies also suggest that submicrometer size aerosols may alter lung function
to a greater degree than larger sized aerosols in asthmatic subjects (CD, p. 11-31; Avol et
al., 1988a,b,) albeit at larger concentrations  than found  to affect adolescent asthmatics
(Koenig  et al., 1983, 1989).
       Changes in clinical status of human subjects are  often accompanied by changes in
airway responsiveness as measured by the sensitivity to challenge by a broncho-constrictive
agent. Airway responsiveness may be a predictor of responsiveness to acid aerosol  exposure
in asthmatic subjects (Utell et al, 1983b; Hanley et al.,  1992).   Accordingly, effects from
exposures to pollutants which increase airway responsiveness may be clinically significant
even in the absence of direct effects on lung function (Godfrey, 1993; Wiess et al.,  1993).
Despite the absence of effects on lung function in healthy subjects,  Utell et al.  (1983a)
observed in healthy nonsmokers an increase in airway responsiveness to carbachol challenge
24 hours (but not immediately) following exposure to 450 //g/rn3 H->SO4 (0.8  ^ m diameter).

-------
                                          V-25
which suggests the possibility of delayed effects.  Other studies which have attempted to
measure airway responsiveness immediately after acid aerosol exposure have reported little if
any effect from low  levels of acid aerosol exposure (CD, p. 11-33,34).
       Studies in humans have suggested an increase in airway  responsiveness to O3
following low concentrations of H2SO4 aerosol exposure in both healthy and asthmatic
subjects  (Linn et al., 1994; Frampton et al., 1995; CD).  Synergistic or interactive effects
between  sulfates and SO2 exposure have not been demonstrated  (CD, p.  11-37).  Indeed,
given the low solubility of SO2 in  acid aerosol, it is unlikely that fine acid particles could
facilitate an interaction through transport of SO, to the deeper regions of the lungs, to which
SO2 alone has difficulty penetrating (U.S. EPA,  1994c).  Reflex broncho-constriction by high
levels of SO2 could,  however, increase the deposition  of particles  in the  tracheobronchial
region by narrowing the conductive airways.
       As described  in the CD,  controlled  human studies of PM are limited as they tend to
use pulmonary function and  symptoms from exposure to acid aerosols as the endpoint of
response, and few have examined  airway inflammation or other more sensitive indicators
related to pulmonary function changes.  No studies have examined effects of particles or acid
aerosol exposure on  airway inflammation in asthmatic subjects (CD, p. 11-30).
       Many laboratory animal studies have also been conducted using acid aerosol
exposures with the most recent studies on effects on pulmonary function presented in Table
11-5 of the CD. In  general,  exposure to H2SO4 at levels ranging above ambient but  < 1000
/xg/m3 does not produce direct changes in pulmonary function in healthy animals except in
guinea pigs (CD, Table 11-5).  Airway hyper-responsiveness (alteration  in the degree of
reactivity to exogenous or endogenous bronchoactive agents resulting in increased airway
resistance at levels of these agents that would not affect airways of normal individuals) from
exposure to  (< 1/xm  diameter) H2SO4  particles has been reported in several studies (Chen  et
al., 1992b; Gearhart and Schlesinger,  1986; and El-Fawal and Schlesinger,  1994).  Hyper-
responsiveness  has also  been  observed to be increased in guinea pigs exposed  to acid-coated
particles  in comparison to pure H2SO4 aerosols of the  same size (Amdur and Chen, 1989;
Chen  et al.,  1992b).  Whatever the underlying mechanism, the results of pulmonary function

-------
                                          V-26
studies indicate that H2SO4 is a broncho-active agent and can therefore alter lung function of
exposed animals via contraction of smooth muscle (CD, p. 11-47).
       b.      Effects Related to Long-Term Exposures
       Table V-8 summarizes effects estimates reported from studies highlighted in the CD
which assess the association between long-term exposure to PM and pulmonary function
changes and symptoms of respiratory disease.  Two initial studies conducted in the Harvard
six cities (Ware et al., 1986, Dockery et al., 1989) demonstrated that there is a statistically
significant association of particulate  pollution with respiratory symptoms in children,  with  no
significant changes in lung function.  As noted in the CD, the absence of significant  findings
in lung function effects in the Six  City comparison  may be due to the inherent variability of
the measure.  To follow-up on the suggestions that respiratory symptoms and probably lung
function were associated mostly with fine particle levels and acidity, a more comprehensive
study of 24  cities across North America using the same questionnaire was conducted
(Raizenne et al.,  1996; Dockery et al., 1996).  The cities were chosen to provide a gradient
in aerosol acidity exposures.  Air  monitoring data was collected for one year.  This study
reported statistically significant positive associations between bronchitis and sulfate
concentration  and acidity as well as  between changes in lung function (FVC) and PMIU.
PM25, sulfate particle concentration, and particle acidity indicators.
       Abbey et al. (1995a,b,c) in California reported elevated but marginally non-significant
associations, which were in the range of the results of the other studies, between sulfate
concentration  and bronchitis well as acute obstructive disease, as defined in the studies. Two
other long-term pulmonary function  studies (presented in  Table  12-22 of the CD) reported
decreases in lung function in children (with no confidence level given) (Spector et al.,  1991)
and  statistically significant decreases in lung function in adults (Ackermann-Liebrich  et al.,
1996) associated with long-term PM exposure.
       The  results from  the long-term respiratory symptom studies are consistent and
supportive of those reported for short-term  studies.  The CD concludes that the results are
consistent with a PM gradient (CD,  p. 12-372), and that while the evidence is suggestive  for
long-term exposure to PM being associated with pulmonary lung function decrements,  it is
more limited  (CD, p. 12-202).

-------
                                     V-26a
TABLE V-8.  MORBIDITY EFFECTS ESTIMATES PER INCREMENTS' IN
ANNUAL MEAN LEVELS OF FINE/THORACIC PARTICLE INDICATORS FROM
U.S. AND CANADIAN STUDIES (After CD, Table 13-5).
Type of Health
Effect & Location
Indicator
Increased bronchitis in children
Six City6
Six City0
24 Cityf
24 Cityf
24 Cityf
24 Cityf
Southern California*
Decreased lung function
Six City"-h
Six City1
24 CityiJ
24 City'
24 City1
24 City'
PM,5/,0
TSP
H +
so:
PM.,.,
PMIO
SO,
in children
PMI5,,0
TSP
H" (52 nmoles/m3)
PM,, (15 Mg/m3)
SO: (7 jig/m3)
PM,0 (17 Mg/m3)
Change in Health Indicator per
Increment in PM*
Odds Ratio (95% CI)
3.26(1.13, 10.28)
2.80(1.17, 7.03)
2.65 (1.22, 5.74)
3.02 (1.28, 7.03)
1.97 (0.85, 4.51)
3.29 (0.81, 13.62)
1.39 (0.99, 1.92)

NS Changes
NS Changes
-3.45% (-4.87, -2.01) FVC
-3.21% (-4.98, -1.41) FVC
-3.06% (-4.50, -1.60) FVC
-2.42% (-4.30, -.0.51) FVC
Range of City
PM Levels
Means (jtg/m3)

20-59
39-114
6.2-41.0
18.1-67.3
9.1-17.3
22.0-28.6
—

20-59
39-114
—
—
—
—
"Estimates calculated annual-average PM increments assume: a 100 ^g/m3 increase for TSP; a 50 jig/m3
 increase for PMIO and PM15; a 25 /xg/m3 increase for PM25; and a 15 /ig/m3 increase for SO^, except where
 noted otherwise; a 100 nmole/m3 increase for H*.
"Dockery et al. (1989)
'Ware et al. (1986)
Dockery et al. (1996)
'Abbey et al.  (1995a,b,c)
hNS Changes  = No significant changes.
'Raizenne et al. (1996)
jPollutant data same as for Dockery et al. (1996)

-------
                                         V-27
       The CD points out that the increased risk for respiratory  symptoms and related
respiratory morbidity reported in the above studies is important not only because of the
immediate and longer-term symptoms produced, but also because of the longer-term potential
for increases in the development of chronic lung disease.  Specifically, recurrent childhood
respiratory illness has been suggested to be a risk factor for later susceptibility to lung
damage (Glezen, 1989; Samet, 1983; Gold et al., 1989).
4.     Morphological Damage
       Traditional  epidemiology  has not been used to evaluate the extent to which PM
directly alters lung tissues and components, although some autopsy studies have found
qualitative evidence of a community air pollution effect on the lung (e.g., Ishikawa et al.
1969).  Evidence of morphological damage from PM exposure has come from animal and
occupational studies for acid aerosols and other  PM components.
       a.     Acid Aerosols
       Morphological alterations associated with exposure to acid aerosols have been most
extensively studied and are outlined  in Table 11-6 of the CD.  Single or multiple exposures
to H2SO4 at fairly high levels (> 1 mg/m3) produce a number of characteristic morphological
responses  (e.g., alveolitis, bronchial and/or bronchiolar epithelial desquamation and edema)
(CD,  p. 11-52).  Chronic exposure to H2SO4 at  concentrations <_  1 mg/m3 produces a
response characterized by  hypertrophy and hyperplasia of epithelial secretory cells.  Gearhart
and Schlesinger (1988), however, show that chronic exposure of H2SO4 (250 ^g/m3, 0.3|um)
also produces an increase in the relative number of smaller airways in  rabbits which can be
an early change relevant to clinical small airway disease (CD, p. 11-52).  Long-term (68
months exposure) studies of combinations of SO2 (1.1 mg/m3) and submicrometer sulfuric
acid (90 ^g/m3) exposure  of dogs found no pronounced effects at the end of exposure, but a
number of morphological changes, including an  increase in interalveolar pores (incipient
emphysema), was found to increase  for up to 3  years following exposure (Hyde et al.. 1978;
Gillespe,  1980).
       Morphologic and cellular damage to the  respiratory tract  following exposure to  acid
aerosols may be determined by methods other than  direct  microscopic observation (CD. p.
11-53). Animal studies of exposure to  fine (0.3 /xm) diameter and  ultrafine  (0.04 /j.m)

-------
                                          V-28
diameter H2SO4 aerosols (300 /zg/m3) have reported lavage fluid to contain increases in
lactate dehydrogenase and protein (markers of cytotoxicity and increased cellular
permeability) following a single exposure to guinea pigs (Chen et al., 1992a).
       In addition, modulation of biological mediators of  inflammatory responses (e.g.
eicosanoids) as well as smooth muscle tone (e.g. prostaglandins and  leukotrienes) could be
involved in damage to the respiratory tract after particle exposure.  Changes  in
prostaglandins (Schlesinger et al; 1990b) have also been observed in lung perfusate after
exposure to H2SO4 and lavage.  Since some of the prostaglandins are involved in regulation
of muscle tone,  changes  in these mediators may be involved in the development of airway
responsiveness  found with exposure to acid sulfates (CD,  p. 11-54).
       b.      Silica^ Crustaj Dusts^and other PM Components
       Silica has long been considered to be a major occupational health hazard, with
exposure to crystalline silica  being associated with pulmonary  inflammation and flbrosis (CD,
p. 11-127). The differing forms of silica (amorphous versus crystalline)  are  thought to have
differential potential for toxicity, but data on amorphous forms is limited (CD,  p.  11-128).
There are limited data on ambient concentrations of silica, which is generally found in the
coarse fraction.   Based on analyses of the silica content of resuspended crustal material
collected from several U.S. cities as part of the last review, staff concluded that the risk of
silicosis at levels permitted by the current long-term PM10 NAAQS was low.  This earlier
conclusion is supported by the CD based on the integration of occupational and autopsy
findings with ambient silica concentrations (CD, p.  13-79).
       The 1982 staff paper (U.S. EPA,  1982b) reported  that  some risk of long-term
exposure to crustal dusts is suggested by autopsy studies of farm workers and residents in the
Southwest (Sherwin et al., 1979), desert dwellers (Bar-Ziv and Goldberg, 1974), and  zoo
animals and humans exposed to various crustal  dusts near  or slightly above current ambient
levels in the Southwest (Brambilla et al, 1979). These studies found evidence of a silicate
pneumonoconiosis, which was related to local crustal  materials.  Responses ranged from the
buildup of particles in macrophages with no clinical significance to possible pathological
fibrotic lesions.   No inferences regarding quantitative exposures of concern could be  drawn
from these studies (U.S.  EPA 19825).

-------
                                          V-29
       Kleinman et al. (1995) have reported increases in alveolar wall thickness as well as
alveolar chord length and cross sectional area from exposure of rats to road dust (900 /xg/m3,
4 /xm diameter), ammonium sulfate (70 ng/m3,  0.2 ^m diameter), and ammonium nitrate
(350 ME/m3> 0.6/xm diameter).  The authors suggest such morphometric changes could lead to
a decrease in compliance or a "stiffening" of the lung.
       Coating the surface of particles with certain transition metals, such as iron, may  have
the potential to enhance pulmonary injury to a variety of environmental particles (CD, p. 11-
92; Costa et al., 1994a,b; Tepper et al., 1994).  These metals can catalyze the oxidative
deterioration of biological macromolecules and  thus could potentially cause oxidative injury
to the respiratory tract (CD, p. 11-92). Silica particles have been reported to be rendered
more toxic when complexed with iron.  Rats fed with iron depleted diets (and thus having
less iron available from body stores to complex intratracheally  instilled  silica particles and to
decrease antioxidant molecules in lung tissue) exhibited less inflammation and fibrotic injury
after such exposures (Ghio et al., 1994; 1992; Ghio and  Hatch, 1993).  However, there is
difficulty in extrapolating the results of experimental paradigms used  in these studies
(intratracheally instillation) to ambient exposure situations.
5.     Effects on Host Defense Mechanisms
       Responses to air pollutants often depend upon  their interaction with respiratory tract
defenses such  as clearance and antigenic stimulation of the immune system.   Furthermore.
either depression or over-activation of these systems may be involved in the pathogenesis of
lung diseases (CD, p.  11-55).  Acid aerosols (H2SO4) alter mucociliary  clearance in healthy
human subjects at  levels as low as 100 /xg/m3 with effects being dependent on the
concentration and duration of the acid aerosol exposure,  the size and distribution of the acid
particles, and  the region of the airways being examined (CD, p. 11-56 to 60,  Leikauf et al.,
1984).  In addition, the acidity of the aerosol has  been reported to affect mucociliary
clearance in animals (CD, p. 11-60).  Acid aerosols have been shown to elicit a slowing in
clearance that lasts several months following multiple exposures (Lippmann et al.,  1981).
Persistent impairment of clearance may lead to the inception or progression of acute or
chronic respiratory disease, and may  be a plausible link between acid aerosol exposure and
respiratory disease  (CD, p.  11-61).

-------
                                         V-30
       Little is known about the effects of particles on humoral (antibody) or cell-mediated
immunity.  Since numerous bioaerosols (potential antigens) are present in inhaled air, the
possibility exists that acid sulfates may enhance immunologic reaction and thus produce a
more severe response with greater pulmonary pathogenic potential (CD, p. 11-67).  There is
evidence that H2SO4 exposure may be a factor in promoting lung inflammation by acting as a
vehicle to increase antigenicity (Pinto et al., 1979; CD, p. 11-69).  Guinea pigs have been
reported  to show increased sensitivity to inhaled antigen (ovalbumin) with concurrent H2SO4
exposure (1,910 ^g/m3 < 1 /*m diameter) as demonstrated by hyper-responsive airways
(Osebold et  al.,  1980).   In addition, Fujimaki et al. (1992) have demonstrated that guinea
pigs have altered mast cell function after exposure to high concentrations  of H2SO4  (1000 and
3000 jug/m3).  These cells are involved in allergic responses including broncho-constriction
(CD, p.  11-69).
       Alveolar macrophages not only play a major role in defense against bacteria, but are
involved in the induction and expression of immune reactions, and are capable of release of
pro-inflammatory cytokines (CD, p.  11-56). In order to maintain the function  of clearance,
macrophages must be competent in a number of other functions including phagocytosis,
mobility, and attachment to a surface  (CD,  p.  11-63).
       Macrophages also produce a number of biologically active chemicals which  are
involved in host defense [tumor necrosis factor (TNF) release activity and production of
superoxide radical]  (CD, p.  11-66).  Exposure to H2SO4 (50 to 500 ng/m*, 0.3^m  diameter)
in rabbits produced reductions in TNF cytotoxic activity as well as   reduction in superoxide
radical in alveolar macrophages recovered by lavage (Zelikoff and Schlesinger, 1992).
However, exposure to H2SO4  (300 /*g/m3, 0.3 and 0.04/xm diameter) in guinea pigs
enhanced TNF and  hydrogen peroxide from alveolar macrophages (Chen et al., 1992a).
Such differences in response may reflect either interspecies differences or differences in
experimental conditions.  Kleinman  et al. (1995) have reported in their study of cellular and
immunological injury  by  PM that respiratory burst activity by macrophages was depressed by
exposure to  fine ammonium  sulfate (70 /ig/m3, 0.2 /xin diameter), ammonium nitrate (350
     3, 0.6^m diameter) particles, and road dust (900 ,Mg/m\ 4 /*m  diameter)

-------
                                          V-31
       Animal infectivity models have been used to examine effects of H2SO4 exposure on
susceptibility to bacterial infection. Exposures of up to 1 mg/m3 of submicrometer H2SO4
aerosols for 30 days alone have not resulted in enhanced susceptibility to bacterially-mediated
respiratory disease in mice (See Table 11-8 in the CD).  However, Zelikoff et al. (1994)
demonstrated an effect of high  concentrations of acid alone in rabbits exposed for 2 h/day for
4 days to 500 to 1000 /ng/m3 H2SO4 and demonstrated reduction of intracellular killing and
uptake of the bacterium Staphylococcus aureus by alveolar macrophages.
       Multi-pollutant exposures  have been shown to elicit changes in infectivity in mice
after short-term exposure.  For example, Gardiner et al. (1977) reported increased
susceptibility to infection by exposing mice to O3 (0.1 ppm) followed by H2SO4 (0.9 ing/m1).
Neither pollutant produced any effect alone.  Although conducted using high acid levels, the
results of this study are of particular  interest given the co-occurrence of Q, and acid sulfates
in summertime episodes over broad regions of North America.
D.     Sensitive Subpopulations
       The recent epidemiologic  information summarized in  the CD provides evidence that
several subgroups are apparently more sensitive (susceptible) to the effects of community air
pollution containing PM. As discussed above, observed effects in these groups range from
the decreases in pulmonary  function reported  in children to increased mortality reported in
the elderly and in  individuals with cardiopulmonary disease.   Furthermore,  the same
individual characteristics which can be described in those who succumbed  to air  pollution
during the more extreme historical episodes are also present  in those  most susceptible to
effects during routine fluctuations in  PM level.  Table V-9 is a qualitative assessment  of the
short-term and long-term PM epidemiologic evidence with regard to subgroups that appear to
be at greatest risk with respect to particular health endpoints.  It is a condensation of results
presented in Tables  13-6 and 13-7 of the CD.  The table summarizes the  findings for  the
indicated health indices in the  specified subpopulations.

-------
                                           V-32

      TABLE V-9.  QUALITATIVE SUMMARY OF RECENT PM COMMUNITY
  EPIDEMIOLOGIC RESULTS FOR SHORT- AND LONG - TERM EXPOSURE***
Age Class

Adults


Children

Adults
and
Children
Subpopulation

Elderly
Pre-existing Respiratory
Disease*
Pre-existing
Cardiovascular Disease
General
Pre-existing Respiratory
Disease
Asthmatics
Mortality
Acute Chronic
(Exposure to PM)
-f 0
+ +
+ +
ID +\-
0 0
0 0
Morbidity**
Acute Chronic
(Exposure to PM)
+ 0
+ 0
+ 0
-f +
+ 0
+ +
Lung Function
Change
Acute Chronic
(Exposure to PM)
0 0
0 0
0 0
+ +\-
+ 0
+ 0
* Note, this includes, those with pneumonia, acute bronchitis and COPD.
** Note, morbidity includes hospitalization and emergency room visits, and communit)  morbidit) and symptoms
reported in table 13-6 of the CD.
*** Note; + indicates positive associations have been reported for this group with PM  exposure; -t \- means
few pertinent studies identified, weight of evidence of PM related effect is somewhat positive but uncertain: 0
means that no pertinent studies have been identified: ID means insufficient data, at least 1 pertinent stud)
identified but inference as to weisht of evidence is not warranted.
       The following section expands upon individual risk factors (including age,  asthma,
COPD, and cardiovascular disease), characteristics of those  factors which may increase
inherent susceptibility to PM effects, and incidence of such risk factors (as well as overall
mortality associated with such factors) to provide some perspective on the scope of
subpopulations at risk from PM exposure.  Table 13-9 of the CD presents more detailed
information  concerning the incidence of selected cardiorespiratory disorders by age and by
geographic region.   In addition, Table 12-1 of the CD shows age-specific and age-adjusted
U.S. death rates for selected causes in 1991 and selected components in 1979, 1990, and

-------
                                         V-33
1991.  Information from these tables is incorporated in the discussion below, and gives some
indication of the relative sizes of sensitive subpopulations.  Such subpopulations may
experience effects at lower levels of PM than the general population, and thus, the
subsequent magnitude of effects may be greater.
1.      Individuals with Respiratory and Cardiovascular Disease
       Both  the early  London episode studies and the most recent community  studies in
North America have found air pollution with elevated particle concentrations to be associated
with increased mortality, hospital admissions, and symptoms in individuals with respiratory
and cardiovascular disease (CD, Chapter 13).  Because smoking is associated with the same
types of cardiopulmonary diseases which characterize individuals also susceptible to PM
exposures, smoking is an important variable to be controlled in epidemiologic studies
attempting to investigate the effects of PM (see CD, p. 13-86 for further discussion).
       COPD is the most common pulmonary cause of death, the fourth leading cause of
death overall (84,000  deaths in 1989, U.S. Bureau of the Census 1992), and a major cause of
disability. COPD incidence increases with age of the population (e.g., excluding asthma, the
incidence rate for those over  75 is approximately twice that as for those under 45 years of
age) (CD, Table 13-9).  Patients with COPD have a larger relative risk of mortality from
PM exposure than the general population (CD, Chapter 12, see Section C of this document).
COPD is a broad disease category used to cover patients with varying degrees of chronic
bronchitis, emphysema and asthma,  etc. COPD is characterized by airway obstruction in
which there  is increased resistance to airflow during forced expiration. According to the
International Classification of Disease definitions and classification codes,  COPD includes
chronic bronchitis, emphysema, asthma, and pneumonitis.  Many epidemiology studies use
these codes  and therefore reported effects such as hospital admissions for COPD include
asthma admissions.  The American Thoracic Society only includes emphysema and chronic
bronchitis in their definition of COPD and,  when referring to COPD, the CD  uses this
definition. Subcategories of COPD, emphysema, and chronic bronchitis may result in
chronic inflammation  of distal airways, destruction of the lung parenchyma, and loss of
supportive elastic tissue leading to airway closure during expiration (CD, p. 13-84).

-------
                                          V-34
       Recent community studies summarized in the previous section also found increased
risk from death and morbidity (increased hospital admissions) due to cardiovascular causes
associated with exposure to increased PM concentration (Tables V-4, V-6).  As with COPD,
the preexisting condition of heart disease occurs at high frequency in the general population
and contributes significantly to total mortality (represents 1/3 of all  causes of mortality for all
ages)  (CD, Table 12-1).  The pathophysiology of many lung diseases is related to cardiac
function, and plausible, but undemonstrated mechanisms have been  advanced that suggest
possible links between effects of air pollution exposure and the presence of cardiovascular
disease [Table V-2, Appendix D, Bates (1992)].
2.     Individuals with Infections
       Individuals with respiratory symptoms are at increased risk of morbidity and mortality
from PM  exposure and are often  those with respiratory  infection. Exposure to PM may
exacerbate illness from infectious agents and increase risk of severe outcomes.  In general,
increased  mortality associated with PM exposure from pneumonia and influenza has been
reported  for the elderly.  Mortality rates from pneumonia and influenza combined are just
somewhat lower than  those for COPD and allied conditions (i.e. asthma) (CD, Table  12-1).
As with COPD,  there is also an increased rate of mortality from pneumonia and influenza
with increasing age.   An increase in  respiratory symptoms in children has also been reported
to be associated with PM exposure (see Section C of this Chapter).
3.     The Elderly
       Although recent epidemiology studies suggest higher relative risks for people  over 65
years  of age,  currently little information suggests how aging in the absence of pathology
might make the elderly more susceptible to the effects of ambient particles (Cooper et al.,
1991). Length of exposure increases the cumulative lung burden (dose equals concentration
times  time) which may be related to  susceptibility to particle effects.  The elderly may be
more  sensitive to respiratory  insult from PM because such exposure may have effects on
pulmonary and cardiovascular function which augment decreases seen with increasing age.
In addition,  cardiorespiratory disease  and  infection (e.g., pneumonia and influenza) are more
prevalent in the elderly which may predispose such  individual to effects of PM exposure. In

-------
                                         V-35
people over 75 years of age, 40% have some form of heart disease, 35% have hypertension,
and approximately 10% have COPD (CD, p.  13-84).
4.     Children
       Increased community morbidity, decreased lung function, and increased respiratory
symptoms have been reported to be associated with PM exposure in children, both as a
general group and in individuals with respiratory illness (CD, Table 13-6).  Children have
the potential to be inherently more susceptible to the  effects of PM as they show a greater
incidence of respiratory and other illness, suggesting  decreased immunological protection,
and higher deposition of particles than adults  (CD, p. 10-77).   Children  may spend more
time outdoors and may have higher ventilation rates due to increased activity and thus have
increased inhalation of outdoor pollutants (CD, Chapter 10).   Infants in particular have been
hypothesized to be a sensitive subpopulation for PM  effects as exposure  may increase the
incidence or severity of acute respiratory infection including bronchitis, bronchiolitis, and
pneumonia (Samet et al., 1995).  However, recent studies in North America have not found
clear evidence of increased mortality or morbidity associated with exposure to PM in infants
or children (CD,  Chapter 12).  The rate of mortality from pneumonia and influenza is
relatively high for children  under 1  year of age (11 times that for children 1  to 4 years, twice
that of adults 45-54 years of age) (  CD, Table 12-1).
5.      Asthmatic Individuals
       Asthma is a lung disease characterized by  (1) airways obstruction that is reversible,
but only partially in some patients,  either spontaneously or with treatment, (2) airways
inflammation, and (3) increased airway responsiveness to a variety of stimuli.  The airways
of asthmatics may be hyper-responsive to a variety of stimuli including exercise, cigarette
smoke, odors, irritating fumes, changes in temperature, humidity, allergens,  pollen, dust, as
well as viral infection (CD, p.  13-86).  [A more complete discussion of the characteristics of
asthma may be found in the SO2 Staff Paper (U.S. EPA,  1994c)].  The heightened
responsiveness of the airways of asthmatics to such substances and conditions raises the
possibility  of exacerbation of this pulmonary  disease  by PM.
       Increases in PM have been associated  with increased hospital  admissions for asthma,
worsening  of symptoms, decrements in lung function and increased medication use (CD,

-------
                                          V-36
Chapter 12, Tables V-6, V-7).  There are approximately 13 million people in the U.S. with
asthma and that number is increasing (National Center for Health Statistics, 1994).  Incidence
of asthma is higher among children and young adults, with asthma being the leading cause of
non-infectious respiratory mortality below age 55.  Approximately 70%  of all asthma-related
deaths occur after age 55 (National Center for Health Statistics, 1993).  The available
studies of PM and mortality do not, however, single out asthma from the larger category of
respiratory-related mortality. Thus, from the available evidence a direct association between
PM exposure and asthma mortality has not been  demonstrated.
E.     Evaluation of the Epidemiological Evidence
       The majority of the evidence concerning health effects of PM exposure comes from
epidemiological studies.  While severe  effects at  the high concentrations of air pollution in
the historical episodes are widely accepted as being causally related,  there  is less consensus
as to the most appropriate interpretation of studies  finding associations of health effects with
ambient levels of PM below the current NAAQS (e.g., Schwartz, 1994b; Dockery et at.,
1995; Moolgolvkar, 1995b;  Moolgolvkar and Luebeck,  1996; Li and Roth, 1995; Samet et
al., 1996a; Wyzga and Lipfert, 1995).  Thus, evaluation and interpretation of the
epidemiological studies is key to assessing the weight of the evidence for causal relationships
between health effects and PM exposures at  ambient levels below the NAAQS.  Evaluation
of the epidemiological evidence for these purposes  requires both assessing  the individual
studies as well as the body of evidence as a  whole  for drawing appropriate conclusions.
       The CD summary of perspectives on  the epidemiology studies is  pertinent here:
              "By far the strongest evidence for  ambient PM exposure health risks is derived
       from epidemiologic studies.  Many epidemiologic studies have shown statistically
       significant associations of ambient PM levels with a variety of human health
       endpoints, including mortality, hospital admissions and emergency room visits,
       respiratory illness and symptoms measured in community  surveys, and physiologic
       changes in mechanical pulmonary function.   Associations of both short-term  and long-
       term PM exposure with most of these endpoints have been consistently observed.
       The general internal consistency of the epidemiologic data  base and available findings
       have led to increasing public health concern, due to the severity of  several studied
       endpoints and the frequent demonstration  of associations of health and physiologic
      effects with ambient PM levels at or below  the current U.S. NAAQS for PM1(I.  The
      weight of epidemiologic evidence suggests that ambient PM exposure has  affected the
      public health of U.S. populations.  However, there remains much uncertainty in the

-------
                                          V-37

       published data base regarding the shapes of PM exposure-response relationships, the
       magnitudes and variabilities of risk estimates for PM, the ability to attribute observed
       health effects to specific PM constituents, the time intervals over which  PM health
       effects are manifested, the extent to which findings in one location can be generalized
       to other locations, and the nature and magnitude of the overall public health risk
       imposed by ambient PM exposure.
              The etiology of most air pollution-related health outcomes is highly
       multifactorial, and the effect of ambient air pollution exposure on these outcomes is
       often small in comparison to that of other etiologic factors (e.g., smoking).  Also,
       ambient PM exposure in the U.S. is usually accompanied by exposure to many other
       pollutants, and PM itself is composed of numerous physical and chemical
       components.  Assessment  of the health effects attributable to PM and its constituents
       within  an already-subtle total air pollution effect is difficult even with  well-designed
       studies. Indeed, statistical partitioning of separate pollutant effects may  somewhat
       artificially describe the etiology of effects which actually depend on simultaneous
       exposure to multiple air pollutants.  Furthermore, identification of anatomic sites at
       which particles trigger end-effects and elucidation of biological mechanisms through
       which these effects may be expressed are still at an early  stage.  Thus, it remains
       difficult to form incisive a priori  hypotheses to guide epidemiologic and experimental
       research.  Lack of clear mechanistic understanding also increases  the difficulty with
       which available findings can  be integrated in assessing the coherence of PM-related
       evidence.
              In this regard, several viewpoints currently exist on how best to interpret the
       epidemiology data:  one sees PM exposure indicators as surrogate measures of
       complex ambient air pollution mixtures and reported PM-related effects  represent
       those of the overall mixture; another holds that  reported PM-related effects are
       attributable to PM components (per se) of the air pollution mixture and reflect
       independent PM effects; or PM can be viewed both as a surrogate indicator as well as
       a specific cause of health  effects.  In any case, reduction  of PM exposure would lead
       to reductions  in the frequency and severity of the PM-associated health effects (CD,
       pp.  13-31)."
       The CD also outlines major criteria useful in evaluating the adequacy and strength of
the epidemiological studies and in interpreting them.  These criteria include quality of the
aerometric data, clear definition of study populations and health endpoints, appropriate
statistical  analysis, adequate control of confounders, and evaluation of the consistency and
coherence of the findings with other known facts (CD,  Chapter 12).  The CD  addresses each
of these issues, including both the strengths and inherent limitations of such  studies. The
discussion below in Section V.E.I  focuses on several key factors  identified in  evaluating the

-------
                                          V-38
individual studies and outlines observations on sensitivity to model specification, exposure
error, and potential confounding by weather and other pollutants.  Individual studies can not
be used by themselves to determining whether attributable health effects are occurring from
current levels of PM because of inherent limitations in any single study.  Thus, to evaluate
the potential for PM to effect public health, the collective weight of evidence from studies
must be evaluated together.  Accordingly,  the interpretation of individual studies is followed
by a discussion of the consistency and coherence of the epidemiological evidence across
studies.
1.     Interpretation of Individual PM Study Results
       a.      Model Selection  and Specification
       The recent epidemiological literature contains extensive discussion of model selection
and specification for short-term  mortality studies (CD. Section 12.6.2.1).  The discussion has
focussed on a number of issues  including distributional assumptions, assumptions about
temporal structure or correlation, assumptions about random and systematic components of
variability, assumptions about the shape  of the relationship between response and covariate,
and assumptions about additivity and  interactions of covariates (CD, 13.4.2.3).  Sensitivity of
the effects estimates to model specification has been explored  by many authors, and an in-
depth discussion of model specification for short-term mortality studies is presented  in
Section 12.6.2 of the CD, where PM10 studies  of mortality are reviewed and analyzed (Pope
et al. 1992a; Ostro et al., 1996; Dockery et al.,  1992; Thurston and Kinney, 1995; Kinney et
al., 1995; Ito et al., 1995; Styer et al., 1995).  Also, importantly, alternative TSP mortality
analyses for the same city, Philadelphia (Moolgavkar et al. 1995b; Li  and  Roth,  1995;
Wyzga and Lipfert,  1995; Cifuentes and Lave, 1996; Samet et al., 1995; Schwartz and
Dockery, 1992b) are reviewed and  analyzed. Based  on these assessments,  the models appear
to be most  sensitive to the following specifications:  adjustments for seasonality and for long-
term time trends; adjustments for co-pollutants; and  adjustments for weather  (CD. p. 13-53).
       While  the CD finds that  model specification is important and can influence the health
effect estimates from PM exposure, it also notes  that appropriate modelling strategies have
been adopted by most investigators  (CD, section  13.4.3.2), that have resulted in consistent
PM effects estimates reported across the studies.  These  strategies include  use of several

-------
                                          V-39
standard models (e.g. GLM, LOESS) and a number of particular specifications.  For
example, it is important to remove long-term trends in the data before evaluating the
association between short-term changes in PM and health effects.   As the CD points out, a
several  different methods used by  the various authors are adequate for carrying out this
adjustment, including nonparameteric detrending, use of indicator variables for season and
year, and filtering (CD, section 13.1.3.2).  The CD concludes that, "the largely consistent
specific results, indicative of significant positive associations of ambient PM exposures and
human mortality/morbidity effects, are not model specific, nor are they artifactually derived
due to misspecification of any specific model.  The robustness of the  results of different
modelling strategies and approaches  increases our confidence in  their validity" (CD, p. 13-
54).
       b.     Measurement Error
       A difficulty in interpretation of the epidemiological studies,  particularly for
quantitative purposes, is the determination of uncertainties and possible biases  introduced by
measurement error in the outdoor  monitors.  In the ecological context of  the daily
mortality/morbidity studies, investigators estimate a population-level index of pollution
exposure for those at risk of dying or experiencing illness.  The variation in
mortality/morbidity is modeled implicitly as a function of the variation  in this index.
Measurement error includes both the error in the measurements  themselves and the error
introduced by using a central  monitor to estimate such population-level exposures.  It  is
important to  examine the possible  effect measurement error may  have on  the reported
associations in the studies, as it may bias the results in either direction.  Unfortunately, most
studies provide only qualitative assessments of this issue, as opposed  to their more formal
treatment of weather  and some other confounders.  The  discussion  that follows is drawn from
the CD assessment of the relationship between the monitored pollutant  levels (using TSP,
PM-10, and  fine particles as indicators) and exposure and on how the error in the
measurements might bias the  reported associations.
        The CD points out  that, although generally useful for qualitative epidemiologic
demonstration of PM effects, TSP measurements can include large coarse-mode particles do
not penetrate to the thoracic region.  Thus, TSP can reasonably be expected to provide

-------
                                           V-40
"noisy" estimates of exposure-effect relationships if such relationships are due to thoracic
particle fractions of the measured TSP mass.  By definition, PM1(I is a better index of
thoracic particles than is TSP, and PM10 may be a better index of ambient fine particle
exposure  than TSP because the smaller paniculate fraction contained  in PM10 is more
uniformly distributed in an urban area or region than are larger coarse particles also indexed
by TSP.  As discussed in  Section 13.2.6, PM25 particles are generally likely to be more
uniformly distributed than coarse particles within an urban airshed.  For example,
measurements of the coarse fraction of PM10 appear to be more variable from  site to site,
while PM2 5 levels have been shown to be particularly well correlated across at least  one
eastern metropolitan region, i.e., Philadelphia (Burton et al., 1996; Wilson and Suh, 1996),
as well as in more limited data from Riverside,  CA (Wallace, 1996).  The use of a spatial
average of multiple TSP or PM10 monitors in some studies (e.g.,  Philadelphia, Minneapolis)
can reduce exposure uncertainties for these less uniform pollutant indicators.
       Even if outdoor levels near population centers are well represented by monitors, the
extent to  which outdoor concentration fluctuations  are found to affect indoor concentrations
and personal exposures to outdoor-origin particles  is still an issue of particular importance.
Some of the sensitive populations in the  short-term mortality and  hospital admissions studies
(i.e., the  elderly and those with pre-existing disease)  can be expected to spend more  time
indoors than the general population. Some commentors have expressed concerns regarding
the lack of correlation shown in some cross sectional studies of outdoor and indoor or
personal exposures, and suggest that confounding by indoor sources of PM might bias  the
effects/outdoor PM  response function towards a linear relationship when a threshold  model
may be more appropriate.3 The CD assessment of this issue, however, found longitudinal
correlations of personal exposure to PM10 can be well correlated with outdoor  measurements.
The CD assessment concluded that  "the exposure to indoor-generated particles will not be
   'Implicit in this suggestion is the hypothesis that indoor- and outdoor-generated particles are essentially the
same with respect to those characteristics important to producing particular health effects of concern.  While
some indoor-generated particles may have composition similar to outdoor PM, there may he significant
differences in the adsorbed components, acidity, and other physico-chemical properties of potential importance
that are more unique to particles that originate in a complex urban atmosphere.  The relative importance ot such
factors is critical to testing the above hypothesis.

-------
                                          V-41
correlated with the concentration of ambient (outdoor-generated) particles, and time-series
epidemiology based on ambient measurements will not identify health effects of indoor-
generated particles" (CD, p. 1-10).  Furthermore, the CD assessment of the literature found
that "the measurements of daily variations of ambient PM concentrations, as used in the
time-series epidemiology studies of Chapter 12, have a plausible linkage to the daily
variations of human exposures to PM from ambient sources, for the populations represented
by the ambient monitoring stations. This linkage should be better for indicators of fine
particles (PM25) than for indicators of fine plus coarse particles (PM,0 or TSP), which, in
turn, should be better than indicators of coarse particles (PMK,-2 s)" (CD, p 1-10).  The
strength of the correspondence between  outdoor concentrations and personal exposure levels
on a day-to-day basis serves to reduce, but not eliminate, the potential error introduced by
using outside monitors as a surrogate for personal exposure.
       The effect of instrument and "representativeness" components of measurement error
of PM and other covariates on the association between PM and effects can  vary with
modeling approach.  Measurement error in the exposure variable, PM, in a univariate
regression can bias the association toward the null.  However, in multivariate regressions.
which  are used in the PM literature, the association is also influenced  by the relationship
between PM and the other covariates  which can bias the association in either direction.  This
issue has been discussed  in two recent analyses,  one of cardiovascular hospital admissions in
Detroit,  (Schwartz and Morris, 1995) and the other of mortality in the six cities of the Six
City Study, (Schwartz et al, 1996).  In the cardiovascular hospital admission study, Schwartz
and Morris  discuss the potential influence of measurement error from  the other covariates,
CO and  weather on the PM/cardiovascular hospital  admissions relationship. High correlation
between the covariates and the exposure of interest  represents potential influence of error in
the covariates on the exposure of interest.  They evaluated the correlation between the
covariates and found the correlations between CO levels and the weather variables, and
between CO and PM levels, were small.  In addition, the correlation between  PM levels and
weather variables was also small.  They conclude that such low  correlations may  imply it is
likely  significant portions of bias do not come from the covariates, but from the  error in
measuring PM, which would decrease the association between PM levels with hospital

-------
                                          V-42
admissions.  The authors point out, however, that this does not mean that the estimated
magnitude of the associations was unbiased.
       This issue is explored further in the short-term mortality study in the  six cities of the
Six City Study (Schwartz et al., 1996). The authors examine the potential influence of
measurement error on the association between excess mortality and PM2 5 levels. They note
that the correlations between PM25 level and the  other covariates,  (e.g., weather) are not
large, and thus not likely to influence the measurement error in the level of PM2 5 itself.
They examine this by leaving weather terms out of the regression  model,  which is similar to
a large measurement error  in these terms, and find a slight decrease in  the effects estimate
for exposure to PM25.  They further test the effects of measurement error in the city of
Boston by creating 10 new PM2S exposure variables each based on the original PM2S
measurement with additional random error.  They then repeat the multivariate regression 10
times using each of the 10  new PM25 variables.   They find the mean coefficient for PM
effects with the added measurement error was reduced by 13 % compared to the original
effects coefficient.  These two results suggest that the net effect of random measurement
error in the multiple regression is  to bias toward  underestimating the particle effect.
       Schwartz et al., 1996 did not, however, assess  either the effect of differential
measurement error among the various particulate components, or the effect of other co-
pollutants.  Because coarse fraction particles occurring at the lower concentrations found in
most of the six-cities are likely measured with less  precision  than are fine particles (Rodes
and Evans,  1985), any effects of coarse particles  would tend to be underestimated relative to
fine particles (CD,  p.  13-52). This does not diminish  the significance of the findings for fine
particles or PM10,  particularly in view of the fact that the association remained  highly
significant even when limited to days with PM2 s  concentrations under 25 ^g/m\
Measurement error would be expected to be greater for fine particles at these lower
concentrations than for the full  data set.
       Although the issue of confounding by other  pollutants (e.g., SO2, CO, O,, NOX
NO2) is addressed  in a subsequent  section, measurement  error clearly has  implications for
separating the effects of individual pollutants from a complex urban mixture. When  collinear
pollutants having different degrees of exposure error are  entered into a regression jointly, the

-------
                                          V-43
variable with the least exposure error will tend to be assigned higher significance, all else
being equal (Lipfert and Wyzga,  1995a).
       While the magnitude of measurement error and its effect on the PM/health effect
associations is unknown, it is possible to test potential influences of measurement error in the
PM measure or the influence of other covariates.  Some aspects of these issues have been
discussed in two recent studies, suggesting — although not conclusively — that the influence
of measurement error is to bias the estimate downward.  Nevertheless, a comprehensive,
formal treatment of exposure misclassification studies of PM and other community air
pollutants is an important research need.  As discussed below, however,  the consistency of
the PM/effects relationship in  multiple locations with widely varying indoor/outdoor
conditions and  a variety of monitoring approaches makes it  less  likely  that the observed
findings  are an artifact of exposure misclassification.
       c.     Potential Influence of other Covariates in Short-Term Studies
       Other factors that vary temporally with  PM may influence the estimated relationship
between  PM and health effects, either independently or through  interaction with PM.
Independent risk factors related to both PM concentrations and the health effect of interest
which could potentially confound the apparent associations between  PM exposure  and  health
effects.  Inadequate control for confounding can result in incorrect interpretations,  e.g.,
regarding the reported effect as being the result of an observed risk factor, when a third
variable  (the confounder) is really responsible.   The estimated relationship between PM and
health effects can also be biased up or down by potential interactions between PM  and other
risk factors, particularly other pollutants.
       Significant attention has been  focused on addressing  potential confounders  in the
short-term studies.  The CD points out that it is preferable to control confounding by
designing a study in such a way that  potential confounders are avoided (CD, Section
12.6.3.4). However, in many studies this is not a feasible option because it is not possible to
avoid some potential  confounders, such as weather, and  in some cases, the levels  of PM and
the confounders are highly correlated.  This can also be a problem for areas in which co-
pollutants are derived from a common mixture of sources, such as combustion.

-------
                                          V-44
The CD discusses the difficulty in conducting studies in enough cities to make the
appropriate number of comparisons.  As discussed more fully in section V.E.2 below,
however,  the observed similarities in  relative risk of health effects from PM exposure across
study areas with large differences in the potential for confounding from copollutants adds
credibility to the conclusion that the PM mortality effects are real (CD, p.  12-331).
       Covariates associated with daily changes in health effects, such as weather, season
and levels of other pollutants (e.g., SO2) potentially associated with PM levels need to be
considered.  Most of the epidemiology studies of PM have considered at least some of the
potential confounders in their analysis. These studies have used a number of methods to
address or reduce confounding, with varying degrees of success.  Less attention  has been
given to effects modification from the interaction  between co-occurring pollutants and PM.
A summary of the major issues discussed in the CD regarding the potential influence  of other
potential risk factors on PM and the most relevant PM  studies is presented  below.
       i.      Weather
       Weather  is an important confounder in short-term PM studies because fluctuations in
weather are associated with both changes in PM  and other pollutant levels and health  effects
reported  in the studies4.  Individual studies  have  used a variety of approaches to  separate the
effects of PM exposure and weather with most treatments appearing to be adequate (CD, p.
13-54). Most studies include temperature and dewpoint as covariates  in their studies  (CD, p.
13-54). In addition, many investigators use statistical  methods to adjust for weather and
season on an annual basis when modeling the PM and health effect relationship.   In several
of these studies  (Schwartz, 1993a,  1994a,  1994d, 1994e, 1994f) nonlinear functions have
been used that can reflect the complex relationship between weather and health effects [e.g.,
the effect of temperature in Birmingham, Alabama (Schwartz.  1993a)].  In other studies,
linear and categorical variables were used (e.g.,  for very high  temperature  days) to adjust for
routine fluctuations in weather  and  extreme conditions (Kinney et al.,  1995; Pope et al.,
1992).  In an  examination of the sensitivity of the associations of exposure to PM^ with
   4The relationship between temperature and health effects over the course ot a year tends to be "U" shaped.
with increasing effects on days with very hot or cold temperatures (Moolgavkar and Luebeck. 1996).

-------
                                         V-45
health effects to control for weather, several studies reported distinct effects of weather on
mortality that were largely separable from the effects of PM exposure  in the areas studied.
Moreover,  elimination of all weather variables from the PM-mortality  models did not
substantially affect the size of the observed associations between PM exposure and excess
mortality (Schwartz et al., 1996; Schwartz and Dockery, 1992a, 1992b).
       Because of the limitations in using temperature and humidity alone to examine the
much more complex changes that accompany various weather patterns, two recent studies of
pollution and mortality associations in Utah Valley  (Pope and Kalkstein,  1996)  and
Philadelphia (Samet et al., 1996b)  further examined confounding by weather through the use
of synoptic weather categories.   In these studies the synoptic weather categories were
defined independently of the health effects information, in an approach first recommended by
Kalkstein (1994).  Both studies show that the reported association  between PM  exposure and
excess mortality was relatively insensitive to the changes in weather.  All of the studies of
daily PM levels and  mortality use  some method to adjust for weather,  and report consistent
associations between PM exposure and health effects.
       The CD concludes that the  PM coefficient is relatively insensitive to different methods
of weather adjustment, as recently demonstrated in the recent studies and  the  reanalysis by
HEI (CD, p.  13-54).  Recent studies have adequately addressed the role of weather-related
variables.  (CD, p. 13-54).  Clearly, weather affects human  health;  however, it is highly
unlikely that weather can explain a substantially greater  portion of  the PM attributable health
effects than has already been accounted  for in the models (CD, p.  13-54).
       ii.      Confounding By Other Pollutants
       One of the concerns raised  by a number of authors conducting  reanalyses of the
mortality studies is whether  the observed PM effects are confounded or modified by other
pollutants commonly occurring in  community air such as SO,, O3,  NO2, and CO (Samet et
al., 1995,  1996a; Moolgavkar et al.,  1995b; Moolgavkar and Luebeck, 1996; Li and Roth,
1995).  Based on successive reanalyses, Moolgavkar has advanced  the contention that PM is
serving as  a surrogate for the general ambient air pollution  mixture and that the reported
health effects are more appropriately attributed  to the mixture rather than to PM alone
(Moolgavkar 1995b; Moolgavkar and Luebeck, 1996).  Much of the support for this
interpretation comes from the recent reanalyses of the Philadelphia data where it has proven

-------
                                          V-46
to be difficult to separate individual effects of multiple pollutants (Samet et al., 1995, 1996a;
Moolgavkar et al.,  19955;  Moolgavkar and Luebeck, 1996; Li and Roth,  1995).  The HEI
investigators concluded that "...a single pollutant of the group TSP, SO2,  NO2, and CO
cannot be readily identified as the best predictor of mortality" based only  on analyses of the
Philadelphia data (Samet et al., 1996a).
       The CD examined the evidence for confounding in these  and other studies in some
detail in Section 12.6.  It concludes that other pollutants can play a role in modifying the
relationship between PM and health effects.   The CD also notes  that some studies have  found
little change in the PM relative risk (RR)  after inclusion of other copollutants in the model
and in analyses where the PM RR estimate diminished, the RR typically remained
statistically significant (CD 13-57). Based on an evaluation of the existing studies and its
assessment of confounding within and across a number of areas  with differing combinations
of pollutants, the CD concludes that the PM health effects associations are valid and, in a
number of studies, not seriously confounded by  co-pollutants (CD,  p.  13-57).  The  role of
co-pollutants in modifying  the apparent RR associated with PM is less clear.  The following
discussion summarizes evidence regarding PM confounding  and  effects modification for each
of several criteria pollutants.
       Sulfur Dioxide (SO2).  SO2, which was present at high concentrations with PM during
the historical episodes, has long been  seen as a potential confounder of the PM effect.
Reanalyses of the extensive London data  (Schwartz and Marcus,   1986) provided some
support for the suggestion of Mazumdar et al., (1981) that at lower SO2 values in London,
mortality effects may be associated with PM alone.  The more recent  studies, in particular
short-term exposure  mortality  studies, have applied several approaches to  address SO2
confounding, including restriction (studies in  areas with low SO, levels) and more direct
means.  The discussion below highlights key  findings from the recent  epidemiological studies
together with other pertinent information  from SO2 and PM air quality relationships and from
studies of the penetration of SO2,  alone and in combination with  particles, to the respiratory
tract described below.
       In areas where  the potential  for confounding from SO2 is  relatively high, investigators
have adjusted for SO2 in the model  (Ostro et al., 1995a; Totilomi et al., 1994; Schwartz and

-------
                                          V-47
Dockery,  1992a).  These studies have also conducted sensitivity analysis of the association
between PM and health effects, by evaluating the association before and after adding SO2 to
the model.  These analyses produced inconsistent results.  Studies conducted  in Santiago
Chile, Philadelphia, PA  and  Sao Paulo, Brazil,  found that the association between PM and
mortality  remained positive and significant after the addition of SO2; whereas, the association
between SO2 and mortality became insignificant (Ostro et al.  1996; Schwartz 1992a; Saldiva
et al., 1995). A similar analysis in Athens, Greece found that after modeling both SO, and
PM, the association with SO2 remained significant and  positive  (Touloumi et  al., 1994). The
estimates  of associations with health effects for both pollutants were  reduced, however.
       The PM/SO2 confounding issue has been thoroughly explored in Philadelphia through
extensive  analysis by several investigators, where SO2 and PM  are highly correlated
(Schwartz,  1992a; Moolgavkar,  1995b; Li and  Roth, 1995; Samet et al., 1995,  1996a).  In
these studies, investigators have been  concerned about  the potential  for confounding from
SO2 in the observed TSP/mortality association.  The original analysis by Schwartz and
Dockery (1992a) evaluated the association between TSP and mortality  in Philadelphia
between  1973-1980.  They found the association between TSP and mortality  remained
significant after adding SO, to their model; whereas, the relationship between SO2 and
mortality  became insignificant.   Moolgavkar et  al. (1995b) evaluated the association between
TSP and  mortality  in Philadelphia between  1973-1988.  In this  study,  they attempted to
account better for modification  of the effect of  air pollution  on  mortality by factors that vary
with  season (e.g., weather, pollutant mix, activity patterns).  The Philadelphia daily air
pollution/mortality data  set is one of those large enough to conduct  such seasonal  analyses
without undue loss of statistical power.  Modeled individually,  both pollutants were found to
be significantly  associated with mortality in each  season.  In models where TSP and SO,
where included  simultaneously, they concluded that TSP was positively associated with
mortality in the summer and fall, and SO, was  positively associated in all four seasons5.
    5In a seasonal analysis of the later years of the Philadelphia data (1983-88), Cituentes and La\e (1996)
 found somewhat different results.  In their analysis, SO, was only significant in the winter, and only without
 TSP in the model, while TSP was significant in spring and summer and the coefficient was stable across all
 seasons (CD, p. 12-53).

-------
                                           V-48
       HEI evaluated both of the Philadelphia data sets discussed above (Samet et al., 1995;
Samet et al., 1996d) and conducted their own analysis on data collected directly from the
National Center for Health Statistics and EPA's AIRS database.  Although the overall results
of the reanalyses were similar to those of the original authors, the new HEI analyses used
techniques that revealed a more complex, non-linear set of relationships among pollutants,
season, and mortality.  The authors concluded that the Philadelphia data showed a
relationship between air pollution and mortality, but  that  it would be difficult to use the
results of this single study to attribute such effects solely to particles. The combined
pollutant mortality relationships are of some interest.  The first HEI analysis explored the
relationship between SO, and TSP in depth. The relationship between TSP and mortality
indicates a monotonically increasing response occurs only at particle levels above 100
TSP.  This result  is consistent with either a no-observed- effects level for TSP at 100
or a reduced association caused by a correlation with SO2 at lower concentrations.
Conversely, SO, displays a monotonically increasing concentration response function from
the lowest levels to about 40-60 ppb, where the curve flattens out.  It is difficult to find a
plausible mechanism for such a concentration-response relationship  for a  single pollutant,
suggesting confounding is likely.
       Dockery et al. (1995) commented on the HEI analysis, suggesting  that TSP  and SO,
are indicators of a more appropriate risk factor, such as fine particles. The facts that fine
particle sulfates and SO, share a common source in  Philadelphia and that the coarse fraction
of TSP is poorly correlated with the fine fraction  (CD, Table 6-15) indicate that either or
both pollutants could  reasonably  serve as a  surrogate for  fine particles.   In this event, SO,
itself might play no direct role in causing effects,  with only a fraction of TSP participating.
Resolution of the  merit of the original investigator's  suggested hypothesis,  however, must
await the  results of subsequent studies that  use fine particle indicators in lieu of TSP.
       In  evaluating the findings in Philadelphia, an  important consideration is the evidence
on the penetration  and deposition of particles in the respiratory system as  compared to SO?.
Although  quantitative support is lacking,  the discussion of controlled human and animal
studies of particles indicates that smaller  particles can more effectively penetrate to the
portions of the lung where irritation or other interactions  with lung  tissues  might produce

-------
                                          V-49
effects.  (See section V.A above). Beyond reflex broncho-constriction observed only at very
high peak levels, however, deep lung effects of SO2 are minimal because gas-phase SO2 is
generally efficiently removed in the extrathoracic region in humans (U.S. EPA, 1994c).
This lack of penetration in the lung greatly reduces the likelihood that SO2 alone could
produce significant cardio-pulmonary effects, particularly for sensitive individuals spending
more of their time indoors where SO2 concentrations are low due to rapid removal by indoor
surfaces.  However, one mechanism by which SO2  can be transported deeper into the lung is
absorption or dissolution onto the surfaces of atmospheric particles (See  Section V.F).   In
this case, the complex results reported by HEI  in regard  to effects associated with SO,
exposure might be partially reflecting varying atmospheric interactions of the two pollutants,
rather than a direct SO2 effect.
       Given the difficulty in ascribing effects to a single pollutant in Philadelphia or similar
cities where elevated particles are associated with SO2, confounding by SO, can be addressed
by assessing the PM/mortality relationship in areas with low levels of SO,.  Dockery et al.,
(1993) found no association between SO, and mortality in Kingston and  St. Louis, areas with
considerably  lower SO2 levels.   While consistent associations between PM and health effects
are observed across the different studies, the reported association between  health effects and
SO2 can vary widely.  In Steubenville, the association between SO, and  mortality  was ten-
fold greater than in Philadelphia (i.e., coefficients of 0.0104 versus 0.00132 per ppb)
(Schwartz and Dockery, 1992a,b) although the two areas have comparable SO, levels.
       In a single city such as Philadelphia, where SO2 and PM levels are highly  correlated.
it is more difficult to ascribe the observed mortality effects to a single pollutant.   In such
cases, consideration of the observed relationships and relevant information on air  quality,
indoor exposures, dosimetry, and mechanisms suggest that it is  unlikely  that an independent
effect of SO2 is  occurring that does not involve PM.  Moreover, given the number of studies
using different methods to correct for potential confounding in areas of high and low SO, that
find an association between PM and mortality,  it is unlikely that SO, is responsible for all of
the observed associations between PM and mortality. Similarly, when the more severe
morbidity endpoints such as respiratory-related hospital admissions are considered, the

-------
                                          V-50
presence or absence of SO2 is also seen to have little effect on observed PM associations (see
Table V-ll, Schwartz, 1995a) in most cases.
       Ozone.  The co-occurrence of episodes  involving high temperatures with elevated
levels of O3 and PM raised the potential for confounding, particularly during the O3 season in
large regions of eastern North America, Los Angeles, and some Oiher cities).  In such cases,
covariate adjustment has often been used to  try to distinguish the effects of multiple
pollutants.  A number of studies using such  methods have found PM to be a stronger
predictor of mortality than O3 (Dockery et al. 1992b; Saldiva et al.,  1995; Kinney et al.,
1995; Ostro et al.,  1996).  Adjusting for the presence of O3 did not significantly affect the
associations with PM and mortality.  For example,  in Los Angeles, which has the highest
concentrations of O3 studied, investigators found a significant association between both PM
and O3 mortality when each pollutant was entered into the model separately, but found no
significant  association between O3 and mortality in models that  included PM (Kinney, 1995).
On the other hand,  the coefficient for PM remained stable when O3 was in the model along
with PM, but the uncertainty in the PM association increased,  making it marginally
significant; this finding suggests that the PM-mortality association was not completely
independent of O3 (CD, p. 13-55).  In Santiago, where a negative correlation exists between
O3 and PM levels, no association was observed between O3 and  mortality  across a full year
even without PM in the model; this was despite summertime values of O,  that were twice the
U.S. standard (Ostro et al., 1996).  In the Utah Valley, O, and PM were  also negatively
correlated,  and the  inclusion of O3 as a covariate strengthened  the estimated  PM effect (Pope
et al. 1995a, Table  V-3).  Furthermore, the  relative risk estimates for PM were  relatively
unchanged  and there was little increase in the width of the confidence interval after inclusion
of O3 in the model, and indicating little evidence of confounding of the PM effect (CD, p.
13-52).
       Samet et al., (1996a) extended their analysis of the Philadelphia mortality data by
examining  combinations of multiple pollutants (TSP, O3,  NO2, SO2, and CO).  This analysis
found a low correlation between PM and O3, indicating independence between the two
pollutants.  Ozone had a  stable and significant association  with mortality that appeared to be
independent of the other pollutants.  The effect estimate for TSP was  lowered, but remained

-------
                                         V-51
significant when O3 was added to the model.  The CD reanalysis of the HEI results suggests
that O3 may be a potential confounder of TSP in the summer, but not in other seasons (CD,
p. 12-297).
       In some locations, the potential for O3 to confound the effects caused by PM is
minimized by the low concentrations of O3 observed during seasons which show a robust PM
effect. Examples include Utah Valley and Santa Clara, where O3 levels are minimal in the
winter when  the PM levels are high (Pope et al., 1992a; Fairley,  1990).  The discussion
above of confounding by weather notes a number of cities with cooler climates, where
particles are  associated with mortality, which would have low O3 levels.
       There is a higher potential for O3 confounding for the risk of respiratory-related
morbidity, because  multiple studies have demonstrated apparent separable associations
between respiratory effects and PM and O3 concentrations.  Moreover, the recent  review of
the O3 criteria found that the biological basis for O3 aggravation of respiratory  symptoms was
supported by controlled human and animal studies (EPA, 1986c). The respiratory-related
hospital admission studies often find O3  and PM are each singularly associated  with
respiratory-related admissions (Schwartz, 1994d; Schwartz, 1996; Burnett et al., 1994).
When both pollutants are modeled together, the association  between PM and respiratory-
related admissions in general remains relatively unchanged, indicating a separable effect
independent  of 03.  The potential for O3 confounding for cardiac-related hospital admissions
appears to be much lower.  Two studies have reported that PM is associated with cardiac
hospital admissions but O3 is not (Burnett  et al., 1995; Schwartz and  Morris, 1995).
       Carbon Monoxide  (CO).   The lethality of high concentrations of CO is well
documented; as such, it must be considered as a potential confounder  in community studies
(U.S. EPA,  1991).  Three of the short-term PM exposure studies examined  the effect of CO
on the PM/mortality relationship.  A study in Athens found a significant association between
mortality and CO and PM when each pollutant was considered separately (Touloumi et al.,
1994).   When considered  together, only PM remained significantly associated with
mortality.  However, there was  a high correlation between  CO and PM making such
separation difficult.  Similarly in Los Angeles,  where CO and PM were also correlated,
positive associations between each pollutant and mortality were reported when both  were

-------
                                          V-52
evaluated simultaneously (Kinney et al., 1995).  However, in Chicago, insignificant
associations were reported between CO and mortality (Ito et al.,  1995).  The recent analysis
by HEI of Philadelphia also  evaluated the role of CO in mortality (Samet et al., 1996a).
Similar to the other studies they found a moderate correlation between TSP and CO
concentrations, and they considered  CO, along with SO2 and NO2 to be interrelated with TSP
because of their common sources. Their results show that the average CO concentration on
current and previous day was never  significantly associated with  mortality, whereas CO
lagged by three and four days, was significantly associated with mortality.  The authors note
that  this finding was not expected given the mechanism of CO toxicity and the half-life of
carboxyhemoglobin.  With TSP and  lagged CO in the model, they find both TSP and lagged
CO  level are each significantly associated with mortality.   Based  on an extended analysis of
these results, the CD finds that TSP effects can be  reasonably distinguished from CO in all
seasons (CD, p. 12-297).
       The results  from these studies are inconsistent with respect to CO.  Because of the
nature of urban sources of CO as well as indoor sources,  exposure misclassification may
introduce significant problems, which reduces the ability of community studies to detect a
CO  effect.  In addition, while cardiovascular effects are plausibly linked to CO, controlled
studies do  not suggest  CO is a respiratory irritant (U.S. EPA, 1991).   It is therefore unlikely
to confound studies reporting respiratory related mortality, hospital admissions, or
aggravation of conditions such as asthma, all of which are linked to PM.
       The potential relationship of CO and PM to  cardiovascular effects was examined in
the Schwartz and Morris (1995) study of hospital admissions for  cardiovascular diseases in
Detroit.  They found an association between CO and PM and ischemic heart disease and
congestive heart failure admissions when evaluating each pollutant separately. When
evaluated together,  CO was no longer associated with ischemic heart disease admissions, but
the association with admissions for congestive heart failure for both pollutants remained
relatively unchanged, suggesting each pollutant had a separable,  independent association with
congestive heart failure.  While significant exposure to CO in microenvironments
characterized by high CO levels may render a hypoxic effect on patients with
cardiopulmonary disease, which may aggravate heart disease (see section B above and

-------
                                          V-53
Appendix D), it is unlikely that most patients would be exposed to such a level of CO.  In
addition, once taken to the hospital or to other places with low CO the carboxy hemoglobin
levels of such patients would rapidly decline.
       Nitrogen Dioxide (NO2).  By comparison, fewer of the mortality studies have directly
assessed  NO2 as a potential confounder of PM10 effects. Several such studies have reported
high correlations between NO2 and PM in Los Angeles, CA;  Toronto,  Canada; and Santiago,
Chile (Kinney, 1991, Ostro et al.,  1996, Ozkaynak et al.,  1994).  Mixed results were
reported  concerning the association between NO2 and mortality.   Kinney and Ozkaynak
(1991) found a statistically significant relationship with NO2 and mortality in Los Angeles,
but reported  that these results were interchangeable with CO and PM, since the correlations
were so high between these pollutants.  In Los Angeles and some other Western U.S. cities,
nitrogen  oxide emissions are  themselves a major source of fine particles and nitric  acid.
The Santiago study found, however,  that NO2 was not associated with mortality when
included in the model of PM and mortality (Ostro et al.,  1996).  Furthermore, the
association between PM and  mortality remained  relatively unchanged after addition of NO2 to
the model.  Similar results were found in the Sao Paulo study, where NO2 was not associated
with mortality in adults after including PMUI in the model (Saldiva et al., 1995). All these
studies were conducted in areas of relatively high NO2 levels; Santiago had the lowest mean
level of 0.0556 ppm.  A study in St.  Louis, with  a lower mean level of 0.02 ppm, found no
significant association between mortality and NO2 (Dockery et al., 1992b).  While  the
association between NO2 and health effects in these  studies is inconsistent, the association
between  PM and health effects remains  positive  and consistent, both  across study areas with
varying levels of NO2 and after controlling for NO2 in  the model (Ostro et al.,  1996; Saldiva
et al., 1995; Schwartz et al., 1994).
       NO2 was also included in the multi-pollutant analyses  of mortality in Philadelphia.
Moolgavkar and Luebeck (1996) found  that, when all co-pollutants were entered
simultaneously into their model, NO2 appeared to emerge as the most important pollutant.
By contrast, the recent HEI multi-pollutant analysis (Samet et al., 1996a) of mortality in
Philadelphia found that with  both TSP and NO,  in the  model, the coefficient and the t-value
for TSP increased. NO2, on the other hand, was not significantly associated with mortality

-------
                                          V-54
when modeled alone, and when TSP or all pollutants combined were included in the model,
the coefficient for NO2 became significantly negative.  In essence, the more limited results
for NO2 and mortality to date do not show a consistent association.
 2.    Consistency and Coherence of the Epidemiological Studies
       While individual studies indicate health effects are associated with PM, a more
comprehensive synthesis of the available evidence is needed to evaluate fully the likelihood
of PM causing effects at levels below the current NAAQS.  Because  individual  studies in
themselves are inherently limited as a basis for addressing causality, the consistency and
coherence of the effects across the studies must be considered. As noted above, it is too
difficult to resolve the question of confounding using these results from any single city
because of the correlation among all the pollutants (Samet et al, 1996a).  The HEI
investigators conclude that "insights into the effects of individual  criteria pollutants can be
best gained by assessing  effects across locations having differing  pollutant mixtures and not
from the results of regression models based on data  from single locations" (Samet et al.,
1996a).  The consistency of the association is evidenced by  its repeated observation by
different investigators, in different places, circumstances and time; and by the consistency of
the associations  with other known facts (CD, Chapter  13; Bates,  1992).  A complement to
consistent associations found for individual endpoints is coherence,  which is the logical or
systematic interrelationship among different health indices, which should be demonstrated
across the studies of different endpoints.  As the CD notes,  the discussion of the consistency
and coherence of the epidemiological  studies must be largely qualitative because it relies on a
series of judgments concerning the reliability of the  individual studies (CD, p.  13-58).  The
consistency and the coherence of the PM epidemiological evidence is discussed  and evaluated
below.
       a.      Consistency
       The CD  summarizes over 80 community epidemiological studies evaluating
associations between short-term PM levels and mortality and morbidity endpoints in tables
12-2 and 12-8 to 12-13.  Over 60 of these have  found consistent, positive, significant
associations between short-term PM levels and mortality and morbidity endpoints. These
studies have been conducted  in a number of geographic locations throughout the world,

-------
                                          V-55
including the US, Canada, Europe and Latin America, using a variety of statistical
techniques, and with varying temporal relationships.  Despite the variations in the
approaches, the effects estimate for each health endpoint is relatively consistent among the
studies. Figure V-2 displays the estimated relative risk per 50 /ig/m3 PM10 increase derived
from the U.S. and Canadian short-term studies of mortality and morbidity effects presented
in Tables V-4, V-6, and V-7.
       Clearly, the relative risk estimates exhibit some variation for particular endpoints.
For example,  the relative risk estimates for mortality associated with a 50 /xg/m3 increase in
PM10 range from 1.02 to 1.08.  The CD observes that this kind of variation in the RR
estimates would be expected for the following reasons: 1)  the relative toxicity of  PM  varies
from region to region; 2) the demographic and socioeconomic characteristics of the
population vary regionally; 3) the health status, and thus the distribution of the sensitive
population vary regionally; and 4) ambient PM levels vary regionally.  Thus,  the CD
concludes that some variation in the RR estimates is  not inconsistent with a real effect of PM
exposure on daily mortality (CD,  Section 13-4.1.1).  Similarly, some variation in the RR
estimates for morbidity endpoints would be expected, as is observed in Figure V-2.
       The large number of studies in a number of different geographic areas, provides an
opportunity to evaluate the consistency and sensitivity of the PM estimates to  different levels
of potential influence by weather and copollutants.  Such an evaluation allows consideration
of both the potential for confounding from these factors and interpretation of whether the
observed health effects are attributable to PM or to the complex air pollution  mixture.  As .
for confounding, the CD notes  generally similar RR  estimates for acute mortality in different
studies with different levels of potential confounding copollutants lend credibility to the
conclusion that the PM mortality effects are real (CD, p.  12-33).
        If PM is acting independently, then a  consistent association should be observed  in a
variety of locations of differing relative proportions of particles and potential  gaseous
pollution confounders.  If, instead, the observed PM effect results from influence from
another pollutant, either through confounding or synergistic interaction, the  associations with
PM would be expected to be consistently high in areas with high concentrations of the
pollutant, and consistently low  in areas with  lower concentrations of the pollutant.  In

-------
FIGURE V-2.
Relationship Between Relative Risk per 50 ug/m3 PM,0 and Specific Causes of Mortality and Morbidity in Adults and
Children.
0
O
s~ 26-
o.
o
^l 2-2-
O)
3
O
u 1.8-
a
2 1.4-
Relative
i
0.6



4 Adults x








< ? J « « <




All Children v Symptomatic/Asthmatic Children








I., „...,„!. I











, >
}{} r i



4

I1




















•*











T

i'Hl


                                                                                                                                                              Ul
                                                                                                                                                              Ul
                             Total
                           Mortality
                    Respiratory
                      Mortality
Cardiovascular     Respiratory     COPD or Ischemic
   Mortality   Hospital Admissions    HD* Hospital
                                     Admissions
 Adults                   	»•
Cough      Lower      Upper
          Respiratory  Respiratory
                                                                                                                                 Children
                         Total. Respiratory and Cardiovascular Mortality
                         1. Pope etal. 11992)
                         2. Schwartz (1993)
                         3. Stveretal. (1995)
                         4. Ostroetal. (1995a)
                         5. Ito and Thurston (1996)
                                              Respiratory Hospital AJnissions
                                              1. Sclwartz (1995) NewHaven, CT
                                              2. Schwartz (1995) Tacorna, WA
                                              3. Schv«rtz (1996) Spokane, WA
                                              4. Thurston et al. (1994) Toronto, Canada
                                                    COPD or Ischerric Heart Disease (HP)
                                                    Hospital AJrrissJons
                                                    1. Schvtartz (1994fl, MmeapoHs, MN
                                                    2. Schwartz (1994c), Bimringham AL
                                                    3. Schwartz (19961, Spokane WA
                                                    4. Schwartz (1994d), Detroit M
                                                    <5. Schwartz and IVbnis (1995), Detroit, Ml, Ischerric HD
                                                     Couoh. Lower Respiratory, and Upper Respiratory
                                                     1. Hoek and Brunekreef (1993)
                                                     2. Schwartz etal. (1994)
                                                     3. Pope etal. (1991), astrrrBtic children
                                                     4. Pope and Dockery (1992), synptomatic chidren

-------
                                            V-56
addition, consistent PM effects across a range of pollutants indicates would indicate that it is
more likely that there is an independent effect from PM, that is not confounded by other
components of the air pollution mix.  Figure V-3 shows the reported relative risk of PM10
effects and associated levels of  SO2, NO2, O3, and CO from studies conducted in the U.S. as
reported in Table V-3.  The relative risks are those reported in each of the studies,
unadjusted for the other pollutants.  The figure indicates that the association with PMIO
remains reasonably consistent through a wide range of concentrations of these  potentially
influential pollutants.  While it is possible that different pollutants may serve to confound or
otherwise influence particles in  different areas6, it seems unlikely that this  would lead to
such similar associations and relative risk numbers for particles.  Within the observed range
of relative risk, however, it is certainly possible that other pollutants might modify the
apparent effects of particles by atmospheric interactions (e.g., through dissolution/adsorption
or aerosol formation reactions) or by independent effects on sensitive populations (e.g.
respiratory function changes  from O3 or SO2) as described in the previous  section.
Moreover, the possibility of exposure misclassification for primary gaseous pollutants (e.g.,
CO, SO2) could diminish  their apparent significance.  Nevertheless, epidemiological  studies
have been conducted in a broad range of areas across the U.S. and  Canada, where
meteorological  and pollution patterns vary distinctly.  These studies find a consistent,
positive association between PM and mortality and morbidity effects.   The  CD has concluded
that the effects  are unlikely to be explained by weather (CD, p.  13-54), that the  PM  effects
are not sensitive to other  pollutants  and the  "findings regarding the PM effects are valid"
(CD, p.  13-57).
       b.     Coherence
       In addition to the consistently observed associations for each effect, this collection of
studies shows coherence in the kinds of health effects associated with PM  exposure.  The CD
       6In this interpretation of the results advanced by Moolgavkar and Luebeck (1996), CO, for example,
would lead to a false association with particles in Utah Valley where SO: was low, and SO, would lead to a
false particle signal in Philadelphia, where CO levels were more modest. Such a serendipitous combination ot
variable confounding would make the more ubiquitous pollutant, particles,  appear to be consistently associated
with the effect.  In this event, at least two other pollutants, or an unidentified substance(s) correlated with them.
would be associated with mortality and other effects.

-------
o
°- 1.2-
w
E_
Dl
 1 3-^

T—
•s.
CL
o 1.2-
E
01
^ 1.1-
o
LO
0)
Q- 1

| 0.9-
•5 0.8



Utah Valley

j StBubenville
j Philadelphia
{:1;i }
Portage Chicago
1 StLouis^
St Louisb
a Boston
Kingston.
Kingston
ff i i
"• A r>K n 1 n 1 K n '



o
a- 1.2-
w
E^
D>
* 1.1-
O
LO
OJ
a T
(0
CC
> 0.9-
Jj
0)
01 O.R





Portage
i IM n i
Kingston LA Boston St Louis8 Steubenvile
Chicago Philadelphia
St Loiisb


1 ft O 1 ft ' I I I I
10 u'ia 0 0.02 0.04 0.06 0.08 0.1 C.
SO (ppm), 2nd 24 Hour Maximum
1 •?







J
LA



o
^f
9- 1.2.
w

^ 1.1-
o
in
Q)
°- 1
^ ~
(A
it
5 0.9_
"5
cr r\ D
> n OK

3
Kingston

St Lexis3
UtahVatey
Portage -
« I ^ l' 1 'i
LA Bitrringham Bostor Steubenvil
Chicago St Louisb } Kingston
Phiadelphia

i i i i i i i
                                                                                                                                                            I
                                                                                                                                                           Ui
                        Ozone (ppm), 2nd Hour Maximum
CO, 2nd 8 hour max (ppm)
FIGURE V-3a. Relationship Between Relative Risk of Mortality Associated with PM10 and Maximum Levels of SO2, CO, NO2 and Ozone
Data on SO2, CO, NOZ and O3 are from EPA's AIRs Database. Pollution concentration value for each city is the mean of the 2nd maximum value observed at all monitors in the study area over the
study time period, which is designed to represent typical high daily values in each city. Cities without recorded levels of a pollutant are not included. The RRs are from the cities referenced in Table
V-3.  Superscripts a and b on St. Louis and Kingston indicate the RR from the Dockery et al. 1992 study and the Schwartz et al. 1996a study respectively.  Chicago RR is from Styer et al. 1995.

-------
  LO
   »_
   0)
   Q.
  j*:
   (O

  if
   0)
   0)
  CC


1.1-

1



0.9-


0.8-
St Louis3
a
Kingston
ill Hi i
] ^
LA* Chicago Boston Philadelphia Stuebenville
Portage St Louisb





c
*-
2
f\
t JL.
w
zv.
o
LO

Q)
Q.
(0
if
0)
"5
0)
CC
            0
0.005
0.01
0.015
0.02
0.02
                                    SO  (ppm). Mean

1.1-
1



0.9-

o a

Stuebenville

I



Boston Philadelphia
I « I
Utah Valley LA
St Louis3
Chicago

K
St Louis"

i i
0 0.01 0.02

I I
0.03 0.04 0.
                                                                                                                                                               I
                                                                                                                                                              l_n
                                                                                                                   NO  (ppm). Mean
FIGURE V-3b. Relationship Between Relative Risk of Mortality Associated with PM,0 and Mean Values of SO2 and NO2.

Data on SO2, and NO2 are from the EPA AIRs Database. The concentration value for each city is the mean of all recroded values observed at all monitors in the study area over the study time period.
Cities without recorded levels of the pollutant are not included. The RRs are from the cities referenced in Table V-3. Superscripts a and b on St. Louis and Kingston indicate the RR from the
Dockery et al. 1992 study and the Schwartz et al. 1996a study respectively.  Chicago RR is from Styer et al. 1995.

-------
                                          V-57
provides a qualitative review of the coherence of the health effects associated with  both
short-term and long-term exposure to PM (CD, Tables  13-6 and  13-7).  Short-term exposure
to PM is related to a number of effects ranging from mortality to morbidity and changes in
lung function and respiratory symptoms.  The association of PM with mortality is mainly
linked to respiratory arid cardiovascular causes, which is consistent with the range of
observed morbidity effects,  from respiratory and  cardiovascular-related hospital admissions to
changes in lung  function.  In addition, the CD tables show a number of similar health effects
are associated with both long-term and short-term exposure to PM.
       This qualitative coherence is further supported by quantitative coherence across
several endpoints as demonstrated in Figure V-2  and Table V-10 which also  provides some
perspective on the baseline incidence for effects of concern.   Observations of increases  in
cardiovascular and respiratory mortality associated with PM should be accompanied by more
frequently  occurring increases in hospital  admissions for the same causes.  Table V-10 shows
this to be the case.   Using the RR estimates developed in Chapter 12, the CD finds about 0.3
respiratory deaths expected per day per million for all age groups attributable to a 50 /ug/m3
increase in PM.  The CD notes  a higher expected increase in respiratory-related hospital
admissions of 2.0 per day per million in the total  population.  Similar results are found  for
cardiovascular deaths, with 0.9 cardiovascular deaths and 2.3 cardiovascular hospital
admissions per million per day associated with a  50 ptg/m3 increase in PM.  There  are some
numerical inconsistencies  in Table V-10, but,  given the diversity of the studies and analytical
methods used  to derive  the estimates, the  coherence between the mortality and morbidity
endpoints is consistent with  expectations (CD, p.   13-64).
       The coherence is further strengthened by multiple studies demonstrating associations
with a range of effects in  the same population. Studies in  Detroit, Birmingham, Philadelphia
and Utah Valley show increased frequency of a variety respiratory and cardiovascular related
health effects associated with PM exposure in the same population (CD, Section  13.4.3.5).
For example, studies in Utah Valley have shown  a number of closely related  outcomes
associated  with PM exposures, including decrements in lung function, increased respiratory
symptoms, increased medication use in  asthmatics, and increased elementary  school absences
(frequently due to upper respiratory illness). Finally, there is coherence  in the sense that the

-------
                                         V-57a
TABLE V-10.  QUANTITATIVE COHERENCE OF ACUTE MORTALITY AND
HOSPITALIZATION STUDIES (CD, Table 13-8)


Age
Group
Whole
All


All


All





Health
Endpoint
Population
Total mortality

Total hospit.
Resp. mortality
Total resp.
hospitalization
Cardiovascular
mortality
Heart disease
hospitalization
Population
Annual Baseline
Per Million
Total Population

8,603'

124,110s
676'
12.1803

3,635'

21,310'

Population Daily
Baseline
Per Million
Total Population

23.6

340.0
1.85
33.4

10.0

58.4


PM,0
Lag
Time

<2d
3-5d
-
3-5d
<2d

3-5d

<2d

Excess
Risk per
50 /ig/m3
PMIO Incr.

0.03-
0.062
-
0.19"
0.06s

0.094

0.046

Possible Number of
PM -Related Events
Per Day Per 1 Mil.
Pop. for 50 /ig/m3
PM,0 Increment

0.7
1.5
-
0.3
2.0

0.9

2.3

Elderly
65 +

65 +





Total mortality
Total hospit.
Total resp.
hospitalization
Pneumonia hospit.
COPD hospit.
Heart disease
hospitalization
6,2017
42,845'
5,101'

2,335'
2,560"
13,502'

17.0
117.4
14.0

6.4
7.0
37.0

2d
-

-------
                                          V-58
observed health effects, which are related to respiratory and cardiovascular causes, are those
that would most likely to be associated with the inhalation route.
       The CD concludes there  is evidence for increased health effects risks associated with
PM exposure ranging in severity from asymptomatic pulmonary function decrements, to
respiratory and cardiopulmonary illness requiring hospitalization, and finally to excess
mortality from respiratory and cardiovascular causes (especially in those older than 65 years
of age) (CD, p. 13-67).  Such a coherence of effect greatly adds to the strength and
plausibility of the association  (Bates,  1992).
F.     Health Effects Associated with Fine and  Coarse Fraction Particles
       The health effects information summarized in previous sections of this chapter and in
the criteria document provides substantial evidence that ambient PM, alone or in combination
with commonly occurring pollutant gases, is associated with small  but significant increases in
mortality and morbidity in some sensitive populations at concentrations below  the levels of
the current ambient standards  for PM.  An examination of potential contbunders and other
methodologic issues associated with these studies suggests that these associations are valid
(Section V.E).  Taken together,  the extensive body of recent epidemiologic studies show both
qualitative and  quantitative consistency suggestive of causality, although supporting evidence
for plausible mechanisms of action that have been hypothesized is lacking in the published
literature.  The purpose of this section is to examine the health effects evidence most useful
in determining  which PM measure(s) are the most appropriate  surrogate(s) or indicators for
those components of PM that  are most likely to be associated with the array of health effects
discussed in the previous sections of this chapter.
       A substantial body of quantitative effects information exists for PMUI,  which is the
indicator most frequently used in recent community studies (CD, Tables  13-3, 13-5).
Particle dosimetry and mechanistic considerations continue to suggest that typically occurring
ambient particles capable of penetrating to the thoracic regions of the respiratory tract (i.e.
< 10/xm diameter) are of greatest concern to health (Section V-B).  As discussed in Chapter
IV, PM10 occurring in ambient atmospheres is composed of two distinct mass fractions (fine
mode and coarse mode fractions). Based on atmospheric chemistry, exposure,  and
mechanistic considerations,  the CD concludes it would be most appropriate to  "consider fine

-------
                                           V-59
and coarse mode particles as separate subclasses of pollutants"  (CD, p. 13-94) and to
measure them separately in order to plan effective control strategies.
       Accordingly, this section  summarizes evidence on the health effects associated with
fine and coarse fraction particles7, with an emphasis on epidemiologic  results the criteria
document judges as most useful in  making quantitative conclusions.  While the
epidemiological data providing a direct comparison of the health effects of fine and coarse
particles are quite limited in comparison to that of PM10 (which contains both  coarse  and fine
mode fractions),  multiple indicators of fine mass and/or its constituents (PM25, SO4,  COH,
KM, BS) have been associated with short term effects in over 15 different cities on three
continents.  In addition, in community studies where PMU> is known to be dominated by fine
(e.g. Philadelphia) or coarse (e.g. Anchorage) particles, some qualitative  inferences can be
made about the dominant fraction.   The following sections review the epidemiologic
evidence presented in the CD for health effects associated with fine and coarse mode
particles and discusses their implications.  The discussion addresses  1) community studies
using fine particle indicators, 2)  community studies directly comparing fine and coarse
fractions, 3) studies of PMU, effects in communities with high coarse particle levels, and 4)
insights from  air quality, toxicology, and controlled human studies on  particle characteristics
as they relate  to  the potential toxicity of the two fractions.
       The focus of this examination  is on  evidence that permits a quantitative evaluation of
the extent to which fine and coarse fractions of PMU, are most  likely to be associated  with the
key health effects categories of mortality, morbidity, symptoms, and functional changes in
sensitive populations.  This is a more meaningful and tractable comparison than that  between
PM10 and the  fine fraction of PM10, which is inherently confounded.   Given the profound
physicochemical differences between the two subclasses of PMU1, it is  reasonable to expect
some differences may exist in both the nature of potential effects and in the relative
concentrations required to  produce similar responses.  In this regard, components within both
pollutant classes could be implicated in causing effects,  but the level and  nature of risk posed
    Tables 13-6 and 13-7 of the CD provide a qualitative summary of the strength of the epidemiologic
evidence for several alternative indicators of PM, including thoracic, fine, coarse, and individual components ot
fine particles (sulfate and acids).

-------
                                           V-60
 may vary between the two.  In that event, the most appropriate protection from the effects of
 particles smaller than 10 ^m would be provided by consideration of more than one indicator
 in developing control strategies. (CD, p.  13-94).
 1.     Epidemiological Studies using Fine Particle Indicators
       This section briefly summarizes the epidemiological evidence on the health effects
 associated with  fine particles as measured by a variety of indicators.  As noted in the CD
 (Tables 13-6, 13-7), community studies have shown fine particles to be associated with a
 range of health  outcomes, including mortality in sensitive population groups,  increased
 hospitalization,  respiratory symptoms, and decreased lung  function.  While a number of the
 studies used an  indicator of fine particle mass, such as sulfates, many of them employed
 PM25 or PM2, instruments.  These studies are listed in Tables V-l 1, V-12 and V-13, with
 key aspects summarized below.
       a.      Short-Term Studies
       Tables V-ll and V-12 lists  18 studies identified in the CD as evaluating short-term
 associations between mortality and  morbidity and a number of different measures  of fine
 particles.  Table V-ll lists studies that used  filter based  optical techniques (BS,  KM, COH,
 see Appendix B), which provide mainly qualitative support  for an association  of mortality
 and fine particles, while Table V-12 lists  quantitative results from studies reporting
 gravimetrically measured components that serve as indicators of particles in the fine fraction
 (i.e. sulfates  and acids), and direct  measures of PM2 5 or PM2,. These tables indicate that
 statistically significant associations have been found between fine particles and mortality in a
 number of cities.  Six of these studies found statistically significant associations with
 mortality and fine particles as measured with filter-based optical techniques (BS,  KM and
 COH), while two others could not separate effects  of particles from potential  confounding by
other pollutants  (Kinney and Ozkaynak, 1991) or the effects of a heat wave (Katsoyanni et
ah, 1993). More quantitative results  on fine particles (PM2,) and mortality are provided by
Schwartz et al (1996a),  which includes 6 cities (Table V-12). This study is reviewed in
detail in the subsection V.F.2 below,  along with other studies that provide direct comparison
of effects associated with fine and coarse  particles.

-------
                                         V-60a
   TABLE V-ll.  SHORT-TERM EXPOSURE EPIDEMIOLOGICAL STUDIES OF
          MORTALITY USING OPTICAL FINE PARTICLE INDICATORS*
City
Study Years
Indicator
Reference
Acute Mortality
London
Athens
Los Angeles
Santa Clara
1963-1972,winters
1965-1972, winters
1975-1987
July, 1987
1984-1988
1970-1979
1970-1979
1980-1986, winters
BS
BS
KM
COH
Thurston et al., 1989
Itoetal., 1993
Katsouyanni et al., 1990
Katsouyanni et al., 1993
Touloumi et al., 1994
Shumway et^al., 1988
Kinney and Ozkaynak, 1991
Fairley, 1990
*BS, KM, COH are optical measurements that are most directly related to elemental carbon concentrations, but
only indirectly to mass (See Appendix B).  Site specific calibrations and/or comparisons of such optical
measurements with gravimetric mass measurements in the same time and city are needed to make inferences
about particle mass. Both the nature of the monitor inlet and the fact that elemental carbon particles are found
in the fine fraction mean such measurements reflect variations in fine particle mass (if calibrated) or in that
portion of fine particles indexed by elemental carbon (largely primary combustion particles).  Comparisons
between the respective optical measurements and mass  measurements were made for the historical London
winters (EPA, 1982a), the Athens studies (Katsouyanni et al., 1995), and  Santa  Clara  (Fairly. 1990).  Such
comparisons were not reported for the Los  Angeles study using KM,  but the same investigators also reported
significant associations between mortality and PM gravimetric mass in Los Angeles, (Kinney et al., 1995).

-------
                               V-60b
TABLE V-12. FINE PARTICLE INDICATOR (PM2S, SO;, H+) EFFECTS STUDIES
            FROM THE U.S. AND CANADA 
-------
                                             V-60c

References:

ASchwartz et al. (1996a)
BBurnett et al.  (1994)
cBumett et al.  (1995) O3
'Thurston et al. (1992, 1994)
ENeas et al. (1995)
FOstro et al. (1993)
GSchwartz et al. (1994)
•"Ostro et al. (1991)
QKoenig et al.  (1993)
          24-h PM indicator level shown in parentheses unless otherwise noted as (± S.D.), 10 and 90 percentlle
 (10,90).
'Change per 100 nmoles/m3.
"Change per 20 jtg/m3 for PM25; per 5 /*g/m3 for PM25 sulfur; per 25 nmoles/m3 for H*.
"**50th percentile value (10,90 percentile).
""** "Coefficient and SE in parenthesis.

-------
                                          V-61
       Nine studies in the U.S. and Canada have found positive associations between short-
term exposure to gravimetrically measured fine particles or components (including sulfates
and acids) and indicators of morbidity, including increased hospital admissions, increased
respiratory symptoms and decreased lung function (Table V-12).   All the studies found a
positive association between PM2 5 and measured health effects; in eight of the studies the
associations were significant.  A particularly informative study was conducted by Thurston et
al. (1994b) in Toronto, which evaluated  the associations of respiratory-related hospital
admissions with a range of particle indicators.  This study is discussed below in subsection
V.F.2.
       b.      Long-Term Studies
       Table V-13 lists the studies the CD finds most useful for presenting quantitative
estimates  of effects associated with long-term exposure to PM (CD, Table 13-5).  Two recent
prospective studies, the Six City Study and the ACS study,  reflect significant methodological
advances  over earlier cross-sectional studies and provide the best evidence of the association
between long-term PM exposure and mortality.  The relative strength of the results for fine
and coarse indicators is discussed below  in subsection V.F.2.
       The designs and approaches of the Six City  and ACS studies are complementary in
nature (See Section V-13). The Six City study provided a more complete consideration of
co-occurring pollutants that might confound the results (CX,, SO2,  NO2), but  lacked some
power due to the limited number of cities and the size of the total population included.   The
ACS study was designed to test the major hypothesis derived from the Six City study.
namely that long-term exposure to fine particles (as PM2, or sulfates) was associated with
increased  mortality.  The  ACS design improved upon the Six City study by evaluating  a
larger population in many more cities across the U.S. (151) but, based on the earlier
findings, did  not include multiple pollutants.  The ACS  study found a significant association
between mortality and both PM25 and sulfates (Table V-13). For reasons discussed  in
Section V.C., the staff concludes the somewhat smaller effects estimates from the ACS study
are likely  more useful for risk assessment of long-term mortality  than those from the Six City
study.  In addition, consideration must be given to the role of earlier exposures to higher
concentrations with respect to the applicability of these estimates  based on a  few years of

-------
                                     V-61a
TABLE V-13. EFFECT ESTIMATES PER INCREMENTS' IN
ANNUAL MEAN LEVELS OF FINE/THORACIC  PARTICLE INDICATORS FROM
U.S. AND CANADIAN STUDIES (CD, Table 13-5).
Type of Health
Effect & Location
Increased total chronic
Six Cityb


ACS Study0
(151 U.S. SMSA)

Increased bronchitis in
Six City"
Six City*
24 Cityf
24 Cityf
24 City'
24 Cityf
Southern California8
Indicator
mortality in adults
PM.S/10
PM25
so:
PM2.5
so:
children
PM15/10
TSP
H+
so:
PMj.,
PM10
so:
Change in Health Indicator per
Increment in PM'
Relative Risk (95% CI)
1.42(1.16-2.01)
1.31 (1.11-1.68)
1.46(1.16-2.16)
1.17(1.09-1.26)
1.10(1.06-1.16)
Odds Ratio (95 % CI)
3.26(1.13, 10.28)
2.80(1.17,7.03)
2.65(1.22,5.74)
3.02(1.28, 7.03)
1.97(0.85,4.51)
3.29(0.81, 13.62)
1.39(0.99, 1.92)
Range of City
PM Levels
Means (Mg/m3)

18^7
11-30
5-13
9-34
4-24

20-59
39-114
6.2-41.0
18.1-67.3
9.1-17.3
22.0-28.6
—
Decreased lung function in children
Six City" h
Six City'
24 City'-1
24 City'
24 City'
24 City'
PM.s/.o
TSP
H+ (52 nmoles/m3)
PM2 , (15 A
-------
                                           V-62
monitoring (CD, P 12-366).  If the effects are the result of long-term exposures,  as opposed
to the sum of episodic or daily effects, then the reported relative risk estimate are apt to be
high.
       Cross-sectional studies conducted by Ozkaynak and Thurston  (1987,  1989) and Lipfert
(1988) provide some additional insights into the relationship between long-term exposure to
fine particle indicators and mortality. Ozkaynak and Thurston's cross-sectional analysis of
various particle measures and  1980 total mortality across US cities found the most consistent
and significant associations with fine particles and sulfates.  In their analysis, TSP and PM1S
were often found to be nonsignificant predictors of mortality.   Lipfert also analyzed  1980
total mortality across US cities in relation to different particle measures (CD, p. 12-15).  In
general,  when evaluating single site TSP or PM,5 and sulfates or PM2 s in  models with the
same covariates, the effects estimates for sulfates and fine particles were generally larger
than those for TSP or PM15.  Some model  specifications also show significant associations
between  mortality and multi-station TSP.  A supplemental analyses of the Lipfert 1980  data
in the CD found that the introduction of numerous potentially confounding variables  (e.g.
water  hardness, sedentary  lifestyle) reduced but did  not eliminate the PM2 s effect on
mortality (CD, Fig 12-7)8.  Clearly there are inherent methodological issues with these
ecological approaches, but  they show evidence of associations between long term measures of
fine particles, including  sulfates,  and mortality  that are quantitatively more consistent with
the lower risk estimates  found in the ACS study (CD, p  12-177).
       Several studies have evaluated the association between long-term fine particle
exposure and  increased respiratory symptoms and decreased lung function most which have
been conducted  in children (Table V-13).  The 24 city studies are of particular interest.
These studies evaluated  the association between  different measures of long term PM  (PMU),
PM2.5, SO4 and  H+) and respiratory symptoms and pulmonary function in children (Raizenne
1996;  Dockery et al.  1996). The one year surveys found a significant increase in bronchitis
8In this example, the PM:s effect was reduced from 0.045 to 0.02 deaths per /ig/m\  While it is likely, that
addition of some of these variables to the Six Cities and ACS cohort studies would reduce the effects estimates
for these two studies as well, the relevance and independence of including all ot their variables (e.g.. sedentary
lifestyle and overweight) can be questioned.

-------
                                          V-63
in children (one episode or more) associated with particle strong acidity and fine paniculate
sulfates.  Elevated, but nonsignificant associations were observed between reporting a
bronchitis and PM2 5 and PMi0.  No other respiratory symptoms, including asthma symptoms,
were significantly associated with any of the pollutants.
       In contrast  to the earlier 6 city results,  annual mean particle strong acidity, total
sulfates,  PM2.5 and PM10 were all significantly associated with FVC and FEVl deficits (Table
V-13).  A slightly  larger FVC decrement was  found for children who were lifelong residents
of their communities, though it was not significantly different.  For the 24 cities, there was a
strong  correlation between particle strong acidity and sulfates (r=0.90) and PM2.1 (r=0.82),
but not with PM10  (r=0.47).  Thus, it is difficult to ascribe the association to any one of the
3 fine particle indicators.
2.     Community Studies Comparing Effects of Fine  and Coarse Fraction PM
       Several studies provide quantitative information directly comparing the association
between  health effects and fine and coarse particles.  They include an examination of short-
term PM exposure  mortality in the Harvard  six cities (Schwartz et al., 1996),  a short-term
exposure hospital admission study (Thurston et al.,  1994b), and the long-term exposure
mortality Six City  Study  (Dockery et al.,  1993).  Supporting  information on long term effects
can also  be found in the data from the  ACS  study (Pope et al., 1995b) and the 24 city  study
reports (Spengler et al, 1996; Dockery et al.,  1996; Razienne et al, 1996).
       a.     Short-Term Comparisons
       A recent analysis  of mortality in six cities by Schwartz et al (1996) evaluated the
association between mortality and 5 different particle measures: coarse fraction particles
(PM15/10  minus PM25); thoracic particles (PM1S/K1), PM2
-------
   TABLE V-14. ESTIMATED INCREASE IN DAILY MORTALITY, 95% Cl, AND t STATISTIC BY CITY AND
        COMBINED ESTIMATE ASSOCIATED WITH A 10 jig/m3 INCREASE IN PARTICULATE MASS
     CONCENTRATIONS. EFFECT OF EACH PARTICLE MASS MEASURE ASSOCIATIONS ESTIMATED
               SEPARATELY, CONTROLLED FOR LONG-TERM TRENDS AND WEATHER.
                     Correlation
Study City              PM25/CM          PM:s                CM                  PM,0
Boston

Knoxville

St. Louis

Steubenville

Portage

Topeka
0.

0.

0.

0,

0.

0.
,23

.44

,45

,69

,32

.29
2.2% (1.
» =
1.4% (0.
i 	
1.1% (0.
« =
1.0% (-0
f 	
5%,
6.31
2%,
2.26
4%,
3.17
.1%,
1.79
1.2% (-0.3%,
t =
1.64
0.8% (-2.0%,
2.9%)

2.6%)

1.7%)

2.1%)

2.8%)

3.6%)
t = 0.53
0.2% (-0.6%,
t = 0.58
1.0% (-0.6%,
t=1.20
0.2% (-0.7%,
t = 0.45
2.4% (0.5%,
t = 2.43
0.5% (-1.2%,
t=0.57
= 1.3% (-3.3%
t=1.32
1.2%)

2.6%)

1.1%)

4.3%)

2.3%)

, 0.6%)

1.

0.

0.

0.

2%

9%

6%

9%

0.7%


(0.7%,
t = 4.86
(0.1%,
t=2.21
(0.1%,
t = 2.42
(0.1%,
t=2.I7
(-0.4%,
t=1.22
-0.5% (-2%,


t = 0.67
1.7%)

1.8%)

1.0%)

1.6%) <
i
Cf
1.7%) £

0.9%)

All Cities Combined

 Total Mortality                      1.5% (1.1%, 1.9%)      0.4% (-0.1%, 1.0%)        0.8% (0.5%, 1.1%)
                                    t=7.13              t=1.48                t = 5,84

 Ischemic Heart Disease                 2.1% (1.5%, 2.7%)
                                    t = 7.12

 Chronic Obstructive                   3.3% (1.0%, 5.7%)
 Pulmonary Disease                       t = 2.79

-------
                                            V-63b
                                  Relative Risk for 50 pa/m1 PM15
                                      in Six Ctty Acute Study



0

Topeka


St. Louis
Boston

»-—( .VS 1
1 rt 1
»—--*» I
v |
f 	 .
                                 0.9
        1
   Relative Risk
                                                          1.1
Relative Risk for 25 pg/m3 Fine Particles
    (PM2.5) in Six City Acute Study
                                                                                  ir-loc  >
              Relative Risk for 25 pg/m3 Coarse Particles
                (PM15-PM2.5) in Six CHy Acute Study
   O
Topeka
Steubenville
St. Louis

Boston


1





— o — |
H-OH
»sv..-^
HO-I
             0.9
1.1
Topeka
Portage

0 St.Louis
Boston
i/s .. i

v I
»>s j
H-0 	 1
                                                         09
                      Relative Risk
                                1
                            Relative Risk
                                                                                   1.1
Figure V-4.  Relative risks of acute mortality in the Six City Study, for thoracic particles
(PM10, PM,5), fine particles (PM2 5) and coarse fraction particles (PM15-PM2 5). The coarse
fraction effects are small and insignificant, except in Steubenville, where there is a high
correlation between fine and coarse particles (R2=0.69).  In Topeka, which has the second
highest level of coarse fraction particles, the association is negative and nearly significant.
Source: CD, Figure 12-33. U.S. EPA graphical depiction of results from Schwartz et al. (1996).

-------
                                           V-64
mortality associated with PM25 was consistently positive in all 6 cities (0.8 to 2.2% for a 10
/ng/m3 PM2.5 increase) and statistically significant in 3 cities.  In contrast,  the relative risks
for mortality associated with coarse particles was inconsistent across the 6 cities (-1.3% to
2.4% for a  10 /ng/m3 increase in coarse particles) (Table V-14). The association with coarse
particles was significant only in Steubenville, but it is difficult to interpret these results given
the high correlation between fine and coarse particles (r=0.69) in this city.  All of the other
cities have r of 0.45 or less. The negative but non-significant association  between PM,0 and
mortality in Topeka noted  above appears  to be driven by the coarse fraction.  Although
Topeka has the highest percentage of crustal particles and the second  highest average coarse
mass, coarse particles have a nearly significant  negative association with mortality, while  fine
particles have a positive but non-significant association. While greater measurement error
for the coarse fraction (see Section V.E above)  could depress a potential coarse particle
effect, this would not explain the results in Topeka relative to other cities. Even considering
relative measurement error, these results  provide no clear evidence implicating coarse
particles in  the reported effects.
       In a combined analysis across the  6 cities,  PM2 _s was significantly  associated with  an
increase in mortality of 2.1%  (CI 1.5% to 2.6% for a 25* to 75 percentile increase in
PM2.5).  In contrast, the coarse particles were associated with a small but  insignificant
increase in mortality, 0.4% (CI -0.1%  to 1.0%, for a 25th to 75m percentile increase in
coarse particles).  To determine whether coarse particles were independently  associated with
mortality, both fine and coarse particles were considered simultaneously in the regression
across all  six cities. The estimated effect for PM2  5 across the interquartile range remained
unchanged with a significant association with mortality (2.1%,  CI 1.5% to 2.6%).
Conversely, the coarse particle estimate was substantially  lowered (-0.2%, CI -0.8% to 0.4%
for the interquartile range). This study provides clear evidence that fine particles are more
likely to be responsible for the numerous observed associations between PMUI and mortality.
The study also evaluated the association with fine particles by  age and cause of death.
Similar to studies of PM10  and  mortality,  a higher RR estimates for deaths from ischemic
heart disease and deaths from chronic obstructive pulmonary disease was  found in their

-------
                                           V-65
analysis (Table V-14).  The authors note that this is a similar pattern to that seen in London
during the 1952 dramatic pollution episode.
       Thurston et al. (1994b) evaluated the association between summertime respiratory and
asthma related hospital admissions and 5 different particle measures: acids, sulfates, fine
particles,  coarse particles and PMi0.  Without adjusting for the risk associated with
concurrent O3 levels, the investigators found a significant association between respiratory-
related hospital admissions and all measures of particles except the coarse fraction.  Only
fine acids and sulfates were significantly associated with asthma admissions in the univariate
models.  When O3 was included in the model, only acids and sulfates remained significantly
associated.  The authors note the high correlations between  the other particle measures and
O3 concentration  make it difficult to select a best indicator,  but these results provide no
evidence of a coarse particle association with respiratory admissions in an area meeting the
PM10 standards.  The authors conclude that, based on  the relative strengths of hospital
admissions associations, the particle indicator, could be ranked as H+  > sulfates > PM2 5
> PM10.
       b.      Long-Term Comparisons
       The Six City study evaluated the relationship between mortality and long-term
exposure to particles using several indicators; total  particles, inhalable particles, fine
particles,  coarse particles, sulfate fine particles and non-sulfate fine particles  (Dockery et al..
1993).  Figure V-5 plots the relationship between mortality  risk and each of the particle
indicators.  Although such  comparisons involving only 6 cities should be viewed  with
caution, there is a trend toward increasing associated of relative risk of mortality with the
particle indicator as the size of the particle indicator decreases (CD, Chapter 13).  Although
some association is apparent for TSP alone, the "super-coarse"  fraction of particles larger
than 10-15 ftm does not appear to be clearly linked with mortality, particularly in areas other
than Steubenville.  This further supports the notion that extrathoracic particles present a
lower risk than thoracic PM. The distinction between PM2.t and coarse fraction  (PM1U.2,)
particles is less clear,  although  — as was the case in the short term  mortality results above -
the relative risk for the city with  the highest proportion of crustal materials (Topeka) appears

-------
                                       V-65a



























Total Particles ia

|
5
10
no

S

H
L
W
P T
30 40 50 60 70 80 90 101
A To««l PvtdM. no*1'
^

Total Particles 12
Divided into Inhalable and *





* 1 1
Non-lnhalable Particles |
10
0 9


S

H
L

W
P T
13

12
i
"11
I
10
no


S

H
L

W
P T
15 20 25 30 35 40 45 50 10 20 30 40 5l
Inhalable Particles, pg/m1 Nor>- Inhalable Particles ps/m1

r ••
Inhalable Particles
Divided into Fine 12
and Coarse Particles %-\i




K
10
n o
!
H
L
W
PT

12
£
•
(K
10
n Q
S
H
L
W
P T
\




10 15 20 25 30 68 10 12 14 16
A Fine Particles pfl/"1 Coarse Particles, n9'ni
x*C"*
Fine Particles
Divided into

Sulfate and «n
M
Non-Sulfate "
10
Particles
0 9

S
H
L

W

TP

4 6 8 10 12
Sulfate Particles pg/m


12
*
* 1 1
j*
*
10
0 9
S
H
L

W

f T

"^»k








5 7 9 11 13 15 17
Notv Sulfate Fine Particles yg'm
Figure V-5. Adjusted relative risks for mortality are plotted against each of seven long-term
average particle indices in the Six City Study, from largest range (total suspended particles,
upper right) through sulfate and nonsulfate fine particle concentrations (lower left).  Note that
a relatively strong linear relationship is seen for fine particles, and for its sulfate and non-
sulfate components. Topeka,  which has a substantial coarse particle component of inhalable
(thoracic) particle mass, stands apart from the linear relationship between relative risk and
inhalable (thoracic) particle concentration.  Some gradient exists for all indicators with respect
to Steubenville and Portage..

Source:  CD,  Figure 12-8.  U.S. EPA replotting of results from Dockery et al. (1993).

-------
                                          V-66
to be more consistent with a fine particle effect.  For the other cities,  there is less difference
between fine and coarse rankings.
       Some additional insight into the Six City results is found in an  ecological analysis of
data from the ACS study (Pope et al., 1995b). Figure V-6 shows scatterplots of adjusted
mortality and PM as indicated by sulfate and TSP taken from the ACS study.  These figures
show a pattern consistent with a sulfate mortality effect across a large number of cities, but
no clear relationship for TSP.  The relative position of the six cities in these figures shows
that, consistent with the original study design  (Ferris et al, 1986), which selected cities to
show gradients in both TSP and sulfur oxides, the mortality risk in the six cities shows an
apparent relationship with both sulfates and TSP.  The similarity in gradients for mortality
for both fine particles (sulfates) and TSP in the six cities is not typical of the full set of 151
cities in the ACS study.   Given the strong significant  association between fine particles and
mortality in the full ACS and Six City cohort  studies and the  lack of significant association
with TSP in the ACS data (Pope  et al., 1995b), the  evidence  for chronic mortality effects
appears to be stronger for fine  particles than for coarse.
       Both the ACS study and the Six City study found the increase in risk of mortality
associated with fine particle matter was mostly attributed to increases  in cardiopulmonary
mortality.  As noted in  Section 5.C, the Harvard Six City study reported a 37 percent
increase in cardiopulmonary  mortality associated  with  PM2 s, and the ACS study reported a
31 percent  increase in cardiopulmonary mortality associated with PM2 s.
       The negative results of the third prospective cohort study (Abbey et al,  1991) do not
diminish the  above conclusions.  As noted in section V-C, despite the  theoretically improved
approach to exposure classification in this study (CD,  p. 12-162),  the  choice of PM indicator
(days >200 jug/m3 as TSP) for a large number of California sites  limits the inferences that
can be made about smaller particles sizes.  Peak TSP in various times and places in
California may be associated with coarse agricultural or  road  dust or high photochemical!y
derived fine particles.  Unlike other national cross sectional comparisons that use mean TSP
from  multiple monitors in metropolitan  areas spanning the East and Midwest U.S.  (e.g.
Lipfert, 1993 ), peak TSP in California is less likely to be a useful surrogate for fine or
thoracic particles. Thus, while neither this study nor the ACS study finds a significant

-------

1000
0*
o
o 900
TA
^ 800
0)
H 700
r
o
3?
*C»
•a
£ 600
(A
^
500
0
n
-
i j
u „
u ^ | ^yjf ' ' " ""
D Cr-^qh CO COD D
^ nQR0* ^ °
o n t'j
6 o a
n
• Metro areas Inclusive of a six -cities study site
- + Reference metro areas near a six-cities study site
n and ' All other metro areas with TSP and Sulfate data
i i i i

^ 1000
o"
0
o
o 900
1
ra 800
« 700
c
o
£ 600
(A
|f
500

n
-
u
a
~ Ci C
" - wjso- ':
aa :
n o D o
n
• Metro areas Inclusive of a six -cities study site
• Reference metro areas near a six-cities study site
il and All other metro areas with TSP and Sulfate data
i i i






1

-------
                                           V-67
mortality effect of long term exposures to TSP, only the ACS study tested this hypothesis
with respect to fine particles using appropriate measurements.
       Staff also further examined the data in the 24 city studies of the effects of PM on lung
function in children  (Raizenne et al., 1996). As noted above, the  authors report significant
associations between lung function and strong acids, sulfates, PM2,, and PM10,  but did not
report on any analyses for coarse fraction particles.  Figure V-7 plots the lung function
results for the 22 cities where  such data were taken against both PM2, and coarse fraction
(PMjQ.21). The lack of any significant association of coarse particles is apparent.  The
careful selection of the cities and  study participants was intended to provide a clear gradient
across regions with elevated fine acid aerosols and areas with lower levels, and to provide
for a separation of potential O3 and PM effects.  Multiple pollutants and indoor conditions
were considered.  The use of children of similar socioeconomic status and race  reduces  much
of the confounding.   This study provides clear evidence of an effect of fine particles  that is
independent of coarse fraction  particles.
       A longitudinal study by Johnson et al. (1990) in five Montana cities evaluated the
association between  lung function and TSP,  fine and coarse particles in school children over
one school year. They found significant decrements in FEV1 for TSP and significant
decrements in FVC for fine particles, but  at best, results were insignificant and inconsistent
in effects for coarse particles.
3.     Epidemiological Studies of Areas Dominated by Coarse  Particles
       The studies discussed in Section V.F.2  above are the only ones cited in the CD to
have evaluated the association  between  directly measured coarse particles and health effects.
In general, such  studies have found equivocal results, suggesting an inconsistent or
insignificant association between  coarse particles and mortality and morbidity.  However,
with the possible exceptions of Steubenville and Topeka, the concentrations of coarse
particles were relatively low and below those of fine particles, and measurement error could
have influenced the results.  The CD identifies only two additional studies as suggesting
morbidity effects associated with short-term episodes of coarse particles (p.  13-47).  In these
cases,  coarse particles were not measured, but ancillary evidence indicates that measured

-------
                                  V-67a
                    22 City Fine Mass vs. ^Children <85% FVC
   00
   V
     3-


     2-


     1 -


     0
            %FVC .85
            Linear (% FVC .85)
                                10
                                      12
                                           14
                                                16
                                                     18
                                                          20
                                                               22
                                                                    24
                                PM2.1 (Mg/m3)
   10  5-
   00
   O
      3-
      2-
1 -
             22 City Coarse Fraction Mass vs. % Children  < 85% FVC
         4  %FVC .85
         — Linear (% FVC .85)
                                                     10
                                                        12
14
                               PM 10-2.1 (M9/m3)
Figure V- 7. % of Children with <85% Normal FVC vs. Annual Fine and Coarse Fraction
Mass in 24 City Study. (EPA graphical depiction of results from Raizenne et al. , 1996;
Spengler et al, 1996). The relationship between fine mass and lung function decrement is
significant. No clear relation is shown for coarse fraction particles, which are generally at
low concentrations in these  cities.

-------
                                          V-68
PMIO is likely to be dominated by coarse particles, at least during significant episodes or
seasons..
       A study in Anchorage, Alaska evaluated the association between PM10 and daily
outpatient visits taken from insurance claims for employees for the State of Alaska and the
Municipality of Anchorage (Gordian et al,  1996).  They collected data on asthma,  bronchitis,
COPD, congestive heart failure, diarrhea and upper respiratory illness ( defined as upper
respiratory problems such as sore throat, sinusitis, earaches, rhinitis, and other nonspecific
upper airway problems).  They were not able to evaluate COPD and congestive heart failure
because of insufficient number of cases.  The investigators report that there  are no industrial
sources of the fine portion of PM10  in Anchorage, and the scanning electron microscopy of
10 random samples found over 80% of the PM10 mass was between 2.5 to 10 /*m in
diameter.  Daily PM10 values ranged from 5 to 565 jig/m3 (corresponding to a volcanic
eruption), with an average over the 22-month study period of 45.5 ^ig/m3.  Gordian et al.,
report a 3-6% increase in visits for  asthma and a 1-3% increase in visits for upper
respiratory illness associated  with 10 /xg/m3 increase in PM,0.  They found no association
with visits for bronchitis. They also found a nonsignificant association with PM10 in the
period immediately after a volcanic eruption, and significant associations in the period
excluding the volcanic eruption.  The authors suggest that personal  intervention minimized
exposure after the eruption.
       Hefflin et al.,  (1994) evaluated the potential influence of dust storms on emergency
room visits for respiratory disorders in three Southeast Washington  State communities.  The
investigators report that particle exposure is mostly from windblown soil and related natural
crustal materials  (the majority volcanic in  origin).  Thus, PM is likely dominated by coarse
particles.  This area also had high levels of PM^, with peak 24-hour values ranging from 1
to 1,689 fj-g/rn3 with an average of 40 ng/m3.  Aside from the periodic dust storms, the
authors provide no additional evidence regarding the size composition of PMU) (e.g. extent of
wood stoves, other sources).  In contrast to Gordian,  Hefflin et al.  found a significant 0.35%
increase in emergency room visits for bronchitis associated with a  10  /xg/m3 increase in
PM10.  They also found a significant 0.45% increase in emergency room visits for sinusitis
for a 10 /ig/m3 increase in PM10 levels over 150 /ig/m3. There was no association with

-------
                                          V-69
asthma.  They found a slight association between emergency room visits and two high dust
storms days where particle concentrations were over 1,035 and 1,689 /xg/m3, but suggested
that the reduced unit risk could have been related to mitigating behavior in these severe
conditions.
       These studies are suggestive of potential associations between high concentrations of
coarse particles and health  effects, but with some inconsistencies.  The effects estimates  for
the Hefflin et al.  study are much smaller than the Gordian et al. study.  In addition, the
Gordian et al. study found  an association between PM-10 and asthma but not with bronchitis,
and the Hefflin study  found the opposite. This contrast should be interpreted cautiously due
to possible difference in disease classifications int he two study areas.   Hefflin et al. (1994)
have found overall asthma  incidences in the region  to be lower than expected, reducing the
power of the study to detect effects.  Both  studies report multiple exceedences of the PM10
standard.  The apparent diminished response of the very highest days suggests that mitigative
measures such as staying indoors on days of perceived dust episodes offered some protection
against the effects of coarse particles on asthma and upper respiratory  illness.  Based on the
Gordian results and the potential  for significant deposition of coarse particles in  the
tracheobronchial  regions of the lung where they may irritate sensitive receptors  in asthmatics,
the CD concludes that particles in the coarse fraction appear to be associated with the
exacerbation of asthma via ambient exposure (CD,  p. 13-51).
4.     Relevant Physicochemical Differences between Fine and Coarse Fraction Particles
       Current understanding of the toxicology of ambient PM suggests that fine and coarse
particles may have different biological effects (CD, p.  13-91).   The discussion  below
summarizes information the CD presents regarding  differences in potential toxicity between
the two fractions based on  composition and size related properties.
       a.      Comparisons of fine and coarse component toxicity in laboratory  studies
       A comparison  of the major components of typical  ambient particles (Table IV-2)  and
the size and composition of particles studied in the  recent toxicologic literature (CD, Chapter
11) suggests that, while substantial work has been conducted on simulated constituents of fine
particles such as  acid aerosols,  trace elements, and  components of diesel particles, very  little
attention has been focused  on health effects from exposure to ambient  coarse particles or

-------
                                          V-70
 their significant components.  The only study in humans of a coarse aerosol (10 /zm diameter
 NaCl, see Table IV-2) cited in Chapter 11 (CD, Table 11-1) was considered to be a control
 for an acid fog exposure. Furthermore, because  of size limitations of particles that can
 appreciably deposit in the tracheobronchial and alveolar region in small laboratory animals,
 most experimental animais studies involve fine particle exposures (CD, p. 13-44).  The most
•clear and relevant comparison between the different constituents typically found in the fine
 and coarse fractions of PM was that of Kleinman et al (1995), who found that the relative
 cellular and immunological toxicity of fine particle components, sulfate (70 /xg/m3, 0.2/zm
 diameter (NH4)2 SO4) and nitrate (350 uglrcv1, 0.6/xm diameter NH4NO,) were greater than
 that of a typical resuspended coarse fraction component - road dust (900 ng/m*,  4/xm
 diameter), in the rat. While it is clear from the results of the study that the road dust elicited
 effects and was present in some concentration in  thoracic region of the rat, the extent of
 deposition was not given in the study and it is possible that some of the differential toxicity
 shown between fine and coarse particle constituents in this study are due to differential
 penetration efficiencies of the particles.
       Chapter  11 of the CD highlights the results of a volcanic ash study (Raub et al,  1985)
 as a comparison of fine and coarse mode particles.  This study used intratracheal instillation
 of large amounts of 12.2 /xm and 2.2 /xm  diameter volcanic ash into rats.  The authors report
 finding a number effects at the higher concentration used, but essentially no difference in
 several measures of toxicity.   While these result are of interest, the 2.2 ^m particles should
 not be characterized as fine mode, but rather as the "tail" of the coarse mode.  Thus, this
 study suggests little or no difference in  the toxicity of coarse mode particles of different
 sizes, but even this conclusion is limited by the artificial nature by which the particles were
 deposited in the animals.
       Raub et al. (1985) also found no differences in toxic responses between normal and
 emphysemic animals inhaling 9600 ptg/m3 submicrometer sized volcanic ash for short
 durations.  Mauderly (1990) found that emphysematous rats had less effects than normal
 animals because of the sparing effects of emphysema to high levels of diesel particles.
 However,  Raabe et al. (1994) exposed rats with induced emphysema to two fine particle
 mixtures intended  to simulate  a London aerosol (ammonium sultates, coal fly ash, lamp black

-------
                                          V-71
carbon) and a California aerosol (ammonium sulfates and nitrate, graphitic carbon, clay, and
trace metal sulfates).  Even at the lowest levels tested (550 -800 Atg/m3), 3 to 30 day
exposures resulted in  significant responses that were greater than those seen  in normal
animals (CD,  p 11-176).
       b.      Toxicity of Fine and Coarse  Mode Chemical Components
       Table IV-2 lists the key differences in chemical composition of fine and coarse
particles.  The CD review  highlights a number of specific components of PM that could be
of concern to  health,  including typically fine components (e.g., acids, certain metals,  diesel
particles, and  ultrafmes), and typically coarse  components (e.g., silica and bioaerosols).  It is
clear that components of both modes can produce responses, although in general, the  fine
mode appears to contain more of the irritant substances potentially linked to the kinds of
effects observed in the epidemiological studies.  The following is a brief summary of the
potential toxicity associated with fine and coarse substances.
       Most of the aerosol acidity is contained in the fine fraction.  Section V-C details a
variety of effects associated with acids in community epidemiology and  at high levels in
laboratory studies.  Acids may produce effects as liquid droplets or surface coatings in
mixtures.  For example, Chen et al. (1990)  exposed guinea pigs to fly ash derived from
either low or high  sulfur coal.  The acidity of the resulting particles was proportional to
sulfur content with the greatest pulmonary functional response noted for the  high sulfur fly
ash.
       Acid aerosol exposure has been associated with changes in airway morphology as well
as airway responsiveness (Gearhart and Schlesinger, 1988; Kleinman  et al.,  1995; Chen et
al., 1992b; Gearhart and Schlesinger 1986;  and  El-Fawal and  Schlesinger, 1994) in
experimental animals. Markers of cytotoxicity and increased cellular permeability, following
a single exposure to fine or ultrafine H2S04 aerosols, have also been reported  (Chen et  al.,
1992a). Levels of biological mediators of inflammatory responses, as well as smooth muscle
tone, have been shown to be altered after exposure to fine  acid aerosols (0.3 ^m diameter)
and lavage. Fine acid aerosol exposure has been shown to alter macrophage function,
production of tumor necrosis factor cytotoxic  activity, and  superoxide radical production, all
of which are related to host defense mechanisms.  Fine aerosols of ammonium sulfate and

-------
                                           V-72
nitrate at relatively low levels have also been shown to alter antigen binding and respiratory
burst activity by macrophages (Kleinman et al., 1995).
       As noted in the 1982 Staff Paper, extractable organic matter from particles with
potential carcinogenic activity is also preferentially derived from the fine fraction.  The CD
'{p. 5-10) notes that the majority of diesel exhaust particles is in the fine mode and both short
'and long  term inhalations of diesel particles are associated with respiratory effects at higher
than ambient levels in experimental animals. Occupational studies report (at levels higher
than ambient concentrations) bronchitis, impaired respiratory function, cough,  and wheezing
(CD, Table 11-11), all of which have been reported in community air pollution studies of
PM.
       Ultrafine aerosols (<0.1 /zm) are a class of fine particles that have the potential to
cause toxic injury to  the respiratory tract as seen  in studies conducted both in vivo and in
vitro (CD, p. 13-76).  An  important aspect of their potential toxicity is their relatively low
solubility (CD, p. 13-77).  Studies on a number of relatively insoluble ultrafine particles
(diesel, carbon black), present in the ambient air  as aggregated ultrafines, indicate that
inhalation exposure to these as well as TiO2 to rats are associated with epithelial cell
proliferation, chronic pulmonary inflammation, pulmonary fibrosis, and induction of  lung
tumors at high concentrations (CD,  p.  13-77).  Ultrafine particle have also been shown to
evade macrophage phagocytosis and penetrate the interstitium more easily than larger sized
particles (Takenaka et al.,  1986; Ferin et al., 1990, CD, p. 13-77). There is also evidence
that some aggregated insoluble ultrafine particles  dissociate into singlet ultrafine particles in
the lung which would facilitate transport across the epithelium (Takenaka et al.,  1986; Ferin
et al.,  1990; Oberdorster et al,  1994; CD, p. 13-77).  Because  of their short lifetime, it is
unclear that unaggregated ultrafine particles make up any  significant fraction of the mass of
fine particles or of PM10, other than in the vicinity of significant  sources of ultrafine
particles.  The relationship between ultrafine numbers (or mass) and the mass of fine or
thoracic particles found in  typical community air pollution has not been established.
Although  the CD provides little direct information,  it might be  expected that penetration and
persistence of unaggregated ultrafine particles to indoor environments would be limited.  For
these reasons, it is questionable whether ultrafine aerosols could be playing  a major role in

-------
                                          V-73
the reported epidemiologic associations between the measured mass of fine or PM10 particles
and health effects in sensitive populations.  Because of the potential toxicity suggested by the
available literature,  however, this an area where significant additional research is needed.
       The only major coarse particle components  highlighted in the CD summary are silica
and bioaerosols.  The majority of silica particle mass is found in the coarse fraction (CD, p.
11-127). Occupational, but not community exposures to crystalline silica has been associated
with pulmonary inflammation and silicosis (pulmonary fibrosis from silica) (Spencer  1977;
Morgan et al 1980; Bowden, 1987).  Although some evidence of long term accumulation of
silicate material at near ambient levels  has been noted (Section V-C), the CD provides no
evidence of any significant short term effects of ambient silica.  Thus, there is no evidence
suggesting that this class contributes  to the observed daily  mortality and morbidity effects.
       Bioaerosols (which includes fungal spores, pollen, bacteria, viruses, endotoxins, and
animal and plant debris) can be distributed in both  fine and coarse fractions and are capable
of producing serious health effects.   Strong sources (e.g., grain elevators)  of these materials
may have obvious effects on allergic individuals.   However, as the CD points out, the
annual variability, relative mass, and distribution of such materials suggests that they too
"appear to be unlikely to account for observed ambient (outdoor) PM effects on human
mortality and morbidity demonstrated by epidemiology studies  reviewed in Chapter 12"  (CD,
p. 11-136).
       c.      Physical Aspects of Fine and Coarse Particles
       Figure IV-2 and Table IV-2 show key differences between fine and coarse particles.
The fine fraction contains by far the  largest number of particles and a much larger aggregate
surface area than the coarse fraction.  As noted above, the size range of particles containing
the largest number of particles (<0.02 /xm) is not  that with  most of the mass of the aerosol
(fine or coarse).  However, most of the aggregate  surface area of the entire size distribution
of typical urban particles is contained in the fine size range of  0.1 to 1.0 ptm diameter (CD,
Figure 13-4; Figure IV-2).  Unlike the case with particle number,  therefore, it is  clear that
the aggregate surface area of PM10 is likely to be strongly  related to the mass of fine
particles (see Figure IV-).  This relationship should be a common property of PM in a
variety of different urban settings.

-------
                                          V-74
       The greater surface area of the fine fraction means this fraction has a substantially
greater potential for absorption of other potentially toxic components of PM (e.g. metals,
acids,  organic materials), as well as for dissolution or absorption of pollutant gases.  It is the
surface of a particle that is primarily in contact with respiratory cells and surfaces (CD, p.
13-68).  The total surface area of a particle may be important in the presentation  of active
groups on the surface of the particle to cell surfaces (CD, p. 13-26).  Biological effects on
epithelial cells or macrophages may depend on the number of cell surface receptors
stimulated or occupied by particles. Consequently, numbers of particles may be relevant to
their toxic effect (CD, 13-27). Therefore, in comparison to coarse mode particles, fine mode
particles will have the greatest probability of interactions with potential  respiratory targets of
toxicity through increased numbers of particles as well as  surface area (see Appendix D).
       The CD notes that the presence of surface coatings can increase  the toxicity of
particles.  Such considerations may be important when trying to ascertain the appropriate
dose metric for evaluation of lower respiratory tract health outcomes (CD, p. 13-24).  For
example, retardation of alveolar macrophage phagocytosis due to particle overload appears to
be better correlated with particle surface area than particle mass (Morrow,  1988;  Oberdorster
eta al  1995a,b, CD, p.  13-24).  Various biological responses (e.g., reduction in lung
.volumes  and diffusion capacity, alteration in biochemical markers, and changes in lung tissue
morphology) in guinea pigs have been reported  after exposure to ultrafme zinc coated with a
surface layer of H2SO4 (CD, Chapter 11, Chen et al.,  1992b,1995).  These responses were
much greater than those following exposure to larger size  H2S04 in pure droplet form yet
having similar mass concentration of acid.  A possible  mechanism for the differential toxicity
of the  two aerosols is the difference in  particle numbers deposited at  target sites.  At an equal
total sulfate mass concentration, H2SO4 existed on many more particles when layered on  the
ZnO carrier particles than when dissolved into aqueous droplets.  In addition, a recent study
by Chen et al., (1995) confirmed that the number of particles in the exposure atmosphere,
not just total mass concentration of acid,  is an important factor in biological responses
following acidic sulfate particle inhalation when aerosols having the same size distribution
were compared (CD, Chapter 11).

-------
                                          V-75
       Coating the surface of insoluble particles with certain transition metals (e.g. iron) has
been shown to enhance pulmonary toxicity (Costa et al., 1994a,b,; Tepper et al., 1994).
Accordingly, fine particles may serve as an efficient carrier of more toxic material  to
respiratory tract targets.  Coating of micrometer-sized particles with formaldehyde has been
shown to increase the delivery of formaldehyde and consequently increase irritant responses
in human  subjects (CD, 13-76). Jakab  and Hemenway (1993) suggest that reaction products
on particle surfaces may  be more toxic than the primary material.  Exposure to O3 was
shown to increase the toxicity of carbon black particles  in mice. The authors hypothesized
that this result was due to a "reaction of O3 on the surface of the carbon black particles in the
presence of adsorbed water, producing surface bound, highly toxicologically reactive oxygen
species" (CD, p. 11-161).
       Increased surface coating of water or the presence of hygroscopic sulfates,  nitrates,
and organic compounds found as droplets in the fine fraction may also increase the potential
for delivery of irritant  species such as  SO2, hydrogen peroxide, and aldehydes to more
sensitive regions of lung, which, when in the gas phase, would normally be removed in the
extrathoracic region  (CD, p 13-9).  The potential  for increasing delivery of pollutant gases
provides some basis for expecting  some interaction among PM as a pollutant and gases
observed in community studies.
       d.      Deposition in Sensitive Individuals
       As shown in Table V-l, both fine and  coarse particles penetrate to and deposit in the
tracheobronchial and alveolar region.  Based on the epidemiological results and deposition
considerations, it is reasonable to expect that high  levels of coarse particles alone could
aggravate asthmatics through tracheobronchial deposition.  However acids and fine particles
have also  been associated with hospital admissions for asthma in areas with relatively low
coarse mass (Thurston et al., 1992). Receptors that have been linked to an asthmatic
response have been demonstrated to be in areas of the lung where both coarse and fine
particles deposit (see Appendix D).  Moreover, certain  insoluble coarse particles can deposit
and remain  for extended  periods in the alveolar region,  although the relation to the chronic
effects observed in epidemiologic studies  is unclear..

-------
                                           V-76
       The epidemiological studies suggest greater mortality and morbidity effects in
individuals with cardiopulmonary disease.  In this regard,  it is of note that fine particles have
been shown to have a greater deposition in the lungs of individual with chronic respiratory
disease than in normal subjects (CD, Chapter 13).  Such individuals also have reduced
clearance for these particles (see Appendix D).  Thus, the potential for greater target tissue
dose in susceptible patients is present (CD, Chapter 11).  Simulations discussed in Chapter
10 of the CD,  suggest that adolescent children (14-18 yrs of age) are predicted to have
greater respiratory tract daily mass deposition of submicron particles than adults.
5.     Summary and Conclusions
       The staff assessment  of the evidence finds substantial quantitative and qualitative
information on the effects of fine particles and its constituents.  Because  of the remarkable
volume of pertinent literature produced in the last 9 years, far more quantitative
epidemiologic data exist today for relating fine particles to mortality, morbidity, and lung
function changes in sensitive populations on a short- and long-term basis than was the case
for PM10 at the conclusion of the last review.9  Like the PMKI studies,  the fine particle
studies consistently  find positive, significant associations between fine particle levels and
mortality and  morbidity endpoints, with over 20 studies conducted in a number of geographic
locations throughout the world, including the US, Canada, and Europe.  This collection of
studies shows qualitative coherence in  the types of  health effects associated with fine particle
exposure including mortality, morbidity, symptoms, and changes in lung function  (Tables V-
11 to V-13).  The association with mortality  is mainly attributable to respiratory and
cardiovascular causes,  which is consistent with the  range of observed respiratory  and
cardiovascular-related morbidity effects, from respiratory and cardiovascular-related hospital
admissions, respiratory symptoms to changes in lung function.
       By contrast,  the CD and this staff assessment  find much less direct evidence in the
recent epidemiologic and toxicologic literature regarding the potential effects of coarse
particles.  The previous staff assessment of occupational and toxicologic  literature (EPA
    9The 1986 staff assessment of the quantitative basis for the standard cited studies conducted in essentially 3
locations for the 24-hour standard and 4 studies involving a total of 10 cities for the annual standard; none measured
PM10 (EPA, 1986).

-------
                                          V-77
1982a,b) as well as the present review have found ample qualitative reasons to be concerned
about elevated levels of coarse particles smaller than 10 /*m. These effects (e.g., asthma) are
consistent with enhanced deposition of coarse particles in the tracheobronchial region (CD, p.
13-51). However, unlike the case for fine particles, the clearest community evidence
regarding coarse particles finds such effects only in areas with numerous marked exceedences
of the current PM,0 standard (CD, p. 13-51).  In this regard, it appears that the weight of the
available evidence allowing direct comparisons suggests that ambient coarse particles are
either less potent or a poorer surrogate for community effects of air pollution  than are fine
particles.
       It is clear, however, that still more quantitative evidence exists today for PMU), which
includes both fine and  coarse particles.  The above assessment does not conclusively
demonstrate that coarse particles  play no role in the effects  associated  with PMU) at levels
below the standard.  The potential role of coarse particles in producing such effects  could  be
masked in community  studies  by  potential differences in measurement error and exposure
patterns between fine and coarse  particles.  As noted in the  CD, fine particles tend to be
more uniformly distributed than coarse mode particles within (and among) urban areas.
Moreover, the apparent greater infiltration ratio (penetration and settling) of fine particles
indoors means that variations in both short- and long-term personal exposures to outdoor  PM
will be more influenced by fine than coarse particles.
       It is also important to note that some of the more important components of ambient
fine particles (e.g. acid sulfates)  have no notable indoor sources, while a substantial fraction
of indoor coarse particles comes  from indoor resuspension of local crustal  (e.g. deposited  or
tracked in on  footwear) and other coarse materials (Wallace, 1996). This  means that any
effects that are potentially produced  by coarse particles (from outdoor air and indoor
resuspension)  are more likely  to be decoupled from outdoor concentrations.   The less even
urban distribution of coarse particles and stronger  indoor sources would tend to diminish  the
power of community studies of outdoor air to detect the effects of such crustally derived
materials as compared to fine particles (CD, p. 1-9).   Viewed from another perspective, this
also suggests that efforts to reduce any such effects by controlling outdoor coarse particles
would be less successful than  a program to reduce outdoor fine particle effects.   Thus, while

-------
                                           V-78
the epidemiologic data are not conclusive with regard to the potential effects of coarse
particles, they more strongly support the notion that fine panicles are a better surrogate for
that fraction of ambient PM that is most clearly associated with the health effects observed in
community air pollution studies at levels below the current standards.  This view is also
supported by qualitative considerations derived from a consideration of the toxicologic
implications of the profound physical and chemical differences associated with components of
these  fractions.

-------
                                           VI-1
VI.    RISK ASSESSMENT
       The objective of this PM health risk assessment is to provide quantitative estimates of
the risks to public health associated with 1) existing air quality levels, 2) projected air quality
levels that would occur upon attainment of the current PM10 standards, and 3) projected air
quality levels that would occur upon attainment of alternative PM2 5 standards.  As an integral
part of this assessment, qualitative and, where possible, quantitative characterizations of the
uncertainties in the resulting risk estimates have been developed, as well as information on
baseline  incidence rates for the health effects considered.  This assessment provides
information most relevant to evaluating alternative levels of PM standards, rather than to
selecting the most appropriate indicator of PM.  This risk information is intended as a tool that
may, together with other information presented in this Staff Paper, assist the Administrator in
selecting primary PM standards that, in her judgment, would reduce risks to public health
sufficiently to protect public health with an adequate margin of safety, recognizing that such
standards will not be risk-free.
       As discussed in section V.E above, the CD concludes that the overall consistency and
coherence of the epidemiologic evidence suggests a likely causal role of ambient PM in
contributing to adverse health effects (CD, p. 13-1).  Also discussed in section V.E. is an
alternative interpretation, suggested by some researchers, that PM may be serving as an index
for the complex mixture of pollutants in urban air.  The risk assessment described here is
premised on the assumption that PM (measured as PM]0 and PM2 5) is causally related to the
health effects observed in the epidemiological studies and/or that PM is a useful index for the
mixture of pollutants that is related to these effects.
       In presenting this risk assessment,  the staff cautions that despite the consistency and
coherence of the epidemiological evidence with respect to the existence  of effects, quantitative
relative risk results derived from these studies include significant uncertainty. Due to the
uncertainties in the concentration-response study results,  as well as the many sources of
uncertainty inherent in the  analyses presented in this chapter, the risk estimates developed in
this assessment should not be interpreted as precise measures of risk.  The major uncertainties

-------
                                          VI-2
and assumptions associated with these analyses are highlighted in the following discussion and
presentation of results.  In addition, some key uncertainties are addressed quantitatively
through individual sensitivity analyses as well as integrated uncertainty analyses which assess
the combined effects of several key uncertainties.
       The following sections summarize the scope of the analyses, key components of the
risk model, and results of baseline risk and sensitivity analyses.  A detailed discussion of the
risk assessment methodology and results is presented in technical support documents (Abt
Associates, 1996a,b).
A.     General Scope
       The PM risk analyses focus on selected health effects endpoints such as increased daily
mortality, increased hospital admissions for respiratory and cardiopulmonary causes, and
increased respiratory symptoms for children.   Although the risk analyses could not address all
of the various health effects for which there is some evidence of association with exposure to
PM, all such effects are identified and considered above in section V.C.  All concentration-
response functions used in these analyses are based on findings from human epidemiological
studies, which rely on fixed-site, population-oriented, ambient monitors as  a surrogate for
actual integrated PM exposures. Measurements of daily variations of ambient PM
concentrations, as used in the time series epidemiological'studies that provide the
concentration-response relationships for these analyses, have a plausible linkage to the daily
variations of exposure from ambient sources for the populations  represented by ambient
monitoring stations, as discussed in Chapter IV.  The CD concludes that this linkage should be
better for indicators of fine particles (e.g., PM25) than for indicators of fine plus coarse
particles (e.g., PM10, TSP), and in turn,  should be better than indicators of inhalable coarse
fraction particles (PM10 - PM2 5) (CD, p.  1-10).  A more detailed discussion of the possible
impact of exposure misclassification on the estimated concentration-response relationships
derived from the community epidemiological studies is presented above in section V.E.

-------
                                          VI-3
       These PM risk analyses feature:
•      analyses of risks under a recent 12-month period of air quality (labeled "as is" air
       quality) and under a situation where air quality just attains various alternative standards
       being considered;
•      estimates of risks for the urban centers of two example cities, one eastern (Philadelphia
       County) and one western (Southeast Los Angeles County),  rather than national
       estimates;
•      estimates of risks only for concentrations exceeding an estimated background level;
       and
•      qualitative and quantitative consideration of uncertainty, including sensitivity analyses
       of key individual uncertainties and integrated uncertainty analyses combining key
       uncertainties.
       More specifically, consistent with the recommendations to the Agency provided in the
January 5,  1996 CAS AC letter to the Administrator (Wolff,  1996b), alternative 24-hr and
annual PM25 standards are examined alone and in combination with the current  PM10
standards.  This  focus also reflects the conclusions drawn in the CD (CD, Chapter 13) that it is
appropriate to consider fine and coarse fraction particles separately, and that for mortality and
some measures of morbidity, the most consistent associations are seen with fine and thoracic
particles (e.g., PM:5, PM10) as compared to coarse  fraction particles (CD, Chapter 13; section
V.F above).  The scope of these analyses initially focuses on developing risk estimates for
portions of two selected urban areas:  Philadelphia County and a portion (roughly the
southeastern third) of Los Angeles County (hereafter referred to as "Los Angeles County").
These areas were chosen based on availability of PM]0 and PM2 5 air quality data, and the
desire to include areas from the eastern and western parts of the United States to reflect
regional differences in the makeup of PM.   Finally, estimates of risks above background PM
concentrations are judged to be more relevant to policy decisions about the level of ambient air
quality standards than estimates that include  risks potentially attributable to uncontrollable
background PM  concentrations.

-------
                                          VI-4
B.     Components of the Risk Model
       In order to estimate the change in health effects incidence corresponding to the
difference  in PM levels between "as is" conditions and just attaining alternative standard
scenarios, the following three key components are required for a given health endpoint and
selected city: 1) air quality information, 2) concentration-response relationships, and 3)
baseline health incidence rates.  Figure VI-1 is a broad schematic depicting the role of these
components in the risk analyses.  The general health risk model which combines changes in
PM air quality concentrations (Ax), the concentration-response relationships for a given health
endpoint (reflected by P, the PM coefficient derived from epidemiology studies), and the
baseline health effects incidence rate (y) for  a given health endpoint is represented by equation
1:
  Equation  1                                          Ay=.y|>pAx-l]

       Estimates  of risk (i.e., health effects  incidences attributable to PM) are quantified for
PM concentrations above background except for those studies in which the range of observed
PM concentrations did not go down to estimated background (e.g., the prospective cohort
mortality studies). For these studies effects  are quantified down to the lowest concentrations
observed in the study.  As indicated in Figure VI-1, sensitivity analyses on various key inputs
to the PM health  risk model are conducted as part of this assessment, as well as an integrated
uncertainty analysis that examines the potential impact of combining several key uncertainties.
Each of these key components is briefly discussed below.
1.     Air Quality Information
       The air quality information required to conduct the PM risk analyses includes:  1) "as
is" air quality data for both PM10 and PM2 5  from population-oriented monitors for the  selected
cities,  2) estimates of background PM concentrations appropriate to that location, and 3) a
method for adjusting the "as is" data to reflect patterns of air quality change estimated to occur
when each  city attains various alternative standards. Table VI-1 provides a summary of the

-------
                                              VI-5
      Figure Vl-1  Major Components of Particulate
                       Matter Health Risk Analysis
 Ambient Population-
 oriented Monitoring
 for Selected Cities
   Air Quality Adjustment
   Procedures
   Alternative Proposed
   Standards
Human Epidemiological
Studies (various health
endpoints)


City-specific (or National)
Baseline Health Effects
Incidence Rates (various
health endpoints)
                                   "As is" Analysis
     Changes in
     Distribution
      of PM Air
       Quality
Concentration
Response
Relationships
                           Health
                            Risk
                           Model
Risk Estimates.
  "As is"
  "Alternative
  Scenarios"
         Sensitivity Analysis: Analysis of effects of alternative assumptions, procedures or data occurs at these points.

-------
                             TABLE VI-1. CITIES EXAMINED IN PM RISK ANALYSIS





City
Philadelphia
County, PA
Los Angeles
County, CA




Population"
(millions)
1.6

3.6






Year
1992-93

1995

% of Days on Which
Air Quality Data are
Available


PMIO
99

59



PM2,
96

59

PM10h


Annual
Average
(/ig/m3)
25

52

Second Max,
24-hr Avg.
(Mg/m3)
77

195

PM2.5b


Annual
Average
(Mg/m3)
17

30

Second Max,
24-hr Avg.
(Mg/m3)
72

129

"Based on 1990 U.S. Census data.
hConcentrations are reported for the monitor with the highest value.
Note: More detailed information about the air quality data in these cities is presented in Section 4 of Abt Associates (1996b).
                                                       VI-6

-------
                                          VI-7
PM10 and PM2 5 air quality data for the two areas included in these analyses. The PM10 and
PM2 5 monitoring information for Philadelphia County are from three monitors used in the
Acid Aerosol Characterization Study during 1992-1993 (network sites described in Suh et al.,
1995). The monitoring information for southeast Los Angeles County comes from two
dichotomous samplers operated during 1995 by the South Coast Air Quality Management
District.  Figure VI-2 presents frequency distributions of the daily PM]0 and PM2 5
concentrations in Philadelphia County based on spatially averaging the reported concentrations
available from the different monitors for each day.  Figures VI-3  and VI-4 show the frequency
distributions of the daily PM10 and PM2 5  concentrations by quarter in southeast Los Angeles
County based on spatially averaging the reported concentrations available from  the different
monitors for each day.
       As discussed above, these ambient concentrations are used as a surrogate for population
exposures in these analyses, a procedure consistent with the health literature but which adds
uncertainty to the risk estimates.  In an effort to limit uncertainties that would result in
combining data across different monitoring methods, only information from these monitors
was used directly in the risk analysis.1
       Background PM concentrations used in these analyses are defined in Chapter IV as the
distribution  of PM concentrations that would be observed in the U.S. in the absence of
anthropogenic emissions of PM and its precursors in North America.  For these analyses, an
estimate of the annual average background level is desired, rather than a daily average (e.g.,
the maximum 24-hour level), since estimated risks are aggregated for each day  throughout the
year. The staff have chosen to use the midpoint of the appropriate ranges of annual average
estimates for PM background presented in Table IV-3 for the base case risk estimates  (i.e.,
    'Although not directly used in the risk analyses, information from the AIRS database for sites in Los
Angeles county was used to help define the region of Los Angeles County included in this analysis (see Abt
Associates, 1996b).

-------
                               VI-8
Figure Vl-2. Daily Average PM Concentration Frequencies
Philadelphia County,  September 1992 - August 1993
                          PM-10
                   Data Available on 358 Days
              03
              025	
              02 --
              015 -
            0)
            (T
              01 —
              005 —
                                 _ bin width = 5 ug/m3 _
                        ILL.
                0  20  40  60  80 100 120 140 160 180 200
                     24-hour Average PM-10 Concentration
                          PM-2.5
                    Data Available on 352 Days

              03	
              025 —
              02 -
        |    £ 015 —
              01
              005 —
                    	  	  	   bin width = 5 ugm3
                      IllL....
                 0  20  40  60  80  100 120 140 160 180 200
                     24-hour Average PM-2 5 Concentration

-------
Relative Frequency
0 0
o -» ro
:L

9 — —



•M^^M




a
1°
SO
0
o -»
II
S-
go>
0 0
3
OD
0
^^^•B
mfm^^m
wm











ro ;
0 I
0



1
























0 0
01 *




















1
1 S"
' 1 5 1
0.1
i g-1
1 i ||
e'i
i| ?i
1 'j->.







O
U
> Sf
£ Q.
Quarter
lable on 86 DC
^z

>0
i
§ °
^ _^
o o
OO
O
o
o





mmmmm
mmmmmm
Umm*mm
IMMHM^
••^•••4
pM^^M











H^B

«»
••••
••MM



























O
ro
































0
w







































0
«k




«rl
3
t
|
II
'""I
C 1
ll
O-f

O
£ 5^
01 r*
I D
CD C
o 0)
-5
w fl)
D "»
01





(








0) -n
igure VI-C
outheast
i — •
o
v> D
"V* 0)
*^ \^*~
(Q ^
— <
w H)
	 	 - ^., ^
O(Q
O (D
C
Relative Frequency
0
*






CT-
5
0,
II '


0
Q)
Q) _
i 3"
01 /^
cr kJ
CD C
o 0)
3 -t
OO ^^r
co (D
o t
0)
(/)

c i
T=,

1




















o


o
ro



"—

T'
|g^— H
a>
18
5s
§s
5-
5 0)
00
g
K)
O








O
Relative Frequency
o o o o
-• ro o> *.







^^^ '



































































s
1
"ll

c
B
UJ




o
01
01 K)
> 3
01 ^^
oT /^
s D
CD gj
S to
^ (D
O ^1
w






CD °
CD Q
01 y

O
(D
2.
CO
o

o








-------
                     Relative Frequency

                  o            o
I1
Oo
o   I
I:
                     Relative Frequency
?§
Oo
                                                         O
                                                         Q)

                                                         I
                                                         >
o

*
cn
O
   w
   -n
   Q.

   o
   C
Relative Frequency
o
o
O 1
NJ
O
ro
•f^
rr o
o


MHHHMM

>g —
Q) QO
(0 0


"0 -*•
?8
ro
"•w
Oo

o
o ^
3°
as

§s
3 _^ :
oo ;
o
to

o
o



O O
-* M O>



•^"

MMM


































































1



































2:
3
>
i
ii
t>i
1
u>



§
tB -
^. -^
^ w
Q> r*
| D
^ ^^
3 <
CQ y>
CD Q)
                                                                                                                                                      {/)   CD
                                                                                                                                                                  I
                                                                                                                                                                  I—*
                                                                                                                                                                  o
                                                          D
                                                          0)
                                                          <
                                                          0)
O


oo
CD

O
Q)
    c
    Q)
                                                                              iro
                                                                              SS





Relative Frequency
o
D -»

	




0 O
NJ Ol


U-. 	 i














_











































































21
1
»1
"H
1
!


D
D)

5 Q.
S- D
to c
O Q)

M i3-
->l (P
B> "^
cn




13 N3
«<" cn
_^ 0
CD O
CD 3
^t
Q)
i-t-
O
-n

—i
CD
o
CD
O
CD
CO



-------
                                         VI-11
eastern values are used for Philadelphia and western values for Los Angeles):
•      For PM10:  5-11 /ig/m3 for Philadelphia, and 4-8 /xg/m3 for Los Angeles
•      For PM2 5:   2-5 fig/m3 for Philadelphia, and 1-4 /*g/m3 for Los Angeles.
Sensitivity analyses have been done using the appropriate lower and upper ends of the above
ranges to characterize the impact of this model input choice on the risk estimates.
       To estimate health risks associated with just attaining alternative PM2 5 standards, it is
necessary to estimate the PM concentrations that would occur under each alternative standard.
When assessing the risks associated with long-term epidemiological studies that use an annual
average concentration level, the annual mean is simply set equal to the standard level. In
contrast, when assessing the risks associated with short-term epidemiological studies, the
distribution of 24-hour values that would occur upon just attaining a given 24-hour PM
standard has to be  simulated. While there are many different methods of reducing daily PM
levels, preliminary analysis found that PM levels have in general historically fluctuated in a
-proportional manner (i.e., concentrations at different points in the distribution of 24-hour PM
values have decreased by approximately the same percentage) (Abt Associates,  1996b).
Therefore, attainment of the current PM]0 and alternative PM2 5 daily  standards  has been
simulated by adjusting the "as is" air quality data using a proportional rollback approach (i.e.,
concentrations are  reduced by the same percentage) for concentrations exceeding the estimated
background level (see Abt Associates, 1996b).  Sensitivity  analyses have been conducted to
examine alternative air quality adjustment procedures (e.g., a method that reduces the top 10%
of daily PM concentrations more than the lower 90%).
2.     Concentration-Response Functions
       The second key component in the risk model is the set of concentration-response
relationships which provide estimates of the relationship between each health endpoint of
interest and ambient PM concentrations.  Table VI-2 summarizes the  selected epidemiological
studies which are judged adequate by the CD to provide estimated concentration-response
relationships  for a  variety of health endpoints associated with elevated PM10 and/or PM2 5
exposures (CD, Tables 13-3, 13-5). Only studies based on either PM10 and/or PM2 5 as a
measure of PM have been used  in these analyses.  Each study provides an estimate of relative

-------
   Table VI-2. Selected Epidemiological Studies and
Associated Relative Risk Estimates Used in Risk Analyses


Health Effect

PM
Indicator


Study Location
Reported PM Levels
(Hg/m1)
Mean (Range)'

Estimated Relative Risk2
(95% Confidence Interval)

Pooled Relative
Risk1
TOTAL MORTALITY
Short-term Exposures

















Long-term Exposures

PM.o










PM25






PM25

Six Cities'1
Portage, WI
Boston, MA
Topeka, KS
St. Louis, MO
Kingston/Knoxville, TN
Steubenville, OH
Chicago, IL"
Utah Valley, UT
Birmingham, ALd
Los Angeles, CAe
Six Cities'1
Portage, WI
Topeka, KS
Boston, MA
St. Louis, MO
Kingston/Knoxville, TN
Steubenville, OH
ACS Study'
(50 U.S. SMSA)

18 (.+ 11. 7)
24 (+12. 8)
27 (+16.1)
31 (+16.2)
32 (+14.5)
46 ( + 32.3)
38 (NR/128)
47(11/297)
48(21,80)
58 (15/177)

11.2 ( + 7. 8)
12.2 (+7.4)
15.7 (+9.2)
18.7 (+.10.5)
20.8 (±9.6)
29.6 (+21. 9)
9-344


.04(0.98, 1.09)
.06(1.04, 1.09)
0.98(0.90, 1.05)
.03(1.00, 1.05)
.05(1.00, 1.09)
.05(1.00, 1.08)
.03(1.02, 1.04)
.08(1.05, 1.11)
.05(1.01, 1.10)
.03(1.00, 1.06)

.03(0.99, .07)
.02(0.95, .09)
.06(1.04, .07)
.03(1.01, .04)
.04(1.01, .07)
.03(1.00, .05)
1.17(1.09, 1.26)

HOSPITAL ADMISSIONS - Short-term Exposures
All Respiratory
Causes
(for Elderly > 64 years)

PM10



Tacoma, WA8
New Haven, CTB
Cleveland, OH"
Spokane, WA1
37(14,67)
41 (19, 67)
43 (19, 72)
46(16,83)
1.10(1.03, 1.17)
1.06(1.00, 1.13)
1.06(1.00, 1.11)
1.08(1.04, 1.14)





1.04(0.99, 1.09)








1.04(1.00, 1.07)



—



1.09(1.02, 1.19)


                           VI-12

-------
Health Effect

PM
Indicator
PM2,
Study Location
Toronlo1
Reported PM Levels
Oig/m1)
Mean (Range)1
18.6(NR/66.0)
Estimated Relative Risk2
(95% Confidence Interval)
1.15(1.02, 1.28)
Pooled Relative
Risk1

HOSPITAL ADMISSIONS -- Short-term Exposures
COPD
(for Elderly > 64 years)
Ischemic Heart Disease
(for Elderly > 64 years)
Congestive Heart Failure
(for Elderly > 64 years)
Pneumonia
(for Elderly > 64 years)
PM10
PM,o
PMIO
PM,0
Minneapolis, MM1
Birmingham, AL1
Spokane, WA1
Detroit, MI1"
Detroit, MI"
Detroit, MI"
Minneapolis, MNk
Birmingham, AL1
Spokane, WA'
Detroit, MI"1
36 (18,58)
45 (19,77)
46 (16,83)
48 (22,82)
48 (22,82)
48 (22,82)
36(18,58)
45 (19,77)
46 (16,83)
48 (22,82)
1.25(1.10, 1.44)
1.13(1.04, 1.22)
1.17(1.08, 1.27)
1.10(1.02, 1.17)5
1.02(1.01, 1.03)
1.03(1.01, 1.05)
1.08(1.01, 1.15)'
1.09(1.03, 1.15)
1.06(0.98, 1.13)
1.06(1.02, 1.10)5
1.14(1.05, 1.31)
—
—
1.07(1.01, 1.14)
RESPIRATORY SYMPTOMS
Lower Respiratory
Symptoms in Children:
Short-term Exposures
Bronchitis in Children:
Long-term Exposures
PMIO
PM2,
PM15/10
Six Cities"
Utah Valley, UP
Six Cities"
Six Cities"
30(13,53)
46(11/195)
18.0 (7.2-37)
20-594
2.03(1.36, 3.04)6
1.28(1.06, 1.56)
1.44(1. 15-1. 82)6
3.26(1.13, 10.28)6
...
—
—
'Kinneyetal. (1995)
'Pope etal. (1995)
8Schwartz (1995)
"Schwartz et al.  (1996b)
'Schwartz (1996)
Thurston et al. (1994b)
"Schwartz (19940
'Schwartz (1994e)
"Schwartz (1994d)
"Schwartz and Morris (1995)
"Schwartz et al. (1994)
Tope etal. (1991)
References:
"Schwartz et al. (1996a)
''Ito and Thurston (1996)
Tope etal. (1992)
"Schwartz (1993a)
Endnotes:
1. Range of  24-hour PM indicator level shown in parentheses is typically either the standard deviation (+ S.D.) or 10th and 90th percentiles.
2. Based on a 50 ng/m* increase for PMIO studies, and a 25 /xg/m1 increase in PM25 studies.
3. See Abt Associates (1996b) for calculation method.
4. Range of city means of PM levels.
5. Only  RR  reported includes other pollutants in model.
6. Odds  ratio.
MDockery et al. (1989)
                                        VI-13

-------
                                          VI-14
risk (P), along with a measure of the uncertainty (95% confidence interval) of the estimate,
associated with specific changes in PM levels (i.e.,  a 50 fig/m3 increase in PM10 or a 25 /*g/m3
increase in PM25).
       As indicated in the CD, the most credible approach to risk analysis would be to use
site-specific relative risk (RR) estimates for PM (CD, p. 13-87). For Los  Angeles County,
site-specific RRs are available from two studies (Kinney et al, 1995; Ostro et al., 1995).
Philadelphia County has been the location of several studies reporting associations between PM
and mortality and hospital admissions, but none of the published reports have used PM10 or
PM2 5.  Since site-specific relative risks are not available for all endpoints  in both locations
(and in the absence of more information concerning which individual studies might most
appropriately characterize the health risk in a risk analysis location), an approach was
employed which combined available information from all the key studies for a health endpoint.
A form of meta analysis (referred to as a "pooled analysis" in this Staff Paper) was conducted
which combined the results of the various studies.   For comparison purposes, Table VI-2 lists
the mean estimate of RR from the pooled analysis along with the  RRs for  the individual studies
comprising the pooled analysis.
       Given differences in population,  particle size distribution,  and other environmental
stressors (e.g., weather variables, co-pollutants), RRs may be expected to vary from location
to location.  The CD notes such variation appears to be observed  in coefficients for mortality
associated with short-term exposures,  and cautions against the application  of a single "best
estimate" relative risk value across various  locations (CD,  p. 13-87).  The  pooled analyses in
this risk analysis have utilized an "empirical Bayes" approach in an effort to  more fully reflect
the range of relative risk estimates,  and accompanying statistical uncertainty, seen from
location to location.  Standard meta analysis techniques, such as a random effects meta
analysis, estimate a mean relative risk and the statistical uncertainty around that mean estimate.
The empirical Bayes approach estimates the underlying distribution of RRs observed across
areas and the likelihood that any relative risk estimate from that distribution will be applicable
to an uninvestigated location.   The empirical Bayes approach uses the  random effects model
framework, in which the relative risks from different locations can be genuinely different,

-------
                                           VI-15
while adjusting the relative risk and statistical uncertainty observed in individual locations to
some degree to reflect the information available from the entire set of studies (see Abt
Associated, 1996b, for further details).  However, the distribution of RRs from the empirical
Bayes approach provides uncertainty estimates ("credible intervals") which are intended to
represent the range of reported RRs (and not simply the uncertainty around a mean estimate)
and is not restricted to assuming a normal distribution (see Abt Associates, 1996b, Exhibit
5.12). As a result, credible intervals from the empirical Bayes approach are typically wider
than confidence intervals from random effects meta analysis2 and are expected to more fully
convey information on both statistical uncertainty  and potential inherent differences (due to
different population characteristics, PM size distributions, etc.) in the RRs for different
geographic locations.2
       In the risk analyses, the 5th and 95th percentile values from the distributions of RRs
estimated by the empirical Bayes approach are provided as a 90% "credible interval" to
characterize uncertainty in the risk estimates for each endpoint. (In Table VI-2, the 95%
credible  interval around the pooled relative risk estimate is provided instead, to facilitate
comparison with the reported RRs from the original studies).  In the risk analyses the mean of
the distribution based on the empirical Bayes approach is also reported as an estimate of the
central tendency of the distribution.  Because a random effects framework was  used for the
empirical Bayes approach, this mean estimate is identical to what would be estimated by a
random effects meta analysis.  A more detailed description of the techniques used to develop
the pooled estimates and the application of the empirical Bayes approach is provided in the
technical support document (Abt Associates,  1996b).
       In the absence of site-specific RRs for all the endpoints of interest (a product of data
limitations that preclude constraining the assessment solely to those areas where both adequate
air quality and concentration-response information are available), pooled analyses using this
     Exhibit 5.10 of Abt Associates (1996b) shows that the credible intervals resulting from the empirical Bayes
approach are wider for cases in which a number (6-10) of location-specific concentration-response relationships are
available (e.g., mortality associated with short-term exposures of PMIO or PM25), but not substantially different for
hospital admissions endpoints for which fewer studies (3-4) were pooled.

-------
                                           VI-16
empirical Bayes approach is one method employed to allow potential differences in RR from
location to location to be reflected in the risk estimates.  As an additional approach, sensitivity
analyses have been performed evaluating the effects of including alternative studies or
excluding studies or groups of studies from the pooled analyses (Appendix F, Table F-4; Abt
Associates, 1996b).
       The CD identifies the interpretation of specific concentration-response relationships as
the most problematic issue for risk assessment purposes at this time due to the absence of clear
evidence regarding mechanisms of action for the various health effects of interest (CD, p. 13-
87).  The reported study results used in these analyses are based on linear models extending
over the range of air quality within the study,  as illustrated in Figure VI-5 (CD,  Figure  13-5)
by Line A. This model implies a possible  linear, no-threshold underlying relationship
potentially extending to zero PM concentrations (illustrated by Line B).  Alternatively, the
existing data do not rule out the possible existence  of an underlying non-linear, threshold
relationship (illustrated by Line C).  Although these alternative interpretations of study results
could significantly affect estimated risks, only very limited information is available to aid in
resolving this issue (CD, pp. 13-87-91). Thus,  the approach taken in this risk assessment is to
address alternative models through sensitivity and integrated uncertainty analyses to develop
ranges of estimated risks, rather than characterizing any of the sets of risk estimates as
representing best estimates.
       To frame the sensitivity analyses of concentration-response models, the results from
various studies have been examined  through a number of alternative approaches to identify
appropriate PM concentration "cutpoints"3  which define the lower end of the range over which
the concentration-response functions would be applied. Table VI-3 summarizes the cutpoints
examined in the sensitivity and integrated uncertainty analyses.  A more detailed discussion of
the basis for selecting these particular cutpoints  is presented  in Appendix E.
    3 "Cutpoint" as used in Chapter VI refers to concentrations determined to be of interest for evaluating the
sensitivity of risk estimates to assumptions about the shape of concentration-response relationships.  This is in contrast
to the use of the term "cutpoint" in Chapter IV, which refers to the aerodynamic diameter of particles being sampled
by a monitor.

-------
                                        VI-17
                      10
                      Particulate Matter Concentration (pg/rn3)
Figure VI-5. Schematic Representation of Alternative Interpretations of Reported
Epidemiologic Relative Risk (RR) Findings with Regard to Possible Underlying PM
Mortality Concentration-Response Functions (CD, Figure 13-5). Published studies
typically only report results from linear models that estimate RR over a range of observed PM
concentrations as represented by Line A (specific PM values shown are for illustrative
purposes only),  compared against baseline risk (RR = 1.0) at the lowest observed PM level.
One alternative  interpretation is that the RR actually represents an underlying linear, no-
threshold PM-mortality relationship (Line B) with the same slope as Line A but extending
below the  lowest observed PM level essentially to 0 //g/m3.  Another possibility is that the
underlying functional relationship may have a threshold (illustrated by Curve C), with an
initially relatively flat segment, not statistically distinguishable from the baseline risk (1.0)
until some PM concentration where it sharply increases (or more likely somewhat less sharply
ascends in the vicinity of the breakpoint as shown by the dashed lines).

-------
                                         VI-18
  Table VI-3. Concentration-Response "Cutpoints" Examined in Uncertainty Analyses
Pollutant
PM10
PM25
PM25
Health Effects
Effects Associated with Short-Term Exposure
Effects Associated with Short-Term Exposure
Effects Associated with Long-Term Exposure
Cutpoints Examined
(Mg/m3)
20
10
12.5
30
18
15
40
30
18
       In conjunction with defining such concentration cutpoints, the slopes of the
concentration-response functions have been increased to reflect the effect of potential
thresholds at the selected levels. This concept that the slope above a cutpoint would be
expected to increase somewhat in a threshold model is illustrated by the comparison of linear
and nonlinear models applied, for example, to the TSP data set from Philadelphia presented in
the CD (CD, Table 13-6; Appendix F, Figure F-l).  Figure VI-6 illustrates the two methods
used to adjust slopes when nonlinear models with cutpoints were applied in the risk analyses.
The first method adjusts the slope of the relationship from the cutpoint to the maximum
concentration observed in the health effects studies so that the area under this line is the same
as the area under the original concentration-response relationship that went down to estimated
background. To compensate for fewer PM-associated health effects at low concentrations (and
no effects below the cutpoint level), the adjusted function must rise more rapidly than the
original function.  The second slope adjustment method assumes that the RR associated with
the maximum concentration observed in the studies is the same as in the original function and,
therefore, the concentration-response  relationship extends from the cutpoint to the RR
observed at the maximum concentration in the original study.  This second method increases
the slope less than the first method.  It is important to recognize that the two adjustment

-------
                               VI-19
    Figure VI-6.  Slope Adjustment Methods Used in
           Sensitivity and Uncertainty Analyses
                    (PM-10 Pooled Mortality Function)
U \j.tt •
I
O)
X. 0.20-
O
S

>
o
JO
(0
2-012-
(0
£
S> 0.08-
1s
"3
t£ 0.04-
"c"
_J

0,
'
j

x~
x j
•^ ;
Slope Adjustment ^ - ' [
Method 1 •^v.^ ^-' ^-^1
^""T^^. x- ' ^^- \
' ' ^"^ ' ^ '''•
' ^^^^ •
_ - ' ^-"/ ^ j
^ ^ '
-• ^"' ^
Original Function ^^ ^ -^ \
\^" -^ :!
^'' \ 1
^' ,'s- Slope Adjustment j
,. -^ .^^ Method 2 !
-^ -V- =
"^ ^ :
- '"' 	 ^^" 	 	 	 ^ 	 	 	 ^, 	 j,
	 T 	 T 	 T T
J 30 125 220 25
     Eastern   Example
     Background Outpoint
PM-10 Levels (M9/m3)
Highest
Observed
PM-10 Level
in Studies
Relative Risks shown are the risks associated with elevated PM-10 levels relative to the
risks associated with the background PM level (8 ug/m3) for Philadelphia County.

-------
                                           VI-20
methods are illustrative and intended to roughly bound the potential impact on concentration-
response relationships if cutpoints or thresholds above background exist.
       Based on this examination of study results, presented in Appendix E, the cutpoints
identified in Table VI-3 have been selected as a basis for a series of sensitivity and uncertainty
analyses. Results of sensitivity and uncertainty analyses involving cutpoint and other
important uncertainties are presented in section VI.C below.
       An additional issue concerning the appropriate interpretation of ambient PM
concentration-response relationships is whether they may represent effects resulting from the
combined exposure to ambient and indoor particles (or some subset of ambient and indoor
exposures,  such as the combined exposure to ambient and indoor combustion source particles).
While total personal exposure to ambient and indoor particles can be substantially higher than
exposure to ambient particles alone4, the CD concludes that additional exposure to particles
indoors from sources independent of ambient sources  (which individuals can be exposed to
when either outdoors or indoors, since particles penetrate residential indoor microenvironments
(CD, p. 1-9)) would not be expected to systematically affect coefficients of ambient
concentration-response relationships (CD, p. 1-10).
3.      Baseline Health Effects Incidence Rates
        The third key component required in the PM risk analyses is an estimate of the baseline
health effects incidence rate corresponding to "as is" PM levels.  Incidence rates express the
occurrence of a disease or event (e.g., asthma episode, hospital admission, death)  in a
specified time period, usually per year.  Health effects incidence rates vary among geographic
    4For example, the PTEAM study found that for a study population in Riverside, CA, during a period in which
 daytime ambient PM10 concentrations measured at a central monitor averaged 91 ^g/m3 and ranged from 37 - 158
 /ig/m3 (10th -90th percentile of daytime concentration distribution), daytime total personal exposure averaged
 approximately 60% higher (150 jig/m3, ranging from 60 - 263 j*g/m3 (10th -90th percentile) (Clayton et al, 1993).
 However, nighttime ambient and personal exposures were highly similar [mean concentrations were identical (77
 /ig/m3) with ambient PM10 values ranging slightly above and below personal exposure values across the group (10th-
 90th percentile range 30-156 /ig/m' ambient; 37-135 /jg/m3 personal)].

-------
                                          VI-21
areas due to differences in population characteristics (e.g., age distribution) and factors
affecting illness or response (e.g., smoking, occupation, income levels,  air pollution levels).
       Tables VI-4 and VI-5 provide a summary of population estimates and baseline mortality
and morbidity incidence rates used in these analyses for Philadelphia and Los Angeles
Counties. Mortality rates are based on county-specific data from the National Center for
Health Statistics.  Morbidity rates for hospital admissions in Philadelphia are based on
Philadelphia County admissions data obtained from the Delaware Valley Hospital Council,
and for Los Angeles County from California's Office of Statewide Health Planning and
Development.  For respiratory symptoms, baseline incidence information on symptoms is not
routinely reported, so for these endpoints the incidence rates from the studies themselves were
used. This would be expected to introduce considerable uncertainty, since baseline symptoms
incidence would be expected to vary across locations, and because many diary studies (e.g.,
Schwartz et al., 1994; Pope et al., 1991) do not record symptoms incidence across an entire
year. Thus, incidence estimates  for respiratory symptoms are particularly uncertain and are
primarily included to provide perspective on the number of effects estimated relative to other
health effects.
       Uncertainty  in baseline incidence rates primarily affects estimates of numerical
incidence (e.g., counts of number of hospital admissions, symptoms). Percent of incidence
estimates can be obtained without the use of baseline incidence health information, since
almost all of the key studies used in the risk analysis report results in the form of RR versus air
quality (the exception being Thurston et al., 1994) which generate the same percent of
incidence estimates  regardless of the baseline incidence rates.  Baseline incidence rates are only
involved in estimating the implication of the estimates of percentage incidence in terms of
numbers of health effects.

-------
                                               VI-22
            Table VI-4. Relevant Population Sizes for Philadelphia County and
                                 Southeast Los Angeles County
Population
Total
Ages > 65
Children, ages 8-12
Children, ages 10-12
Asthmatic Children,
ages 9-11
Asthmatic African- American
Children, ages 7-12
Philadelphia County
1,590,000
241,000 (15.2%)
103,000 (6.5%)
62,000 (3.9%)
3,900* (0.3%)
—
Southeast Los Angeles
County
3,640,000
322,000 (8.9%)
282,000 (7.8%)
166,000 (4.6%)
10,700* (0.3%)
1,800* (0.05%)
Incidences for asthmatic children were obtained using the national asthma prevalence among children (6.3%).  The
incidence of asthmatic African-American children ages 7-12 in Southeast L.A. County, for example, is 3,640,000
multiplied by {0.0937 (the proportion of the population that is ages 7-12) x 0.085 (the proportion of the population
that is African-American) x 0.063 (the proportion of the national population of children that are asthmatic)}.

-------
                                                   VI-23
Table VI-5.  Baseline Health Effects Incidence Rates
Health Effect
Mortality13
(per 100,000 general population/year)
Philadelphia
County
1280
Southeast Los
Angeles County
667
National
Average8
830
Morbidity:
A. Hospital Admissions (per 100,000 general population/year)
Total respiratory hospital admissions0 (all ages):
ICD codes 466, 480-482, 485, 490493
Total respiratory hospital admissions (65 and older):
ICD codes 460-519
COPD admissions (65 and older): ICD codes 490-496
Pneumonia admissions (65 and older): ICD codes 480-487
Ischemic heart failure (65 and older): ICD codes 410-414
Congestive Heart Disease (65 and older): ICD code 428
816
650
202
257
614
487
427
428
116
205
307
197
~
504
103
229
450
231
B. Respiratory Symptoms (percent of relevant population)
Lower Respiratory Symptoms (LRS) in children, ages 8-12
(number of cases of symptoms per day)
Lower Respiratory Symptoms (LRS) in asthmatic children,
ages 9-11 (number of days of symptoms)
(Doctor diagnosed) acute bronchitis in children ages 10-12
per year
0.15%*
16%*
6.5%*
0.15%*
16%*
6.5%*


—
All incidence rates are rounded to the nearest unit.
a. National rates for hospital admissions for patients over 64 years of age were obtained from Vital and Health
Statistics, Detailed Diagnoses and Procedures, National Hospital Discharge Survey, 1990.  June, 1992.  CDC.
Hyattsville, Md. Each rate is based on the number of discharges divided by the  1990 population of 248,709,873 .
b. Mortality figures exclude suicide, homicide, and accidental death, which corresponds to the measures used in the
epidemiological studies employed in this analysis.
c. Although a baseline incidence rate is not needed for calculating the incidence of total respiratory hospital
admissions associated with PM (because the concentration-response function is linear), it is needed for calculating the
percent change in incidence associated with PM.

*Baseline incidence rates for respiratory symptoms were taken from the original studies:  Schwartz et al. (1994):
percent of all child-days  on which there were respiratory symptoms, as defined in the  study;  Pope et al. (1991): for
number of days of LRS in asthmatic children ages 10-12; and Dockery et al. (1989), for acute bronchitis in white
children ages 10-12.

-------
                                          VI-24
4.     Limitations and Uncertainties
       This PM health risk assessment involves substantial uncertainties given the nature of the
pollutant, limited data on population exposures, and the nature of the epidemiological evidence
of effects.  The major uncertainties include:
•      Limited information on air quality and on human activity patterns (e.g., how they vary
       over tune and location compared to the original studies) add uncertainty to the analyses.
        Errors in measurement of relevant air quality, both instrument error in monitored
       concentrations and errors resulting from using averages of population-oriented monitors
       to represent population exposure, are potentially important sources of uncertainty.
•      Modeled air quality simulations of attainment of alternative PM standards introduce
       potentially significant uncertainties, particularly in assessing the impact of alternative
       standards with regard to the pattern of reductions that would be observed across the
       distribution of air quality values.
•      The use of uncertain estimates of annual average background PM concentration for
       each location results in uncertainties with regard to estimates that are representative of
       risks in excess of those potentially attributable to uncontrollable background PM levels.
•      Insufficient information exists to fully assess the extent to which PM concentration-
       responses functions reflect the best estimates of risk associated with PM, as well as
       whether such functions are transferable across cities due to (1) variations in PM
       composition across cities, (2) the possible role of associated  copollutants in influencing
       PM risk, and (3)  variations in the relation of total  exposure to ambient monitoring in
       different locations.  There also is the additional  uncertainty concerning the
       transferability of health functions to future PM aerosol mixes.
•      The use of pooled concentration-response functions from studies in several locations to
       represent the overall effect of particles on a particular health endpoint in any  one
       location introduces uncertainty.
•      The impact of historical air quality on estimates of health risk from long-term PM
       exposures is not well understood, nor is the duration of time that a reduction  in particle

-------
                                          VI-25
       concentrations must be maintained in a given location in order to experience the
       predicted reduction in health risk.
•      Normalizing the health risk experienced or reduced in different locations due to
       differences in the completeness of the air quality data sets introduces uncertainty.
•      Additional uncertainty is related to baseline health effects incidence information,
       particularly  where location specific information is not available and must be estimated
       either by scaling national incidence rates or using reported rates from the original
       studies.  Uncertainties in baseline health information would be expected to affect
       numerical estimates of total incidence more than estimates of the percentage of
       incidence.
Sensitivity and uncertainty analyses addressing many of these uncertainties are presented along
with the PM risk estimates in the following  section and in Appendix F.
C.     Risk Estimates for Philadelphia and I^>s Angeles Counties
       In the sections below risk estimates are first presented for the two locations analyzed
using base case assumptions associated with "as is" PM levels.  Risk estimates are then
presented for Los Angeles County with PM  levels adjusted to just attain the current PM]0
standards using base case assumptions.  Finally, risk estimates are presented associated with
attainment of alternative PM2 5 standards.  For each of these cases, the potential impacts of
alternative assumptions and uncertainties inherent in the risk assessment are examined in
sensitivity analyses  of individual key uncertainties and in an integrated uncertainty analysis that
looks at the combined effect of several uncertainties.
1.     Base Case Risk Estimates Associated with "As Is" PM Levels
       The estimated health risks associated with exposure to short- and long-term ambient
particle concentrations in Philadelphia County and Los Angeles County have been estimated
using base case assumptions,  as discussed in Section VI-B, for recent 12 month periods.
Estimates for health risks posed by ambient  particles measured both as PM10 and PM2 5 are
provided.  The risk estimates for PM]0 and PM2 5 should be viewed as providing alternative
estimates of the total health impacts of particles for the health endpoints listed in the Tables.
The risk estimates for the two different measures of PM should not be  summed.   The estimates

-------
                                          VI-26
are for annual health risks from particle concentrations above estimates of annual background
concentrations (8 /*g/m3 PMi0 and 3.5 /zg/m3 PM25 in Philadelphia County, 6 /-ig/m3 PM10 and
2.5 /ig/m3 PM25 for Los Angeles County).
       These risk estimates of effects associated with particles have been restricted to those
endpoints where associations between particles and health endpoint have been demonstrated in
U.S. and Canadian cities  (CD, p. 13-36).   Risk estimates for other health endpoints reported
to be associated with short-term PM10 concentrations,  such as emergency room visits for
asthma (Schwartz et al., 1993), respiratory hospitalization in children (Pope,  1991), school
absences (Ransom and Pope, 1992), symptoms of cough (Schwartz et al., 1994; Ostro et al.,
1991; Pope and Dockery, 1992), and asthma medication usage (Pope et al.,  1991),  or
associated with short-term PM2 5 concentrations, such  as respiratory-related restricted activity
days and work loss days in adults (Ostro and Rothschild, 1989) have not been developed.  Risk
estimates also have not been developed for some health endpoints reported to be associated
with long-term PM concentrations, such as chronic bronchitis in  adults (Abbey et al., 1995a)
and decreased lung function in children  (Raizenne et al., 1996)   In addition, risk estimates
have not been extended to different age groups from those in the original study,  even though
this means often estimating risks for only narrow age groups of children.5
       a.     Philadelphia County
       Base case  risk estimates presented in Table VI-6 suggest that PM  is associated with
between  1.1-1.8% (90% credible intervals (CrI) = 0.8-1.4% to 1.1-2.5%) of total mortality
for short-term exposures and with about 4.6% (CrI 2.8-6.2%) of total mortality  for long-term
exposures in Philadelphia County.  The risk estimates associated  with long-term exposure are
likely to reflect both a component of mortality from short-term exposures as well as mortality
not tightly linked to daily changes in PM concentrations.   Expressed in terms of number of
deaths, the mortality incidence in Philadelphia County estimated  to be associated with PM
    However, for studies of respiratory symptoms in Caucasian children which were restricted to exclude racial
differences for analytical purposes (Schwartz et al., 1994; Pope et al., 1991; Dockery et al., 1989) the resulting
concentration-response relationships were applied to the whole population of children in the pertinent age group
(children 8-12, 0-11, and 10-12 years old, respectively) in the two cities examined for the risk analysis.

-------
                                         VI-27
ranges from 220 deaths (CrI 160-290) associated with short-term exposures to 920 deaths (CrI
580-1260) associated with long-term exposures.
       Base case morbidity risk estimates associated with "as is" PM levels in Philadelphia
county are approximately 2.4%  (CrI 1.5-3.3%) of total respiratory hospital admissions for
individuals over 64 based on a pooled analysis of studies using PM10 as the pollutant indicator.
This compares to an estimated risk of 2.0% (CrI 0.5-3.5%) of total respiratory hospital
admissions for all ages in Philadelphia County based on a single study using PM2 5 as the
pollutant indicator. Risks associated with PM exposure range from 0.7-1.4%(CrI 0.3-1.2 to
0.7-2.1%) of cardiac hospital admissions among individuals over 64 years of age for ischemic
heart disease and congestive heart failure.
       Risks associated with short-term exposures to PM range from 6.8% (CrI  2.4-10.9%) to
20.1 % (CrI  10.3-28.3 %) of the lower respiratory symptoms reported in children 8-12 years in
age, depending on PM indicator and the exact ages and asthma status of the children.  Long-
term exposure to PM over the course of the year was estimated to be associated with a 0.3%
(CrI 0-0.6%) increase in incidence of doctor diagnosed acute bronchitis among 10-12  year
olds.
       b.     Los Angeles County
       Base case risk estimates associated with "as is"  PM levels in Los Angeles  County are
presented in Table VI-7.  The PM10 and PM: 5 annual concentrations are approximately double
the PM concentrations in Philadelphia (annual mean concentration of approximately 52 ptg/m3
PM10 and 30 /*g/m3 PM25 in Los Angeles County versus 25 /xg/m3 PMj0 and 17 ^g/m3 PM25
for Philadelphia).  Risks associated with "as is" particle levels  in Los Angeles County are
estimated to range from 1.6-3.7% (CrI 0.2-3.1% to 0.8-6.3%) of total mortality for short-term
exposure and to be approximately 11.9% (CrI 7.5-16.0%) of total mortality for long-term
exposure.  The estimate of 1.6% of total mortality is based on a study of mortality in Los
Angeles County (Kinney et al., 1995).  This lower estimate of mortality incidence may be due
in part to the fact that this study employed the shortest averaging time (1 day) of those
included in the pooled estimate (CD, p. 12-72).

-------
                                                                                      VI-28
Table VI-6. Estimated Annual Health Risks Associated with "As Is" PM Concentrations
in Philadelphia County, September 1992- August 1993 (for base case assumptions)
Health Effect*'
Mentality (all ages)
Hospital Admissions
Respiratory
Hospital Admissions
Cardiac
Lower Respiratory
Symptoms
in Children""
(A) Associated with short-term exposure
(B) Assoc with long-term exposure
(51 locations)
(C) Total Respiratory
(all ages)
(D) Total respiratory
(>64 years old)

(G) Ischemic Heart Disease "•
(>64 years old)
(E) COPD
(>64 years old)
(F) Pneumonia
('64 years old)

(H) Congestive Heart Failure —
(>64 years old)
(I) Lower Respiratory Symptoms (* of cases)
(8-12 year olds)
(J) Lower Respiratory Symptoms (* of days)
(9-1 1 year ok) asthmatics)
(K) Doctor-diagnosed Acute Bronchitis assoc-
iated with long-term exposure (10- 12 year olds)
Health Effects Associated with PM-10 Above Background"
Incidence
220
(160-290)
	
:::
250
(150-340)
120
(80- 150)
80
(50- 100)
80
(30 - 120)
110
(50- 160)
< 10000 >
(8000 - 1 1000)
< 16000 >
(6000 - 25000)
< 190 >
( 20 - 370 )
Percent of Total Incidence
1 1%
(08-14)
- .. „
::::
24%
(15-33)
37%
(25-47)
19%
(13-26)
08%
(03-13)
1 4%
(07-21)
175%
(153- 196)
68%
(24-10 9)
03%
(00 -06)
Health Effects Associated with PM-2 5 Above Background--
Incidence
370
(220-510)
920
(580- 1260)
260
(70 - 450)
- - ~
:::
:::
70
(30 - 120)
100
(50- 150)
< 11000>
(6000-15000)
	
	
Percent of Total Incidence
1 8%
(11-25)
46%
(28-62)
20%
(05-35)
:::
:::
- - —
07%
(03-12)
13%
(06-20)
20 1%
(10 3 - 26 3)
:::
_ _, _
*  Health effects are associated with short-term exposure to PM. unless otherwise specified
" Health effects incidence was quantified across the range of PM concentrations observed in each study, when possible, but not
     below background level  Background PM-10 Is assumed to be 8 ug/m3. background PM-2 5 is assumed to be 3 5 ug/m3
•" PM-2 5 results based on using PM-2 5 mass as PM-10 mass in the PM-10 functions
""Angle brackets <> indicate incidence calculated using baseline incidence rates reported in studies, with no adjustment for
     location-specific incidence rates  This increases the uncertainty in (he incidence estimates

The numbers in parentheses for pooled functions are NOT standard confidence intervals
All the numbers in parentheses are interpreted as 90% credible intervals based on uncertainty analysis
that takes into account both statistical uncertainty and possible geographic variability
Sources of Concentration-Response (C-R) Functions
(A) PM-10 C-R function based on pooled results from
   studies in 10 locations, PM-2 5 C-R function based on pooled
   results from studies in six locations
(B) Pope et al. 1995
(C)Thurston, etal, 1994
(O) PM-10 C-R based on pooled results from 4 functions
(E) PM-10 C-R based on pooled results from 4 functions
(F) PM-10 C-R based on pooled results from 4 functions
(G) Schwartz & Motrls. 1995
(H) Schwartz & Moms, 1995
(I)Schwartz, etal. 1994
(J) Pope etal. 1991
(K)Dockeryetal, 1989

-------
                                         VI-29
       The estimated mortality risks in Los Angeles County based on the pooled, short-term
mortality functions and the long-term mortality functions expressed in either percentage terms
or as number of deaths are roughly two to three times the risks estimated applying the same
functions in Philadelphia County.  The population of the Los Angeles County area used in the
analysis is more than twice as large as Philadelphia  County (3.6 million versus 1.6 million),
however, the death rate is half of that observed in Philadelphia (667 versus 1280 per 100,000).
The differences in population size and death rate between the two study areas  are largely off-
setting in terms of the risk calculations, but Los Angeles County PM annual levels are nearly
double those observed in Philadelphia county.  Thus, the differences in risk estimates between
the two study areas appears to be largely due to differences in PM levels.
       With respect to morbidity health endpoints, short-term exposures to PM concentrations
in Los Angeles County are estimated to be associated with approximately 6.9% (CrI 4.2-
9.4%) to 7.7% (CrI 2.1-13.4%) of total respiratory hospital admissions (all ages and
individuals over 64, respectively).  PM also is estimated to be associated with between 1.4%
(CrI 0.6-2.3%) to 4.1%(CrI 2.0-6.1%) of cardiac hospital admissions among  individuals over
64 years of age for ischemic heart disease and congestive heart failure.
       Short-term exposure to PM in Los Angeles County is estimated to be associated with
between 18.4% (CrI 6.9-28.0%) and 41.4% (CrI 37.2-45.2) of the lower respiratory
symptoms reported in children 8-12 years in age, depending on PM  indicator and the ages,
races, and asthma status of the children.  These incidences seem high, and EPA staff notes that
questions can be raised about the transferability of concentration-response functions derived in
eastern U.S. locations to Los Angeles.  Therefore, risk estimates based on a recent study of
asthmatic symptoms among African-American children in central  Los Angeles are provided for
comparison (Ostro et al., 1995).  Estimates based on this study indicate that daily variations in
PM concentrations are associated with 19.3% (CrI 6.4-29.2%) of the reported incidence of
shortness of breath, which is similar to that derived  from the other studies.  Long-term
exposure to PM over the course of the year is estimated to be associated with  a 3.1% increase
(CrI 0.4-4.7%) in incidence of doctor diagnosed acute bronchitis among 10-12 year olds.

-------
                                                                                         VI-30
Table VI-7.  Estimated Annual Health  Risks Associated with "As Is" PM Concentrations
in Southeast  Los Angeles County, 1995* (for base case assumptions)

Mortality (all ages)
Hospital Admissions
Respiratory
Hospital Admissions
Cardiac
Lower Respiratory
Symptoms
in Children 	
Health Effects'*
(A) Associated with
short-term exposure
(B) Associated with short-term exposure
(study done In Los Angeles)
(C ) Associated with long-term exposure
(51 locations)
(D) Total Respiratory
(all ages)
(E) Total Respiratory
(>64 years old)
(F) COPD
(>64 years old)
(G) Pneumonia
(>64 years old)
(H) Ischeme Heart Disease""
(>64 years old)
(I) Congestive Heart Failure""
(>64 years old)
[J) Lower Respiratory Symptoms (* of cases)
(8- 12 year olds)
(K) Lower Respiratory Symptoms (» of days)
(9-1 1 year old asthmatics)
(L) Days of shortness of breath (7- 1 2 year old
African American asthmatics In Los Angeles)
[L) Doctor-diagnosed Acute Bronchitis assoc-
iated wKh long-term exposure (10-12 year olds)
Health Effects Associated with PM-10 Above Background1"
Incidence
800
(570-1020)
400
(40 - 750)
:::
—
1,070
(660-1460)
440
(310-560)
420
(290-550)
260
(100-420)
290
(140-430)
< 62000 >
(56000 - 68000)
< 115000 >
(43000 - 175000)
<7200>
(2400 - 10900)
<5090>
(680 - 7750)
Percent of Total Incidence
33%
(23-41)
1 6%
(02-31)
:::
- —
69%
(42-94)
103%
(73-131)
56%
(39-73)
23%
(09-37)
4 1%
(20-61)
41 4%
(37 2 - 45 2)
184%
(6 9 - 28 0)
193%
(6 4 - 29 2)
3 1%
(04-47)
Health Effects Associated with PM-2 5 Above Background*"
Incidence
900
(200-1560)
:::
2.920
(1850-3930)
1,200
(330 - 2080)
:::
	
:::
160
(60 - 260)
180
(90 - 270)
< 51000 >
(28000 - 68000)
	
:::
	 	
Percent of Total Incidence
37%
(08-63)
	
119%
(7 5 - 16 0)
77%
(21-13 4)
	
:::
:::
1 4%
(06-23)
25%
(12-38)
344%
(19 1 - 45 7)
:::
:::
	
   ' Southeast Los Angeles County was not in attainment of current PM-10 standards (50 ug/m3 annual average
      standard and 150 up/m3 daily standard) in 1995 Figures shown use the actual reported concentrations
  ** Health effects are associated with short-term exposure to PM, unless otherwise specified
 "' Health effects incidence was quantified across the range of PM concentrations observed in each study, when possible, but not
      below background level Background PM-10 is assumed to be 6 0 ug/m3 and background PM-2 5 is assumed to be 2 5 ug/m3
 "" PM-2 5 results based on using PM-2 5 mass as PM-10 mass in the PM-10 functions
	Angle brackets <> indicate incidence calculated using baseline incidence rates reported in studies, with no adjustment for
      location-specific incidence rates  This increases the uncertainty in the incidence estimates
The numbers in parentheses for pooled functions are NOT standard confidence intervals.
All numbers in parentheses are interpreted as 90% credible intervals based on uncertainty
analysis that takes into account both statistical uncertainty and possible geographic variability.
Sources of Concentration-Response (C-R) Functions
(A) PM-10 C-R function based on pooled results from
   studies in 10 locations, PM-2 5 C-R function based on pooled
   results from studies in six locations
(B)Kinneyetal,199S
(C) Pope etal. 1995
(D)Thurston. etal, 1994
(E) PM-10 C-R based on pooled results from 4 functions
(F) PM-10 C-R based on pooled results from 4 (unctions
(G) PM-10 C-R based on pooled results from 4 functions
(H) Schwartz & Morris, 1995
(I) Schwartz & Morris, 1995
(J) Schwartz, etal, 1994
(K) Pope etal, 1991
(L)Dockeryetal, 1989

-------
                                         VI-31
       C.     Key Uncertainties
       There are additional uncertainties about the risk estimates for both locations beyond
those reflected in the credible intervals.  These additional uncertainties include but are not
limited to the degree of transferability of concentration-response functions and measurement
error in air quality values for each location.  Because national or community gathering of
respiratory symptoms information is not routinely performed, the numbers of days or cases of
symptoms is estimated by applying the percentage of incidence associated with PM to the
baseline incidence rates reported in the health studies, which are from locations different than
those being analyzed, with the exception of the Ostro et al. (1995) study. Baseline incidence
may be considerably different from that observed in the cities analyzed,  resulting in additional
uncertainty pertaining to the numerical estimates of incidence reported in Tables VI-6 and VI-
7. The estimates of percent incidence are less uncertain than the estimates of incidence counts
for respiratory symptoms risk estimates in both Philadelphia  and  Los Angeles.
2.     Base Case Risk  Estimates Upon Attainment of Current Standards
       For comparisons with alternative standards it is desirable to estimate health risks
associated with PM air quality that does  not include  the effects of concentrations in excess of
those allowed by the current national PM standards.  For Philadelphia county, Table VI-6 also
represents the estimated health risks associated with  PM at or below the  current PM10
standards, since the monitors used in estimating Philadelphia's air quality are already in
attainment of the current PM10 standards. For Los Angeles County, however, the estimates
given in Table VI-7 include contributions from concentrations in excess  of those allowed by
the current PM10 standards.  The PM10 concentrations for the monitors used in the risk analysis
in Los Angeles County have an annual mean controlling value of 52 /zg/m3 and a 2nd-daily
max controlling value of 195 p.g/m\ versus the current PMJO standards of 50 /zg/m3 annual
mean and 150 /ig/m3, 24-hr average.  Adjusting PM air quality for Los Angeles County to
simulate attainment of the current PM10 standards introduces additional uncertainty into the risk
estimates, but is required in order to compare risks associated with attaining the current PM,0
standards with risks associated with meeting alternative PM: 5 standards.

-------
                                          VI-32
       The method chosen to simulate attainment of the current PM10 standards is to apply a
proportional rollback to both PM10 and PM2 5 concentrations (preserving the PM2 5/PM10 ratio)
to air quality concentrations that "just attain" current standards (under current interpretation,
this means reducing annual mean concentrations to 50.4 jug/m3, and the second daily max
concentration6 to 154 /zg/m3, to reflect rounding conventions used to judge attainment). This
modeling of attainment in Los Angeles County through proportional rollback contains two
analytic assumptions.  First, it assumes that the general shape of the distribution of PM air
quality concentrations  in Los Angeles County will remain the same as observed under the "as
is" situation and that PM levels will be reduced proportionately based on the controlling
standard. For Los Angeles County the 24-hr second daily max concentration of 195 /*g/m3 is
the controlling value and needs to be reduced 21 % to bring it into attaintment.  Thus, the
amount of each PM concentration above estimated background for the 1995 year in Los
Angeles  County was reduced by 21%. The second assumption is that the relationship between
PM;,5 and PM10 (PM25/PM10 ratio = 0.58) would be preserved as PM,0 concentrations are
reduced.  If control strategies are used to reach attainment that preferentially controls coarse
particles relative to fine particles (as has been observed in some areas, see Chapter IV), or that
preferentially controls  fine particles relative to coarse particles, this simplifying assumption
introduces some inaccuracy. If the error is in the direction of not adequately reflecting a
preferential control of  coarse particles,  then PM2 5 concentrations in the "just attain PM10
standards case" would  be expected to be higher than those estimated in this analysis. In this
case, larger reductions in PM health risks would be expected than those reported later in the
alternative standards risk analysis.
       The results for  Los Angeles County based on simulating attainment of the current PM10
standards are shown in Table VI-8. The reduction in PM concentrations results in an
approximately 18-28% reduction in the risk estimates associated with short-term PM exposures
compared to "as is" levels. This provides an example of how the estimated change in health
     The current 24-hr standards are applied to the 4th highest daily concentration in a three year period.  Since we
are only examining a year of air quality concentrations in the risk analysis, the second daily max concentration was
chosen as an approximate surrogate for the 4th highest concentration in three years value.

-------
                                                                                   VI-33
 Table VI-8.  Estimated Annual Health Risks Associated with Attainment of Current Standards
 in Southeast Los Angeles County,  1995* (for base  case assumptions)
Health Effects"
Mortality (all ages)
-tospital Admissions
Respiratory
Hospital Admissions
Cardiac
Lower Respiratory
Symptoms
in Children 	
(A) Associated with
short-term exposure
(B) Associated with short-term exposure
(study done in Los Angeles)
(C) Associated with long-term exposure
(51 locations)
(D) Total Respiratory
(all ages)
(E) Total Respiratory
(>64 years old)
(F)COPD
(>64 years old)
(G) Pneumonia
(>64 years old)
(H) Ischemlc Heart Disease""
(>64 years old)
(I) Congestive Heart Failure""
(>64 years old)
(J) Lower Respiratory Symptoms (0 of cases)
(8- 12 year olds)
(K) Lower Respiratory Symptoms (* of days)
(9-1 1 year old asthmatics)
(L) Days of shortness of breath (7-12 year old
African American asthmatics in Los Angeles)
(L) Doctor-diagnosed Acute Bronchitis assoc-
iated with long-term exposure (10-12 year olds)
Health Effects Associated with PM-10 Above Background'"
Incidence
630
(450- 800)
290
(30 - 550)


	
840
(520- 1160)
350
(240- 440)
330
(230- 430)
200
(BO - 330)
230
(110- 340)
< 52000 >
(46000-57000)
< 93000 >
(34000- 143000)
< 5200 >
(1700- 8100)
< 3760 >
(470-6190)
Percent of Total Incidence
26%
(18-33)
1 2%
(01-22)
	


54%
(33-74)
8 2%
(58- 105)
4 4%
(31-58)
1 8%
(07-29)
32%
(15-48)
348%
(31 0-384)
14 9%
(55-230)
14 1%
(46-21 8)
23%
(03-37)
Hearth Effects Associated with PM-2 5 Above Background"*
Incidence
710
(430-970)
	
2,110
(1330- 2860)
940
(250- 1630)
	
:: :: ::

130
(50 - 200)
140
(70-210)
< 43000 >
(23000 - 58000)
	
	

	
Percent of Total Incidence
29%
(17-39)
	
86%
(54-117)
6 1%
(16- 105)
:: :: ::
:: :: ::
	
1 1%
(04-18)
20%
(10-30)
28 7%
(154-390)
	
:: :: ::
	
   • Southeast Los Angeles County was not in attainment of current PM-10 standards (50 ug/m3 annual average
      standard and 150 ug/m3 daily standard) in 1995  "As is" daily PM-10 concentrations were first rolled
      back to simulate attainment of these standards  "As is" daily PM-2 5 concentrations were rolled back
      by the same percent as daily PM-10 concentrations  See text in Chapter VI for details
  " Health effects are associated with short-term exposure  to PM, unless otherwise specified
 '" Health effects incidence was quantified across the range of PM concentrations observed in each study, when possible, but not
      below  background level Background PM-10 Is assumed to be 6 0 ug/m3 and background PM-2 5 is assumed to be 2 5 ug/m3
 "" PM-2 5 results based on using PM-2 5 mass as PM-10 mass  in the PM-10 functions
	Angle brackets <> indicate incidence calculated using baseline incidence rates reported in studies, with no ad|ustment for
      location-specific incidence rates  This increases the uncertainty in the incidence estimates

The numbers in parentheses for pooled functions are NOT standard confidence intervals
All numbers in parentheses are  interpreted as  90%  credible intervals based on uncertainty
analysis that takes into account both statistical uncertainty and possible geographic variability
See text in  Chapter VI for details
Sources of Concentration-Response (C-R) Functions
(A) PM-10 C-R function based on pooled results from
   studies In 10 locations, PM-2 5 C-R function based on pooled
   results from studies in srx locations
(B) Kmney etal.1995
(C)Popeetal, 1995
(D)Thurston, etal, 1994
(E) PM-10 C-R based on pooled results from 4 functions
(F) PM-10 C-R based on pooled results from 4 functions
(G) PM-10 C-R based on pooled results from 4 functions
(H) Schwartl & Morris, 1995
(I) Schwartz & Morris. 1995
(J) Schwartz, et al. 1994
(K) Pope etal, 1991
(L) Dockeryetal, 1989

-------
                                         VI-34
risks associated with PM is approximately equal to the amount of proportional air quality
reduction required (for Los Angeles County, a reduction of 21% in air quality concentrations
results in a 18-28% reduction in health risks associated with short-term exposures). This
correspondence results from the shape of the concentration-response relationships reported in
the literature and in the base case analysis, which are essentially linear over most of the range
of concentrations considered here.   For risks associated with long-term exposures, the
reduction is greater than the relative change in PM levels because estimated health risks
associated with long-term exposures are quantified relative to lowest observed annual mean
concentrations in the health studies used in the risk analysis which are considerably in excess
of background.
       Although there are substantial uncertainties in predicting annual health risks associated
with attainment of the current standards in Los Angeles County, the estimates in Table VI-8
suggest that short-term exposure to PM could be associated with approximately 1.2% (Cri 0.1-
2.2%) to 2.9% (Cri 1.7-3.9%) of mortality, 5.4% (Cri 3.3-7.4%) of respiratory hospital
admissions for those over 65, 1.1% (Cri 0.4-1.8%) to 3.2% (Cri 1.5-4.8%)  of cardiac
hospital admissions for ischemic heart disease and congestive heart failure, and from 14.9%
(Cri 5.5-23.0%) to 34.8 (Cri 31.0-38.4%) of respiratory symptoms in children upon
attainment of the current PM]0 standards.  Estimated mortality associated with long-term
exposure is about 8.6% (Cri 5.4-11.7%) and doctor-diagnosed acute bronchitis associated with
long-term exposure is about 2.3% (Cri 0.3-3.7%) upon attainment of the current NAAQS.
However, in considering  such estimates it is important to consider the substantial uncertainties
that may affect these estimates.  The next section summarizes the results of several sensitivity
analyses to provide some insight into the magnitude of the uncertainties associated with the PM
risk estimates. Additional uncertainties, not captured by the sensitivity analyses, were
discussed previously in Section VLB and VI.C.I.e.

-------
                                          VI-35
3.     Uncertainty Analyses of Estimated Risks Associated with "As Is" PM Levels in
       Philadelphia County and Attaining Current PM10 Standards in Los Angeles County
       a.      Sensitivity Analyses of Individual Key Uncertainties
       A number of sensitivity analyses of the health risk model have been conducted to
provide some perspective on the impact of various uncertainties and assumptions on the health
risk estimates presented in this Staff Paper.  These sensitivity analyses are presented in
Appendix F and in the technical support document (Abt Associates, 1996b).  Table VI-9
summarizes the results of a number of these sensitivity analysis indicating the effects of
alternative specifications for several important air quality and concentration-response
parameters (background, cutpoint concentrations,  averaging time for mortality functions,  and
the effects of reduced slopes for long-term mortality functions resulting from the potential
effects of inadequately considered confounders or previous air quality).  The results are
presented as  a range of estimates of the percent of mortality and respiratory hospital
admissions incidence associated with PM under "as  is" air quality in Philadelphia County.
       From Table VI-9 it can be seen that the estimates of health risks show particular
sensitivity to assumptions concerning the use of appropriate cutpoint concentrations for
quantifying risk.7   The cutpoints used in the analysis can be used to inform judgments
concerning the potential  effects of nonlinear concentration-response relationships resulting
from potential biological considerations, copollutant effects, or exposure misclassification
associated with the use of ambient monitors as a measure of population exposures.
       Disaggregating the pooled PM10 mortality  analysis into subsets of studies with effects
estimates based on more homogenous averaging times also can make substantial differences in
the estimates of PM10 mortality health risk; for example, when studies with the shortest (1-day)
and longest (3-5 day) averaging times are contrasted.  As would be expected, assuming lower
than reported coefficients for long-term mortality risk from PM exposures reduces  risk
    To quantify risks above various cutpoints, two alternative slope adjustment methods have been used to examine
the potential impact of a concentration-response function having a steeper slope (i.e., larger RRs per /xg) above
specified cutpoints. See Figure VI-6 and discussion in Appendix F for further details.

-------
                                          VI-36
estimates by an amount equal to the reduction in the coefficient. The estimates of health risks
associated with PM also show some degree of sensitivity to alternative specifications of
background concentrations.
       One important uncertainty that is not included in Table VI-9 concerns the effect of
copollutants on the estimated risks associated with PM. The base case estimates risk resulting
from concentration-response relationships developed without inclusion of copollutants. Since
not all of the studies included in the base case analysis controlled for copollutants by
simultaneously incorporating them in the analysis, it is not possible to directly estimate the
sensitivity of the base case results by taking into account the effect of simultaneous inclusion of
all copollutants in all studies.  However, an examination of the sensitivity of risk estimates
from individual studies that did include  copollutants is provided in Appendix F,  Table F-5b.
The results for most, but not all, of the  studies are consistent with the assessment in the CD
that the magnitude of PM effects and their statistical uncertainty in many studies showed little
sensitivity to the adjustment for copollutants (CD, p. 13-55).  As discussed in Section V.E.,
however, reanalyses of Philadelphia using TSP data by the HEI (Samet et al.,  1996a) and
Mooglavkar et al.  (1995a,b) have reported a potential for more significant interaction by
copollutants when multiple pollutants are entered into the concentration-response model.  The
implications of the perspective that PM  may be serving as an index reflecting the effects  of
several pollutants in combination is discussed below in section VI.C.4 and is an  area of
uncertainty that needs to be investigated further.
       Similar sensitivity  analyses to the ones summarized above for Philadelphia County were
performed for Los Angeles County.  A primary point of interest is that the Los Angeles
County risk estimates show less sensitivity to the choice of cutpoint than  the Philadelphia
County results, since a larger proportion of days in  Los Angeles County have PM
concentrations above some or all of the  cutpoints analyzed (see exhibits 7.17 - 7.20 in Abt
Associates, 1996b).

-------
Table VI-9.  Summary of Selected Sensitivity Analyses on Estimates of Risk Associated with PM in Philadelphia County
HEALTH
ENDPOINT
MORTALITY
Short-Term
Exposure
MORTALITY
Long-Term
Exposure
HOSPITAL
ADMISSIONS
Total
Respiratory7
PM
Indicator
PM10
PM25
PM25
PM.o
PM25
BASE
CASE
Central
Estimate
1.1%
1.8%
4.6%
2.4%
2.0%
SENSITIVITY ANALYSES
Central Estimates
BACKGROUND1
(Low-High
Concentration)
1.3-09%
2.0- 1.6%
No change5
29-1 .9%
2.3-1 8%
CUTPOINT2
Method I
(Low-High)
0.4-0.1%
1.1 -0.1%
CUTPOINT2
Method II
(Low- High)
0.4-0.1%
1.0-0.1.%
2.4 - 0%6
1 3 - 0.4%
14-0 4%
1.0-0.2%
1.2-0.2%
AVG TIME3
(5 day-1 day)
1.8-0.4%
—
...
—
—
SLOPE
REDUCTION4
Long-Term Study
—
—
3.4 - 2.3%
—
—
1  Low = 5 /xg/m1 PM,0, 2 /ig/m3 PM2,; High =  11 uglm PM,0, 5 ftg/m1 PM2,; Base Case = 8 /ig/ m1 PM,0, 3.5

2  Low = 20 ^g/nVPM,,,, 10 /ig/m1 PM2,; High = 40 ug/m1 PM10, 30 ^g/m1 PM2,;  Base Case = linear relationship above background. Method I and Method II
refer to methods of adjusting the slope of the concentration-response relationship above the cutpoint upwards to different extents to reflect the anticipated effect of a
"hockey stick"-style threshold concentration response function. See Appendix F for further details..
1  5 day = results using 3-5 day averaging time studies; 1 day  = result using single day averaging time study;  Base Case used 2 day averaging time.
4  First number represents effect of 33% reduction in slope; second number represents effect of 50%  reduction in slope; Base Case used relative risk as reported
in study (i.e., no adjustment). Slope Reduction intended to  roughly model potential effects of previous air quality or uncontrolled confounding.
5  Background concentration sensitivity analyses make no difference in the risk estimates for mortality from long-term exposure since the lowest observed
concentrations in this studies (the limit to which the concentration-response function was applied) was well above background.
6 Low =  12.5 /ig/m' PM2,; High = 18 /xg/m1 PM2,;  Base  Case = linear relationship above the lowest observed concentration in study (9 ^g/m3). No slope
adjustment was made to the long-term mortality concentration-response relationship when applying the cutpoints.

7 Total Respiratory Hospital Admissions for those > 64 yrs of age for PMin; for all ages for PM2,
                                                                       VI-37

-------
                                           VI-38
       In general, these sensitivity analyses indicate that alternative analytic choices within the
range of those considered in this analysis may lead to sizable differences in risk estimates.
However, these are also primarily intended as bounding exercises to characterize the
magnitude of potential uncertainty, and as such do not reflect judgments concerning the
likelihood of specific alternative cases tested.
       b.     Integrated Uncertainty Analysis
       In addition to individual sensitivity analyses discussed above, an integrated uncertainty
analysis has been conducted for mortality associated with short-term exposures to PM2 5 to
assess the potential combined  effects of several  key uncertainties simultaneously.  Through
Monte Carlo sampling approaches, a distribution of values for several key parameters in the
model has been estimated or specified, and 90 percent credible intervals have been generated
representing the probability that the risk estimates fall within a particular range once the
combined effect of these uncertainties  have been considered.  An advantage of this approach is
that it allows the combined effect of several uncertainties to be quantitatively estimated.  A
major difficulty of the approach, however, is that the method inherently requires  an estimate
of the distribution of values for each uncertainty included, even if little  empirical evidence is
available to inform what is an appropriate choice for each distribution.  Since there is little
information on which to base  some of the distributions and/or  weightings chosen  to represent
certain key parameters in the integrated uncertainty analyses, the results of this analysis should
be viewed as illustrative in character.  The purpose of the  analysis is to  show the  potential
sensitivity of the risk estimates when several uncertainties, rather than just a single
uncertainty, are considered simultaneously.
       As discussed earlier in this Chapter, there are a number of uncertainties encountered as
one attempts  to estimate health risks associated with PM levels for a given city or location.
Given the availability of specific data for baseline health effects incidence and daily PM air
quality data for the two locations examined (i.e., Philadelphia and Los Angeles Counties),
staff judges that the uncertainties associated with these two inputs to the risk model are
relatively small compared to the uncertainties associated with what is the appropriate
concentration-response function for these locations. Therefore, the integrated uncertainty

-------
                                          VI-39
analysis is primarily focused on the concentration-response uncertainties, since this is judged
to be the largest source of uncertainty in the health risk model. In addition, uncertainty about
background levels and uncertainty about how PM air quality distributions might change upon
attainment of alternative standards also is included in the analysis.
       Table VI-10 below summarizes how each of the uncertainties incorporated into the
integrated uncertainty analysis is treated.  As outlined in Appendix E, there is substantial
uncertainty concerning whether cutpoint concentrations above background exist based on  a
review of the available data.  As discussed previously in this Chapter and in Appendix E,
various approaches have been used to derive cutpoints  of interest from the available data. The
current data does not provide strong evidence concerning where a cutpoint concentration might
exist (CD).  To account for this state of uncertainty, the integrated uncertainty analysis use
several illustrative weightings to assess the possible effects of this important uncertainty in
combination with other key uncertainties (i.e., estimated background  levels, air quality
rollback approach).  Each of the key uncertainties were incorporated sequentially into the
analysis to illustrate the impact of each uncertainty on the risk estimates.
       Figure VI-7 displays the results of the integrated uncertainty analysis for mortality
associated with short-term exposure to PM2 5 for Philadelphia County  under the "as is"
scenario.  The risk estimates are expressed in terms of both number of deaths  over a 1-year
period and as a percent of total mortality. Each vertical bar represents a set of risk estimates
that  includes the uncertainties identified below the bars.  The mean estimate is given, as well
as the 5th, 25th, 75th, and 95th percentiles. The first vertical bar includes only uncertainty in
the RR and assumes that background equals 3.5  /Ltg/m3.  The second vertical bar incorporates
uncertainty in RR and in the PM2 5 background concentration for Philadelphia, with the
cutpoint set  equal to the background concentration.  The final  three vertical bars incorporate
uncertainties about RR, background, and three weighting schemes differentially weighting the
likelihood that various cutpoint (or threshold) concentrations exist.  The three weighting
schemes are indicated in the box below Figure VI-7. Case I represents a judgment that
concentration-response functions are more likely to exist down to background or  10 /ig/m3;
Case III represents a judgment that concentration-response functions are more likely to  have a

-------
                                        VI-40
        Table VI-10. Summary of Uncertainties Incorporated Into Integrated
                                 Uncertainty Analysis
               Uncertainty
               Distribution
Coefficient (P) in concentration-response
function
Based on distribution of p's obtained from
pooled results of PM2 5 mortality studies in
six locations
Cutpoints in concentration-response function
Four cutpoints (background, 10, 18, 30
/ig/m3) with three discrete weighting
schemes and two slope adjustment methods
Background PM2 5 concentration
Uniform distribution on the intervals [2,5]
and [1,4] (jug/m3) for Philadelphia County
and Los Angeles County, respectively, based
on the estimated ranges identified in the CD
for the Eastern and Western sections of the
United States
Shape of PM2 5 air quality distribution upon
attainment of alternative standards
Based on distribution of regression slope of
linear rollback over background to ratio of
second high 24-hr PM2 5 values for 129 pairs
of site-years of data (see Section 8.2 in Abt
Associates (1996b))

-------
                                VI-41
         Figure VI-7.  Effect of Several Uncertainties on
     Mortality Risk Associated With Short-Term Exposure
                 to PM-2.5 in Philadelphia County

                     September 1992 - August 1993
                       (Population:  1.6 Million)
30V,
= 25*
I
"s
£ 20%
e
e
fc
91 1 .371
1
i
,< 1.0%

05%
0.0%

^
C

^


i
•
95th % lie
75th % lie
) Mean (

, 25th % lie

5th % lie
i i
i
i

)



C



1
1





)

c
<
1
•
•



i


)
t
C
(
4
• 1


•

j '
Uncertainty in Uncertamn in Case I Case II Case in
Just RR RR and Background |
(Background (cutpoint = background)
= 3 5 ug m')

Uncertamn in RR, Background,
DUU
500
Mortality
400 Risk "
Associated
with PM-2.5:
.^ Number of
Deaths

200

100
0
J
                                     and Cutpomt Weightings

Uncertainty in background concentration enters into these calculations only when
the cutpoint is set equal to background. The other cutpomts are greater than the
highest background concentration considered.

          Cutpoint Weighting Schemes

Background
10(jg/m3
18pg/m3
30ng/m3
Case I
05
0.3
0.15
0.05
Case II
0.2
0.3
0.3
0.2
Case III
0.05
0.15
0.5
0.3

-------
                                           VI-42
cutpoint at 18 or 30 /ig/m3; and Case II represents a judgment that concentration-response
functions are somewhat more likely to have cutpoints in the 10-18 /ig/m3 range.8  Figure VI-8
shows a similar figure for Los Angeles County where attainment of the current PM10 standards
is simulated.
       The results of the integrated uncertainty analysis illustrate the impact on the mortality
risk estimates of whether or not one judges there to be a likely cutpoint or threshold above
estimated background levels.  If one assumes no cutpoint above background, mortality
associated with short-term exposure in Philadelphia County under the "as  is" scenario is
estimated to be about 1.8 (CrI 1.2-2.7) percent of total mortality or 375 (CrI 225-525) excess
deaths.  Allowing for the possibility of a cutpoint above estimated background levels, three
alternative cutpoint weighting schemes reduce the mean risk estimates to about 1.3, 0.8, and
"0.5 percent of total mortality for Cases I, II, and III, respectively. For Cases I and II the 90
percent credible intervals also become considerably wider than the risk estimates incorporating
only uncertainty  in the  RR slope and estimated background concentration and all three cutpoint
weighting schemes indicate a lower bound of the 90 percent credible interval of about 0.2-0.3
percent of total mortality.  For Los Angeles County under the just attaining the current PMi0
standards, the mean mortality risk estimates assuming no cutpoint is about 2.8 percent (CrI
1.7-3.8). The alternative cutpoint weighting schemes reduce the mean mortality risk estimates
to about 2.2, 1.6, and 1.2 percent for Cases I, II, and III, respectively. The higher risk
estimates in Los  Angeles County are due mainly to the higher PM2 5 levels,  since Philadelphia
County air quality is lower (i.e., better) than the current PM10 standards.
    In the sensitivity analysis described previously in the Chapter two different methods for adjusting the slope of
the concentration-response function were examined when various cutpoints (or thresholds) were analyzed. In the
integrated uncertainty analysis, the two slope adjustment methods were given equal weight.

-------
                                         VI-43
Figure VI-8.  Effect of Several Uncertainties on Mortality Risk Associated
       With Short-Term Exposure to PM-2.5 After Meeting Current
                   PM-10  Standards in Los Angeles County

                               (Population: 3.6 Million)
£.3.5%

1
S 3.0%
"3


f 2.5%

f

t 2.0%

W
fe.1.5%


  1.0%


  0.5%


  00%
            T 95th % ile
              75th % lie
            ( )Mean
            ' ' 25th % ile
            * 5th % lie
          Uncertainty in
            Just RR
          (Background
          = 25 ug,™1)
                                                                             1000
                                                                             900
                                                                             800
                                                                             700
                                     Mortality
                                     Risk
                                     Associated
                                     with
                                 600  PM2.5:
                                     Number of
                                 500  Deaths

                                 400

                                 300

                                 200

                                 100
                              Uncertainty in
                             RR and Background
                             (cutpoint = background)
Case I
 I	
                                                                     Casein
                                                  Uncertainty in RR, Background,
                                                  and Cutpomt Weightings
     Uncertainty in background concentration enters into these calculations only when
     the cutpoint is set equal to background. The other outpoints are greater than the
     highest background concentration considered.

                     Cutpoint Weighting Schemes

Background
10fjg/m3
18>ig/m3
30Mg/m3
Case I
0.5
0.3
0.15
0.05
Case II
0.2
0.3
0.3
0.2
Caselll
0.05
0.15
0.5
0.3

-------
                                           VI-44
4.     Risk Estimates Associated with Alternative PM2 5 Standards
       This section presents risk estimates associated with just attaining several alternative
PM2 5 standards for the Philadelphia and Los Angeles County study areas.  In addition to risk
estimates using base case assumptions, individual sensitivity analyses and integrated
uncertainty analyses also are presented, analogous to the approach used for the "as is" risk
estimates. The additional uncertainty introduced primarily by adjusting air quality to reflect
future attainment of alternative standards also is discussed.
       a.     Base Case Risk Fstimates
       Table Vl-lla summarizes the air quality information indicating which monitor in each
location has the "controlling value" for a rollback to attain 24-hr or annual mean alternative
standards.9 Table Vl-llb shows the amount of reduction in air quality required to attain the
alternative PM2 5 standard, and which standard of the combination, daily or annual, is
"controlling" (i.e.,  requires the larger reduction in concentration).  To model attainment of
alternative PM2 5 standards, a proportional rollback approach is used as the base case.
Although it is extremely difficult to predict what patterns of air quality would be observed in
these two locations upon attaining alternative PM2 5 standards, a preliminary investigation of
changes in PM2 5 air quality observed  over the past 15 years of limited monitoring reported to
the AIRS database  finds that the  general pattern of air quality changes observed is a
proportional change in both daily and annual mean concentrations  (Abt Associates,  1996b).
The estimated effects of alternative assumptions concerning patterns  of air quality rollback are
presented in Table  VI-14.
       Tables VI-12a and VI-12b show the risk estimates for just attaining alternative PM25
standards in Philadelphia County, and Tables VI-13a and VI-13b show the risk estimates for
just attaining alternative PM2 5 standards in Los Angeles County using base case assumptions.
Similar to the approach used to model attainment of the current PM10 standards in Los Angeles
    Q
     The terminology of "controlling value" and "controlling monitor" are used here as synonyms for the well-known
terms "design value" and "design value monitors". The monitors used in the risk analysis are not genuine design
value monitors established for particular air sheds, and thus the alternative terminology is used to avoid confusion.

-------
                                              VI-45
     Table VI-1 la. Controlling Monitors for Rollbacks to Attain Alternative PM-2.5
                                           Standards
Monitor Site
Weighted Annual
Average PM2 5
Concentration*
Second Daily
Maximum 24-Hour
PM25
Concentration*
Controlling Monitor
Philadelphia County
N/E
PBY
TEM
16
17
17
65
72
70

For daily standard
For annual standard
Southeast Los Angeles County
Central LA
Diamond Bar
24
22
91
102
For annual standard
For daily standard
All concentrations are given in jig/m3 .
*Both weighted annual averages and second daily maximum concentrations at the two monitors in Southeast Los
Angeles County were adjusted to reflect attainment of the current PM10 annual standard of 50 /ig/m3 and the current
PM10 daily standard of 150 /ig/m3.  These standards are currently attained in Philadelphia County.
     Table Vl-llb. Controlling Standards and Percent Rollbacks Necessary to Attain
                                  Alternative PM2S Standards
Alternative PM-2.5
Standards
Annual Avg.
Standard
20 alone
20
20
20
15 alone
15
15
15
24-Hour
Standard

65
50
25

65
50
25
Philadelphia County
Controlling Standard and
Percent Rollback*
—
Daily - 10.4%
Daily ~ 32.3%
Daily -68.7%
Annual - 15.5%
Annual - 15.5%
Daily -32.3%
Daily -68.7%
Southeast Los Angeles County
Controlling Standard and
Percent Rollback**
Annual - 18.8%
Daily - 37.0%
Daily -52.1%
Daily -77.3%
Annual - 42.0%
Annual - 42.0%
Daily -52.1%
Daily -77. 3%
All concentrations are given in /tg/m3 .
*Based on controlling values for Philadelphia County of 17 /tg/m3 for the annual standard and 72 /ig/m3 for the daily
standard.
** Based on controlling values for Southeast Los Angeles County of 24 jig/m3 for the annual standard and 102 /ig/m3
for the daily standard.

-------
                                                                                            VI-46
  Table Vl-12a.  Estimated Changes in  Health Risks Associated with Meeting Alternative PM-2.5 Standards
  in Philadelphia County, September 1992 - August 1993  (for base  case assumptions)

Mortality (all ages)
Hospital Admissions
Respiratory
Hospital Admissions
Cardiac
Health Effects*
(A) Associated with short-term exposure
Percent Reduction In PM-Assoclated Inckttnce:***
Percent Redaction in Total Incidence:""
(B) Associated with long-term exposure
Percent Reduction In PM-Attociated Incidence:
Percent Redaction In Total Incidence:
(C) Total Respiratory
(all ages)
Percent Reduction In PM-A**ocMed Incidence:
Percent Reduction In Total Incidence:
(0) Ischemic Heart Disease""*
(>64 years old)
(E) Congestive Heart Failure"*"
(>64 years old)
Range of Percent Reductions In PM-Assoclated Incidence:
Range of Percent Reductions In Total Incidence:
[F) Lower Respiratory Symptoms (8-1 2 yr olds) """

PM-2 5-Associated
Incidence
associated with
current standards**
370
(220 -510)

920
(580 -1260)

260
(70 -450)

70
(30 -120)
100
(50 - 150 )

< 11000 >
(6000 -15000)
Percent Reduction In PM-Astoelated Incidence:
Percent Reduction In Total Incidence:
Incidence Associated with Meeting Alternative Standards
20 ug/m3 annual
370
(220 -510)
0.0%
00%
920
(580 -1260)
00%
00%
260
(70 -450)
00%
00%
70
(30 -120)
100
(50 - 150 )
0.0% -00%
0 0% - 0 0%
< 11000 >
(6000 - 15000)
d.o%
00%
20 ug/m3 annual
330
(200 -460)
10.6%
02%
750
(440 -960)
185%
0 8%
230
(60 -400)
115%
02%
60
(30 -110)
90
(40 - 130 )
10.0% -14,3%
0 1% -0 1%
< 10000 >
(5000 - 13000)
9.1%
1 8%
20 ug/m3 annual
250
(1 50 - 340 )
32.4%
06%
390
(230 -490)
576%
26%
180
(50 -300)
30.6%
06%
50
(20 -80)
70
(30 - 100 )
28.6% • 30 0%
0 2% - 0 4%
<7000>
(4000 - 9000 )
36.4%
73%
20 ug/m3 annual
110
(70 - 160 )
703%
1 3%
0
(0-0)
1000%
4 6%
60
(20 -140)
692%
1 4%
20
(10 -40)
30
(20 - 40 )
70.0% -71.4%
0 5% - 0 9%
<3000>
(2000 - 4000 )
727%
146%
   * Health effects are associated with short-term exposure to PM, unless otherwise specified
   ' Health effects incidence was quantified across the range of PM concentrations observed in each study, when possible, but not below background
        PM-2 5 level  Background PM-2 5 is assumed to be 3 5 ug/m3 in Philadelphia County
   1 The percent reduction in PM-associated incidence achieved by attaining alternative standards as opposed to the current standards is the reduction in
       incidence divided by the incidence associated with current standards  For example, the percent reduction in PM-associated incidence of mortalit
       associated with short-term exposure to PM-2 5 achieved by meeting both a 15 ug/m3 annual and a 65 ug/m3 daily standard is (370-330)/370=10 8%
   * The percent reduction m total incidence achieved by attaining current or alternative standards is the reduction in incidence achieved by attaining
       the standard divided by the total (not only PM-associated) incidence
   '* PM-2 5 results based on using PM-2 5 mass as PM-10 mass in the PM-10 functions
   '"Angle brackets <> indicate incidence calculated using baseline incidence rates reported in studies, with no adjustment for location-specific
      incidence rates This increases the uncertainty in the incidence estimates
Sources of Concentration-Response (C-R) Functions
(A) C-R function based on pooled
   results from studies in six locations
(B) Popeetal, 1995
(C)Thurston. etal.1994
(D) Schwartz ft Moms, 1995
(E) Schwartz & Moms, 1995
(F)Schwartz, etal, 1994
The numbers in parentheses for pooled functions are NOT standard confidence intervals All the numbers in parentheses are interpreted as 90% credible Intervals
based on Monte Carlo analysis that takes into account both statistical uncertainty and possible geographic variability See text in Chapter VI for details

-------
                                                                                          VI-47
Table Vl-12b.  Estimated Changes in Health Risks Associated with Meeting Alternative PM-2.5 Standards
in  Philadelphia County,  September 1992 - August 1993 (for base case assumptions)
Health Effects*
Mortality (a» ages)
Mortality (al ages)
Hospital Admissions
Respiratory
Hospital Admissions
Cardiac
(A) Associated with short-term exposure
Percent Reduction In PM -Associated Incidence:—
Percent Reduction In Totel Incidence'****
(B) Associated with long-term exposure
Percent Reduction In PM -Associated Incidence:
Percent Reduction In Total Incidence*
(C) Total Respiratory
(all aoeft)
Percent Reduction In PM -Associated Incidence:
Percent Reduction In Total Incidence'
(D) Ischemic Heart Disease*****
(>64 years old)
(E) Congestive Heart Failure*****
(>64 years old)
Range of Percent Reductions In PM -Associated Incidence:
Rentte of Percent Reductlone in Total Incidence*
(F) Lower Respiratory Symptoms {8-12 yr olds) •**•••

Percent Reduction In PM -Associated incidence:
Percent Reduction In Totel Incidence*
PM-2 S-associated
Incidence
associated wth
current standards**
370
(220 -510)

920
(580 - 1260)

260
(70 -450)

70
(30 -120)
100
(50 -150)

< 11000 >
(6000 -15000)

Incidence Associated with Meeting Alternative Standards
15 ug/m3 annual
310
(190 -430)
162*
03%
660
(390 -850)
283%
1.3%
220
(60 -380)
154%
0.3%
60
(30 -100)
80
(40 -130)
14.3% -20.0%
0.1% • 0 3%
< 9000 >
(5000 -12000)
18.2%
36%
15 ug/m3 annu***t
and 65 ug/m3 daiv
310
(190 -430)
16.2%
0.3%
660
(390 -850)
28.3%
1.3%
220
(60 -380)
15.4%
0.3%
60
(30 -100)
80
(40 -130)
14. 3% -200%
0.1%- 0.3%
< 9000 >
(5000 -12000)
18.2%
3.6%
1 5 ug/m3 annual
and 50 uo/m3 daiy
250
(150 -340)
32.4%
0,6%
390
(230 -490)
57.6%
2.6%
180
(50 -300)
30.6%
0.6%
SO
(20-80)
70
(30-100)
28.9% - 30.0%
0.2%- 0.4%
< 7000 >
(4000 -9000)
36.4%
73%
1 5 ugmi3 annual
and 25 ug/m3 daily
110
(70 - 160 )
703%
1 3%
0
(0 -0)
1000%
4.6%
80
(20 -140)
692%
1.4%
20
(10 -40)
30
(20-40)
70.0% -71.4%
05% -0.0%
< 3000 >
(2000 -4000)
727%
14.6%
  * Hearth effects are associated with short-term exposure to PM, unless otherwise specified
 ** Health effects incidence was quantified across the range of PM concentrations observed in each study, when possible, but not below background
       PM-2 5 level Background PM-2 5 is assumed to be 3 5 ug/m3 m Philadelphia County
 **• The percent reduction in PM-associated incidence achieved by attaining alternative standards as opposed to the current standards is the reduction in
       incidence divided by the incidence associated with current standards  For example, the percent reduction in PM-associated incidence
       of mortality associated with short-term exposure to PM-2 5 achieved by meeting both a 15 ug/m3 annual and a 65 ug/m3
       daily standard is (370 - 310)/370 =162%
"'* The percent reduction in total incidence achieved by attaining current or alternative standards is the reduction m incidence achieved by attaining
       the standard divided by the total (not only PM-associated) incidence
*•••* PM-2 5 results based on using PM-2 5 mass as PM-10 mass in the PM-10 functions
••"••Angle brackets <> indicate incidence calculated using baseline incidence rates reported in studies, with no adjustment for location-specific
      incidence rates  This increases the uncertainty in the incidence estimates

The numbers in parentheses  for pooled functions are NOT standard confidence intervals. All the numbers in parentheses are interpreted as 90% credible intervals
based on  Monte Carlo analysis that takes into account both statistical uncertainty and possible geographic variability.  See text in Chapter VI for details.
Sources of Concentration-Response (C-R) Functions
(A) C-R function based on pooled
   results from studies m six locations.
(B) Pope etal, 1995
(C)Thurston, etal, 1994
(D) Schwartz & Morris, 1995
(E) Schwartz & Morris, 1995
(P) Schwartz, etal, 1994

-------
                                                                                            VI-48
 Table Vl-13a.  Estimated Changes in  Health  Risks Associated with Meeting Alternative PM-2.5  Standards
 in  Southeast Los Angeles  County, 1995* (for  base case assumptions)

Mortality (alt ages)
Hospital Admissions
Respiratory
Hospital Admissions
Cardiac
Health Effects
(A) Associated with short-term exposure
Percent Reduction In PM -Associated Incidence:*"
Percent Reduction In Total incidence:*"*
(B) Associated with long-term exposure
Percent Reduction In PM-Atsoclated Incidence:
Percent Reduction In Total Incidence:
(C) Total Respiratory
(ell ages)
Percent Reduction In PM -Associated Incidence:
Percent Reduction In Total Incidence:
(D) Ischemrc Heart Disease 	
(>64 years old)
(E) Congestive Heart Failure ""*
(>64 years old)
Range of Percent Reductions In PM-Assoctated Incidence:
Range of Percent Reductions in Total Incidence:
F) Lower Respiratory Symptoms (8-12 yr olds)**""

Percent Reduction In PM-Assoclsted Incidence:
Percent Reduction In Total Incidence:
PM-2 5-Related Incidence
associated with
current standards**
710
(430 -970)

2110
(1330 -2860)

940
(250 -1630)

130
(50 -200)
140
(70 -210)

< 43000 >
(23000 - 58000 )

Incidence Associated with Meeting Alternative standards
20 ug/m3 annual
560
(350 - 780 )
21 1%
06%
1540
(980 - 2060 )
270%
23%
750
(200 -1320)
202%
1.2%
100
(40 -180)
110
(60 - 1 70 )
21.4% -231%
0.3% • 0 4%
< 32000 >
(18000 - 43000)
25.6%
73%
20 ug/m3 annual
and 65 ug/m3 daily
430
(270 -900)
394%
1 1%
940
(600 - 1260)
55.5%
4 8%
570
(160 - 1030)
394%
24%
80
(30 -120)
80
(40 - 130)
38 5% -42 9%
0.4% • 0 8%
< 23000 >
(14000 -31000)
46.5%
133%
20 ug/m3 annual
and 50 ug/m3 daily
310
(210 -480)
56,3%
1 6%
480
(310 -640)
77.3%
68%
410
(120 -780)
564%
34%
60
(20-90)
60
(30 -100)
538% -57.1%
0,6% • 1 1%
< 16000 >
(10000 - 22000 )
628%
180%
20 ug/m3 annual
and 25 ug/m3 daily
120
(100 -220)
631%
24%
0
(0-0)
100.0%
86%
160
(50 -370)
830%
50%
20
(10 -40)
20
(20 - 40 )
84 6%- 85.7%
1 0% - 1.7%
<6000>
(5000 -9000)
860%
247%
Health effects are associated with short-term exposure to PM. unless otherwise specified
   * Los Angeles County was not in attainment of current PM-10 standards in 1995  Figures shown assume actual PM-10 concentrations
      are first rolled back to simulate attainment of these standards, and that actual PM-2 5 concentrations are rolled back by the same
      percent as PM-10 See text in Chapter VI for details
  ** Health effects incidence was quantified across the range of PM concentrations observed m each study, when possible, but not below background
       PM-2 5 level  Background PM-2 5 is assumed to be 2 5 ug/m3 in Southeast Los Angeles County
 "•The percent reduction in PM-asscoated incidence achieved by attaining alternative standards as opposed to the current standards is the reduction in
       incidence divided by the incidence associated with current standards For example  the percent reduction in PM-associated incidence
      of mortality associated with short-term exposure to PM-2 5 achieved by meeting both a 20 ug/m3 annual and a 65 ug/m3
      daily standard is (710 - 420J/710 = 40 8%
*"* The percent reduction in total incidence achieved by attaining current or alternative standards is the reduction in incidence achieved by attaining
      the standard divided by the total (not only PM-associated) incidence
***** PM-2 5 results based on using PM-2 5 mass as PM-10 mass in the PM-10 functions
"""Angle brackets «• indicate incidence calculated using baseline incidence rates reported in studies, with no adjustment for location-specific
        incidence rates This increases the uncertainty m the incidence estimates

The number* in parentheses for pooled studies are NOT standard confidence intervals.  All the numbers in parentheses are interpreted as 90% credible Intervals
based on Monte Carlo analysis (hat takes into account both statistical uncertainty and possible geographic variability. See text in Chapter VI for detail*.
Sources of Concentration-Response (C-R) Functions-
(A) C-R function based on pooled results from
   studies in 6 locations
(B)Popeetal, 1995
(C)Thurston, etal, 1994
(D) Schwartz & Morris, 1995
(E) Schwartz & Morris, 1995
(F)Schwartz, etal, 1994

-------
                                                                                          VI-49
Table Vl-13b.  Estimated  Changes  in Health Risks Associated with Meeting  Alternative PM-2.5 Standards
in  Southeast Los Angeles County, 1995*  (for base case assumptions)
Health Effects
Mortality (all ages)
Hospital Admissions
Respiratory
Hospital Admissions
Cardiac
(A) Associated with short-term exposure
Percent Reduction In PM -Associated Incidence:*"
Percent Reduction In Total incidence'****
(B) Associated with long-term exposure
Percent Reduction In PM -Associated Incidence:
Percent Reduction In Total Incidence'
(C) Total Respiratory
(all Met)
Percent Reduction In PM -Associated Incidence:
Percent Reduction In Total Incidence*
(D) (scheme Heart Disease 	
(>B4 years old)
(E) Congestive Heart Failure *****
(>64 years old)
Ringe of Percent Reductions In PM-Assoclated Incidence:
R*m0*fc of percent Reductions Ift Total incidence:
F) Lower Respiratory Symptoms (8-12 yr. olds)****'*

Percent Reduction In PM -Associated Incidence:
Percent Reduction In Total Incidence*
PM-2 5-Related Incidence
associated with
current standards**
710
(430 - 970 )

2110
(1330 -2860)

940
(250 - 1630)

130
(50 - 200 )
140
(70 - 210)

< 43000 >
(23000 -58000)

Incidence Associated with Meeting Alternative Standards
15 ug/m3 annual
390
(250 - 560 )
451%
1 3%
810
(520 - 1090)
61.6%
53%
520
(140 - 950 )
44 7%
2.7%
70
(30 - 110)
80
(40 - 120)
42 9% - 46 2%
0.5% - 0.8%
< 21 000 >
(13000 -28000)
51 2%
14.7%
15 ug/m3 annual
and 65 ug/m3 daily
390
(250 - 560 )
451%
1 3%
810
(520 - 1080)
81.6%
5.3%
520
(140 -950)
44 7%
2.7%
70
(30 - 110)
80
(40 - 120)
42.9% - 46 2%
0.5% - 0.8%
< 21000 >
(13000 -28000)
51.2%
147%
15 ug/m3 annual
and 50 ug/m3 daily
310
(210 -460)
58.3%
1.8%
480
(310 -840)
77.3%
8.8%
410
(120 - 780 )
56 4%
3.4%
60
(20-90)
60
(30 -100)
S3 8% -57.1%
0.8%- 1.1%
< 18000 >
(10000 - 22000 )
8X8%
18,0%
15 ug/m3 annual
and 25 ug/m3 daily
120
(100 - 220 )
83.1%
24%
0
(0-0)
100.0%
8.6%
180
(50 • 370 )
830%
50%
20
(10 -40)
20
(20-40)
84.6% - 85 7%
1.0% -1.7%
< 6000 >
(5000 -0000)
86 0%
247*
Health effects are associated with short-term exposure to PM, unless otherwise specified
   * Los Angeles County was not in attainment of current PM-10 standards in 1995 Figures shown assume actual PM-10 concentrations
      are first rolled back to simulate attainment of these standards, and that actual PM-2 5 concentrations are rolled back by the same
      percent as PM-10 See text in Chapter VI for details
**  Health effects incidence was quantified across the range of PM concentrations observed in each study, when possible, but not below background
     PM-2 5 level  Background PM-2 5 is assumed to be 2 5 ug/m3 in Southeast Los Angeles County
*** The percent reduction in PM-associated incidence achieved by attaining alternative standards as opposed to the current standards is the reduction in
     incidence divided by the incidence associated with current standards For example, the percent reduction in PM-associated incidence
     of mortality associated with short-term exposure to PM-2 5 achieved by meeting both a 15 ug/m3 annual and a 65 ug/m3
     daily standard is (710-390)/710 = 45 1%
**** The percent reduction m total incidence achieved by attaining current or alternative standards is the reduction in incidence achieved by attaining
     the standard divided by the total (not only PM-associated) incidence
***** PM-2 5 results based on using PM-2 5 mass as PM-10 mass in the PM-10 functions
	Angle brackets <>  indicate incidence calculated using baseline incidence rates reported m studies, with no adjustment for location-specific
      incidence rates  This increases the uncertainty m the incidence estimates

The numbers in parentheses for pooled studies are  NOT standard confidence  intervals.  All the numbers in parentheses are interpreted as 90% credible intervals
based on Monte Carlo analysis that takes into account both statistical uncertainty and possible geographic variability.  See text in Chapter VI for details.
                                                                                                                                                   Sources of Concentration-Response (C-R) Functions
                                                                                                                                                   (A) C-R function based on pooled results from
                                                                                                                                                      studies In 6 locations
                                                                                                                                                   (8) Pope et el, 1995
                                                                                                                                                   (C) Thurston. et el, 1994
                                                                                                                                                   (0) Schwartz & Morris, 1995
                                                                                                                                                   (E) Schwartz & Morris, 1995
                                                                                                                                                   (F) Schwartz, et al, 1994

-------
                                                     VI-50
Table VI-14.  Sensitivity Analysis:  Effect of Alternative Rollback Methods on Mortality Estimates
Short-term Exposure (Pooled Function) and Long-term Exposure PM-2.5 Mortality Functions
Philadelphia County, September 1992 - August 1993

Initial Air Quality: 16.3 ug/m3 annual average, 69.3 ug/m3 2nd daily maximum


(A) Mortality associated with
short-term exposure
(B) Mortality associated with
long-term exposure
Alternative Standard
15 ug/m3 annual
50 ug/m3 daily
15 ug/m3 annual
50 ug/m3 daily
Percent Change in PM-Associated Incidence
All PM concentrations rolled
back equally
10.6%
29.7%
19.4%
54.1%
Higher PM concentrations
reduced more
9.2%
18.6%
19.4%
39.3%
Portion of Proportional
Rollback Incidence Reduction
Achieved by Alternative
Rollback
86.4%
62.6%
100.0%
72.6%
* Health effects incidence was quantified across the range
   of PM concentrations observed in each study, but not below
   background PM-2.5 level, which is assumed to be 3.5 ug/m3.
(A) C-R function based on studies in 6 cities
(B)Popeetal., 1995

-------
                                   VI-51
County, alternative PM2 5 standards have been modeled based on the amount of air
quality  reduction required to meet the numerical value of the controlling standard.
Rounding conventions to be applied to any PM2 5 standards have not been determined
yet, and so the effect of rounding conventions has not been incorporated into this
analysis of alternative standards.  Several points from these Tables are of particular
interest:
Daily standards control the air quality reduction,  and thus the estimated health risk
reductions observed, for almost all of the alternative standards scenarios (Table VI-
llb). In Philadelphia, which has an "as-is" annual mean concentration close to 15
/ig/m3,  an annual standard of 20 /xg/m3 has no effect on reducing estimated incidence of
health effects  (Table VI-12a).  Attaining an annual standard  of 15 /-tg/m3 without a
daily standard is estimated to result in reductions in air quality concentrations and
health risks (about 14-20% reduction for effects associated with short-term exposures
and about 28% reduction for mortality associated with long-term exposure). However,
the estimated reductions in health risks associated with attaining the 50 /ig/m3 24-hr
standard are significantly higher (e.g., about 29-36% reduction in mortality and other
health effects associated with short-term exposures and about 58%  reduction in
mortality associated with long-term exposure upon attaining a 50 Mg/rn3 24-hr
standard).  Attaining a 25 ng/m3 24-hr standard in Philadelphia County is estimated to
result in the largest risk reductions (e.g., about 69-73% reduction in mortality and
other health effects associated with short-term exposures and  100% reduction in
mortality associated with long-term exposures  to PM).
In Los Angeles County, an annual standard of 20 /xg/m3 is estimated to reduce air
quality  concentrations about 19%, with all three of the 24-hr  alternative standards (65
/xg/m3,  50 /xg/m3, and 25 ptg/m3) requiring considerably greater reductions.  A 15
/xg/m3 annual standard controls the amount of air quality reduction and estimated health
risk reduced for the case involving a 65/xg/m3 alternative 24-hr standard, but not for
cases involving a 50 /xg/m3 or 25 /xg/m3 alternative 24-hr standard.  An annual standard
of 15 /xg/m3 alone reduces estimated health risks associated with PM about 43-51 % for

-------
                                   VI-52
mortality and other health effects associated with short-term exposure and about 62%
for mortality associated with long-term exposure relative to just attaining the current
PM10 standards in Los Angeles County. Attaining a 50 )ug/m3 24-hr standard reduces
estimated health risks associated with PM about 54-63 % for mortality and other health
effects associated with  short-term exposure and about 77% for mortality associated with
long-term exposure.  Attaining a 25 /ig/m3 24-hr standard is estimated to further reduce
health risks relative to the current PM10 standards, with about a  83-86% reduction in
mortality and other health effects associated with short-term exposure and a 100%
reduction in mortality associated with long-term exposure.  As expected, the estimated
health risk reductions are larger  for Los Angeles County than Philadelphia County due
to the higher PM air quality levels associated with meeting the current PM10 standards
(i.e., baseline air  quality  in Philadelphia is below the level required to meet the current
standards).
The  proportion of  estimated risk associated with reductions in PM2 5 under alternative
standard scenarios can  be considered either as a percentage in the PM-associated
incidence reduced  or as a percentage of total incidence of that health endpoint due to
PM  and all other causes.  As  an example,  standards of 15 /ig/m3 and 50 /itg/m3 24-hr in
Philadelphia County lead to an estimated 32% reduction in mortality associated with
short-term exposures to PM and a  29-36% reduction in morbidity (hospital  admissions
and  respiratory symptoms) associated with short-term exposures to PM.  These
changes result in reductions in the  overall incidence rates of these endpoints that are
considerably smaller.  For example, a 32% reduction in mortality associated with
short-term PM exposures leads to an estimated 0.6% reduction  in the total mortality
incidence.
Estimates of the reduction in total annual incidence of mortality upon attainment of
alternative standards are more uncertain than estimates of the reduction in total annual
incidence of other health effects,  as a consequence of uncertainties in the extent of
mortality displacement (shortening of life) that may be associated with PM (see Section
V C. 1 .c; CD, pp. 13-44-45). These uncertainties concerning the  degree of mortality

-------
                                            VI-53
       displacement are not as salient for estimates of reductions in annual mortality incidence
       associated with long-term PM exposures compared to short-term PM exposures, since
       the type of study design that produced the long-term exposure concentration-response
       functions provides findings that indicate effects on annual mortality rates (Utell and
       Frampton, 1995).  However, depending on assumptions concerning the biological lags
       and cumulative effects of air pollution involved in these long-term exposure studies,
       uncertainty is involved concerning how long an area would need to be  in attainment of
       an alternative standard in order for the full measure of estimated mortality rate
       reduction to be realized.
  •     Greater percent reduction of PM-associated risks is estimated for mortality associated
       with long-term exposures to PM than from short-term exposures.  This is the
       consequence of quantifying increases in mortality associated with long-term exposures
       only at concentrations considerably above background (PM2 5 concentrations > 9
       Mg/m3 based on Pope et al. (1995)).
       b. Individual Sensitivity Analysis Concerning  Air Quality Rollbacks
       The estimates of risk reductions in Tables VI-12 and VI-13 particularly depend on what
•inherently must be assumptions about the pattern of air quality reductions that will be observed
in the future in attaining the alternative standard cases.  While the base model used assumes a
proportional reduction would be  observed in all PM2 5 concentrations above background as a
consequence of control strategies intended to meet a controlling annual mean or 24-hr
standard, it is quite possible that substantial differences in PM25 air quality reductions could
occur across the PM2 5 distribution.!0 An attempt to bound the potential effects of these
possible alternative rollbacks has been examined in a  sensitivity analysis of PM-associated
    1 Information on past reductions of PM2 5 concentrations as a direct result of NAAQS is not available, given that
prior and current ambient standards for panicles regulated larger particle indicators (TSP, PM10).  Existing
monitoring information can be examined instead, although it is uncertain how much of the variation observed will
reflect actual control strategies versus more general year-to-year variability. In a preliminary examination of changes
in the distribution of PM, 5 concentrations from sites with multiple years of data (from AIRS and CARB data sets),
Abt Associates found that while a proportional rollback was a reasonable approximation of the central tendency of
variation observed, considerable variation in this relationship was observed (see Abt Associates, 1996b for more
information).

-------
                                          VI-54
mortality risks by choosing alternative assumptions for modeling PM2 5 rollbacks.  The results
of this sensitivity analysis are presented in Table VI-14.  The alternative reduction approach
provided for illustration decreases the upper 10% of PM2 5 24-hr air quality concentrations by
a larger amount (a ratio of 1.6) than the reductions in the remaining 90% of the distribution of
PM air quality concentrations and is intended to model a control strategy that preferentially
targets peak PM levels.
       The results of the sensitivity analysis in Table VI-14 indicate that estimated mortality
risks reduced by annual PM2 5 standards are largely  insensitive to the pattern of rollbacks in
PM2 5 concentrations, whereas estimates of risk associated with alternative 24-hr PM2 5
standards are somewhat more sensitive to the choice of rollback methodology.
       C.      Integrated Uncertainty  Analysis
       Using the same approach described previously in Section Vl.C.S.b, an illustrative
integrated uncertainty analysis was prepared for estimating the reduction in mortality risk
associated with short-term exposures upon attainment of example alternative PM2 5 standards in
Los Angeles County. These risk reductions were calculated relative to the scenario where Los
Angeles County just attains the current PM10 standards.  Figure VI-9 displays the results of the
integrated uncertainty analysis for attaining example PM2 5 standards of 15 ^g/m3, annual
average and 50 ^g/m3, 24-hour average in Los Angeles County. Several sources of
uncertainty were progressively included from left to right in the figure.  The first vertical line
reflects only uncertainty in the RR. The second vertical line  includes uncertainty in RR and
estimated background concentration, but no cutpoints are included. The next three vertical
lines incorporate uncertainty about cutpoints, using  the same  three cutpoint weighting schemes
discussed previously in Section Vl.C.S.b and employs a proportional  rollback method to
simulate attainment of the PM2 5 standards.  The last three vertical lines also incorporate
uncertainty about cutpoints, but use a non-proportional rollback approach to simulate
attainment of the PM2 5 standards.
       As was observed in the earlier integrated uncertainty analysis, uncertainty about
cutpoints has the largest impact on the estimated risk reduction associated with alternative
standards. In contrast, the use of a proportional or non-proportional rollback method appears

-------
                                               VI-55
          Figure VI-9.  Effect of Several Uncertainties on Reductions in Mortality
      Risk Associated With Short-Term Exposure to PM-2.5 Upon Attaining PM-2.5
         Standards of 15 jig/m3 Annual and 50 (ig/m3 Daily in Los Angeles County
S5%
2o%
w
c
00%
                                        (Population   3.6 Million)
         T 95th % lie   T
          5th % lie
o
                                                o
                                                                                             600
                                                                                             500
                                                                                             400
                                                                                             300
                                                                                             200
                                                                                  Mortality
                                                                                    Risk
                                                                                  Reduction
                                                                                  Associated
                                                                                 with PM-2.5:
                                                                                  Number of
                                                                                   Deaths
                                                                                             100
       Just RR
     (background =
      .2 5 pg/nf)
 RRand
background
                           Case I
                                     Case I
                                              Case I
            Cutpomt =
            Background
                RR, background, and
                Cutpomt Weightings
                                     RRand
                                    background
                                    Cutpomt =
                                    Background
                                                                  Case I
                                                                            Case I
                                                                                     Case I
RR, background, and
Cutpomt Weightings
                       Proportional Rollback
                                                                Non-Proportional Rollback
                                        Culpomt Weighting Schemes

Background
1 0 yg/m3
1 8 iig/m3
30 ng/m'
Case I
05
0.3
0 15
0.05
Case II
02
0.3
03
0.2
Case III
0.05
0.15
0.5
03

-------
                                          VI-56
to have only a slight impact on the estimated risk reduction for mortality associated with short-
term exposure to PM2 5 when placed in the context of the other uncertainties that also affect our
ability to predict risk reductions from alternative PM2 5 standards.
       In addition to the uncertainties inherent in estimating risks for the as is scenarios, such
as the relative risk, background, and cutpoint uncertainties assessed in the integrated
uncertainty analyses, estimates of reductions in risk resulting from attainment of alternative
PM2 5 standards are subject to uncertainties related to the projection of air quality that would
occur when alternative standards are attained.  These uncertainties relate in part to the
potential that PM2 5  may be serving in varying degrees as an index for air pollution (either by
indexing the effects of other gaseous copollutants in addition to PM2 5, or by indexing
relatively more harmful constituents within PM2 5).   Such uncertainties may serve to alter
estimates  of risk reduction associated with attainment of alternative PM2 5 standards, and the
anticipated effects of potential strategies used to reduce  PM concentrations.
       Figure VI-10 displays the results of the integrated uncertainty analysis for Los Angeles
County associated with attainment of several alternative PM2.5 standards.  Four sets of
standards are included: an annual standard alone set at 15 /xg/m3, and three pairs of standards
with an annual standard set  at 15 /xg/m3 accompanied by a 24-hour standards set at 65, 50, or
25 /ig/m3. In this figure, each set of four vertical lines  represents the estimated risk reduction
where uncertainties about background, RR,  and cutpoint, and form of rollback have been
included.  The first vertical  line in each group, labeled "background", assumes  a cutpoint set
equal to background, while  the next three lines represent the  three different cutpoint weighting
schemes described previously and listed in the table at the bottom of the figure.
              The estimated risk reduction associated with the 15 jig/m3 annual standard alone
is the same as that associated with this annual standard coupled with a 65 /xg/m3 daily standard,
because the annual standard is the controlling standard.  The greatest risk reduction is
associated with the 15  ^ig/m3 annual, 25 /ig/m3 daily standards pair.  For this standard
combination, the estimated mean risk reduction is about 2.2% (CrI 1.3-3.0) of total mortality
or about  500 (CrI  300-700)  excess deaths avoided when the cutpoint is set equal to the
estimated background concentration level.  Under the alternative cutpoint weighting schemes,

-------
                                  VI-57
Figure VI-10.  Effect of Several Uncertainties on Reductions in Mortality
  Risk Associated With Short-Term Exposure to PM-2.5 Upon Meeting
          Alternative PM-2.5 Standards in Los Angeles County

                          (Population: 3 6 Million)

o
f-<3 0%
0
£25%
u
£
j»20%
•g
sh 5%
Q.
.= 0 5%
41
Ct
0 0%









c




S









* <













5 «
1

> C

* J











€
)




D p O O C
a Q> (U 0) 1
£ 8 8 !

j
1
3
i


— :

I1









5 (3












i
^ «

•
5 C

™ i



95th H lie '



Mean (<


<
) J


) .
V



J
A



J
•

•
•


)
/
V

Slh % lie *
)
•
•


N
V
C


•

•


)
u

•

.

) •

> ~"


BOO
700

600

500 M ortallty
Risk
400 Reduction
Associated
300 with PM-2.5:
Number of
200 Deaths

100


u
POOO O7OOO SPPOP
0)0)0) oTntcuai O> fu 0) Ai
8 8- S S ?
1 cc
- 1
C
3
!


— :

1 IR
» x-(S(S(S S- S JS SS

- 3 — ^ B — ^

1 i a 1 1 R

15 ng/m3 annual
no daily
15 ng/m1 annual
65 ng/m1 daily
15 ug/m3 annual
50 ng/m3 daily
15 ng/m3 annual
25 ug/m3 daily
                        Cutpoint Weighting Schemes

Background
1 0 ne/rn'
ISjig/m1
30 ng/m!
Case I
0 5
0 3
0.15
005
Case II
0 2
0.3
03
0 2
Case III
0.05
0.15
0.5
0.3

-------
                                          VI-58
the estimated mean risk reduction for this same suite of standards is reduced to about 1.2 to
1.8% of total mortality (or about 290-430 excess deaths avoided) depending on the weighting
scheme used.  As discussed previously, the percent reduction in total mortality can be
expressed as either a percentage of total mortality due to all causes as shown on Figures VI-9
and VI-10 or as a percent reduction in the PM-associated mortality.  For example, a reduction
of 1.5% in total mortality (or 400 deaths) corresponds to a 56% reduction in PM-associated
excess mortality and a 1.0% decrease in total mortality (or 300 deaths) corresponds to a 42%
reduction  in PM-associated mortality.
       5.      Key Observations from the Risk Analyses
       This Chapter has presented a summary of a PM health risk assessment that quantifies
health risks associated with 1) existing  air quality  levels, 2) projected air quality levels that
would occur upon attainment of the current PM10  standards, and  3) projected air quality that
would occur upon attainment of several alternative PM2 5 standards in two urban areas.
Summarized below are key observations resulting from the risk analyses, as well as several
important caveats and limitations:

1)     Fairly wide ranges of risk estimates result for mortality and morbidity health effects in the
       two locations analyzed when  the effects of key uncertainties and alternative assumptions
       are considered

2)     In  the staff s judgment,  estimates of mortality and morbidity risks remain significant from a
       public health perspective when the current PM10 standards are attained.

       These points are illustrated below for mortality risks using base case and alternative
assumptions as well as for morbidity  risks using base case assumptions. For  example, risk of
mortality from short-term PM2 5 exposures upon attainment of the current standards  was
estimated to range from approximately 400 to 1,000 deaths a year in Los Angeles County
(population = 36 million) under base case assumptions, and from approximately 100 to 1,000
deaths across alternative assumptions considered in the integrated uncertainty analysis For
Philadelphia County (population = 1  6 million), a city with more moderate air quality already well

-------
                                         VI-59
below the current standards, mortality risk associated with short-term PM2 5 exposures ranged
from approximately 200 to 500 deaths under base case assumptions, and from approximately 20
to 500 deaths under alternative assumptions. In addition, risks of morbidity effects associated
with exposures to PM2 5 are estimated to center around approximately a thousand hospital
admissions and many thousands of cases of respiratory symptoms in children per year for Los
Angeles, with several hundred hospital admissions and thousands of cases of respiratory
symptoms estimated for Philadelphia (mean estimates of base case assumptions).

3)     Attainment  of the range of alternative PM2S standards considered was estimated to lead to
       essentially no changes in PM-associated risk to very substantial changes, depending on the
       city and the levels of the standards
       Mortality and morbidity risks associated with short-term PM exposures in Los Angeles
County  are estimated to be reduced by roughly 20-25% upon attainment of an annual PM2 5
standard of 20 ug/m3 and 45-50% for an annual standard of 15 ug/m3 beyond the risks associated
with attainment of the current PM10 standards when base case assumptions are used.  Under
alternative assumptions, a greater proportion of PM-associated risk would be expected  to be
reduced (although reductions in the absolute incidence of health effects may be less).  Daily
standards ranging from 65 ug/m3 to 25 ug/m3 would reduce PM-associated risks from roughly
40% to 85% beyond those associated with attainment of the current PM]0 standards when base
case assumptions are used  For an area already within attainment of the current standards
(Philadelphia County), risk reductions are estimated upon attainment of an annual standard of 15
ug/m3 (of roughly 15-20%) and attainment of 24-hr standards of 65  to 25 ug/m3 (ranging from
10-70%, respectively), for base case assumptions.

4)     Based on the results from the sensitivity analyses of key uncertainties and the integrated
       uncertainty  analyses, the single most important factor influencing the uncertainty
       associated with estimates of PM health risk is whether or not a cutpoint concentration
       exists below which PM health risks are not likely to occur.
       Alternative cutpoint concentrations considered for these analyses could result in as much
as a 3  to 4-fold difference in estimated risk associated with PM exposures in Los Angeles County

-------
                                           VI-60

(Figure VI-8, see also Exhibits 7.19 and 7.20, Abt Associates, 1996b) depending on the degree of

confidence one imputed to the likelihood that a PM2 5 cutpoint concentration existed at the highest

concentrations evaluated relative to the base case assumptions. In an area with PM

concentrations well below the current PM standards (e.g., Philadelphia County), differences in "as

is" risk for alternative cutpoint assumptions may be even greater, since these locations would be

expected to have a greater proportion of air quality values below the cutpoint concentration.


5)     Based on results from the sensitivity analysis of key uncertainties and/or the integrated
       uncertainty analyses, quantitative consideration of the following uncertainties have a much
       more modest impact on the risk estimates  inclusion of individual copollutant  species
       when estimating PM effect sizes, the choice of approach to adjusting the slope in
       analyzing alternative cutpoints; the value chosen to represent average annual background
       PM concentrations;  and the choice of rollback adjustment approaches for simulating
       attainment of alternative PM standards.

6)     Risk analyses of alternative standard scenarios incorporate several additional sources of
       uncertainty, including: uncertainty in the pattern of air quality concentration reductions
       that would be observed across the distribution of PM concentrations in areas attaining the
       standards ("rollback uncertainty") and uncertainty concerning the degree to which current
       PM risk coefficients may reflect contributions from other pollutants, or the particular
       contribution of certain constituents of PM2 5, and whether such constituents would be
       reduced in similar proportion to the reduction in PM2 5 as a whole.

       To the extent concentrations of other combustion source copollutants are reduced more or
less than PM25 concentrations in attaining alternative PM25 standards, estimates of health risk

reduced by alternative PM2 5 standards would be expected to vary in proportion to the degree to

which such copollutants have a genuine role in producing, or modifying the ability  of PM to

produce, some of the health effects associated with PM in current concentration-response

relationships.  Similarly, if specific constituents of PM25 mass have differing potencies in

producing health effects relative to other PM2 5 constituents, estimates of risk reduced would be

expected to vary if these constituent concentration are reduced to different  degrees by control

strategies designed to attain alternative PM2 5 standards.

-------
                                          VII-1
VII.   STAFF CONCLUSIONS AND RECOMMENDATIONS ON PRIMARY NAAQS
       This chapter presents staff conclusions and recommendations for the Administrator to
consider in deciding whether to retain, revise, and/or supplement the current primary PM
NAAQS. Drawing from the synthesis of information and analyses contained in both the Criteria
Document (CD, Chapter 13) and in Chapters IV, V, and VI herein, this chapter begins with staff
findings on the overall adequacy of the current primary standards for PM, going on to address
each of the major components needed to specify ambient standards: pollutant indicator, averaging
time, form, and level. Staff conclusions and recommendations on each of these interrelated
components for the current and alternative primary standards are based on considering how both
the components of an individual standard and a suite of standards operate together to protect
public health with an adequate margin of safety.
       In recommending a range of options for the Administrator to consider, the staff notes that
the final decision is largely a public health policy judgment. A final decision must draw upon
scientific information about  health effects and risks, as well as judgments about how to deal with
the range of uncertainties that are inherent in the scientific evidence and analyses. The staffs
approach to informing these judgments is based on a recognition that the available health effects
evidence generally reflects a continuum consisting of levels at which scientists generally agree that
health effects are likely through lower levels at which the likelihood and magnitude of the
response become increasingly uncertain. This approach is consistent with the requirements of the
NAAQS  provisions of the Clean Air Act and with how EPA and the courts have historically
interpreted the Act.  These provisions do not require the Administrator to establish a NAAQS at a
zero-risk level but rather at a level that avoids unacceptable risks and, thus, protects public health
with an adequate margin of safety.
       In addition, the staff notes that especially where considerable uncertainty exists with
regard to appropriate policy  choices based on the  scientific information and analyses, it is
appropriate to consider the risk management implications of alternative approaches that  represent
scientifically sound options.  For example, if the Administrator concludes that  the current
standards should be revised  to provide greater health protection, it is appropriate to consider
whether it would be more effective and efficient to do so by tightening the current PM10  standards

-------
                                          VII-2
or by establishing new PM2, standards. Thus, staff has considered risk management implications
together with the scientific evidence in assessing whether alternative approaches to establishing
PM standards would provide both the requisite  level of protection and an effective and efficient
basis for pollution control strategies that will result in the attainment and maintenance of adequate
public health protection.
A.     Adequacy of the Current Primary Standards for Paniculate Matter
       As discussed in Chapter II, the Clean Air Act calls for periodic review of the criteria and
the NAAQS. The overarching issue in such reviews is whether revision of the existing standards
is appropriate to reflect advances in scientific knowledge.  The information presented in the
Criteria Document and this Staff Paper is intended to provide a scientifically sound and policy-
relevant basis, in accordance with sections 108 and 109 of the Clean Air Act, for the
Administrator to reach conclusions with respect to whether the existing standards should be
revised and, if so, what revised or new standards, are appropriate.  The concluding section of the
integrative summary of health effects information in the PM Criteria Document provides the
following cogent summary of the science with  respect to this issue for the current review of the
PM  standards:
              "The evidence for PM-related effects from epidemiologic studies is fairly
       strong, with most studies showing increases in mortality, hospital admissions,
       respiratory symptoms, and pulmonary function decrements associated with several
       PM indices. These epidemiologic findings cannot be wholly attributed to
       inappropriate or incorrect statistical methods,  misspecification of concentration-
       effect models, biases in study design  or implementation, measurement errors in
       health endpoint, pollution exposure, weather,  or other variables, nor confounding
       of PM effects with effects of other factors.  While the results of the epidemiology
       studies should be interpreted cautiously, they  nonetheless provide ample reason to
       be concerned that there are detectable health effects attributable to PM at levels
       below the current NAAQS" (CD, p 13-92).
       This finding from the review of the scientific criteria clearly calls into question the
adequacy of the current NAAQS. The extensive PM epidemiologic database provides evidence of
serious health effects (e.g., mortality, exacerbation of chronic disease, increased hospital
admissions) in susceptible population groups (e.g., the elderly older adults with chronic
cardiopulmonary disease). Although the increase  in individual relative risk is small for the most

-------
                                           VII-3
serious outcomes, it is likely significant from an overall public health perspective, because of the
large number of individuals in susceptible population groups that are exposed to ambient PM
(CD, p 1-21).  While the lack of demonstrated mechanisms that explain the range of
epidemiologic findings is an important caution which limits conclusions as to causality, qualitative
information from laboratory studies of the effects of particle components at high concentrations
and dosimetry considerations suggest that the kinds of effects observed in community studies
(e.g., respiratory- and cardiovascular-related responses) are at least plausibly related to paniculate
matter. Indeed, the CD points to the consistency of the results of the epidemiologic studies from
a large number of different locations and the coherent nature of the observed effects as being
suggestive of a likely causal role of ambient PM in contributing to the reported effects. Given the
evidence that such effects may occur at levels below the current standards, as well as the nature
and potential magnitude of the public health risks involved, the staff believes that revision of the
current standards is clearly appropriate. Thus, the principal recommendation of this staff
assessment is that the current standards should be revised.
       The remainder of this chapter focuses on developing a range of alternative standards for
the Administrator to consider in determining what revised or new standards are appropriate to
protect public health. In formulating alternative approaches to establishing adequately protective,
effective, and efficient PM standards, staff concurs with the important conclusion from the CD
that fine and coarse fractions of PMIO should be considered  as two separate pollutants (CD, p 13-
93).  As discussed in Section V.F., the staff  assessment finds sufficient evidence to support
establishment of separate standards relating to these two fractions of PM10. On the other hand,
the staff also notes the larger body of epidemiologic evidence and air quality information related
to undifferentiated PM,0.
       Therefore, staff concludes that it is reasonable to consider two alternative approaches for
revising the standards:  1) adopt more protective standards using PM,0 as the sole indicator
combining fine and coarse fractions; and 2)  develop separate standards for fine and coarse
fractions of PM,0 using appropriate indicators for each fraction.  Conceptually, the first approach
is precautionary and gives significant weight to recent findings using PM10 as a surrogate for both
fine and coarse fraction particles, with less consideration of the evidence that suggests that the

-------
                                          VII-4
current standards provide adequate protection for coarse fraction particles. Because the PM10
monitoring network is in place, it also would result in more immediate implementation of revised
standards.  The second approach is based on the view that in the long run, more effective and
efficient protection can be provided by separately targeting appropriate levels of controls to fine
and coarse particles. Because of the need to develop and install additional monitors, this
approach would provide additional time to consider significant new scientific information before
any such standards were actually implemented.
       The relative merit of these two alternative approaches are considered in the next section,
which also summarize staff conclusions and recommendations regarding indicators for thoracic
particles, fine particles, and coarse fraction particles. Subsequent sections focus on identifying
alternative averaging times, forms, and levels for the recommended approach.
B.     Alternative PM Indicators  and Risk Management Implications
1.      PMIO as Surrogate Indicator for Fine and Coarse Fraction Particles
       The most recent summary  of scientific information in the CD and outlined in Chapters IV
and V continues to support past staff and CASAC recommendations regarding selecting size
specific-indicators for PM standards.  More specifically, the staff believes that the following
conclusions reached in the 1987 assessment remain valid:
1)     Health risks posed by inhaled particles are influenced both by the penetration and
       deposition of particles in the various regions of the respiratory tract and by the biological
       responses to these deposited materials.
2)     The risks of adverse health effects associated with deposition of ambient fine and coarse
       fraction particles in the thorax (tracheobronchial and alveolar regions of the respiratory
       tract) are markedly greater than for deposition in the extrathoracic (head) region.
       Maximum particle penetration to the thoracic region occurs during oronasal or mouth
       breathing.
3)     The risks of adverse health effects from extrathoracic deposition of general ambient PM
       are sufficiently low that particles which deposit only in that region can safely be excluded
       from the standard indicator.

-------
                                           VII-5
4)     The size specific indicator(s) should represent those particles capable of penetrating to the
       thoracic region, including both the tracheobronchial and alveolar regions.

       Based upon the above considerations as well as the  available information on human
dosimetry of particles, in the previous review the staff and CASAC recommended a size specific
indicator that included particles less than or equal to a nominal 10 urn cut point, termed PM,0.
The recent information on human particle dosimetry contained in the CD provides no basis for
changing 10 urn as the appropriate dividing line for particles capable of penetrating to the thoracic
regions. The recent epidemiologic literature also provides some evidence that thoracic particles
can be somewhat more closely linked to effects than can the "super coarse" (>10 um) fraction of
TSP (e.g. Dockery et al., 1993).  The CD concludes that "recent analyses have substantiated the
previous selection of PM|0 as an indicator of particle-related health effects" (CD, p. 13-93).
       In selecting the most appropriate indicator(s) for the PM standards, the staff believes that
consideration should be given to protecting public health through the use of standards that are as
effective and efficient as possible. An effective set of standards would capture all of the most
harmful constituents of PM10 and target them such that an appropriate level of control occurs for
the harmful components. Conceptually, a broad based PM  indicator such as TSP set  at a stringent
enough level can provide effective protection for the most harmful components.  However,
because such a standard would set unnecessarily stringent controls on extrathoracic constituents
unlikely to be most harmful, it would not be an efficient standard. As staff concluded in the
previous review, a PM10 indicator provides more efficient as well as more effective health
protection than would TSP (U.S. EPA, 1982b).  In the present review, it is important to make use
of the current state of knowledge to  select an  indicator(s) that not only captures all of the most
harmful components (i.e., an effective indicator), but also places greater emphasis for control on
those constituents or fractions that are most likely to result  in the largest risk reduction (i.e., an
efficient indicator).
       Therefore, consideration of the available evidence regarding the components of PMIO most
likely responsible for the observed health effects categories  at various levels is critical to
maximizing the effectiveness and efficiency of health protection strategies. The indicator is used

-------
                                          VII-6

to target and monitor health protection strategies, and the choice is key to overall health

protection provided by the PM NAAQS.  Given these concerns and the expanded information, the
staff believes it is appropriate to reexamine the question of whether the PM10 indicator should

undergo additional refinement to reflect new scientific understandings of fine and coarse fraction

particles as separate pollutants.
       The staff assessment of the progress made through implementing the current PM,0

standards is instructive in this regard (Section IV.D).  Figure IV-4 and Table IV-5 summarize

how the States and EPA characterize the  major sources of PM,0 and the extent of progress to

date. In essence, the lessons learned from past TSP and PM10 programs can be summarized as

follows:
•      PM,0 is  generally viewed as a local rather than a regional problem.  This is clearly
       appropriate in most Western areas with the highest PM10 levels.  However, even in the
       eastern U.S., where high regional  levels of transported fine particles make significant, but
       not dominant, contributions to PM10 mass, programs tend to focus on control of local
       sources, in part because of the difficulty in developing multi-jurisdictional strategies. This
       means that abatement programs will generally focus on the most readily available local
       sources of primary particles, leaving secondary or regional options as a last resort.

•      In areas where local fine particle sources are overwhelmingly dominant, for example in
       areas with high woodsmoke contributions (e.g., Klamath Falls, OR), PM,0 controls have
       led to significant reductions in fine particles. Historically, TSP-based local programs have
       also resulted in significant reductions in local primary fine particle emissions from coal
       combustion and industrial sources (e.g., New York City, Pittsburgh, PA).

•      In areas where fugitive sources of crustal materials are clearly dominant (e.g., Coachcella
       Valley, CA), PM10 programs focus on measures that reduce road dust, construction, and
       related sources.  These programs  have had limited success to date. Local sources of
       precursor gases contributing to fine particles generally are not  addressed.

•      In areas dominated by local point  source complexes (industrial emissions), both coarse and
       fine controls are applied, and sources sometimes may trade reductions between the two on
       a mass basis. Where source complexes are located in a zone of high transported fine
       particles, the transported component is treated as background, increasing the need for
       local controls; this likely results in greater relative control for coarse particles than fine.

-------
                                          VII-7
•      In Western areas having "mixed" contributions, including significant local secondary
       particle formation, three areas (SCAB1; Provo, UT; Denver, CO) have begun to require
       controls of gaseous precursors (SOX, NOX) in addition to fugitive dust and other controls.
•      Any reductions in fine particles related to regional sulfur oxides emissions that have taken
       place to date are not related to implementation of the PM10 or TSP standards, but the SO2
       NAAQS and other mandated requirements of the CAA, such as the acid rain program.
       This experience is a useful guide for a qualitative examination of the potential
effectiveness and efficiency of alternative revised protective standards using PM)0 as the sole
surrogate for the harmful components of PM.  To provide a basis for such examination. Table
VII-1 presents a set of increasingly more protective alternative PM10 standards drawn from the
staff analysis of potential PM10 effects "cutpoints" developed in Appendix E for the risk
assessment. These alternatives do not reflect staff recommendations, but are examples presented
for the purpose of the present assessment of the PM,0 indicator.  The table indicates the regional
distribution of the percentage of counties (meeting a 50% data completeness criteria) that would
not attain the listed alternatives. The table also notes the characteristic regional contribution of
coarse fraction particles to PMIO mass, which, like total mass, is generally highest in the West.
       Looking first at annual PM,0 standards alone, the table suggests that a moderate reduction
from the current level (to 40 ug/m3) would result in few controls in eastern areas, but would
approach the combined effect of the current 24-hour and annual  standards in the West. A more
substantial  reduction in an annual standard to 30 ug/m3 would affect  about half of the Western
areas and also  begin to prompt additional controls in the East. By comparison, a revised 24-hour
PM,0 standard of 100 ug/m3 (alone or in combination with a 40 ug/m3 annual) would have  effects
similar to a 30 ug/m3 annual standard alone in the East, but affect still more (approximately 55 to
over 75%) Western areas. Based on the implementation experience outlined above, the eastern
areas would likely develop control programs to achieve such standards with an initial focus on
local sources of PMIO, which would tend to result in a proportionally greater reduction for  coarse
        South Coast Air Basin of California.

-------
                                           VII-8

TABLE VIM.  PERCENTAGE OF COUNTIES NOT MEETING ALTERNATIVE
                                   PM10 STANDARDS*

County Total
Annual



24-hr



Combined
Standards



Coarse/PM10***
Level of
Alternative
Standards**

50
40
30
All
482
2.3
7.3
29
SW
60
13.
22
45
NW
80
3.8
15
48
CE
68
0
7.4
26
SE
99
0
1.0
16
NE
175
0
2.3
23

150
100
50
12
35
97
27
55
97
34
76
98
8.8
32
90
2.0
25
100
3.4
16
98

50/150
40/100
30/50
12
35
97
26
55
97
34
76
98
8.8
32
90
2.0
25
100
3.4
16
98

~
0.44
0.55
0.60
0.37
0.44
0.37
       Based on 1991-1993 data, using 50% data completeness criteria and the Appendix K missing data adjustment
       to account for less than every day sampling frequencies. See staff analyses (Fitz-Simons et al., 19%).

       Based on current 1 -expected-exceedance form of the 24-hour PM „, NAAQS and current expected annual
       average of annual PM,0 NAAQS, at the highest monitor for each standard.
       Regional median ratio of coarse fraction mass to PMM) mass all seasons, based on available data from few sites
       (SAI, 1996).

-------
                                            VII-9
 fraction particles than fine.  Without a more detailed assessment beyond the scope of this paper,
 it is not clear whether or how much PM,0 standards set at these levels would also prompt more
 balanced reductions in fine and coarse fraction particles in the East.  In the West, however,
 widespread nonattainment resulting from such PM10 standards would clearly prompt much more
coarse particle control, based on the prevailing high coarse fraction content of PM,0.
        This analysis suggests that, nationwide, progressively reducing the level of the PM10
 standards alone to the middle levels in the table would place relatively more  emphasis on
 additional controls for coarse fraction particles than for fine.  On a regional basis, relatively less
 impetus for additional control would be placed on the East, which has the highest regional
 concentrations of fine particles, than on the West, which has the highest localized concentrations
 of coarse fraction particles.  Clearly, PMIO standard levels somewhere in the range below the
 middle levels shown in Table VII-1 would also result in relatively more control of fine particles in
 the East. Such standards would inevitably increase the number of areas needing to address coarse
 fraction particles in the West.
        One view of the risk management implications of the recent epidemiology holds that a
 single PM,0 indicator is most appropriate because more studies have used PMIO and it would
therefore be more prudent to prompt proportional reductions in the major components of PM10.
Even accepting such a view, however, our analysis indicates that reduced PMIO standards would
not result in proportional reductions in fine and coarse fraction particles in the  very areas from
which most of the epidemiological results are derived (see cover figure). Selecting levels  that
would achieve such proportional reductions in the East through a PM,0 indicator alone would still
result in significantly disproportional coarse particle control in the West2.  In essence, the  above
analysis is consistent with the admonition in the CD that more effective PM10 programs can be
achieved by establishing separate targets for fine and coarse fraction particles (CD, 13-94).
        The acid rain program should result in some additional regional SOX reductions in the East. However, much
of the improvement has already been realized with more gradual reductions over the next 15 years due to the banking
and trading components. The existence of such a program, however, provides no justification for establishing
inappropriate PM NAAQS targets, nor for the potential over control of coarse fraction particles, particularly in the West.

-------
                                          VII-10
       From this analysis, then, a decision to provide increased health protection through
standards indexed by undifferentiated PM,0 alone would have to be based on two additional
premises:  1) fine and coarse fraction particles are likely to produce similar health effects at
equivalent concentrations, i.e. be of relatively comparable toxicity; and 2) control strategies for
fine and coarse fraction particles would produce roughly equivalent reductions in exposure in
sensitive populations shown to be at increased risk of PM effects. Yet, the staff analyses of the
available information as summarized in Section V.F provides little support for either premise.
While the relative toxicity of fine and coarse fraction particles is not clearly established, both
physical and chemical toxicologic considerations suggest that fine particles are likely to be more
toxic for several, although not necessarily all, of the relevant effects categories than are coarse
fraction particles (Section V.F).  Based on the direct comparisons in epidemiological studies and
on exposure considerations, the staff further concludes that - - whatever the relative toxicity of
fine and coarse fraction particles - - control of sources of ambient fine particles is likely to be
more effective in reducing exposure to sensitive subpopulations than  is control of sources of
ambient coarse fraction particles.
       Given the available evidence, a uniform reduction in the levels of the PM,0 standards could
provide effective health protection from the effects of the most harmful components of PM,0, but
only at concentrations  that appear to be unnecessarily stringent with respect to coarse  fraction
particles.  Limited, but important epidemiological evidence as well  as mechanistic considerations
suggest that coarse fraction particles are linked to effects in areas that exceed the current PM10
standards (CD, p. 13-51). Given the lack of evidence with respect to coarse particle effects at
concentrations at or below the level of the current PM10 standards, however, little justification
exists for proportional, much less disproportional, reductions in coarse fraction particles beyond
those afforded by the current standards. By contrast, a number of epidemiological  studies have
used fine  particles as an indicator.  The available evidence comparing the two fractions suggests
that fine particles are a better surrogate for those components of PM10 that are associated with
adverse effects at levels below the current standard (sectionV.F).  For these reasons, staff
concludes that a single PM10 indicator would not provide the most effective and efficient
protection from the health effects of paniculate matter.  Instead, the  data available in  this review

-------
                                          VII-11
suggest that the most effective and efficient approach would be to control PM,0 through separate
standards for fine and coarse fraction particles.
2.     Alternative Surrogate Indicators for Fine and Coarse Fraction Particles
       The large number of recent community epidemiologic studies and improvements in human
exposure and air quality presented in the CD and outlined in Chapters IV and V above have
greatly expanded the information regarding associations between contemporary community air
pollution containing particles and morbidity and mortality in sensitive subpopulations as compared
to the previous review. Even with the presence of other pollutants in the communities studied,
PM is independently associated with the observed health effects.  While earlier studies mainly
relied on BS, TSP, and sulfates as particle indicators, the recent work-has added a much larger
body of quantitative and qualitative information on PM10, with a lesser but still substantial number
of community studies that provide specific information on fine particles, including sulfate and acid
aerosol components, and to a still lesser extent, coarse fraction particles (CD, p 1-21).
       The CD concludes that the indices most consistently associated with health endpoints are
thoracic  ( PM10 or PM,5) and fine particle indicators. Less consistent relationships have been
observed for TSP and the coarse fraction  of PMIO (CD, p 1-21). Based on an examination of
relevant information  in the CD on fine and coarse fraction particles (Section V.F), the staff
concludes that the weight of the available evidence allowing direct comparisons suggests that
ambient coarse fraction particles are either less potent or a poorer surrogate for community
effects of air pollution than are fine particles.  This assessment finds that the limited evidence
suggestive of independent coarse particle effects was found in areas that significantly exceed the
current standards, while reported associations with fine particles frequently occur at levels well
below the current standards.
       The staff concurs with the CD recommendation that "it would be appropriate  to consider
fine and coarse mode particles  as separate subclasses of pollutants" (CD, p 13-94). The staff also
concludes that sufficient information exists to do so.  The analysis in the preceding section
indicates that establishing distinct targets for fine and coarse fraction  particles would provide
more effective and efficient health protection strategies for PM. Therefore, the staff recommends

-------
                                          VII-12
that separate standards be established for the fine and coarse fractions of PM]0. The discussion
below outlines staff conclusions and recommendations for selecting indicators for such standards.
       a.      Surrogate Indicators for the Fine Fraction of PMIO
       Although fine mode particles consist of several distinct chemical classes (Table IV-2), they
share a number of important characteristics  related to size and formation mechanisms.  The CD
concludes that none of these subclasses can  be specifically implicated as the sole or even primary
cause of specific morbidity and mortality effects (CD, p. 13-93). In essence, fine particle mass is a
surrogate for whatever components appear to be causing the mortality and morbidity effects in
community air pollution.
       In examining the potential effectiveness of fine particles as a surrogate, it is useful to
consider the results of various analyses of air pollution and mortality in Philadelphia as discussed
in Section V.E (Moolgavkar et al., 1995; Wyzga and Lipfert, 1995; Samet et al., 1995, 1996a;
Cifuentes and Lave,  1996).  The CD evaluation of these multiple investigations concludes that for
this single city example, it appears most difficult to separate independent effects of PM (as TSP)
and SO2, concluding that the relationship between these pollutants and mortality may be
inherently non-linear (CD, p 13-57).  Several clearly hypothetical explanations have been
advanced to explain these results. The following qualitative assessment of several speculative, but
plausible hypotheses (in italics), outlines the potential implications of these alternatives for the
effectiveness of fine particle control as a surrogate:
•      The complex relationship is a statistical artifact and only one of the pollutants is causally
       related.  If the pollutant is PM, then fine particle control would clearly be beneficial.  //
       the pollutant is SO2,  which occurs at moderate levels in Philadelphia, reductions in local
       and transported  SO2 precursor control prompted by  a fine particle standard would reduce
       health risk.3
•      The relationship is real and due to increased penetration of an SO2 complex carried on
       carbonaceous or other non-acidic particles.  Then local controls of primary fine particle
       combustion sources would likely reduce risks, because reducing the aggregate particle
       surface area (by reducing fine mass) is more likely to reduce dose than SO2 reductions.
       3  As noted in section V.E, the evidence across multiple areas shows that PM is consistently associated with
 mortality in areas with high and low S02, making the second explanation unlikely.

-------
                                           VII-13
•       The relationship is due to the association between SO2 and acidic sulfates, which are the
        active agent. In this case, fine particle controls are clearly beneficial.
•       The relationship is due to the combined interactions ofSO2 and particles in different
        regions of the respiratory tract.  Again, control of fine particles would be beneficial.

        The staff does not have to accept any one of these hypothesized explanations as more
likely to conclude that control of fine particles as a class appears to be a reasonable approach to
reducing health risks in this particular example of potential confounding. It is also useful to note
that, because of their relatively low surface area and origin, such a conclusion would not be as
applicable to control of coarse fraction particles.
        Although the above examples of alternative consequences of the use of fine particles as a
surrogate are limited to PM and  SO2 interactions, some of these outcomes would extend to PM
interactions with other pollutants as well.  Given the large surface area of aqueous droplet and/or
dry fine particles, as well as the multiplicity of similar effects caused by common gaseous
pollutants such as ozone and related photochemical products and precursors, and NO2 in addition
to SO2, direct or indirect interactions among these pollutants would not be unexpected (Section
V.F.; CD, p 13-9.).  Because ozone precursors, including NO2 and volatile organic compounds,
are also secondary particle precursors, it is reasonable to expect that the control of fine particles
could also prompt control of local and regional sources of some of these precursors as well as
SO2. On the other hand, beyond the possibility of effects modifications in the body, the potential
for gas/particle interactions between PM and CO is limited.  It is also less clear that fine particle
control  would prompt significant additional CO control, the major contributors of which, mobile
sources, are already subject to significant national reduction requirements.  The rationale for
concluding that the existence of PM effects is unlikely to be due to confounding by other
pollutants is discussed in Section V.E.
       The above examples also illustrate  why, based on current information, it is more
appropriate to control fine particles as a group, as opposed to singling out particular classes. The
qualitative literature has found various effects of high concentrations of fine sulfuric acid,
ammonium sulfates  and nitrate, carbonaceous materials, and transition metals, alone or in some

-------
                                          VII-14
cases, in combination with gases (CD, Chapter 11; Section V.C).  Community studies have found
significant associations between fine particles or PM10 and health in areas with significant mass
contribution of these fine components, including sulfates (6 cities), wood smoke (Santa Clara),
nitrates (Los Angeles and Utah Valley), secondary organics (Los Angeles), and acid sulfate
aerosols (24 City Study). As noted above, it is not possible to rule out any one of these
components as contributing to fine particle effects.
       The most substantial laboratory and epidemiologic data for any single class of fine
particles exists for sulfates and associated acids.  The data for acids, which are more difficult to
measure, is less consistent than for sulfates.  For example, the recent 24 City Study data suggest
that regionally high exposures to acids in modest sized communities in the "sulfate belt" are
associated with bronchitis and decreased lung function in children (Dockery et al., 1996; Raizenne
et al., 1996).  Yet relatively strong correlations exist between acids, sulfates, and fine particles,
making it difficult to single out any factor with confidence (CD, p 13-93).  Indeed, the staff
considers sulfates useful as an indicator of fine particles for assessing the health effects literature.
This literature suggests that reductions of regional  sulfates as part of a fine particle standard
control program would likely reduce mortality and morbidity risks for the large segments of the
sensitive population who reside in the East.  It would be inappropriate, however, to extend this
finding to establishing a separate sulfate standard, alone or in combination with fine particle
standards. A sulfate standard, even if understood as an indicator of all fine particles as suggested
by Lippmann and Thurston (1996), would be less likely to lead to controls of the other potentially
harmful components of fine particles.
       A number of monitoring approaches have been used as indicators for fine particles
(Appendix B). All of them have inherent strengths and weakness (CD, pp. 1-6 to 7). In selecting
an indicator for a fine particle NAAQS, the staff places great weight on providing consistency
with the largest segment of the epidemiologic data, and to a lesser extent, on making use of the
existing fine particle data in the U.S.  Staff have submitted their recommendations regarding the
most appropriate monitoring approach for a fine particle standard to the CASAC Technical

-------
                                          VII-15
Subcommittee for Fine Particle Monitoring4.  The staff rejected the use of filter based optical
approaches because they are more sensitive to variations in carbon and require mass calibration
(CD, p 1-6).  Although direct optical (e.g., nephelometry) and other continuous methods can offer
significant advantages and are often well correlated with gravimetric mass measurements, under
some circumstances they are less well linked, in part because of losses of semi-volatile
components (CD, p 1-6).  Further development of such approaches for routine use is an
important need.   Because most of the quantitative epidemiological data for fine particles and
PM)0 were based on gravimetrically determined mass, staff recommends that this measurement
principle be adopted for fine particle standards.  Although some loss of nitrate and other semi-
volatile mass can occur with such methods, gravimetric approaches are most directly related to
the available epidemiology, and they can be used to provide composition information helpful for
developing control strategies. Again, improved continuous approaches that could be used as
equivalent methods for fine particles are an important development need.
?-      Staff also recommend the use of a sharp 2.5 urn cutpoint for a fine particle indicator. As
discussed in Chapter IV and Appendix A, the minimum particle diameter between the fine and
coarse modes lies between 1 and 3 um, and the scientific data support a cut point to delineate fine
particles in this range. Because of the potential overlap of fine and coarse particle mass in this
intermodal region, specific cut points are only an approximation of fine mode particles. Thus, the
decision within this size range is largely a policy judgement. The  staff recommendation for a 2.5
um cutpoint is based on considerations  of consistency with health data, the limited potential for
intrusion of coarse fraction particles into the fine fraction, and availability of monitoring
technology.  Therefore, the staff recommends using PM2 5 as the fine particle indicator.  The
definition will be further specified in the Federal Reference Method and equivalency program.
       PM2 5 encompasses all of the potential agents of concern in the fine fraction, including
most sulfates, acids, fine particle metals, organics, and ultrafine particles and includes most of the
aggregate surface area and particle number in the entire PM distribution.  PM2, has been used
        The Subcommittee met to review these recommendations as well as specifications for a possible Federal
Reference Method and Monitoring Guidance at a public meeting on March 1, 1996.

-------
                                          VIM 6
directly in health studies as described in the CD and Chapter V. Although a number of studies
have used PM2,, in most locations there should be little difference in mass.  The more
widespread use of  PM25 measurement technologies since the 1970s has resulted in the generation
of relatively more data for this cutpoint than for other cutpoints for fine fraction particles.
       PM2 s does have some potential for intrusion of the "tail" of the coarse mode during
episodes of fugitive dust concentrations (See Appendix A). Staff recommends a sharp inlet for
the FRM to minimize this potential intrusion of coarse mode particles. Such intrusions into PM2 5
measurements is not anticipated to be significant in most situations; nevertheless, if subsequent
data reveal problems in this regard, this issue can and should be addressed on a case-by-case basis
in the monitoring and implementation programs.  Because the purpose of a PM2 5 standard is to
direct controls toward sources of fine mode particles, it would be appropriate to develop
analytical procedures for identifying those cases where a PM2, standard  violation would not have
occurred in the absence of coarse mode particle intrusion.5 Consideration should be given to a
policy similar to the natural events policy (See Chapter IV) for addressing such cases.
       Some commentors have recommended use of a smaller cutpoint at 1  urn (PM,) to further
reduce coarse particle intrusion. PM,  has not been used in health studies, although in most cases
mass should be similar as for cutpoints of 2.1 or 2.5.  While this indicator could reduce intrusion
of fugitive dust, it might also omit portions of hygroscopic acid sulfates in high humidity episodes.
PM, sampling technologies have been developed; however, the PM, samplers have not been
widely field-tested to date. Of some concern is the theoretical possibility that different flow
velocities for the smaller cut might increase the loss of semivolatile materials relative to a larger
cut. Thus, the staff recommends the use of PM2 5 as the fine particle cutpoint.
       b.     Surrogate Indicators for the Coarse Fraction of PM10
       The CD and staff assessment finds that epidemiologic information, dosimetry and
toxicology support the need  for a particle indicator that addresses the health effects of coarse
fraction particles smaller than nominal 10 urn. Coarse fraction particles deposit in both the
         Analytical procedures could involve measurements of chemical components related to local coarse mode
particles as a basis for developing a coarse mode intrusion estimate. Lundgren et al. (1996) have submitted a paper
suggesting one such approach

-------
                                          VIM 7
tracheobronchial and alveolar region.  Although the role of coarse fraction particles in much of
the recent epidemiological results is unclear, studies where coarse fraction particles are the
dominant fraction of PM10 suggest that the major short-term effects include aggravation of asthma
and increased upper respiratory illness. Such effects are supported by dosimetric considerations
•(CD, p 13-51).  Children, who spend more time in outdoor activities, may encounter higher
exposures and doses of coarse fraction particles than other potentially sensitive populations.
Long-term deposition of insoluble coarse fraction particles in the alveolar region may have the
potential for enhanced toxicity, in part because clearance from this region of the lung is
significantly slower than from the tracheobronchial region. Qualitative support for this concern is
found in autopsy studies of animals and humans exposed to various ambient crustal dusts at  or
slightly above ambient levels typical in the Western U.S. (Section V.C).
       In selecting an indicator for coarse fraction particles, it is important to note that the
existing ambient data  base for coarse fraction particles (PM](W_S) is smaller than that for fine
particles, and that the only studies of clear quantitative relevance have used undifferentiated  PM,0.
-However, it is possible  to consider PM,0 itself as a useful surrogate for coarse fraction particles,
when used in conjunction with PM2 5  standards. As noted above, in many areas with high fugitive
dust, this is already the case with respect to control strategies.  Because coarse fraction particles
in such areas contribute significantly more mass than smaller particles, risk managers have
incentives to focus reduction measures on particle sources that contribute the most by mass. The
monitoring network already in place for PM10 is large. Therefore, if a fine particle indicator were
chosen, the staff would recommend retention of PM,0 as the indicator to protect against the  risks
of coarse fraction particles.
3.     Staff Conclusions and Recommendations for Particle Indicators
       Based on the above assessments and the scientific information in the CD, the staff draws
the  following conclusions and recommendations:
1)     Ambient particles capable of penetrating to the thoracic  region represent the greatest risk
       to health. Previous staff and CAS AC recommendations for 10 urn as the  appropriate cut
       point for such particles remain valid. In examining alternative approaches to increasing
       the protection afforded by PM,0 standards, the staff finds that  reducing the levels of the

-------
                                          VII-18
       current standards would not provide the most effective and efficient protection from the
       health effects of paniculate matter.
2)     The recent health evidence, the fundamental differences between fine and coarse fraction
       particles, and implementation experience with PM10 have, however, prompted the staff to
       consider separate standards for the fine and coarse fractions of PM10.
3)     The staff finds that the available information is sufficient to support separate indicators for
       these pollutant classes.  While it is difficult to distinguish the effects of fine or coarse
       fraction particles from those of PM10, consideration of comparisons between fine and
       coarse fraction particles suggests that fine particles are a better surrogate for those particle
       components linked to mortality and morbidity effects at levels below the current standards.
       Coarse fraction particles are most clearly linked with certain  effects at levels above those
       allowed by the current standards.
4)     In selecting an indicator for fine particles, staff recommends use of a 2.5 urn cut point for
       fine particle mass.  Adoption of sulfate or other chemical class indicators is not advisable
       during this  review. In selecting an indicator for coarse fraction particles, the staff
       recommends use of PM,0 .
C.     Alternative PM: ^ Standards for Control of Fine Fraction Particles
1.     Averaging Time
       The current primary PM NAAQS include both a 24-hour standard, with no more than one
expected exceedance, and an annual standard with an expected arithmetic mean form. These
standards were  intended jointly to protect the public against the health effects associated with
both short-term and long-term exposures to PM based on epidemiological and other health studies
available at the conclusion of the last review. The recent health effects information includes
reported associations with both short-term  (from less than 1 day to up to 5 days) and long-term
(from generally a year to several years) measures to PM.  This information, summarized in
Chapter V, provides increased support for consideration of both short-term and long-term
standards, as discussed below.

-------
                                          VIM 9
       a.     Short-term PM;_5 Standard
       The current 24-hour averaging time is consistent with the majority of the results from
community epidemiological studies, which have reported associations of 24-hour concentrations
of PM,0, fine particles, and TSP with an array of health effects. Nevertheless, because some such
studies have found a stronger association with a multiple day average (Pope et al., 1992; Ostro et
al., 1995; Pope and Dockery et al., 1992), the staff considered whether a multiple day averaging
time would be more appropriate.  The above results are also consistent with the existence of a
lagged single exposure effect of PM, which may not be due to multiple day exposures. Moreover,
some studies have found health effects to be associated with same day or previous day PM
concentrations. For example, such associations are shown by mortality studies in Los Angeles,
CA; Birmingham, AL; St. Louis, MO; Toronto, Canada; Santiago, Chile; Athens, Greece; and
London, England. Further, most hospital admissions studies show associations with same day
concentrations. In any case, a 24-hour standard can effectively protect against episodes lasting
for several days while also protecting sensitive individuals who may experience effects after a
single day of exposure. Thus, the staff concludes that the complexity in adopting a multiple day
averaging time, e.g. 3 to 5 days, would not provide more effective protection than a 24-hour
average.
       The staff  has also considered the evidence regarding effects associated with PM
exposures of durations less than 24 hours. Some investigators prior to the 1987  review (Lawther
et al., 1970) speculated that the observed health effects might be largely due to short-term peaks
on the order of an hour. Controlled human and animal exposures to specific components of fine
particles, such as  acid aerosols, also  suggest that some effects, such as bronchoconstriction, can
occur after exposures of minutes to hours. Some epidemiological studies of exposures to acid
aerosols have also found changes in respiratory symptoms in children using averaging times less
than a 24-hour period (e.g., 12 hours).  However, the majority of effects have been associated
with daily or longer exposure to PM. Moreover, limitations in current mass monitoring devices
make shorter durations less practical at present.  A 24-hour average can be expected to provide
significant protection from potential effects associated with  short duration peaks  in most urban
atmospheres.  Thus, although some study results may be suggestive of short duration effects, the

-------
                                           VII-20
staff does not believe that the reported results provide a satisfactory quantitative basis for setting
a general particle standard with an averaging time of less than 24 hours. The staff believes that
additional research is needed to examine short duration exposures.
       The staff recommends that consideration be given to retaining the current 24-hour
averaging time as a means of controlling short-term ambient PM25 concentrations, especially peak
concentrations, and thus providing protection from health effects associated with short-term (from
less than 1-day to up to 5-day) exposures to PM25.
       b.      Long-term PM-, 5 Standard
       As summarized in Chapter V, community epidemiological studies have reported
associations of annual concentrations of PM2 „ sulfates, PM,0, and TSP with an array of health
effects, notably increased mortality (Dockery et al., 1993, Pope et al., 1995), respiratory
symptoms and illness (e.g., bronchitis and cough in children), and reduced lung function. The
relative risks associated with such exposures, although highly uncertain, appear to be larger than
those associated with short-term exposures.  Based on the available epidemiology and
consideration of relevant toxicologic and dosimetric information, staff concludes that significant,
and potentially independent, health consequences are associated with long-term PM exposures
(CD,  p!3-34)6.
       The staff notes that some health endpoints may better reflect the cumulative effects of PM
exposures over a number of years (CD, p. 1-13). In such cases, an expected annual average
standard would provide effective protection against long-term exposures to PM that exceed
several years. Requiring a much longer averaging time would complicate and unnecessarily delay
control strategies and attainment decisions.
       In addition, an annual standard would have the effect of controlling air quality across the
entire yearly  distribution of 24-hour PM2, concentrations to varying degrees, although such a
standard would not as effectively limit peak 24-hour concentrations as would a 24-hour standard.
        The seasonality of wintertime smoke and summertime regional acid sulfate and ozone suggest that an
intermediate averaging time might also be appropriate in future reviews.  Annual effects associated with acids, such as
those observed by Dockery et al (1996) and Raizenne et al (1996) might be interpreted as the result of repeated
seasonally high exposures.

-------
                                          VII-21
Thus, an annual standard could also provide protection from health effects associated with short-
term exposures to PM2 5.
       Based on the above considerations, the staff recommends consideration be given to
retaining an annual averaging time as a means of controlling both long- and short-term ambient
PM2 5 concentrations, and thus providing protection from health effects associated with both long-
and short-term exposures to PM2.5.
2.     Form ~ General  Approaches
       a.      24-Hour  PM: 5 Standard
       As part of the last review, the 24-hour standard was changed from a deterministic form, in
which the standard was not to be exceeded more than once per year, to a statistical form. The
statistical form selected permits no more than one expected-exceedance, averaged over 3 years.
The basis for this change in the form of the standard was that a statistical form can offer a more
stable target for control programs and, with reasonably complete data, is less sensitive to truly
unusual meteorological conditions than  the deterministic form (U.S. EPA,  1982b). The staff
continues to believe that this rationale is sound, but could be extended to consider alternatives
that have been developed in conjunction with the ongoing review of the ozone standard.  These
genera] approaches to defining the form of a 24-hour standard include multiple exceedances and
concentration percentile  forms, as discussed more specifically in the next section in conjunction
with the level of alternative standards.
       One additional approach that is also being considered for the ozone standard is some form
of averaging across multiple monitors.  In a previous review of the PM NAAQS, staff
recommended  consideration of a multiple monitor spatial average form in its earlier
recommendations for a secondary fine particle standard (EPA, 1982b). Such a form would better
focus risk management activities on reductions in area or regionwide fine particle concentrations.
Because the health effects information (as well as the risk assessment in Chapter VI) is keyed to
fluctuations in areawide  fine particle concentrations, such a form would also be more directly
related to reduction in population risk. Such an approach would not have to require multiple
monitors in all areas, assuming location  criteria specified sites representative of areawide
population exposures. If such an approach were adopted, consideration should be given to the

-------
                                         VII-22
extent to which peak localized exposures might result in unacceptable individual risk. Limits on
localized peak exposures might be provided through the 24-hour PM]0 NAAQS, if retained, which
is applied at each monitor individually. Appropriately located PM10 monitors would likely limit
not only coarse fraction particle levels but also fine particle levels that result from highly localized
emission sources.
       b.      Annual PM2 5 Standard
       As part of the last review, the annual standard was changed from a geometric mean to an
expected arithmetic mean of the daily measurements. This change in the form of the standard was
based on an arithmetic mean being 1) more directly related to dose, which is associated with
observed health effects, 2) more sensitive to repeated short-term peaks,  and 3) more consistent
with other annual NAAQS (U.S. EPA, 1982b). The staff continues to believe that this rationale is
sound and, thus, recommends that an expected arithmetic average form  be adopted for an annual
PM2 5 standard. Further, as discussed above for a 24-hour standard, staff recommends
consideration be given to adopting a spatial averaging approach for an annual PM2 5 standard.
3.      Level and Specific Forms
       In developing an approach to formulating recommendations on appropriate ranges of
levels and specific forms for 24-hour and annual PM2 5 standards, staff has taken into account the
following considerations:
1)     Recent new epidemiological studies are noteworthy in their scope and efforts to account
       for potential confounding and other uncertainties (e.g., characterization of exposure).
       However, each individual study has inherent and methodological limitations and
       interpretation of these findings is the subject of ongoing debate within the scientific
       community. Thus, the staff views its assessment of each individual study in the context of
       the overall body of epidemiological evidence (with mechanistic support from toxicological
       and dosimetry studies) and the consistency and coherence of results across studies and
       effects.
2)     As  noted in the last review, it continues to be the case that even  the best epidemiological
       studies have inherent limitations. Further, the available studies do not provide clear
       evidence of population thresholds of response. Thus, the staff recognizes that attempting

-------
                                           VII-23
       to identify "lowest observed effects levels" and adding margins of safety below such levels
       is not an appropriate approach in this case.  Instead, the staff has attempted to assess the
       nature of health effects and risks, and the associated uncertainties, along a continuum of
       exposures using the full range of available health and exposure data from studies identified
       in the CD as being appropriate for quantitative assessments.
 3)     Relative to other single pollutants for which NAAQS have been set, establishing
       appropriate ranges of levels for PM2 5 standards involves unusually large uncertainties.
       While recent studies help to reduce the uncertainties that were present in the last review,
       they do not change this basic observation relative to other NAAQS. To better address
       these uncertainties over time, the staff believes that research should continue into the more
       difficult problem of identifying and assessing potential health effects that may be
       associated with specific chemical and physical characteristics within the fine and coarse
       fractions of thoracic  particles.  However, even without any additional chemical-specific
       evidence, the staff believes that the large uncertainties inherent in setting PM2 5 standards
       do not preclude our identifying appropriate ranges of policy alternatives from which
       specific standards  can be selected to effectively and efficiently protect public health with
       an adequate margin of safety.
       Taking these considerations into account, the staff's approach to formulating
 recommendations on appropriate ranges of standard levels and forms for the recommended PM2 5
 indicator and averaging times is based on: 1) quantitative results from studies showing statistically
 significant associations between  ambient concentrations of fine fraction particles and health
 effects; 2) information on  U.S. air quality distributions and estimated background levels of PM25;
 3) examinations of the quantitative concentration-response relationships suggested by  specific
 epidemiological studies identified in the CD as appropriate for quantitative assessment purposes;
 4) quantitative risk analyses  that provide estimates of risk associated with air quality under "as is"
 conditions and attainment of current and alternative new PM2S standards; and 5) quantitative and
 qualitative consideration of the sensitivity of the risk estimates to key assumptions and inherent
 uncertainties in these analyses that affect the margins of safety associated with ranges  of standard
levels. This approach recognizes that final  decisions about appropriate PM standard levels and

-------
                                         VII-24
forms must draw not only on scientific information about health effects and risks, but also on
policy judgments about avoiding unacceptable risk from a public health perspective, addressing
the uncertainties inherent in the evidence and assessments, and establishing health protective
standards that serve as a meaningful guide to action in developing strategies to reduce
unacceptable health risks associated with anthropogenic contributions to ambient PM2 5 levels.
       These staff assessments and considerations are discussed below for both 24-hour and
annual PM2, standards. The following discussions are based on information in the CD and in
Chapters IV, V, and VI, and associated appendices, of this Staff Paper.
       a.      24-Hour PM2, Standard
       Several key observations discussed below frame the staffs thinking in defining a range of
24-hour PM2 5 levels and specific forms for the Administrator to consider in selecting an
appropriate standard that protects public health with an adequate margin of safety from adverse
health effects associated with ambient levels of PM2 v
•      Staff notes, based on consideration of the body of evidence as a whole as discussed
       throughout this Staff Paper, that PM2 s concentrations occurring in areas that attain the
       current PM,0 standards are  likely to be associated with increased risks of mortality,
       hospital admissions, and respiratory symptoms in various sensitive  subgroups.
       As a result, staff concludes that an appropriate range of 24-hour PM2 5 levels should result
in reductions in health risks relative to the risks associated with the current PM,0 standards.
Results estimated for the highest 24-hour PM2 5 level considered in the quantitative risk
assessment done for two example cities, 65 ug/m3, suggest that this level would result in some
reductions in risks relative to the current standard, with the amount of reductions likely to vary
from city to city.
       As would be expected from these risk results, a PM2, level of 65 ug/m3 is below the PM2 5
level that corresponds, based on a national average ratio, to  the current PM,0 standard level of
150 ug/m3 (i.e., a PM2 5 level of approximately 75 ug/m3). Staff notes that the use of a national
average ratio does not take into account the highly regional  nature of the ratio between PM2 5 and
PM,0.  In some Eastern areas, a PM2, level as high as about 100 ug/m3 could  correspond to the
current 24-hour PM10 standard level, whereas in some Western areas the corresponding PM25

-------
                                          VII-25
level could be as low as about 50 ug/m3. Thus, there is no "equivalent" level that applies
nationally based on information on ratios between PM25 and PM10. Alternatively, "equivalence"
with the current NAAQS could be considered on the basis of determining the PM2, standard level
that would result in approximately the same number of counties that would not be in attainment.
Consistent with the information provided in Table VII-1 for alternative PM,0 standards, Table
VII-2 presents the predicted total and regional distribution of the percentage of counties that
would not attain the listed alternative PM2 5 standards defined in terms of the current forms.7 By
comparison with Table VII-1, it can be seen that, based on the 1991-1993  PM,0 data used to
develop the two tables, a PM2, level of greater than 75 ug/m3 but well less than 100 ug/m3 is
predicted to result in approximately the same number of nonattainment counties as for the current
24-hour and annual NAAQS combined.
       Based on the above discussion, although there is no clear point at which "equivalence"
with the current NAAQS would be achieved, in staff s judgment consideration should be given to
a PM2 5 standard set below a level reflecting any type of approximate equivalence with the current
NAAQS.  Thus, staff recommends consideration be given to bounding the upper end of the range
below 75 ug/m\ at approximately 65 ug/m3.
•      Epidemiological studies reporting statistically significant associations were conducted in
       areas in which the mean 24-hour PM25 concentrations ranged from approximately 16 to
       30 ug/m3 for mortality studies, with hospital admissions and respiratory symptoms studies
       falling within this range (Table VI-2).
       Staff notes that these concentrations are relevant to considering a range of a standard, in
that these studies are generally interpreted as providing risk estimates for which there is greatest
confidence around the mean of the air quality  data.  However, as discussed in section V.E, there
are significant uncertainties in any given study due to model specification, exposure
misclassification, confounding, and other issues. Thus, staff believes that no one PM2 5
         The predicted comparison of counties not meeting alternative PM2 5 standards in Table VII-2 is derived from
an analysis that estimates PM2, air quality from the much larger PM1(1 data base in AIRS (Fitz-Simons et a)., 1996). As
such, these estimates are highly uncertain and are presented here for rough comparative purposes only.

-------
                                       VII-26
TABLE VII-2. PREDICTED PERCENTAGE OF COUNTIES NOT MEETING
                      ALTERNATIVE PM2.5 STANDARDS*

County Total
Annual




24-hr





Combined
Standards


Level of
Alternative
Standards**

25
20
15
10
All
482
2.5
8.7
36
84
SW
60
5.0
15
27
52
NW
80
3.8
8.8
28
65
CE
68
4.4
15
48
93
SE
99
0
4.0
26
95
NE
175
1.7
6.9
43
94

100
75
65
50
25
6.8
15
23
42
98
13
28
38
58
97
24
41
59
78
98
4.4
15
21
35
96
1.0
2.0
8.1
38
100
1.1
6.3
10
25
98

25/75
20/65
15/50
15
24
56.
28
38
58
41
59
78
16
24
56
2.0
10
50
6.3
11
50
  These estimates are based on a methodology that uses the PM,,, data in AIRS, together with more limited
  information on PM25/PM10 relationships, to predict which monitors might exceed a given PM25 alternative
  standard. Such estimates are highly uncertain and should be interpreted with caution. More speifically, the
  estimates are based on 1991-1993 data, using a 50% data completeness criteria, and applying the Appendix K
  missing data adjustment to account for less than every day sampling frequenciew. See staff analyses (Fitz-
  Simons et al., 1996) which discusses methodology for calculating estimated PM25 values.
  Based on current 1-expected-exceedance form of the 24-hour PM,0 NAAQS and current expected annual
  average of annual PM,,, NAAQS. at the highest monitor for each standard.

-------
                                          VII-27
concentration derived from any particular study should appropriately serve as the basis for the
level of a standard.
•      Results from the quantitative risk assessment presented in section VI.C suggest a pattern
       of a continuum of decreasing risk with lower levels of alternative PM2, standards,
       extending over and likely below the range of 65 to 25 ug/m3 PM2 5 included in the risk
       analyses.
       Based on the limited risk analyses for two example cities, using base case assumptions, a
24-hour PM2 5 standard of 25 ug/m' is estimated to reduce PM-related risks associated with short-
term exposures for the effects considered by roughly 70% - 85%, relative to risks associated with
attaining the current standards. Alternatively, at a 24-hour PM2 s level of 65 ug/m3, risks are
estimated to be reduced by roughly 10% and 40% for the Philadelphia and Los Angeles study
areas, respectively.  Putting these risk estimates into a broader perspective, these PM-related risk
reductions translate into  much smaller reductions relative to the total incidence of such effects
from any cause. Relative to total incidence, a PM2 5 standard of 25 ug/m3 may reduce total
mortality risk by roughly 1% to 2%, total hospital admissions by roughly 1% to 5%, and
respiratory  symptoms in  children by roughly 15% - 25%.  Alternatively, at a level of 65 ug/m3,
total mortality risk may be reduced by roughly 1% or less, total hospital admissions by roughly
2% to less than 1%, and  respiratory symptoms in children by roughly  2% to 13%.
       In terms of total incidence of effects upon attainment of alternative PM2 5 standards,
mortality incidence  associated with short-term PM exposures is estimated to range from roughly
300 to 400 events per year for the Philadelphia (population 1.6 million) and Los Angeles
(population 3.6 million) study areas, respectively, with a PM2 5 standard of 65 ug/m\  At a level
of 25 ug/m\ mortality incidence is estimated  to be roughly on the order of 100 events per year in
each study area. Estimated incidences of hospital admissions for respiratory and cardiac causes
are up to 70% greater than those of mortality events. Respiratory symptom incidence is judged to
be considerably more uncertain than estimates for the other effects, with roughly 10 to over 20
thousand events per year in the Philadelphia and Los Angeles study areas, respectively, at a level
of 65 ug/m\ and from roughly 3 to 6 thousand events per year, respectively, at a level of 25

-------
                                          VII-28
ug/m3. Thus, under base case assumptions, rough estimates of incidences are appreciably lower,
but not eliminated in going from a PM2 5 standard of 65 to 25 ug/m3.
       Staff emphasizes that these estimates are based on only two cities, include significant
uncertainties, and are sensitive to a number of assumptions that have been considered in the
integrated uncertainty analyses discussed in Chapter VI.  Thus, policy judgments that are based in
part on a consideration of such results should also take into account these uncertainties, critical
assumptions, and the public health implications of the estimated incidence rates.
•      Sensitivity analyses designed to address alternative assumptions in the risk analyses
       presented in section VI.C. suggest that estimated risks are sensitive to a number of
       assumptions, including in particular assumptions about the shape of concentration-
       response relationships and the ranges of air quality to which they are applied. The
       examination of concentration-response relationships that helped to frame the sensitivity
       and integrated uncertainty analyses provides information useful in identifying an
       appropriate PM2, range for consideration.
       For several alternative assumptions examined in the sensitivity and integrated uncertainty
analyses, relatively small to moderately large differences in estimated risks were predicted across
the range of alternative assumptions considered.  In examining relevant concentration-response
relationships using a variety of approaches, staff identified  alternative cutpoints for the lower end
of the range of air quality over which it may be appropriate to calculate increased risk from the
studies. From the short-term PM25 studies, staff identified concentrations of 10,  18, and 30 ug/m3
as potential cutpoints reflecting increased uncertainties in this lower range of observed
concentrations and inherent limitations in the data to detect any potential effects thresholds that
may be present within that range.  Relative to base case risk estimates, which do not assume any
effects threshold or cutpoint within the range of the data, mortality risks estimated from the
integrated uncertainty analysis are lower by as much as a factor of 2 across the range of
alternative assumptions considered. Thus, alternative assumptions, most notably about the shape
of the concentration-response relationship, can have significant impacts in lowering the estimated
total PM-related risk for "as is" air quality as well as for attainment of the current NAAQS and
alternative PM2, standard cases.
•      Several epidemiological studies reporting statistically significant effects include ranges of
       air quality that may approach estimates of background levels in some locations.

-------
                                          VII-29
       To serve as a meaningful guide to action in developing strategies to reduce unacceptable
health risks associated with anthropogenic contributions to ambient PM25 levels, staff believes
that a standard should be set at a level sufficiently above estimated background levels. As
discussed in Chapter IV, while estimated annual average PM2 5 background levels range from
approximately 2 to 5 ug/m3 in the East and 1 to 4 ug/m3 in the West, maximum annual 24-hour
fine particle concentrations of 15 to 20 ug/m3 are possible from background sources particularly
in Eastern areas. Further, staff notes that on a daily basis exceptional natural events such as forest
fires can result in even higher background concentrations, but such excursions are dealt with
through the natural events policy in implementing the standards.
       In taking into account the above observations, staff believes that the lower end of a range
of PM2 5 levels for the Administrator to consider in selecting an appropriate standard level should
be less than  25 ug/m3 but greater than 15 to 20 ug/m3.  While at 25 ug/m3 significant reductions
in risk may result, mortality studies show significant associations even when the observed means
of 24-hour PM2 5 concentrations in each of the study locations are approximately at or below 20
Mg/m3. Further, an assessment of concentration-response relationships below these levels
suggested consideration of possible thresholds at concentrations of 18 and 10 ug/m\ On the
other hand, staff believes an appropriate standard should be  sufficiently above estimated
background  levels so as to meaningfully facilitate the design and implementation of realistic air
quality management strategies.  Further, staff is mindful that the Act does not require that
NAAQS be  set at a zero-risk level, but rather at a level  that  avoids unacceptable risks and, thus,
protects public health with an adequate margin of safety.
•      With regard to specific alternative forms of 24-hour PM2 5 standards, staff analyses of
       predicted PM2 s concentrations provide an illustrative comparison of the impact in terms of
       the number of counties that would not attain alternative forms for an example standard
       level (Table VII-3).
       Table VII-3 compares the predicted impact of alternative exceedance-based forms
(ranging from 1 to 5 exceedances per year) and concentration percentile forms (including the
average nlh concentration percentile, with n ranging from the 95th to the 99lh percentile) for an
example 24-hour PM2,  standard level held constant at 50 pg/m3 (in conjunction with an annual

-------
                                          VII-30

PM25 standard set at 15 ug/m3).8 As can be seen from the table, the form of the standard can

result in significant differences in the number of areas that would not attain a given standard, such

that the degree of health protection provided by a standard is a function of both the level and form

of the standard.
      TABLE VII-3.  PREDICTED COMPARISON OF ALTERNATIVE FORMS
                           FOR A 24-HOUR PM2 5 STANDARD
                   (For counties meeting a 15 ug/m3 annual PM2, standard)
Alternative Forms
of Standard
1 Exceedance
2 Exceedance
3 Exceedance
4 Exceedance
5 Exceedance

Avg 99th percentile
Avg 98th percentile
Avg 95th percentile
Number of Counties
Projected to Meet 24-
hour Standard of
50Aig/m3
210
229
268
274
280
Number of Counties Not
Projected to Meet 24-
hour Standard of
50//g/m3
99
80
41
35
29

277
292
303
32
17
6
NOTE:  Of the 482 counties with at least 50% data completeness per quarter 1991-93, 309 meet the PM2.5 annual
       standard, and 173 do not.  Exceedance forms include the Appendix K missing data adjustment to account for
       less than every day sampling frequencies. See staff analyses in Fitz-Simons et al. (1996).
         As for Table VII-2, these staff estimates are based on predicting PM2, concentrations based on the available
PM|0 data base, and are highly uncertain. See staff analyses in Fitz-Simons et al. (1996).

-------
                                          VII-31
       In weighing all these factors and considerations outlined above, staff offers the following
conclusions and recommendations:
1)     The lower end of the range of consideration for a new 24-hour PM2 5 standard should be
       20 ug/m3. Considering a standard at this level would place significant weight on the
       consistency and coherence of the body of evidence as a whole, and on the results of
       quantitative analyses of concentration-response information and risks, even in light of
       inherent uncertainties in the analyses and alternative interpretations possible for each study
       considered independently. The staff believes that a 24-hour PM25 standard set at this
       level, while not likely to be risk-free, would be precautionary in nature in protecting
       against a full range of short-term effects associated with the identified sensitive subgroups
       of the population. A standard set at this level would give less weight to concerns that the
       relied-upon studies may not have completely controlled for all potential confounding
       variables nor fully accounted for all limitations in the exposure data.  Staff notes that this
       level is at the upper end of the range of uncertainty for peak 24-hour PM2 5 background
       concentrations.
2)     The upper end of the range of consideration for a new 24-hour PM2 5 standard should be
       approximately 65 ug/nr\ A standard set at or near this  level  would give significant weight
       to both the qualitative and quantitative uncertainties inherent in the most recent
       epidemiological studies, and, conversely, little weight to the quantitative assessments of
       the  evidence and associated risks.  Such a standard would likely provide increased
       protection relative to the current standard.
3)     In selecting a level for a 24-hour PM2 5 standard within this range, the staff suggests that
       the  Administrator also take into account the degree and nature of protection that would be
       afforded by a new annual PM2 5 standard. The joint protection provided by a suite of
       standards that includes both 24-hour and annual PM2 5 standards may be an important
       consideration in selecting the levels for each standard. One possible policy approach
       would be to view an annual PM2 s standard, as discussed below, as serving as the target
       for control programs designed to effectively lower the entire distribution of PM2 5
       concentrations, thus protecting not only against long-term effects but also short-term

-------
                                          VII-32
       effects as well.  With this approach, the 24-hour PM2 5 standard could be set so as to
       protect against the occurrence of peak 24-hour concentrations that would likely not be
       controlled in areas attaining a new annual PM2 5 standard. Thus, in conjunction with an
       annual PM2 5 standard, the Administrator may judge that the 24-hour standard should be
       set so as to limit only those peak 24-hour concentrations that are likely to persist  upon
       attainment of the annual standard.
4)     In selecting a form for a 24-hour PM2 5 standard within the range of alternative forms
       analyzed, the staff suggests that the Administrator give primary consideration to a
       concentration percentile form.  Concentration percentile forms are more stable and better
       take into account differences in sampling frequencies than the single (i.e., the current
       form) and multiple exceedance forms.  Further, consideration should be given to the
       relative health protection provided by alternative forms at a given level, considering the
       relative impact of alternative forms on the number of counties affected by a particular
       form, and, thus, the number of areas likely to experience reduced risks to public health as
       a result of attaining a given  standard level and form.
       b.     Annual PM: 5 Standard
       Similar to the approach outlined above for a 24-hour standard, the following observations
frame the staffs thinking in defining a range of annual PM2 5 levels:
•      Staff notes that annual PM2 5 concentrations occurring in some areas that attain the current
       PMIO  standards are likely to be associated with increased risk of mortality beyond that
       associated with short-term mortality effects, as well as possibly increases in doctor-
       diagnosed cases of acute bronchitis in children.
•      Further, as discussed above in the section on averaging times, an annual standard would
       have the effect of controlling air quality across the entire yearly distribution of 24-hour
       PM2, concentrations to varying degrees, such that an annual standard set an appropriate
       level could also provide protection from health effects associated with short-term
       exposures to PM25.
       Based on the above considerations, the staff recommends consideration be given  to use of
an annual averaging time as a means of controlling both long- and short-term ambient PM25

-------
                                         VII-33
concentrations, and thus providing protection from health effects associated with both long- and
short-term exposures to PM2 5.
       By comparing information in Tables VII-1 and VII-2, it can be seen that for the 1991-
1993 data presented in the two tables, an annual PM25 level of 25 ug/m3 is estimated to result in
approximately the same number of nonattainment counties as the current PM,0 NAAQS.  In
staffs judgment consideration should be given to an annual PM2 5 standard set below a level
reflecting approximate equivalence with the current annual NAAQS.  Thus, staff recommends
consideration be given to bounding the upper end of the range below 25 ug/m3, at approximately
20 ug/m3.
       Alternatively, in viewing an annual standard as creating a target for control programs
designed to effectively lower the entire distribution of PM2 s concentrations, staff concludes that
an appropriate range of annual PM2 5 levels for such a standard should result in reductions in
health risks relative to the risks associated with the combination of current 24-hour and annual
PM,0 standards. Under this approach, a comparison of Tables VII-1 and VII-2 suggests that an
annual PM2 5 standard level of less than 20 ug/m3 would be needed to result in the same number of
predicted nonattainment counties as for the combination of current 24-hour and annual PM10
NAAQS.
•      Based on the long-term mortality study used in the quantitative risk assessment (Pope et
       al., 1995), a statistically significant association was observed across 151 cities in which the
       annual PM2 5 concentrations ranged from approximately 9 to 34 ug/m3 (Table VI-2); a
       somewhat similar range is estimated from the long-term studies of lung function
       decrements and  doctor-diagnosed bronchitis in children (Table V-13).
       Staff notes that these concentrations are relevant to considering a range for an annual
standard, although, as discussed in Chapter VI and Appendix E, staff recognizes that uncertainty
in the concentration-response relationships increase at the lower end of the range of data due in
part to inherent limitations in discerning any potential effects threshold that may actually be
present. In examining the concentration-response relationships for long-term mortality from the
Pope et al. (1995) study, as well  as from the more uncertain Dockery et al. (1993) study, possible
concentration cutpoints  at which effects threshold may potentially exist were identified (Chapter
VI and Appendix E). The lowest such cutpoint was 12.5 ug/m3, based on inherent limitations of

-------
                                         VII-34
the data for discerning effects thresholds, and a cutpoint of 15 |ag/m3 was identified based on
visual inspection of the data. The minimum mean concentrations in these two studies were 18
ug/m3.
•      The body of evidence from long-term exposure studies, together with results from the
       quantitative risk assessment presented in section VI.C, suggests a pattern of a continuum
       of decreasing risk with lower levels of alternative annual PM2, standards, likely extending
       below the range of concentrations included in the analyses, 15 and 20 ug/m3 PM25 annual
       average.
       Based on these limited analyses for two example cities, and applying only base case
assumptions, the analyses estimate that an annual PM25 standard of 15 ug/m3 may reduce PM-
related risks for mortality associated with long-term exposures by roughly 30 and 60% relative to
risks associated with attaining the current NAAQS for Philadelphia and Los Angeles study areas,
respectively. Alternatively, at a PM2 s level of 20 ug/m3, reduction in risks associated with long-
term exposure in Los Angeles county are estimated at 30%; staff notes that this level does not
result in any estimated risk reduction in Philadelphia county because the current annual mean in
Philadelphia is below this level. Putting these risk estimates into a broader perspective, these PM-
related risk reductions translate into much smaller reductions relative to the total incidence of
such effects from any cause. Relative to total incidence, an annual PM25 standard of 15 ug/m3
may reduce total mortality risk associated with long-term exposures by roughly land 5% for the
Philadelphia and Los Angeles study areas, respectively. Alternatively, at a level of 20 ug/m3, total
mortality risk for Los Angeles county may be reduced by roughly 2%.
       In terms of total incidence of effects upon attainment of alternative annual PM2 5
standards, mortality incidence associated with long-term exposures to PM is estimated to range
from roughly less than 1000 to about 1500 events per year for the Philadelphia and Los Angeles
study areas, respectively, with an annual PM2 5 standard of 20 ug/m3, to roughly on the order of
half as many events per year for each study location at a level of 15 ug/m3.  Thus, under base case
assumptions, rough estimates of incidences are appreciably lower, but not eliminated, in going
from an annual PM2, standard of 20 to 15 ug/m3.
       Staff again emphasizes that these estimates are based on only two cities, include significant
uncertainties, and are sensitive to a number of  assumptions that can not be fully addressed by

-------
                                           VII-35
additional analysis of available data. Thus, policy judgments that are based in part on a
consideration of such results should also take into account these uncertainties, inherent limitations
in available data and analyses, and the public health implications of the estimated incidence rates.
•      Sensitivity analyses designed to address alternative assumptions in the risk analyses,
       presented in section VI.C. suggest that estimated long-term risks are sensitive to a number
       of assumptions, including in particular assumptions about the shape of concentration-
       response relationships and the ranges of air quality to which they are applied and historical
       air quality information used in the analysis. The examination of concentration-response
       relationships and historical air quality that helped to frame these particular sensitivity
       analyses provides information useful in identifying an appropriate PM2.5 range for
       consideration.
       Based on  an analysis of long-term mortality using the alternative  cutpoints discussed in
Chapter VI, staff notes that estimated risk for Philadelphia County is roughly 50% lower than the
base case estimate if a  12.5 ug/m3 cutpoint is  applied.  Similarly, applying a cutpoint of 15 iig/m*
reduces estimated long-term mortality risk by over 75%, while applying  a cutpoint of 18  ug/m3
•results in an estimate of no long-term mortality risk for "as is" air quality in Philadelphia County.
Further, by assuming higher historical PM2 5 concentrations than were reported in the Pope et al.
(1995) study, estimated risk would be significantly lower than the base case estimate (Appendix
F). Thus, alternative assumptions about the shape  of the long-term PM concentration-response
relationships  and  historical air quality can have very significant impacts on  the estimated risk
reductions associated with attaining alternative PM25 standards.
       In taking into account the above observations, staff believes that  the lower end of a range
of PM2 5 levels for the Administrator to consider in selecting an appropriate annual standard level
should be consistent with the lowest cutpoint for a possible threshold derived from an examination
of the long-term mortality concentration-response relationships, 12.5 ug/m\  Staff believes that
such an annual level is sufficiently above estimated annual PM2 5 background levels as to serve as
a meaningful standard to facilitate the design and implementation of realistic air quality
management  strategies. Further, as  noted above, staff is mindful that the Act does not require
that NAAQS  be set at a zero-risk level, but rather at a level that avoids unacceptable risks and,
thus, protects public health with an adequate margin of safety.

-------
                                          VII-36
       In weighing all these factors and considerations outlined above, staff offers the following
conclusions and recommendations:
1)     The lower end of the range of consideration for a new annual PM2 5 standard should be
       12.5 ug/m3. Considering a standard at this level would place significant weight on the
       consistency and coherence of the body of evidence as a whole, and on the results of
       quantitative analyses of concentration-response information and risks, even in light of
       inherent uncertainties in the analyses and alternative interpretations possible for the
       relevant studies.  The staff believes that an annual PM2 5  standard set at this level, while
       not likely to be risk-free, would be precautionary in nature in protecting against long-term
       mortality effects and other long-term morbidity effects such as lung function decrements
       and doctor-diagnosed bronchitis in children. A standard set at this level would give less
       weight to concerns that the relied-upon studies may not have completely controlled for all
       potential confounding variables nor fully accounted for all limitations in the exposure data.
2)     The upper end of the range of consideration for a new annual PM2 5 standard should be 20
       ug/m\ A standard set at or near this level would give significant weight to both the
       qualitative and quantitative uncertainties inherent in the long-term epidemiological studies,
       and, conversely, little weight to the quantitative assessments of the evidence and
       associated risks.  Such a standard would likely provide some increased protection relative
       to the current annual standard.
3)     As discussed above, in selecting a level for an annual PM2 5 standard within this range, in
       conjunction with a 24-hour PM25 standard, staff suggests that the Administrator take into
       account the joint protection likely to be afforded by both standards.  In an approach that
       viewed the annual PM2, standard as the primary target for control programs designed to
       effectively lower the entire distribution of PM2 5 concentrations, the Administrator may
       choose to consider an annual standard from the lower end of this range.  Correspondingly
       a 24-hour PM2,  standard could be set so as to protect against the occurrence of peak 24-
       hour concentrations that would likely not be controlled by areas attaining a new annual
       PM2, standard.  For example, an annual PM2, standard at 15 ug/m1 may be expected to
       result in substantially reduced 24-hour levels, potentially limiting the second highest 24-

-------
                                          VII-37
       hour levels to less than about 50 ug/m3 in approximately 90% of the areas, thus adding to
       the protection against short-term effects afforded by a 24-hour standard (SAI, 1996).
D.     Alternative PMIO Standards for Control of Coarse Fraction Particles
1.     Averaging Time
       If fine particle standards are adopted, the major function of the PM)0 standard would be to
protect against the known and anticipated effects associated with coarse fraction particles in the
size range of 2.5 to 10 urn.  As noted above, coarse fraction particles are plausibly associated with
certain effects from both long and short-term exposures.  Some epidemiologic evidence suggests
increased asthma and upper respiratory infections may be associated with daily increases in PM^
that was dominated by coarse fraction particles (Gordian et al, 1996), while another study
suggests smaller relative risks of bronchitis symptoms after daily episodes of very high fugitive
dust (Hefflin et al, 1994).  Both studies reported multiple exceedences of the current 24-hour
NAAQS  with PM!0 peaks exceeding 900 ug/m3.  The potential build up of insoluble coarse
fraction particles in the lung after long-term exposures to high levels should also be considered.
       These studies show an important characteristic of significant coarse particle events.  In a
number of Western areas, multiple exceedences occur in relation to high winds increasing
emissions from naturally occurring or human-disturbed surfaces. In the Gordian et al. (1996)
study, the worst levels occurred in relation to a volcanic eruption. In a  number of cases, such
excursions are exempted from control by the natural events policy. In some areas, variations in
annual rainfall or windspeed cause year-to-year changes in dust emissions, making implementation
and assessment of control strategies more difficult.   It is therefore appropriate to consider which
combination of averaging time and form might provide a more robust target for practical coarse
particle controls.  In this regard, basing control on an annual standard alone or in combination
with a 24-hour standard with multiple exceedences may provide adequate protection from
potential  long- and short-term effects of coarse fraction particles.

-------
                                          VII-38
2.     Level and form for alternative averaging times
       a.      Annual PMIO Standard
       The nature of the more limited information for coarse fraction particles means the
approach for selecting a level of the standard should be less like the recommended approach for
fine particles, and more related to the approach taken in the last review for PM,0. In that
approach, evidence from limited quantitative studies was used to select a range, with support
from the qualitative literature used to support decisions within the range (EPA, 1982b, 1986).
       The major quantitative basis for the level of the current annual PMIO standard was a study
of children by Ware et al. (1986), conducted as part of the Harvard Six City series.  This study
has been supplemented in the recent literature by a follow-up  long-term cohort study of acute
bronchitis in children (Dockery et al., 1989).  This study found somewhat better associations with
PM|«j than with PM25 over the entire cohort, but  a direct comparison with coarse fraction particles
was not presented.  However, still more recent studies found bronchitis symptoms in a larger
cross sectional comparison to be unrelated to somewhat lower coarse particle concentrations than
found in some of the six cities (Dockery et al, 1996). It is possible, but not conclusive, that
coarse fraction particles, in combination with fine particles, may have influenced the observed
effects,  at least at the levels in the three most polluted cities in the study. From an
exposure/deposition perspective, it is possible that cumulative deposition of coarse fraction
particles could be elevated in children, who are more prone to be active outdoors than sensitive
adult populations.  Based on the original study by Ware et al. (1986), in the  last review, staff
recommended consideration  that the lower bound of the range for the annual standard be set at 40
ug/m3 (EPA, 1986).
       Qualitative evidence of other long-term coarse particle effects, most notably from long-
term buildup of silica containing materials, supports the need  for a long-term standard, but does
not provide  evidence of effects below this range (CD, p 13-79). Staff concludes that the
qualitative evidence with respect to biological aerosols (13-79) also supports the need to limit
coarse materials, but should not form the major basis for a national standard. The nature and
distribution  of such materials, which vary from endemic fungi (e.g. valley fever) to pollens larger
than 10 urn are not appropriately addressed by traditional air pollution control programs.

-------
                                          VII-39
       A PM10 standard in the range of 40 to 50 ug/m3 (current level) would also provide
substantial protection against the effects of 24-hour exposures associated with asthma and upper
respiratory infections. The national mean ratio for the second highest 24-hour concentration in a
year to the annual mean is 2.41 (SAI, 1996).  This indicates that the mean second highest 24-
hour concentrations associated with such a range (about 95 to 120 u.g/m3)  would be well below
the current standard. Peak levels at the worst sites could still exceed the level of the current 24-
hour standard.  Additional information on the relative short-term protection afforded by the
current  annual standard is summarized in the discussion below.
       Staff recommends that consideration be given to adopting an annual PM)0 standard in the
range of 40 to 50 |ug/m3 to protect against the long- and short-term effects of coarse fraction
particles. Such a standard would provide a more robust target for coarse particle controls that
would be less sensitive to episodic natural events.
       b.     24-Hour PM10 Standard
       Consideration should also be given to a 24-hour standard for coarse fraction particles as
measured by PM10.  The level of the current 24-hour PM10 standard (150 |ag/m3) was based in
large measure on the London mortality and morbidity studies (EPA, 1982b).  As noted above,
staff believes that fine particles are  a better surrogate for such effects. The main quantitative basis
for a short-term standard is provided by the two fugitive dust studies referenced above.  Because
these studies reported multiple large exceedences of the current 24-hour standard they suggest no
need to lower the level of the standard below 150 ug/m3.
       If a 24-hour PM10 standard is retained in conjunction with a fine particle standard,
consideration should be given to maintaining  the current level and revising the PM10 standard to a
more robust form.  Such forms would be less sensitive to naturally occurring episodes.  Staff have
conducted analyses of several alternative forms for a PM10 standard, similar to the analyses for
alternative forms for a PM2 5 standard as discussed above.  Table VII-4 compares the impact of
alternative exceedance-based forms (ranging  from 1 to 5 exceedances per year) and concentration
percentile forms (including the average n1*1 concentration percentile, with n ranging from the 95th
to the 99th percentile) for an example 24-hour PM10 standard level held constant at 150 ug/m3 (in
conjunction with an annual PM10 standard set at 50 ug/m3). As can be seen from the table, the

-------
                                         VII-40
analysis suggests that a 50 jag/m3 annual standard would limit 24-hour exceedences in all but nine
of the sites to 5 or less (i.e., only nine sites would not attain a standard with a 5-exceedance
form). Staff is examining alternative analytical  approaches to provide additional insight into the
relative protection afforded by these forms.
       Because of the episodic nature of coarse particle excursions, the staff recommends that if a
24-hour standard is adopted, consideration should be given to one of the alternative more robust
forms presented in Table VII-4, with or without an accompanying annual PM|0 standard.
3.     Summary of Coarse Fraction (PM]0) Standard Conclusions and Recommendations
       Staff conclusions and recommendations are as follows:
1)     As an indicator for coarse fraction particles, in conjunction with a PM25 standard, the
       basis and purpose for the PM10 standards have been altered.
2)     Staff recommends consideration of an annual PM10 standard in the range of 40 to 50
       ug/m3 to protect against both the short- and long-term effects of coarse fraction particles.
       An annual standard would provide a robust target for effective coarse particle control and
       monitoring strategies.
3)     Consideration should also be given to a 24-hour PMIO standard of 150 ug/m3 with a
       revised, more robust form selected from the range of alternatives presented in Table VII-
       4.  Additional analyses of these forms are needed before more definitive recommendations
       can be made.

-------
                                         VII-41
       TABLE VII-4.   COMPARISON OF ALTERNATIVE FORMS FOR A
                              24-HOUR PMIO STANDARD
                   (For counties meeting a 50 ug/m3 annual PM,0 standard )
Alternative Forms
of Standard
1 Exceedance
2 Exceedance
3 Exceedance
4 Exceedance
5 Exceedance

Avg 99th percentile
Avg 98th percentile
AvgJJSth percentile
Number of Counties
Projected to Meet 24-hour
Standard of 150//g/m3
425
433
451
455
462
Number of Counties Not
Projected to Meet 24-hour
Standard of 150/zg/m3
46
38
20
16
9

455
467
471
16
4
0
NOTE:  Of the 482 counties with at least 50% data completeness per quarter 1991-93,471 meet the PMIU annual
       standard, and 11 do not. Exceedance forms include the Appendix K missing data adjustment to account for less
       than every day sampling frequencies. See staff analyses in Fitz-Simons et a). (1996).
E.     Summary of Key Uncertainties and Research Recommendations
       Staff believes it is important to emphasize the unusually large uncertainties associated
with establishing standards for PM relative to other single component pollutants for which
NAAQS have been set. The CD and this Staff Paper note throughout a number of unanswered
questions and uncertainties that remain in the scientific evidence and analyses as well as the
importance of ongoing research to address these issues.  Prior to summarizing staff
recommendations on the primary PM NAAQS in the next section, this section summarizes key
uncertainties and related staff research recommendations.

-------
                                         VII-42
1)     One of the most notable aspects of the available information on PM is the lack of
       demonstrated mechanisms that would explain the mortality and morbidity effects
       associated with PM at ambient levels reported in the epidemiological literature.  The
       absence of such mechanistic information limits judgments about causality of effects and
       appropriate concentration-response models to apply in quantitatively estimating risks.
       Building on promising preliminary findings from ongoing research involving more
       representative animal models and particle mixes and levels, staff believes there is an urgent
       need to expand ongoing research on the mechanisms by which PM, alone and in
       combination with other air pollutants, may cause health effects at levels below the current
       NAAQS.
2)     Uncertainties and possible biases introduced by measurement error in the outdoor
       monitors, including both the error in the measurements themselves and the error
       introduced by using central monitors to estimate population  exposure, contributes to
       difficulties in interpreting the epidemiological evidence. To  address these concerns,
       additional research into improved continuous sampling and analyses methods, together
       with the use of a research-oriented ambient monitoring network and personal monitors to
       better characterize relationships between personal exposure  and outdoor/indoor air
       quality, is needed for PM components as well as for other criteria pollutants. For example,
       monitoring techniques that allow new epidemiological studies to address not only size
       fractionation and improved measurements of semi-volatile particles but also particle
       number and surface area  will be important to isolate key components of fine and coarse
       fraction particles.  Further, examination of potential exposure to ultrafine particles near
       highways and other possible sources, for example, is important to determine the extent to
       which these materials persist long enough to present significant exposure to sensitive
       population groups.
3)     Inherent in epidemiological studies such as those cited in this review is the question as to
       whether or to what extent the observed effects attributed to PM exposures are
       confounded by other pollutants commonly occurring in community air, such as SO2,
       ozone, NO2, and CO. In  particular, a number of authors conducting reanalyses of

-------
                                          VII-43
       mortality studies within a given city, most notably for Philadelphia, have demonstrated that
       it may not be possible to separate individual effects of multiple pollutants when those
       pollutants are highly correlated within a given area. Based on its assessment of available
       information regarding potential  confounding within and across a number of areas with
       differing combinations of pollutants, as recommended in the HEI reanalysis report, the CD
       concludes that in general the reported PM effects associations are valid and not likely to
       be seriously confounded by copollutants.  Nevertheless, additional research and analyses
       are important to better characterize the  extent to which PM-related effects may be
       modified by the presence of other copollutants in the ambient air.
4)     Although staff has concluded that it is more likely than not that fine fraction particles play
       a significant role in the reported health effects associations, identification of specific
       components and/or physical properties of fine particles which are associated with the
       reported effects is very important for both future reviews of the standards and in
       development  of efficient and effective control strategies for reducing health risks.
       Epidemiological and toxicological research is needed to isolate key components (e.g.,
       nitrates, sulfates, organics, metals, ultra fine particles) and/or characteristics of fine
       particles, as well as to identify the nature  and extent of subpopulations most susceptible to
       the adverse effects associated with such components and/or characteristics.  Such research
       is critical in addressing uncertainties in estimating risk reductions likely to be achieved by
       alternative fine particle standards and new implementation strategies.
5)     Uncertainties in the shape of concentration-response relationships, most specifically
       whether linear or threshold models are more appropriate, significantly affects the
       confidence with which risks and risk reductions can be estimated.  Mechanistic and
       epidemiological research highlighted above would likely help reduce such uncertainties.
6)     Unaddressed  confounders and methodological uncertainties inherent in epidemiological
       studies of long-term PM exposures limit interpretations and conclusions that can be drawn
       with regard to associations between PM and chronic health effects. Additional research
       and analysis are needed to reduce the uncertainties related to the appropriate exposure

-------
                                          VII-44
       periods and historical air quality to consider in evaluating such studies, and to better
       address life-style and other potentially important cofactors.
7)     An important aspect in characterizing the nature of the mortality risk associated with
       short- and long-term exposures to PM, from a public health perspective, is the extent to
       which lifespans are being shortened. Available epidemiological evidence provides a very
       limited basis for testing hypotheses as to whether and to what extent lifespans are
       shortened by only a few days or by years. More research is needed to quantitatively
       characterize the degree of prematurity of deaths associated with exposures to PM.
8)     The characterization of annual and daily background concentrations likely to occur across
    '   the U.S. contains significant uncertainties. Additional air quality monitoring and analyses
       that improve these background characterizations would help to reduce the uncertainties in
       estimating health risks relevant to standard setting, i.e., those risks associated with
       exposures to PM in excess of background levels.
9)     Despite long-standing staff recommendations for a comprehensive examination of the
       effects associated with exposures to coarse fraction particles, there continues to be a  lack
       of animal, clinical, and community studies in this area. Such research would potentially
       provide both qualitative and quantitative information that could allow for the
       establishment of a coarse fraction particle standard rather than continued reliance on  a
       PMIO standard as the means to control exposures to coarse fraction particles.
F.     Summary of Staff Recommendations on Primary PM NAAQS
       The major staff recommendations and supporting conclusions from sections  VII. A-D are
briefly summarized below:
1)     The current PM standards should be revised.  As the Criteria Document concludes,
       current evidence provides ample reason to be concerned that there are detectable health
       effects attributable to PM at levels below the current NAAQS. Given the nature and
       potential  magnitude of the public health risks involved, staff believes revision of the
       current standards is clearly appropriate. The health effects reported, ranging from
       premature mortality to various  measures of morbidity, including increased hospital

-------
                                          VIMS
       admissions, aggravation of existing respiratory disease, including asthma, and decreased
       lung function, include effects that are clearly adverse to public health.
2)     Ambient particles capable of penetrating to the thoracic region, including both the fine and
       coarse fractions of PM,0, should continue to be the focus of PM standards. Staff
       concludes that these thoracic particles represent the greatest risk to health, and that the
       previous recommendations for 10 urn as the  appropriate cutpoint for such particles remain
       valid.
3)     The fine and coarse fractions of PM)0 should be considered as two separate pollutants
       based on the recent health evidence, the fundamental differences between fine and coarse
       fraction particles, and implementation experience with PM!0. The staff concludes that the
       available information is sufficient to support separate indicators for these separate
       pollutants.  Further, while it is difficult to distinguish the effects of fine or coarse fraction
       particles from those of PM,0, consideration of comparisons between fine and coarse
       fractions suggests that fine fraction particles are a better surrogate for those particle
       components linked to mortality and morbidity effects at levels below  the current standards.
       In contrast, coarse fraction particles are more likely linked with certain effects at levels
       above those allowed by the current PMIO  standards.  In examining alternative approaches
       to increasing the protection afforded by PM,0 standards, the staff concludes that reducing
       the  levels of the current PMIO standards would  not provide the most effective and efficient
       protection from these health effects.
4)     A 2.5 urn cutpoint (i.e., PM2 5) should be used  as the indicator for fine fraction particles,
       and the current PM10 indicator should now be used as the indicator for the coarse fraction
       particles. A PM2 5 indicator for fine fraction particles is specifically recommended based
       primarily on consistency with the health effects literature and the suitability and availability
       of ambient  monitors.  The recommendation for PM10 as the indicator for coarse fraction
       particles is based on the very limited data base and monitoring capabilities directly for
       coarse fraction particles, as well as the applicability of the existing PM)0 monitoring
       network. Further, staff concludes that use of sulfate or other chemical class indicators is
       not  advisable on the basis of this review.

-------
                                         VII-46
5)     Staff recommends that new PM2 5 standards be established for two averaging times.
       a)      Annual and 24-hour PM2 5 standards should be established as the most appropriate
              standards to address health effects associated with both short-term (from less than
              1 day up to 5 days) and long-term (from months to years) exposures to fine
              fraction particles.
       b)      Staff recommends consideration of more robust forms for a 24-hour standards
              (especially concentration percentile forms), averaged over three years. In addition,
              staff recommends consideration be given to using the average of multiple monitors
              representative of population exposure as part of the form of the annual and/or 24-
              hour standards.  Staff also recommends the retention of the current expected
              arithmetic average form of the annual standard.
       c)      Staff recommends that the Administrator consider selecting the level of a new 24-
              hour PM2 j standard from the range of 20 ug/m3 to approximately 65 ug/m3, and
              the level of a new annual PM25 standard from the range of 12.5 ug/m3 to
              approximately 20 ug/m3. These recommended ranges are based primarily on
              quantitative results from epidemiological studies, examinations of concentration-
              response relationships suggested by these studies, quantitative risk assessment,
              including consideration of the sensitivity of the  risk estimates to key assumptions
              and inherent uncertainties in the underlying data and analytic approaches, and
              relevant policy considerations based on air quality analyses. In recommending
              these ranges, staff is mindful that the Clean Air Act does not require that NAAQS
              be set at zero-risk levels, but rather at level that avoid unacceptable risks to public
              health,  thus protecting public health with an adequate margin of safety. Further, in
              selecting specific levels for PM2, standards, staff recommends  that the
              Administrator consider the joint protection afforded by both the 24-hour and
              annual  standards.  The recommended approach is to view an annual PM2 5 standard
              as the primary target for control programs designed to effectively lower the entire
              distribution of PM2 s concentrations, with a corresponding 24-hour PM2 5 standard

-------
                                         VII-47
              set so as to protect against the occurrence of peak 24-hour concentrations that
              would likely not be controlled by areas attaining such a new annual PM2 5 standard.
6)     Staff recommends that an annual PM10 standard be retained, alone or in combination with
       a 24-hour PM,0 standard.
       a)      Staff recommends that the Administrator consider selecting the level of an annual
              PM,0 standard from the range of 40 ug/m3 to 50 ug/m3, with an expected
              arithmetic mean form. Such a standard would reflect the range considered in the
              last review, and would protect against the principal effects of concern, including
              effects associated with both short- and long-term exposures to PM such as
              aggravation of asthma, upper respiratory infections, and bronchitis in children, as
              well as the long-term build-up of insoluble coarse fraction particles in the lung.
       b)      Further, if a 24-hour PM,0 standard is retained, staff recommends retention of the
              current level of 150 ug/m3, but with a revised, more robust form to better address
              the episodic nature of coarse particle excursions.

-------
                                        VIII-1
VIII.  CRITICAL ELEMENTS IN THE REVIEW OF THE SECONDARY STANDARD
       FOR PARTICIPATE MATTER
A.     Introduction
       This chapter presents critical information for the review of the secondary NAAQS for
participate matter drawing upon the most relevant information contained in the CD and other
significant reports. The welfare effects of most concern for this review are visibility
impairment,  soiling, damage to man-made materials, and damage to and deterioration of
property.  For each category of effects, the chapter presents (1) a brief summary of the
relevant scientific information and (2) a staff assessment of whether the available information
suggests consideration of secondary standards different than the recommended primary
standards. Staff conclusions and recommendations  related to the secondary standard  for PM
are presented at the end of the chapter.
       It is important to note that the discussion of fine particle effects on visibility in
chapter 8 of the CD is  intended to only include information complementary to several other
significant reviews of the science of visibility.  These reports include the 1991 report of the
National Acid Precipitation Assessment Program, the  National Research Council's Protecting
Visibility in National Parks and Wilderness Areas (1993),  and EPA's 1995 Interim Findings
on the Status of Visibility Research. Where appropriate, this chapter of the staff paper will
cite the above reports directly.
       The chapter does not address the effects of particles on climate change.  As discussed
in the criteria document, particles (in the submicrometer size range) can result in
perturbations of the radiation field that are generally expressed as radiative forcing.
Radiative forcing due to aerosols has a cooling effect  on climate through the reflection of
solar energy.  This is in contrast to "greenhouse gas"  that produces a positive long wave
radiative forcing which has a warming effect. Given  the complex interaction of these two
phenomena and the present state of the science,  it is the staff's judgment that these effects
should not be addressed in this paper, but should instead be considered in the broader context
of global climate change.

-------
                                         VIII-2
B.     Effects of PM on Visibility
1.      Definition of Visibility and Characterization of Visibility Impairment
       Visibility can be defined as the degree to which the atmosphere is transparent to
visible light (NRC, 1993; CD, 8-3).  Visibility effects are manifested in two principal  ways:
(1) as local impairment (e.g., localized hazes and plumes) and (2) as regional haze.  These
distinctions are significant both to the ways in which visibility goals may be set and air
quality management strategies may be devised.
       Local-scale visibility degradation has been generally defined as impairment that is
"reasonably attributable" to a single source or group of sources.  A localized haze may be
seen as a band or layer of discoloration appearing well  above the terrain, and may result
from complex local meteorological conditions. "Reasonably attributable" impairment may
include contributions to local hazes by individual or several identified sources.  Plumes are
comprised  of smoke, dust,  or colored gas that obscure the sky or horizon relatively near
sources.  Sources of locally visible plumes, such as the plume from an industrial facility or a
burning field, are often easy to identify. Overall,  visible plumes appear to be minor
contributors to visibility impairment in Class I areas (i.e., certain national parks, wilderness
areas, and  international parks as described  in section 162(a) of the Clean Air Act) (NRC,
1993).
       The second type of impairment, regional haze,  is produced from a  multitude of
sources and impairs visibility in every direction over a large  area, possibly over several
states.  Regional haze masks objects on the horizon and reduces the contrast of  nearby
objects.  The formation, extent, and intensity of regional  haze is a function of meteorological
and chemical processes, which sometimes cause fine particle loadings to remain suspended in
the atmosphere for several  days and to be transported hundreds of kilometers from their
sources (NRC,  1993).  It is this second type  of visibility degradation that is principally
responsible for impairment in national parks and wilderness areas across the country (NRC,
1993). Visibility in urban  areas may  be dominated by local  sources,  but may be significantly
affected by long-range transport of haze as well.  Fine particles transported from urban areas
in turn may be significant contributors to regional-scale impairment in Class I areas.

-------
                                          VIII-3
2.     Significance of Visibility to Public Welfare
       Visibility is an air quality-related value having direct significance to people's
enjoyment of daily activities in all parts of the country. Survey research on public awareness
of visual air quality using direct questioning typically reveals that 80% or more of the
respondents are aware of poor visual air quality (Cohen et al.,  1986).  Individuals value good
visibility for the well-being it provides them directly, both in the places where they live and
work, and in the places where they enjoy recreational opportunities. Millions of Americans
appreciate the scenic vistas in national parks and wilderness areas annually. Visibility is also
highly valued  because of the importance people place on protecting nationally significant
natural areas,  both now and in the future (i.e., preservation value).  Many individuals want
to protect such areas for the benefit of future generations, even if they personally  do not visit
these areas frequently  (Chestnut et al., 1994). Tracking changes in visibility provides one
measure of the success of efforts to  protect such areas from environmental degradation.
Society also values visibility because of the significant role it plays in air transportation.
Serious episodes of visibility impairment  can lead to increased  risks in the  air transportation
industry, particularly in urban areas with high traffic levels (U.S. EPA, 1982b).
       Many contingent valuation studies have been performed in an attempt to quantify
benefits (or individuals' willingness  to pay) associated with improvements in current  visibility
conditions.  The results of several studies are presented in CD  table 8-5 (CD, 8-83), table 8-
6 (CD, 8-85), and in table VIII-1 (Chestnut et al., 1994).  Past studies by Schultze (1983)
and Chestnut and  Rowe (1990b) have estimated the preservation values associated with
improving the visibility in national parks  in the Southwest to be quite significant,  on the
range of approximately $2-6 billion annually (CD, 8-84).  Another recent study estimates
visibility benefits primarily in the eastern U.S. due to reduced  sulfur dioxide emissions under
the acid rain program  also to be quite significant, in the range  of $1.7 - 2.5 billion annually
by the year 2010 (Chestnut et al., 1994).
3.     Mechanisms of and Contributors to Visibility Impairment
       Visibility impairment has been considered the  "best understood and most easily
measured  effect of air pollution" (Council on Environmental Quality, 1978).  It is caused by
the scattering and  absorption of light by particles and gases in the atmosphere.  It is  the most

-------
                                  VIII-3a
     TABLE VIII-1.
COMPARISON OF RESIDENTIAL VISIBILITY
VALUATION STUDY RESULTS
Study
Eastern CVM
Studies

McClelland et al.


Tolley et al.


Tolley et al


Tolley et al


Tolley et al


Tolley et al


Tolley et al


Tolley et at


Rae
California CVM
Studies

Brooksire et al.


Loehman et al
California Property
Value Study
Trijonis et al
Trijonis et al
City


Atlanta and
Chicago

Chicago


Atlanta


Boston


Mobile


Washington,
DC


Cincinnati


Miami


Cincinnati



Los Angeles


San Francisco


Los Angeles
San Francisco
Mean WTP
($1990)


Unadj. $39
Partial $25
Full $18
-$318
$305
$379
-$265
$255
$381
-$1%
$187
$231
-$212
$227
$266
-$314
$323
$410
-$78
$77
$86
-$134
$120
$141
$175



$115
$294
$161
-$186
$109




Starting VR
(miles)


17.6


9
9
9
12
12
12
18
18
18
10
10
10
15
15
15
9
9
9
13
13
13
11.4



2
2
12
18.6
16.3




Ending VR
(miles)


20


4
18
30
7
22
32
13
28
38
5
20
30
10
25
35
4
19
29
8
19
29
16.4



12
28
28
16.3
18.6




b
coefficient


305
196
140
367


414


372


275


560


106


226


531



105


1172





WTP for
20% changes
VR(3)


$56
$36
$26
$67


$75


$68


$68


$102


$17


$41


$97



$19


$214


$216-$579
$437-$487

Note: VR - Visual Range
                                                       Source: Chestnut et al., 1994.

-------
                                          VIII-4
 noticeable effect of fine particles present in the atmosphere.  Air pollution degrades the
 visual appearance of distant objects to an observer, and reduces the range at which they can
 be distinguished from the background.  Ambient particles affect color of distant objects
 depending upon particle size and composition, the scattering angle between the observer and
illumination, the properties of the atmosphere, and the optical properties of the target being
 viewed.
        Fine particles can be emitted directly to the atmosphere through primary emissions or
 formed secondarily from gaseous precursors.  The fine particles principally responsible for
 visibility  impairment are sulfates, nitrates, organic matter, elemental carbon (soot), and soil
 dust.  The efficiency of particles to cause visibility impairment depends on particle size,
 shape, and composition. Fine particles are effective per unit mass concentration in impairing
 visibility  because their mean diameter is usually comparable to the wavelength of light, a
 condition that results in maximum light scattering. In the size range from 0.1 to  1.0 jon in
 diameter, fine particles are more effective per unit mass concentration at impairing visibility
 than either larger or smaller particles (NAPAP, 1991). Coarse particles (i.e., those  in the
 2.5  to 10 urn size range) also impair visibility, although less efficiently than fine  particles.
 All  particles scatter light to some degree, whereas only elemental carbon plays  a significant
 role in light absorption. In all regions of the country, annual average light extinction is
 dominated by light scattering  as opposed to light absorption  (NRC, 1993).  Appendix G
 provides a detailed discussion of several atmospheric optical indices that are used in
 characterizing visibility impairment and  light  extinction, including the light extinction
 coefficient, visual range, and  deciview.
        Most sulfates, nitrates, and a portion of organics begin as  gaseous emissions and
 undergo chemical transformation in the atmosphere (NAPAP, 1991;  CD, 3-2).  These
 particle constituents can readily absorb water  from the atmosphere (i.e.,  are hygroscopic)  and
 grow in size in  a nonlinear fashion as relative humidity levels increase.   In general, soluble
 organics are considered to  be  less hygroscopic than sulfates and nitrates (Sisler, 1993).  The
 relationship between humidity and particle size is a significant factor in visibility  impairment
 in the East, where in many locations average  relative  humidity exceeds 70% on an annual

-------
                                          VIII-5
average basis and can surpass 80% on many days, particularly in the summer (see more
detailed discussion of humidity in section 5).
       Light absorption is caused mainly by elemental carbon, a product of incomplete
combustion from activities such as the burning of wood or diesel fuel.  Light absorption by
nitrogen dioxide typically accounts for a few percent of total light extinction in urban areas
and is typically negligible in remote areas (CD, 8-13).  It contributes to the yellow or brown
appearance of urban hazes  since it absorbs blue light more strongly than other visible
wavelengths.  Nitrogen dioxide also may be a factor in isolated plumes from industrial
sources in remote locations.
       Atmospheric  transport of fine particles is a critical factor affecting regional visibility
conditions.  Fine particles and their precursors can remain in the atmosphere for several days
and can be carried hundreds or even thousands of kilometers from their sources to remote
locations,  such as national parks and wilderness areas (NRC, 1993).
4.     Background Levels of Light Extinction
       The light extinction coefficient represents the summation of light scattering and light
absorption due to particles  and gases in the atmosphere.  (See Appendix G for a discussion of
visibility metrics and the relationship between the light extinction coefficient, visual range,
and deciview.) Both anthropogenic and non-anthropogenic sources contribute to light
extinction.  The light extinction coefficient is represented by the following equation:
where        asg =  light scattering by gases (also known as Rayleigh scattering)
              
-------
                                         VIII-6
       a.      Rayleigh Scattering
       Rayleigh scattering represents the degree of natural light scattering found in a
particle-free atmosphere, caused by the gas molecules that make up "blue sky" (e.g., N2, Oj,
CO2).  It accounts for a relatively constant level of light extinction nationally, between 10-12
Mm'1 (NAPAP, 1991; U.S. EPA,  1979).  The concept of Rayleigh scattering can be used to
establish a theoretical maximum horizontal visual  range in the earth's atmosphere.  At sea
level, this maximum visual range is approximately 330 kilometers. Since certain
meteorological circumstances can result  in visibility  conditions that are close to "Rayleigh,"
it is analogous to a baseline or boundary condition against which other extinction components
can be compared.
       b.      Light Extinction Due to Background Paniculate Matter
       Light extinction caused by PM from non-anthropogenic sources can vary significantly
from day to day and location to location due to natural events such as wildfire, dust storms,
and volcanic eruptions.  It is useful to consider estimates of background concentrations of
PM on an annual average basis, however, when evaluating the relative contributions of
anthropogenic and non-anthropogenic sources to total light extinction.
       Chapter 4 of the staff paper addresses annual average and 24-hour estimates of
background  concentrations of PM.  Table IV-3 describes the range for annual average
regional background  PM:5 mass in the East as 2-5 /*g/m3, and in the West 1-4 ng/m3. For
PM10, the estimated annual  average background concentrations range from 5-11 /zg/m3 in the
East, and 4-8 ^g/m3  in the West.  For background 24-hour PM25  values,  present day
observed peak to mean ratios of 2 to 4 can be assumed to apply to the background annual
values in table IV-3.  This approach suggests that the highest background 24-hour PM25
levels over the course of a year could be on the order of 15-20 /ug/m3.
       Table VIII-2 from the NAPAP report includes estimates of annual average
background  concentrations of PM by aerosol constituent, as well as their related contributions
to light extinction, expressed in inverse megameters (Mnr1) (NAPAP,  1991). On an hourly
or daily basis background concentrations will vary considerably depending on seasonal,
meteorological, and geographic factors.  The table illustrates that estimated extinction
contributions from Rayleigh scattering plus background levels of fine and coarse particles, in

-------
                                               VIII-6a
              TABLE VIII-2.
AVERAGE NATURAL BACKGROUND LEVELS OF
AEROSOLS AND LIGHT EXTINCTION
Average Concentration
East West
^g/m3 Aig/m3
Fine Particles (<2.5/^m)
Sulfates (as NH,HS04)
Organics
Elemental Carbon
Ammonium Nitrate
Soil dust
Water
0
1.
0.
0.
0.
1.
.2
5
02
1
5
0
0
0
0
0
0
0
.1
.5
.02
.1
.5
.25
Error
Factor
2
2
2-3
2
1.5-2
2
Extinction
Efficiencies3
m2/g
2.
3 .
10.
2.
1.
5
5
75
5
5
25
Extinction
Contributions
East West
Mm'1 Mm"1
0
5.
0.
0.
0.
5.
.5
6
2
2
6
0
0
1.
0.
0.
0.
1.
.2
9
2
2
6
2
Coarse Particles  (2.5-10/jm)  3.0
  3.0
             1.5-2
                        0.6
                                        1. 8
                                               1.8
                                        Rayleigh Scatter
                                                         Total
                                         12

                                       26 + 7
 11

17 + 2 .5
'The extinction efficiencies are based  on the literature review by Trijonis et al.  (1986 & 1988).
All the  extinction efficiencies represent particle scattering,  except  for elemental carbon where
the 10.5 m2/g value  is assumed to consist of 9 m2/g absorption and 1.5  m2/g scattering.   Note that
the 0.6  m2/g value for coarse particles is a "pseudo-coarse  scattering efficiency"  representing t
total  scattering by all  ambient coarse particles  (2.5 /jm) divided by the coarse particle mass
between  2.5 and 10 /im.

-------
                                         VIII-7
the absence of anthropogenic emissions of visibility-impairing particles, are 26 plus or minus
7 Mm"1 in the East, and 17 plus or minus 2.5 Mm"1 in the West. These equate to a
naturally-occurring visual range in the East of 150 plus or minus 45 kilometers, and 230 plus
or minus 40 kilometers in the West.   Excluding light extinction due to Rayleigh scatter,
annual average background levels of fine and coarse particles are estimated to account for 14
Mm"1 in the East and about 6 Mm"1 in the West.  Major contributors that reduce visibility
from the Rayleigh maximum to the ranges noted above are naturally-occurring organics,
suspended dust (including coarse particles), and water.  In these ranges of fine particle
concentrations, small changes have a  large effect on total  extinction.  Thus, one can see from
table VIII-2 that higher levels of background fine particles and associated humidity in the
East result in a fairly significant difference between naturally-occurring visual range in the
rural East and West.
5.     Overview of Current Visibility Conditions
       Annual  average visibility conditions (i.e., total light extinction due  to anthropogenic
and non-anthropogenic sources) vary  regionally across the U.S.  The rural East generally has
higher levels of impairment than remote sites in the West, with the exception of the San
Gorgonio Wilderness, Point Reyes National Seashore, and Mount Rainier, which have annual
average levels comparable to certain sites in the Northeast.  Higher averages in the East are
due to generally higher concentrations of anthropogenic fine  particles and precursors, higher
background  levels of fine particles, and higher average relative humidity levels.
       Visibility conditions also vary significantly by season of the year.   With the exception
of remote sites in the northwestern U.S., visibility is typically worse in the summer months.
This is particularly true in the Appalachian region, where average extinction in the summer
exceeds the  annual average by 40% (Sisler et al., 1996).
       Figures VIII-1 and VIII-2 present 3-year (March 1992 - February 1995) averages of
monitored visibility levels for 44 sites in the  IMPROVE (Interagency Monitoring of
PROtected Visual Environments) network.  (See Appendix G for a description of the aerosol,
optical, and scene measurements taken in the IMPROVE network.)  The regional variation  in
current conditions is quite apparent from these figures. Figure VIII-1 expresses conditions  in
terms of the extinction coefficient.  The highest annual average levels are found in the rural

-------
                    VIII-7a
   60
       60
           40
   23 Dendi N.P.
FIGURE VIII-1. AVERAGE LIGHT EXTINCTION COEFFICIENT
(IN MM'1) FOR EACH OF THE REPORTED SITES IN THE
IMPROVE NETWORK, 1992-1995.  (Sisler et al., 1996)
        J20
        17
           15
     8 Dendi N.P.
FIGURE VIII-2. ANNUAL AVERAGE VISIBILITY IMPAIRMENT
IN DECIVIEWS CALCULATED FROM TOTAL LIGHT
EXTINCTION (RAYLEIGH INCLUDED), IMPROVE NETWORK,
1992-1995. (Sisler et al., 1996)

-------
                                         VIII-8
East, where the coefficient ranges from about 100-160 Mm'1 (about 23-39 kilometers visual
range) for several rural sites south of the Great Lakes and east of the Mississippi River.
This means that in certain eastern sites, 3-year average light extinction due to anthropogenic
sources is 4 to 6 times natural light extinction levels.
       The 3-year average extinction coefficient for many western sites ranges from about
30-70 Mm"1 (about 55-150 kilometers visual range), with the lowest extinction found in the
intermountain west and Colorado plateau regions.  Most of this difference between East and
West is due to greater sulfate concentrations and the effect of higher humidity levels on this
sulfate in the East (NAPAP, 1991).  Studies of historical visibility trends have shown a fairly
strong correlation between long-term light extinction levels and sulfur dioxide emissions.
This correlation is illustrated for the northeast and southeast U.S. in  figure IV-8 and is
further discussed in section IV.B. of the staff paper.
       Figure VIII-2, which expresses 3-year average visibility conditions in terms of
deciviews, shows the same regional variability.  Pristine or Rayleigh conditions are
represented by a deciview of zero, whereas the highest 3-year average  level of impairment  in
a remote site is 28 deciview in Alabama's Sipsey Wilderness.  Under many circumstances,  a
change of one deciview represents a change perceptible to the average person. By using the
deciview scale, the effect of aerosol extinction on human perception  is  portrayed as a linear
scale of visibility  degradation.  Most of the sites in the intermountain west and Colorado
Plateau have  impairment of 12 deciviews or less.  The northwest and eastern half of the U.S.
have values greater than 15 deciviews, with much of the east having  values exceeding 23
deciviews.
       Figures VIII-3 and VIII-4 present multi-year averages for PM2 5 and PM10 at
IMPROVE sites.  Analyses of aerosol constituents from these data are  used in determining
the light extinction coefficient  and deciview.  Again, regional variability is apparent, with 3-
year average  PM2 5 levels for most rural western sites in the 2-5  ng/m3 range, and levels in
the rural East in the 9-15 /xg/m3  range.  Figure VIII-5 compares PM2.5  mass to PM10 mass
for each  IMPROVE site.   It illustrates that fine PM comprises a larger fraction of PM10 in
remote eastern (60-70%) versus western (40-50%) locations.

-------
                       VIII-8a
   1.8 Dendi N.P.
FIGURE VIII-3.  AVERAGE PM2 5 MASS CONCENTRATION (IN
ug/m3) FOR EACH SITE IN THE IMPROVE NETWORK, 1992-
1995. (Sisler et al., 1996)
        10
  8.5
   4.2 Denoli N.P.
FIGURE VIII-4.  AVERAGE PM10 MASS CONCENTRATION (IN
ug/m3) FOR EACH SITE IN THE IMPROVE NETWORK, 1992-
1995. (Sisler et al., 1996)

-------
                             VIII-8b
FIGURE VIII-5. FINE MASS AS A PERCENT OF PM10 FOR EACH SITE IN
THE IMPROVE NETWORK, 1992-1995. (Sisler et al., 1996)

-------
                                        VIII-9
       Figures VIII-6a, 6b, 7a, and 7b show the seasonal variability of visibility impairment,
expressed in terms of the deciview.  One can see  that in the rural East, seasonal averages are
generally highest in the summer, with values exceeding 30 deciview at Shenandoah National
Park and the Sipsey Wilderness in Alabama, and they are generally lowest in the winter. In
the Southwest, impairment is slightly higher in the summer and winter, ranging from 10-13
deciview. In the Northwest and northern Rockies, impairment is highest in the autumn and
winter.  The following subsections further explain significant reasons for the regional
variability in visibility impairment.
       a.     Role of Humidity in Light Extinction
       As mentioned previously, humidity plays a significant role in the impairment of
visibility by fine particles,  particularly in the East, where annual average relative humidity
levels are 70-80% as compared to 50-60% in the  West (Sisler et al.,  1993). Table VIII-2
accounts for relative humidity effects by assigning a separate extinction efficiency  for water
associated with aerosols. Table VIII-3 illustrates  the extinction efficiencies used in a 1996
analysis of data from the IMPROVE network. Total light extinction  for sulfate and nitrate is
calculated by multiplying the extinction efficiencies by a relative humidity  correction factor.
        TABLE VIII-3.  DRY PARTICLE LIGHT EXTINCTION EFFICIENCY
              VALUES USED IN 1996 ANALYSIS OF IMPROVE DATA
Aerosol
Constituent
Sulfates
Nitrates
Organics
Soil dust
Coarse particles
Extinction
Efficiency
(in m2/g)
3.0f(RH)
3.0f(RH)
4.0
1.0
0.6
f(RH) is the relative humidity correction factor. It is
the ratio of wet scattering divided by dry scattering.
                                 Source: Sisler et al.,  1996

-------
                      VIII-9a
   15
           13
    7  Denoli N.P.
FIGURE VIII-6a. AVERAGE WINTER VISIBILITY
IMPAIRMENT IN DECIVIEWS CALCULATED FROM TOTAL
LIGHT EXTINCTION (RAYLEIGH INCLUDED), IMPROVE
NETWORK, 1992-1995. (Sisler et al., 1996)
        20
 20
       20
           15
   9 Dendi N.P.
                      17
FIGURE VIII-6b. AVERAGE SPRING VISIBILITY IMPAIRMENT
IN DECIVIEWS CALCULATED FROM TOTAL LIGHT
EXTINCTION (RAYLEIGH INCLUDED), IMPROVE NETWORK,
1992-1995. (Sisler et al., 1996)

-------
                     VIII-9b
        20
   10 Dendi N.P.
FIGURE VIII-7a. AVERAGE SUMMER VISIBILITY
IMPAIRMENT IN DECIVIEWS CALCULATED FROM TOTAL
LIGHT EXTINCTION (RAYLEIGH INCLUDED), IMPROVE
NETWORK, 1992-1995.  (Sisler et ah, 1996)
   7 Dendi N.P.
FIGURE VIII-7b. AVERAGE AUTUMN VISIBILITY
IMPAIRMENT IN DECIVIEWS CALCULATED FROM TOTAL
LIGHT EXTINCTION (RAYLEIGH INCLUDED), IMPROVE
NETWORK, 1992-1995. (Sisler et al., 1996)

-------
                                         VIII-10
The correction factor represents 1) the hygroscopic nature of the aerosol constituent, and 2)
the average annual humidity for the relevant location (Sisler et al., 1993).  Light absorption
by fine particles can be measured directly by the  Laser Integrating Plate Method, or it can be
estimated by multiplying elemental carbon mass by an extinction efficiency of 10 m2/g (Sisler
etal.,  1996).
       Because annual average relative humidity  is higher in the East, the  same ambient
concentration of sulfate, for example, will on average lead to greater light extinction in an
eastern location rather than a western one. The top map in figure VIII-8 illustrates the
regional  variability of annual mean relative humidity nationwide.  The bottom map depicts
the variability of the relative humidity correction factor used for sulfates in an analysis of
IMPROVE data (Sisler et al.,  1993).   For example, when corrected for humidity, the overall
extinction efficiency for sulfates in the East may  exceed  11-12 m2/g, whereas the extinction
efficiency for  sulfate in the West may be  one-third to one-half of that.
       b.     Significance of  Anthropogenic Sources of Fine Particles
       On  an annual average basis, the concentrations of background fine particles are
generally small when compared with concentrations of fine particles from anthropogenic
sources (NRC, 1993).   The same relationship holds true when one compares annual average
light extinction due to  background fine particles with light extinction due to background plus
anthropogenic sources.  Table VIII-4  makes this  comparison for several locations across the
country by using background estimates from table VIII-2 and light extinction values derived
from monitored data from the IMPROVE network.  These data indicate that anthropogenic
emissions make a significant contribution  to average light extinction in  most  parts of the
country,  as compared to the contribution from background fine particle levels.  Man-made
contributions account for about one-third of the average extinction coefficient in the rural
West and more than 80% in the rural East (NAPAP, 1991).
       It is important to note that even in those areas with relatively low concentrations of
anthropogenic fine particles, such as the Colorado plateau, small increases in anthropogenic
fine particle concentrations can lead to significant decreases in visual range.  This is one
reason why Class I areas have been given special consideration under the Clean Air Act.
This relationship is illustrated  by figure VIII-9, which relates changes in fine particle

-------
                                 VHI-lOa
             TO
                                                          eo
                       (a) Annual mean relative humidity.
           2.1
                (b) Sulfate relative humidity correction factor FT.

FIGURE VIII-8. SPATIAL VARIATION IN AVERAGE RELATIVE HUMIDITY
(NOAA, 1978) AND THE SULFATE RELATIVE HUMIDITY CORRECTION FACTOR
FT. (Sisleretal., 1993)

-------
                        VHI-lOb
TABLE VIII-4.
COMPARISON OF TOTAL LIGHT EXTINCTION TO
ESTIMATED BACKGROUND LIGHT EXTINCTION FOR
SEVERAL EASTERN AND WESTERN LOCATIONS.
REGION

Eastern U.S., estimated
background light extinction
Appalachian
Boundary Waters
Northeast
Washington, D.C.
Western U.S., estimated
background light extinction
Colorado Plateau
Cascades
Southern California
Northern Rockies
TOTAL LIGHT
EXTINCTION
1988-1994
(in Mm'1)
Annual
26 +/- 7
126
62
77
177
17 +/- 2.5
32
74
74
57
Summer
NA
182
63
95
207
NA
33
73
87
48
VISUAL
RANGE .
(in km)
Annual
150 +/- 45
31
63
51
22
230 +/- 40
122
53
53
69
Summer
NA
21
62
41
19
NA
119
54
45
82
Sources: Sisleretal., 1996; NAPAP 1991.

-------
>
o
Q)
_C

.5
I

't
   40
    35
    30
    25
   20
   10
               Current annual average: 11.1 ug/m3 in the Appalachian
               Region.
                                                                Appalachian Region (ave. extinction efficiency 10.2 m2/g)

                                                              *. Colorado Plateau Region (ave. extinction efficiency 5.5 m2/g
                  Current annual average: 3.3 ug/m3 in the Colorado Plateau
                  Region.
                             I	I	I	I	I	I     I	I
                            10
                                                  20                     30
                                               Fine Particle Concentration (ug/m3)
40
                                                                                                                      50
                                                                                                                             o
                                                                                                                             n
       FIGURE VIII-9. PERCEPTIBLE CHANGE IN VISIBILITY AS A FUNCTION OF FINE MASS CONCENTRATION
              Note:  Average extinction efficiencies are calculated from IMPROVE monitoring program data, March 1988 -
              February 1994. Changes in total fine particle concentration reflect current mix of constituents.

              Appalachian region: Great Smokies, Shenandoah, Dolly Sods.
              Colorado Plateau region: Grand Canyon, Bryce Canyon, Canyonlands, Mesa Verde, Bandolier, Petrified Forest.

              Under many circumstances, a change of one deciview represents a change perceptible by the average person.

-------
                                          VIII-11
concentrations to perceptible changes in visibility (represented by the deciview metric). The
graph shows that in cleaner areas, such as the West, perceptible visibility changes are more
sensitive to existing fine particle concentrations than is the case in more polluted areas. In
other words, to achieve a given amount of perceived visibility  improvement, a larger
reduction in fine particle concentration is required in areas with higher existing
concentrations, such as the East, than would be required in lower concentration areas.  This
figure also illustrates the relative importance of the overall extinction efficiency of the
pollutant mix at particular locations. At a  given ambient concentration, areas having higher
average extinction efficiencies (expressed in m2/g in figure VIII-9) due to the mix of
pollutants would have higher levels of impairment.  In the East,  the combination of higher
humidity levels and a greater percentage of sulfate as compared to the West causes  the
average extinction efficiency for fine particles to be  almost twice that in the Colorado
Plateau.
       c.      Regional Differences in  Specific Pollutant Concentrations
       As total light extinction levels vary  significantly across  the country, so does the mix
of visibility-impairing pollutants from region to region.  Table VIII-5, taken from the  1993
National Research Council study on visibility, shows the estimated contribution of various
anthropogenic pollutants to visibility impairment for three main regions of the U.S.  The
table takes into account relative emissions levels of each pollutant type within each  region.
This and other analyses (Sisler et al.,  1993) show  that sulfates  are a significant cause of
visibility impairment in all parts of the country, but  particularly in the East, where  they are
responsible  for about two-thirds of overall  light extinction.  In the Southwest and Northwest,
organics play a larger role, as does elemental carbon.  Suspended dust is also a major
constituent in the Southwest.  The main categories of sources responsible for visibility-
impairing fine particle and precursor emissions are listed in table VIII-6 (NRC, 1993).
       d.      Regional Variation in Urban Visibility
       Visibility  impairment has been studied in several major cities in the past decade (e.g.
Middleton,  1993) because of concerns about fine particles and  their potentially significant
impacts (e.g., health-related  and aesthetic)  on the residents of large metropolitan areas.
Urban areas generally have higher loadings of fine paniculate matter than monitored Class I

-------
                                       VHI-lla

TABLE VIII-5.      VISIBILITY MODEL RESULTS:
                    ANTHROPOGENIC LIGHT EXTINCTION BUDGETS"
                          East"      Southwest0      Northwest"1
Sulfates
Organics
Elemental carbon
Suspended dust
Nitrates
Nitrogen dioxide
65
14
11
2
5
3
39
18
14
15
9
5
33
28
15
7
13
4
"Percentage contribution by specific pollutant to anthropogenic light extinction in three
regions of the United States.

"Based on Table 9, Table  18, Figure 45, Appendix A, and Appendix E of NAPAP Visibility
SOS/T Report (Trijonis et al., 1990). It is assumed that sulfates (3% natural) account for
60% of non-Rayleigh extinction, organics (33% natural) account for 18%, elemental carbon
(3% natural) accounts for 10%, suspended dust (50% natural) accounts for 4%, nitrates
(10% natural) account for 5%, and nitrogen dioxide (10%  natural) accounts  for 3%.

'Based on Table 9, Table  18, Figure 45, Appendix A, and Appendix E of the NAPAP
Visibility SOS/T Report (Trijonis et al., 1990). It is assumed that sulfates (10% natural)
account for 33% of non-Rayleigh extinction, organics (33% natural) account for 20%,
elemental carbon (10%  natural) accounts for 12%, suspended dust (50% natural) accounts for
23%, nitrates (105 natural) account for 8%, and nitrogen dioxide (10% natural) accounts for
4%.

dExtinction efficiencies  (relative to organics are chosen as 1.5 for sulfates, 2.5 for elemental
carbon, 0.3 for fine crustal materials, and 1.5  for nitrates (Trijonis et al.,  1988, 1990).
Coarse dust extinction is assumed to be three times fine dust extinction (Trijonis et al., 1988,
1990). Natural aerosol particle fractions are assumed to be one-tenth for sulfates, one-third
for organics one-tenth for elemental carbon, one-half for crustal materials, and one-tenth for
nitrates.  These assumptions are applied using  the fine mass concentrations in Trijonis et al.,
(1990).  The percentage contribution for nitrogen dioxide is assumed to be 4%.

Source:  NRC,  1993.

-------
TABLE VIH-6.
           VHI-llb
PERCENTAGE CONTRIBUTION BY SOURCE CATEGORY
TO FINE PARTICLE (AND PRECURSOR) EMISSIONS
IN THE EAST, SOUTHWEST, AND NORTHWEST
EAST
Electric utilities
Diesel-fueled mobile sources
Gasoline vehicles
Petroleum and chemical industries
Industrial coal combustion
Residential wood burning
Fugitive dust (on-road/off-road traffic)
Feedlots and livestock waste mgmt.
Miscellaneous
SOUTHWEST
Electric utilities
Diesel-fueled mobile sources
Gasoline vehicles
Petroleum and chemical industries
Copper smelters
Fugitive dust (on-road/off-road traffic)
Residential wood burning
Feedlots and livestock waste mgmt.
Miscellaneous
NORTHWEST
Electric utilities
Diesel-fueled mobile sources
Gasoline vehicles
Petroleum and chemical industries
Residential wood burning
Forest management burning
Fugitive dust (on-road/off-road traffic)
Feedlots and livestock waste mgmt.
Primary metallurgical process
Organic solvent evaporation
Miscellaneous
SOx
78.0
1.5
1.0
4.5
7.0
_
..
—
8.0
SOx
33
12
5
22
19
-
-
-
9
SOx
30
12
4
19
-
-
-
--
8
-
27
Organic
Particles
—
-
34
~
-
20
-
-
46
Organic
Particles
_
5
38
-
-
-
8
-
49
Organic
Panicles
_
_
15
_
22
45
-
-
-
-
18
VOC's
_
-
31
11
-
13
--
--
45
VOC's
—
—
42
12
-
-
5
-
41
VOC's
—
—
31
10
25
13
-
-
15
15
6
Elemental
Carbon
—
47
29
-
-
15
-
~
9
Elemental
Carbon
—
52
31
-
-
—
6
-
11
Elemental
Carbon
—
37
16
—
22
20
-
-
-
—
5
Suspended
Dust
-
-
-
-
«
-
100
-
-
Suspended
Dust
—
-
—
-
-
100
-
-
-
Suspended
Dust
—
—
-
-
-
-
100
-
—
—
~
NH3
—
--
'--
-
--
-
--
66
34
NH3
—
-
-
-
-
—
-
75
25
NH3
—
—
—
—
-
-
-
81
—
--
19
NOx
39
16
26
-
-
-
-
-
19
NOx
19
23
32
~
-
~
--
-
26
NOx
8
29
36
-
-
-
-
-
--
-
27
Source: NRC, 1993.

-------
                                       VIII-12
areas, suggesting that visibility impairment in urban areas is typically greater than in rural
areas.  Monitored annual mean and second highest maximum 24-hour fine particle levels for
selected urban areas are listed in Table IV-4. These levels are generally higher than those
found in the IMPROVE database for rural Class I areas.
       The degree to which different aerosol constituents contribute to overall light extinction
in urban areas can vary significantly. Table VIII-7 illustrates the difference between
percentage contributions of aerosol constituents to annual average total light extinction in the
Washington, DC urban area and the southern California areas. The dominance of sulfate in
Washington, DC exhibits a regional effect stemming from sulfur dioxide emissions outside
       TABLE Vffl-7.
PERCENTAGE CONTRIBUTIONS OF AEROSOL
CONSTITUENTS TO ANNUAL AVERAGE TOTAL LIGHT
EXTINCTION IN THE WASHINGTON, DC AND
SOUTHERN CALIFORNIA AREAS.
Location
Wash, DC
Southern
Calif.
Sulfate
49
14
Nitrate
16
44
Organics
16
18
Elemental
Carbon
12
9
Soil and
Coarse
7
14
                Source:  Sisler etal., 1993

the metropolitan area.  In contrast, nitrate plays the greatest role in the overall light
extinction levels in the mountainous areas just outside Los Angeles, with most of the nitrate
formation in this area coming from nitrogen dioxide emissions within the urban area.
6.     Policy Considerations Pertaining to the Effects of PM on Visibility
       Impairment of visibility in multi-state regions, urban areas, and Class I areas is
clearly an effect of paniculate matter on public welfare. The staff has considered a number
of factors in assessing appropriate regulatory responses.
       An initial question is whether the range of recommended primary standards for fine
PM would provide adequate protection against visibility impairment across the country.  The

-------
                                         VIII-13
range being considered for an annual PM-fme standard is 12.5 /zg/m3 to less than 20
and the range under consideration for a 24-hour standard is 18 ptg/m3 to less than 65 fig/m3.
Table IV-4 presents monitored fine particle annual averages and second highest maximum
levels for several major U.S. cities. Analysis of these data suggests that adoption of an
annual fine particle standard in the lower half of the recommended range, in combination
with adoption of a 24-hour standard in the lower half of the recommended range, would be
expected to  lead to reductions in annual average fine particle concentrations in many urban
areas nationally.  Additionally, reductions could be achieved in broader areas in the East if
regional attainment strategies are carried out.  To examine expected regional visibility
improvements resulting from these reductions requires an understanding of the various
factors affecting the relationship between fine particle loadings and visibility, such as
background  levels,  humidity, and pollutant mix, as described  in section 5 above.
       Expected reductions in fine particle concentrations resulting from adoption of the
primary fine particle standards in the lower half of the recommended range is likely to result
in maintained or improved visibility in many urban areas and  in a broader area in the East.
As with reductions  in fine particle concentrations noted above, improvement of visibility
would be  greater if regional fine particle attainment strategies are carried out.  In its 1993
Report to Congress on the effects of Clean Air Act programs  on visibility in mandatory
federal Class I areas,  EPA examined the impact of expected regional sulfur dioxide
reductions under the acid rain program (U. S. EPA, 1993). This report estimated that
regional annual average sulfate levels would be reduced over a wide area in the eastern U.S.
by the year  2010, resulting in potential improvements in visibility for the region.  The
analysis projected no expected improvement in the rural West. Moreover, despite projected
improvements in visibility, there is no evidence that adoption  of the primary fine particle
standards  in the lower half of the recommended range  will eliminate visibility impairment.
       The staff has also considered whether the adoption of a national secondary standard
would provide adequate and appropriate protection of public welfare across the country.  Due
to the regional variability in visibility conditions created by background fine particle levels
and humidity, the staff has concluded that a national secondary standard would not be the
most appropriate means to achieve  this objective.  The data presented in table VIII-4

-------
                                         VIII-14
indicates that current annual average light extinction levels on the Colorado Plateau
(reflecting effects of anthropogenic and background sources of PM) are about equal to
background levels (i.e., those levels  representing an absence of anthropogenic contributions)
in the East.  Thus, a national secondary standard set to maintain or improve visibility
conditions on the Colorado Plateau would have to be set at or below natural background
levels in the East, effectively requiring elimination of all anthropogenic (and  some
nonanthropogenic) emissions.  Conversely, a national secondary standard that would be both
attainable and improve visibility in the East would permit further degradation in the West.
       An approach which would be more responsive to visibility protection  goals, while
recognizing these significant regional variations, would be to establish a regional haze
program under  section 169A  of the Clean Air Act.  This program, while designed to address
the existing adverse effects of fine particles on visibility in Class I areas, would further
contribute to visibility improvement in non-Class I areas as well. Section 169A established a
national goal  of "the prevention of any future, and the remedying of any existing, manmade
impairment of visibility in mandatory Class I areas."  The EPA  is  required to establish
programs to ensure reasonable progress toward the national goal. These programs are to be
implemented by the States and can be regionally specific.  Concern with regional visibility
impacts to highly valued national parks and wilderness areas in the U.S. led to the inclusion
of specific  language in section 169B  of the 1990 Clean Air Act Amendments, requiring EPA
to form the Grand Canyon Visibility Transport Commission.   In June 1996, the Commission
provided the Administrator with recommendations for regional approaches to protecting
visibility. The  work of the Commission  will be useful to development of a regional haze
program under  section 169A  of the Act.
       Much progress has been made in  technical areas important to the successful
implementation of a regional haze program, including areas such as visibility monitoring,
regional scale modeling, and scientific knowledge of the regional effects of particles on
visibility. The  National Academy of Sciences 1993 report on visibility protection confirmed
this point:
       Current  scientific knowledge  is adequate and control technologies are  available
       for taking regulatory action to improve and protect visibility.  However,

-------
                                          VIII-15
       continued national progress toward this goal will require a greater commitment
       toward atmospheric research, monitoring, and emissions control research and
       development.
       In addition, as noted  above, it is expected that the development of a regional haze
program would  have associated benefits outside of mandatory Class I areas.  The National
Academy of Sciences concluded the following:
       Efforts to improve visibility in Class I areas also would benefit visibility
       outside these areas.  Because most visibility impairment is regional in scale,
       the same haze that degrades visibility within or looking out from a national
       park also degrades visibility outside it.  Class I areas cannot be regarded as
       potential islands of clean air in a polluted sea.
       Based on the above considerations, the staff recommends that the Administrator
consider establishing a regional haze program under section 169A of the Act, in conjunction
with the recommended fine particle primary standards, as the  most effective  means of
addressing the welfare effects associated with visibility impairment. Together, the two
programs and associated control strategies should adequately protect against the effects of
fine particle pollutants on visibility and make reasonable progress toward the national
visibility goal for Class I areas.

C.     Effects of PM on Materials Damage and  Soiling
       The deposition of airborne particles can become a nuisance, reducing the aesthetic
appeal of buildings and culturally  important articles through soiling, and contribute directly
(or in conjunction with other pollutants) to structural damage by means of corrosion or
erosion.  These  potential effects are discussed more fully below.  The  relative importance of
particle size, composition,  and other environmental factors (i.e., moisture, temperature,
sunlight, and wind) in contributing to the effects is also considered.
1.      Materials Damage
       Particles  affect materials principally by promoting and  accelerating the corrosion of
metals, by degrading paints,  and by deteriorating building materials such as concrete and
limestone.  Particles contribute to  these effects because of their electrolytic, hygroscopic, and
acidic properties, and their ability  to sorb corrosive gases (principally sulfur  dioxide).  The

-------
                                         VIII-16
staff review suggests that only chemically active fine mode or hygroscopic coarse mode
(mainly sea or road salt) particles contribute to such effects (U.S. EPA, 1986b).  While
particles have been qualitatively associated with damage to materials, there are insufficient
data at present to relate such effects to specific particle pollution levels. The following
discussion briefly outlines the available information on PM-related effects associated with
each category of material presented in the criteria document.
       a.    Effects on Metals
       The  rate  of metal corrosion depends on a number of factors, including the deposition
rate and nature of the pollutant; the influence of the metal protective corrosion film; the
amount of moisture present;  variability in the electrochemical reactions; the presence and
concentration of other surface electrolytes; and the orientation of the metal surface (CD,
Chapter 9).  This section briefly discusses the factors affecting metal corrosion  set forth  in
the criteria document.
       Nriagu (1978) and Sydberger (1977) conducted studies that highlighted the ability
metals have to form a protective film that slows corrosion rates.  Metals initially exposed to
low concentrations of SOX corroded at a slower rate than did samples continuously exposed to
higher concentrations.  This  protective corrosion layer may, however, be affected by either
dry or wet deposition (CD,  Chapter 9).
       The rate  of metal corrosion decreases in the absence of moisture (CD, Chapter 9).
Moisture influences corrosion rates by providing a medium of conduction paths for
electrochemical  reactions and a medium for water soluble air pollutants.  Schwartz (1972)
established that  the corrosion rate  of a metal could increase by 20 percent for each one
percent increase in relative humidity above  the minimum  atmospheric moisture  content that
allows corrosion to occur (i.e., critical relative  humidity). Later  studies by Haynie and
Upham (1974) and Sydberger and Ericsson (1977) supported Schwartz's theory.
       While particles alone have some effect on the early stages of metal corrosion, there is
insufficient evidence to relate such effects to specific particle levels. One study (Goodwin et
al. (1969))  reported damage to steel,  protected with nylon screen, exposed to quartz particles
larger than 5 ^on; but the exposure time and concentration were not reported. Barton  (1958)
also found that  dust contributed to the early stages of metal corrosion.   A number of the

-------
                                        VIII-17
studies evaluated concluded that paniculate matter increased the corrosion rate of sulfur
dioxides (Sanyal and Singhania, (1956); Yocom and Grappone, (1976); Johnson et al.,
(1977); Russell, (1976); Walton et al., (1982)).  Laboratory studies show mixed results as to
whether catalytic species or conductance of the thin-film surface electrolyte is the cause of
the increases in corrosion rates (Walton et al., 1982; Skerry et al., 1988 a,b; Askey et al.,
1993).
       Results of actual field studies have not established a quantitative relationship between
particles and corrosion.  Thus, the independent effect of particles is not evident since SO2 is
the controlling factor for determining corrosion rate (U.S. EPA, 1986b).  Edney et al. (1989)
exposed galvanized steel panels to actual  field conditions in Research Triangle Park, NC and
Steubenville, OH between April 25  and December 28, 1987. The panels were exposed under
the following conditions:  (1) dry deposition only; (2) dry plus ambient wet deposition; and
(3) dry deposition plus deionized  water.  The average concentrations for SO2 and paniculate
matter was 22 ppb and 70 /-ig/m3  and < 1 ppb and 32 ^g/m3 for Steubenville and Research
Triangle Park, respectively.  The runoff from the steel panel was analyzed and it was
concluded that the dissolution  of the steel corrosion products for both sites was likely the
result of deposited gas phase SO2 on the metal surface and not paniculate matter.
Another study conducted by Butlin et al.  (1992) also demonstrated that the corrosion of mild
steel and galvanized steel was  SO2-dependent. Butlin et al. monitored the corrosion of steel
samples by SO2 and ozone under  artificially fumigated environments, and NO2 under natural
conditions.  Annual average SO2 concentrations ranged from 2.1 /ig/m3 in a rural area to 60
jug/m3 in one of the SO2-fumigated locations.   Annual average NO2 concentrations ranged
from  1.5 to 61.8 ^g/m3. The study concluded that corrosion of the steel samples was
primarily dependent on the long-term SO2 concentration and was only minimally affected by
nitrogen oxides.
       b.     Effects on Paint
       Paints undergo natural  weathering processes from exposure to environmental factors
such as sunlight, moisture, fungi, and varying temperatures.  In addition to the natural
environmental factors, studies  show paniculate matter exposure may give painted surfaces a
dirty appearance (CD, Chapter 9).  Several studies also suggest that panicles serve  as

-------
                                        VIII-18
carriers of other more corrosive pollutants, allowing the pollutants to reach the underlying
surface or serve as concentration sites for other pollutants (Cowling and Roberts, 1954).
       A number of studies have shown some correlation between paniculate matter and
damage to automobile finishes. Fochtman and Langer (1957) reported damage to automobile
finishes due to iron particles emitted from nearby industrial facilities.  General Motors
conducted field tests in Jacksonville, Florida to determine the effect of various meteorological
events, the chemical composition of rain and  dew, and the ambient air composition during
the event, on automotive paint finishes.  Painted (basecoat/clearcoat technology)  steel panels
were exposed  for varying time periods, under protected  and unprotected condition.  The
researcher concluded that calcium sulfate formed on the painted surface by the reaction of
calcium from dust and sulfuric acid contained in rain or dew.  The damage to the paint finish
increased  with increasing days of exposure (Wolff et al., 1990).
       Paint films permeable to water are also susceptible to penetration by acid forming
aerosols (U.S. EPA, 1995).  Baedecker et al. (1991) reviewed studies dealing with solubility
and permeability of SO2  in paints  and polymer films.  These studies showed permeation and
absorption rates varied depending on the formulation of the paint.
       Studies reported in the criteria document (Spence et al., (1975); Campbell et al.,
(1974); Haynie and Spence, (1984); Yocom and Grappone, (1976); and Yocom and Upham,
(1977)) support the conclusion that gaseous pollutants contribute  to the erosion rates of
exterior paints.
       c.      Effects on Stone
       Damage to calcareous stones (i.e., limestone, marble and carbonated cemented stone)
has been attributed to deposition of acidic particles.  Moisture and salts are considered the
most important factors in building  material damage (CD, Chapter 9).  However, many other
factors (such as normal weathering and microorganism damage) also seem to play a part  in
the deterioration of inorganic building materials.  The relative importance of biological.
chemical, and physical mechanisms has not been studied to date.  Thus, the relative
contribution of ambient pollutants to the damage observed in various building stone is not
well quantified.

-------
                                         VIII-19
       Baedecker et al. (1991) reported that 10 percent of chemical weathering of marble and
limestone was caused by wet deposition of hydrogen ions from all acid species.  Dry
deposition of SO2 between rain events caused 5 to 20 percent of the chemical erosion of
stone, and dry deposition of nitric acid was responsible for 2 to 6 percent of the erosion
(Baedecker et al.,  1991).  Under high wind conditions, particulates result in slow erosion of
the surfaces, similar to sandblasting (Yocom and Upham, 1977).
       d.     Effects on Electronics
       Exposure to ionic dust particles can contribute significantly to the corrosion rate of
electronic devices, ultimately leading to failure.  Particles derived from both natural and
anthropogenic sources and ranging in size from tens of angstroms to one /xm can cause
corrosion of electronics because many are sufficiently hygroscopic and corrosive, at normal
relative humidities, to react directly with non-noble metal and passive oxides, or to form
conductive moisture films on insulating surfaces to cause electrical leakage.  The effects of
particles on electronic components were first reported by telephone companies who reported
that particles high  in nitrates caused corrosion, cracking, and ultimate failure of wire spring
relays (Hermance, 1966; McKinney and Hermance,  1969).  More recently, Sinclare  (1992)
and Frankenthal (1993) have reported  that anthropogenically-derived particles penetrating  into
indoor environments can contribute to the corrosion of electronics.
2.     Staff Considerations Pertaining to the Effects of PM on Materials Damage
       While particles, particularly in conjunction with sulfur dioxide, have been
qualitatively associated with damage to materials, there is insufficient  data available to relate
such damage to  specific particle levels in the ambient air.  Absent better  quantitative data,
the staff does not believe the Administrator should consider a separate secondary standard
based on materials damage.
3.     Soiling
       Soiling is the accumulation of particles on the surface of an exposed material  resulting
in the degradation of its appearance.  When such accumulation produces  sufficient changes in
reflection from opaque surfaces and reduces light transmission through transparent materials,
the surface will  become perceptibly dirty to the human observer.  Soiling can be remedied by
cleaning or washing, and depending on the soiled material, repainting.

-------
                                         VIII-20
       Determination of what accumulated level of paniculate matter leads to increased
cleaning or repainting is difficult.  For example, Carey (1959) found that the appearance of
soiling only occurred when the surface of paper was covered with dust specks spaced 10 to
20 diameters apart.  When the contrast was strong, e.g., black on white, it was possible to
distinguish a clean surface from a surrounding dirty surface when only 0.2 percent of the
areas was covered with specks, while 0.4 percent of the surface had to be covered with
specks with a weaker color contrast.
       Hancock et al. (1976) found that with maximum contrast, a 0.2 percent surface
coverage (effective area coverage; EAC) by dust can be perceived against a clean
background.  A dust deposition level  of 0.7 percent EAC was needed before the object was
considered unfit for  use. The minimum perceivable difference between varying gradations of
shading was a change of about 0.45 percent EAC.  Using the information on visually
perceived dust accumulation, Hancock et al. (1976) concluded that dustfall rates of less  than
0.17 EAC/day would be tolerable to the  general public. Similar studies have not been
reported for other soiling effects.
       Despite the observation that airborne particles soil a wide range of man-made
materials, there  is only limited information available with respect to size and composition of
the culpable particles.  In general, the soiling of fabrics and vertical surfaces has been
ascribed to fine  particles, particularly dark, carbonaceous materials.  Soiling of horizontal
surfaces may result from deposition of a wide range of particles, including coarse mode
dusts.
       An important consideration in assessing soiling potential is deposition velocity, which
is defined as flux divided by concentration.  Deposition velocity is a function of particle
diameter, surface orientation and roughness, wind speed,  atmospheric stability,  and particle
density. As a result, soiling is expected to vary with the  size distribution of particles within
an ambient concentration, whether the surface is positioned horizontally  or vertically, and
whether the surface  is rough or  smooth (CD, Chapter 9).
       Theoretically, coverage of horizontal surfaces will be related to particle surface areas
and deposition velocity.  Particle surface areas per unit mass decreases linearly  with diameter
(assuming spherical  particles), while, under quiescent  conditions, deposition velocity

-------
                                          VIII-21
 increases with the square of the diameter.  Under such conditions, large particles would
 result in more soiling than an equivalent mass of smaller particles.  Although second order
 effects may enhance fine particle deposition relative to larger particles,  deposition velocity
 data still suggest substantially higher deposition on horizontal surfaces for particles larger
 than  10 (im than for smaller particles (U.S. EPA, 1982b).
       The increasing soiling potential associated with increased particle size is mitigated by
.lighter particle color, effects of rainfall,  smaller transport distance from sources and
 markedly lower penetration of larger particles to indoor surfaces (relative to smaller
 particles). Because these conflicting factors have not been fully evaluated, it is not possible
 to make clear particle size divisions with respect to soiling of horizontal surfaces.
       The time interval that it takes to transform horizontal and vertical surfaces  from clean
 to perceptibly dirty  is generally determined by particle composition and rate of deposition.
 The process is influenced by the location (sheltered or unsheltered) and  spatial alignment of
 the material, the texture and color of the surface relative to the particles, and  meteorological
 variables such as moisture, temperature,  and wind speed.
       Haynie and Lemmons (1990) conducted a soiling study in a relatively rural
 environment in Research Triangle Park,  North Carolina.  The study was designed to
 determine how various environmental factors contribute to the rate of soiling of white painted
 surfaces,  which are  highly sensitive to soiling by dark particles and represent a large fraction
 of all man-made surfaces exposed in the  environment.  Hourly rainfall and wind speed, and
 weekly data for dichotomous sampler measurements and TSP concentration were monitored.
 Gloss and flat white paints were applied  to hardboard  house siding surfaces and exposed
 vertically and horizontally for 16 weeks, either sheltered or unsheltered  from rainfall.
 Measurements, including reflectance, were taken at 2, 4, 8, and 16 weeks.  Based on the
 results of this study, the authors concluded that: (1) coarse mode particles initially contribute
 more to soiling of both horizontal and vertical surfaces than fine mode particles; (2) coarse
 mode particles, however, are more easily removed by rain than are fine mode particles;  (3)
 for sheltered surfaces, reflectance changes are proportional to surface  coverage by particles,
 and particle accumulation is consistent with deposition theory; (4) rain interacts with particles
 to contribute to soiling by dissolving or desegregating particles and leaving stains; and (5)

-------
                                         VIII-22
very long-term remedial actions are probably taken because of the accumulation of fine
rather than coarse particles (Haynie and Lemmons, 1990).
       Creighton et al.  (1990) reported that horizontal surfaces soiled faster than vertical
surfaces and that large particles  were primarily responsible for the soiling of horizontal
surfaces not exposed to rainfall.  Soiling was related to the accumulated mass of particles
from both the fine and coarse fraction.  Fine mode black smoke and motor vehicle exhaust
have been associated with the soiling of building material and facades (Tarrat and Joumard,
1990; Lanting, 1986).
       Ligocki et al. (1993) studied the potential soiling of art work in five Southern
California museums.  The authors concluded that a significant fraction of fine elemental
carbon and soil dust particles had penetrated to the indoor atmosphere of the museums
studied and may constitute a soiling hazard to displayed art work.  The seasonally averaged
indoor/outdoor ratios for paniculate matter mass concentrations ranged from 0.16 to 0.96 for
fine particles and from 0.06 to 0.53 for coarse particles, with lower values observed for
building with  sophisticated ventilation systems that include filters for paniculate removal.
4.     Societal Costs
       a.      Soiling/Property Value
       The effect of particles on aesthetic  quality depends in part on human perception of
pollution.  The reduction of aesthetic quality may arise from the soiling of buildings or other
objects of historical or social interest from the mere dirty appearance of a neighborhood.  A
number of studies have indicated that such perceptions  of neighborhood degradation are
revealed indirectly through effects on the value of residential property.  That is, when
residential properties similar in  other respects are compared, the properties in the more
highly polluted areas typically have  lower  value.
         Freeman (1979), reporting  on  14  property value studies that used paniculate matter
or dustfall as  one of their pollutant measures, noted that the  results  generally supported the
premise that property values are affected by the full range of particle pollution.  He
cautioned, however, that direct  comparison of the monetary  results  is not possible since the
studies cover  a number of cities and use different data bases, empirical techniques, and
model specifications.

-------
                                         VIII-23
       The extent to which the city-specific results represent soiling as opposed to
perceptions of the effects of particles on health and visibility is not clear.  Therefore, the
results of these studies cannot provide reliable quantitative estimates of the effects of soiling
on property values (U.S. EPA, 1982b).
       b.     Soiling/Materials
       Airborne particles soil a wide range of materials in all sectors of the economy.
Assuming that these sectors are not as well off in a dirtier state as a cleaner one, soiling will
result in an economic cost  to society.  While the household sector has been examined by a
number of investigators, their results have been questioned because of methodology
problems and their failure to appropriately address particle size, composition, and deposition
rates.   As a result, no single study has produced a completely satisfactory estimate of soiling
costs for the  household sector. It is unfortunate that little or no effort has been expended to
account for soiling costs in the commercial, manufacturing, or public sectors.  Results from
MathTech, Inc. (1983) suggest that soiling costs for the manufacturing sector alone could be
significant.
       In the review of effects of household soiling, the staff paper has relied principally on
Booz, Allen  and Hamilton, Inc.,  (1970); Watson and Jaksch,  (1978,  1982) [which was cited
in the CD and discussed in more  detail in the  1982 criteria document]; and MathTech, Inc.,
(1983) to derive estimates of household soiling costs.  For the year 1970, the estimate for
amenity loss  due to exterior household soiling was estimated to range form 1 to 3.5 billion
dollars (1978 dollars).  The 14 ^g/m3 reduction in U.S. annual TSP levels between 1970 and
1978 was estimated to have resulted in an annual benefit for the year 1978 of 0.2 to 0.7
billion dollars or  14 to 50 million dollars for each ^g/m3  of reduction (U.S. EPA, 1982a).
MathTech, Inc. (1983) estimated  household soiling costs in the range of $88.3 million to
$1.2 billion (1980 dollars)  for attaining the primary PM10 standard nationwide.  Gilbert
(1985) used a household production function framework to design and estimate the short-run
costs of soiling.  The results  were comparable to those reported by MathTech (1983).
Finally, McClelland et al. (1991) concluded that households were willing to pay $2.70 per
     3 change in particle level to  avoid soiling effects.

-------
                                         VIII-24
       Haynie (1989), using fine and coarse mode particle levels calculated from 1987 EPA
AIRS data for PM10 and TSP, estimated that $1.74 billion of annual national residential
repainting costs could be attributed to soiling (using national average painting costs and
frequencies). Haynie and Lemmons (1990) estimated that the national soiling costs associated
with repainting the exterior walls of houses probably were within the range of $400 to $800
million a year in 1990. This lower estimate, as compared to Haynie (1989), reflects that
households in dirtier areas  may not respond with average behavior but mitigate their behavior
by (1) accepting greater reductions in reflectance before repainting, (2) washing  surfaces
rather than painting as often, or (3) selecting materials or paint colors that do not tend to
show dirt.  Haynie and Lemmons (1990) extrapolated their  findings for houses to all exterior
paint surfaces and produced a range from $570 to $1,140 million per year.
5.     Staff Considerations Pertaining to the Effects of PM on Soiling
       It is clear that, at high enough concentrations, particles become a nuisance and result
in increased cost and  decreased enjoyment of the environment.  The available  data are
limited, however, and do not permit any definitive findings with respect to societal costs or
provide clear quantitative relationships between ambient particle loading and soiling.  Absent
sufficient data, the staff concludes that there is not a sufficient basis to set a separate
secondary standard based on soiling effects alone.  The recommended suite of primary
ambient air quality standards and the regional haze program should reduce the soiling and
nuisance effects associated with particle pollution.   The effects associated with dustfall are
likely to be very localized  and thus, more  appropriately addressed at the local  level.
D.     Summary of Staff Conclusions and Recommendations on  Secondary NAAQS
       This summary of staff conclusions and recommendations for the PM secondary
NAAQS draws from the discussions contained in the previous  sections of this  Staff Paper.
The key findings are:
1)     Anthropogenic fine particles impair visibility. The level of this impairment varies
       greatly from East to West, in terms of total loadings, pollutant mix, and  the resulting
       total light extinction. Background levels  of fine particles, humidity, and resulting
       total light extinction vary regionally as well, with the East having generally higher
       levels than the West.

-------
                                         VIII-25
2)     The levels recommended in this staff paper for protection of public health from the
       adverse effects of fine particles will not completely address the visibility impairment
       of fine particles on visibility or fully achieve the national visibility goal across the
       country.
3)     Because of regional variations  in visibility conditions created by background levels of
       fine particles, annual average humidity, pollutant mix, and resulting total light
       extinction, the staff concludes that a national secondary standard to protect visibility
       would not be an appropriate approach for addressing visibility impairment due to fine
       particles.  Therefore, to address the impairment of visibility from  fine particles and to
       make reasonable progress towards the national visibility goal,  the  staff recommends
       that the Administrator consider establishing regional haze regulations under section
       169A of the Act.
4)     The available data assessed in the CD does not provide an adequate basis to  establish
       a unique national secondary standard to protect against soiling and materials  damage
       effects.  The staff recommends setting  a secondary standard equivalent to the primary
       standards  for the purposes  of addressing soiling and materials  damage.

-------
                                         A-l
                                    APPENDIX A
   CONSIDERATIONS IN SELECTING PARTICLE SIZE CUT POINT FOR FINE
                                     PARTICLES

       An important decision relating to the choice of indicator is the choice of measurement
which in a sense serves as an operational definition of fine particles.  The CD concludes that
the minimum of mass between  the fine and coarse modes lies between 1  and 3  /un, and that
the scientific data support a cut point to delineate fine particles in this range (CD, Chapter 3-
5).  Because of the overlap of fine and coarse particles in  this intermodal region, specific cut
points are only an approximation of fine particles. Thus, the decision within this range is
largely a policy judgement.  Although most fine particle (accumulation mode) mass is below
1.0/xm, some hygroscopic particles  in conditions of high relative humidity may gain water
and grow above this size.  However, energy considerations normally limit coarse mode
particle sizes to greater than 1.0 ^irn in diameter (CD, 3.1.2).
       The main policy choice centers on  two options: PM2 s and PM,.  Staff recommend
the three primary factors to consider in  selecting a cut point are consistency with health data.
potential for intrusion of mass from  the other mode, and availability of monitoring
technology.
       From a public health perspective, use of a PM2 _, cutpoint will  result  in the capture of
all of the potential agents of concern in  the fine fraction.   For example,  the cutpoint of PM?.,
captures most sulfates, acids, fine particle metals, organics, and ultrafine particles and
accounts for most of surface area, and particle number.  Although the CD outlines some
conditions (e.g., relative humidity near  100 percent) under which it is possible that
hygroscopic particles may  grow above 2.5/xm, use of  the PM25 cutpoint  is still better  at
capturing the constituents of concern than  PM,.
       PM2 5 has been measured directly in many health studies as described in the CD and
Chapter V, Section F above.  Significant associations  have been reported between PM2 5
concentrations and mortality, hospital admissions, cough, upper respiratory infection,  lower
respiratory infection, asthma status,  and pulmonary function changes.

-------
                                         A-2
       PM2 5 measurement technologies are widely available and have been in routine use in
the field since  the early 1980s.  For example,  the EPA AIRS database contains PM2 5 data
from the Inhalable Particle Network (1982-1984), the IMPROVE network (1987 - present),
and the NESCAUM network (1988- present).  In addition, the California Air Resource Board
(CARB) dichotomous sampler network has been collecting PM25 data routinely since 1980,
and many  other special studies measuring PM25 have been conducted across the country.
Furthermore, dichotomous samplers allow the coincident measurement of PM10 and PM25,
increasing the  certainty of comparability between the two measurements.
       Measurement of fine particle mass using a 1 pm  (PM,), on the other hand, has not
been used  in health studies primarily due to lack of available monitoring data.  Comparisons
between PM, and other measurements that were used in  the health studies (e.g., PMUI) are
also not widely available due to lack of available PM, monitoring data.  Furthermore, PM,
may not capture  as much of the hygroscopic substances  such as sulfates which health studies
report as having  statistically significant associations between sulfate measurements and
endpoints  including increased mortality and hospital admissions.
       PM, sampling technologies have been developed  and  some limited validated data are
available from  locations such as Phoenix, Arizona.  However,  the PM, samplers have not
been widely field-tested to date.
       Proponents of the PM, option are concerned that  the intrusion of particles generated
by grinding or crushing (i.e., coarse mode particles) into the daily PM^ measurement could
create spurious NAAQS exceedances.  Given the lack of PM, data currently available, it is
difficult to determine how much intrusion might occur or what areas  might be affected during
the implementation of a PM2 5 NAAQS.  The available data show that typically only 5-15
percent (on the order of 1 to 5 /xg/m3) of the PM25  mass is attributable to soil-type sources
even in dusty areas such as San Joaquin  Valley, California, and Phoenix, Arizona.
However,  this  percentage may  increase during events such as high winds.
       The staff judges that in  typical urban areas, the potential for this type of intrusion
may be smaller,  but without sufficient data these determinations remain very uncertain.  A
sharper inlet for  the Federal Reference Method may help to minimize the intrusion of coarse
mode particles into the PM2 5 measurement.  Although intrusion of coarse mode particles into

-------
                                          A-3
daily PM2 5 measurements is not anticipated to be significant in most situations, if in light of
more data a problem is identified, this issue  might be better addressed on a case-by-case
basis in the monitoring and implementation programs.
       Finally, the staff concludes that PM2.5 measurements are more appropriate than some
of the measurements historically used in the  epidemiological studies (e.g.,  BS, CoH)
although these measurements have been useful in advancing the state of scientific knowledge
of particle effects. British Smoke (BS) readings vary more with darkness of particles (i.e.,
carbon content) than with mass,  making associations with mass highly site- and time-specific.
  The BS method emphasizes control of primary elemental carbon emissions; however,
elemental carbon is a minor contributor to fine and total mass in current U.S. atmospheres.
Furthermore, lack of consistent relationships between BS reflectance and PM mass
measurements diminishes one of the major advantages:   BS is not related to the available
quantitative health data from U.S. cities with as much certainty as the PM25 mass
measurements although BS is used in many other countries.  Using a  similar principle to BS,
the principle of coefficient of haze (COH) is that visible light is transmitted through (or
reflected from as in the case of BS) a section of filter paper before and after ambient air is
drawn through it.  Thus,  COH associations with mass are also highly site- and time-specific.
       Thus, because of the consistency with health data, small  potential for intrusion, and
availability of monitoring technology and existing air quality database, the  staff judges that
the PM25 measurement is more appropriate for regulatory purposes than PM^ or historical
measurements such as BS or COH.

-------
                                          B-l
                                     APPENDIX B
          MEASUREMENT METHODS FROM EPIDEMIOLOGY STUDIES
       The CD and Chapter V of this Staff Paper summarize health studies which have
reported associations between various indicators of PM and health effects.  The main mass
concentration indicators are TSP, PM,0, and PM2 5. In addition to PM2 s mass measurements,
fine particles have been measured in the U.S. and abroad using a variety of techniques
including British or black smoke (BS), coefficient of haze (COH), carbonaceous material
(KM), and estimates from visibility measurements (CD, Section 4.2.8).
       Studies have also reported associations between health effects and exposure to
fractions found predominantly in the fine fraction  such as sulfate (SO4=) and strong acidity
(H+).  The CD describes measurement techniques in detail; this section highlights relevant
information about other indicators of fine particles (i.e., BS, COH, and  KM).
         In the past,  it was noted that visibly black plumes were emitted by industrial
sources; thus, light absorption was adopted as a measure of PM pollution (Chow, 1995).
Measurements of the optical properties of particles may be related to gravimetric  mass
measurements on a site- and time-specific basis with on-site calibrations.
       BS preferentially measures elemental carbon particles found in the fine  fraction  (CD,
Section 4.2.8; Baily and  Clayton 1980).  In addition, the BS inlet design, taken together with
its other operating  parameters,  restricts the size of particles that are sampled.   For example,
it has been shown  in wind tunnel tests that the best estimate of the cut point for BS is 4.5 urn
(CD,  page 4-52; Waller, 1980; McFarland, 1979). Most particles larger than  the cut point
of 4.5 /xm are either rejected at the inlet or lost in the inlet line (U.S. EPA,  1982a).
Furthermore, the BS reading varies  more with darkness of particles (i.e., carbon content)
than with mass, thus making associations with mass highly case-specific.  Because elemental
carbon is found predominantly in the fine mass (less than  1.0 /urn range), variations in BS are
more  closely  related to fine mass and unlikely to be sensitive to coarse mode particles (NAS,
1980; U.S. EPA, 1982b).
         Using a similar principle to BS,  COH measures visible light transmitted through
(compared to reflected from in the case of BS) a section of filter paper before  and after
ambient air is drawn through it.  The amount of light transmitted  is measured by a photocell

-------
                                          B-2
(Chow, 1995; Fairley, 1990).  In addition, this sampler uses a funnel inlet and a small
diameter transport tube nearly identical to the BS  sampler.  Although the two samplers
operate at different flow rates, the particles reaching the filter tape could be expected  to have
a size range similar to that of the BS instrument (U.S.  EPA, 1982a, see Figure 3A-12).
       Prior to the 1980s, PM was measured in California by optical reflectance of particles
collected on a sample tape (KM).  Similar in principle  to BS, KM has been shown to  be
closely related to elemental carbon content in Los Angeles (Kinney and  Ozkaynak, 1990).
Similar to BS, KM is also a fine particle measurement.
       Visibility measurements can also be used as a reasonable surrogate to estimate fine
particle concentrations because the extinction coefficient is directly related to fine particle
mass (CD, page 6-216).

-------
                                           C-l
                                      APPENDIX C
     PM10 NATIONAL CONCENTRATION MAPS AND DEFINITIONS OF REGIONS

       Current U.S. PM10 levels are illustrated in Figures C-l and C-2.  Figure C-l shows the
fourth highest 24-hour PM10 concentration recorded in a county and Figure C-2 depicts highest
annual mean PMj0 concentration using 1992 to 1994 AIRS data in each county for which data
completeness criteria were met.  Counties not represented with a  monitor are left  blank.
       The following methods were used to calculate the values depicted in the maps.  The
current single  exceedance form of the PMi0 daily standard allows for an average of one
exceedance per year over a three-year period. Thus, the fourth highest concentration is  of
interest because this value is used to determine attainment with the current daily standard.  Seven
hundred and twelve counties met the data completeness  criterion of at least 75 percent complete
data for the period 1992 to 1994. For these counties, all daily concentrations were ordered
largest to  smallest and the fourth highest PM10 concentration was  determined for each site.  If a
county had only one site, then the fourth  highest concentration for that site was reported. If a
county had more than one site,  the site with the maximum fourth  highest concentration was used
to represent the county.
       Figure  C-2 shows the maximum annual mean concentration in each county over the three-
year period using an average weighted by calendar quarter.  Three hundred and eighty counties
met the 75 percent data completeness criterion by quarter  for 1992 to  1994. Means were
calculated for  all four calendar quarters for each year in the 3-year period and annual values
were calculated based on the quarterly means.  The three yearly means were then  averaged  to
obtain  one value for each site.   If a  county had only one site, then the annual mean for that site
was reported.  If a county had more than one  site, the site with the maximum annual mean was
used to represent the county.
       Figure  C-3 shows the regions of the country used in some air quality analyses.  Note that
state boundaries were used except that California and Texas were split.
       Figure  C-4 illustrates that a total of 87 different  sites reported PM2.S data to AIRS from
1983 to 1993.  Over the 11 year period, less than 50 sites reported data to  AIRS in any  given
year.  Additional special studies have also monitored PM, 5, but these data are not reported in
AIRS.

-------
Figure c-i.    PM-10 Air Quality Concentrations, 1992-94
                     Maximum 4th Highest Daily Concentration
  170 -



  160 '



  ISO -



  140 -



  130 -



  120 '



  1 '0 -

(A
C
o
I 100 -


c
c 90 -
5
15

R so -

o.


ffi ™-
50 -
30 '




20-
                                                                                          n
                                                                                          K)
                   r
                    Concentration (ugMJ)    <     i <55
                                                 «T1 55-104
105-154
>=155

-------
   Figure c 2.   PM-10 Air Quality Concentrations, 1992-94

                             Maximum Annual Mean
  I 'C






  160






  150






  140






  130






  IPO






  110






S 100
c  90

2
  r,o -
  40 -
  io -
                                                                                      9
                                                                                      UJ
                       Concentration (u^rrrt)
                                       : <=30
31-40
41-50
>50

-------
                                   C-4
Figure C-3.  Regions Used in Air Quality Analyses in this Staff Paper
Region
States
 SW    NV UT CO NM  AZ TX(West) CA(South)
 NW    OR WA ID WY  MT CA(North)
 CE    OK MO KS NE IA SD ND MN  WI IL
 NE    IN KY OH  MI VA WV  PA NY  MD NJ  CT RI  MA VT NH ME  DE  DC
  SE    FL GA  AL MS LA TX(East) AR  TN NC  SC

-------
                 Figure C-4. PM2.5 AIRS Data Summary, 1983-1993
Figure C-4a.  Geographic Distribution of Sites
                  AIRS Sites Reporting PM2.5 Data,  1983-93
                          SLAMS (14)    ASPM(49)     OTHER (24)
Figure C-4b.  Number of Sites and Frequency of Sampling
           Number of Sites Reporting, 1983-93
Sampling Frequency, 1983-93
                                                Days sampled
                                                in one year
                                                      0-29
                                                     30-59
                                                    60-119
                 144
                                                     > 120 H20
             10      20      30
                    Number of Sites
                                   40      50
 0     50    100    150
        Number of Sites
200

-------
                                          D-l
                                     APPENDIX D
             I.  HYPOTHETICAL MECHANISMS OF ACTION FOR PM
1.     Dosimetric Considerations
       Dosimetric considerations formed the principle basis of the approach used for
selecting PM10 as the indicator of the current standard (pp.23-39, U.S. EPA, 1982b).
Exposure can be described,  in the context of regulating PM, as the concentration of particles
available in the ambient air  that  a human or animal breathes over a relevant period of time.
Dose is the amount of this material that is inhaled and available for deposition at various
target sites (e.g., regions of respiratory tract) (CD, p. 10-1).  It is the dose that  the target
site or organ receives upon  which  manifestation  of toxicity depends.  The amount of particles
deposited or retained in each region of the respiratory tract is governed by exposure
concentration, particle diameter and distribution, physico-chemical properties of  the inhaled
particle (e.g. hygroscopy and solubility), and duration of relevant exposure.  In the previous
review, such dosimetric considerations, health effects of concern, and aerosol physico-
chemical characteristics prompted the Staff with  CASAC concurrence to determine that the
major risk of commonly occurring outdoor PM was presented by particles of 10 micron or
less aerodynamic diameter.  Particles of this size are able to penetrate the presumptive targets
of PM (tracheobronchial and alveolar regions of the human respiratory tract) (CD, Chapter
10).
       The human respiratory tract can be divided into three main regions:  (1) extra-thoracic,
(2) tracheobronchial, and (3) alveolar regions as shown in Table 10-5 of the CD. They
differ markedly in structure, function, size, and  sensitivity or reactivity to deposited particles
(U.S. EPA, 1982b).  Disposition and retention of initially deposited particles depends  on
clearance and translocation mechanisms that vary with each region of the respiratory tract.
Coughing, mucociliary transport, endocytosis by macrophages or epithelial cells, and
dissolution and absorption into the blood or lymph are important mechanisms of clearance in
the tracheobronchial region.  Endocytosis by macrophage  or epithelial cells and dissolution of
absorption into the blood or lymph are the dominant mechanisms of clearance in the alveolar
region.

-------
                                           D-2
       In essence, ambient particles of 10 pm diameter or less deposit with varying
efficiencies in tracheobronchial and alveolar regions of the respiratory tract.  Simulations of
deposition show that alveolar deposition is fairly uniform for particle between 0.5 and 4.0
fim diameter.  Table V-l of Chapter V is derived  from Tables 10-21 and 10-23 of the CD
and shows the deposition patterns in the human  lung for typical particle distributions found
the cities of Philadelphia and Phoenix.  This table represents the general population of adult
males with normal breathing.  The  table shows not only do all size fractions below 10 /xm
diameter have the potential for some deposition  in both tracheobronchial and alveolar regions
but deposition patterns of the types of particles found in urban areas can be similar in these
lung regions under specific conditions.
       In regard  to sensitive sub-populations, increased deposition and altered clearance may
play a role in susceptibility to PM.  A detailed discussion of these individuals is presented in
section 5-D.  Model simulations have suggested that deposition efficiency of particles will be
increased  in people with COPD and asthma (Anderson,  1990; Miller et al., 1995;
Svartengren et al., 1994).  Kim et al (1988) demonstrated much greater particle deposition in
COPD patients using aerosol re-breathing tests.  A compromised lung with greater deposition
has a greater probability of interaction of PM with potential targets of PM toxicity and thus
increased  effects.  However, the contribution of such differential deposition of particles to
mortality and morbidity has not been elucidated or quantified.
       Similarly, differences in dosimetry between animals and humans may be a
contributing factor for the apparent differences  in animals and human study results.  Rodents
have a greater deposition of particles in the upper  respiratory  tract than humans.  In  addition,
models show that humans retain a greater fraction  of particles deposited in the alveolar
region than do rats or  mice.  Thus, the differences in deposition patterns of particles between
species and between susceptible and nonsusceptible subpopulations could be a contributing
factor for the necessity of using relatively high concentrations of larger diameter particles to
elicit effects seen in experimental animal studies (CD, Chapter 10).

-------
                                           D-3
2.     Possible Mechanisms of Action for Health Effects Associated with Ambient Levels of
       PM Exposure
       This discussion focuses on more specific possible mechanisms by which airborne
particles may be exerting their effects.  Upon deposition, substantial uncertainty still exists as
to how particles,  alone or in combination with other atmospheric pollutants, produce
physiological and ultimately pathological effects.  Because both the population affected and
PM are heterogenous, the mechanism(s) of action may also be diverse.  As shown in the CD
(Chapter 13), exposure to paniculate matter has been identified as causing a variety of health
effects including respiratory symptoms, mechanical changes in lung function, alteration  of
mucociliary clearance, pulmonary inflammatory responses and morphological alteration  in the
lung.  In addition, from epidemiological studies PM has been reported to be associated with
increases in respiratory illness, hospital admissions, and  daily mortality.
       Consequently, the increasing body of community epidemiological studies finding
associations between PM and mortality and morbidity in recent years have prompted a
number of authors to advance potential mechanisms of PM toxicity.  One major area of
interest is pulmonary inflammation.  Potential mechanisms for induction of an inflammatory
response have been described for:  (1) aerosol acidity (Lippmann, 1989a), (2) presence  of
ultrafine particles (Seaton et al.,  1995), and (3) transition metal ions (Tepper et al.,  1994).
A second area of renewed interest includes examination of the ways particles may affect
individuals with preexisting  conditions.  Frampton et al.  (1995) list potential causes of PM
induced mortality as being:  (1) premature death (i.e., hastening of death  for individuals near
death  within hours or days); (2) increased susceptibility to infectious disease; and (3)
exacerbation of chronic underlying cardiac or pulmonary disease.  Also of significant interest
are new approaches for controlled exposures  to particles which are closest to those found
under ambient conditions than have been possible in past toxicologic studies (Sioutas et  al.,
1995).  The opportunity  to study  such  particles may be particularly valuable in studying the
effects from and potential mechanisms of action for PM  exposure.  The issue of
discrepancies between experimental doses and ambient PM in terms of composition and
magnitude of administer  dose may be resolved.  However, early results of such studies  while
promising are preliminary and may be valuable for future reviews. A brief summary of

-------
                                          D-4
potential mechanisms of toxicity is discussed below.  Further discussion is provided in
Chapters 11 and 13 of the CD.
       The most serious effects associated with community studies of PM appear to be found
in individuals who have preexisting conditions.  Even in the London episodes,  the total
amount of inhaled PM by mass eliciting a response in humans  was small.  Therefore, it is
likely that the effect of PM exposure is amplified in conjunction with preexisting conditions
that increase risk for PM effects. Given that immunological responses  can be  quite rapid,
consistent with  the period between increased PM exposure and an acute effect  such as
mortality, it is plausible that inflammatory processes  can amplify and spread the response
from small amounts  of PM.
       Preexisting inflammation (e.g., from an  ongoing infection) of the lung can amplify the
inflammatory response to residual fly ash in emphysemic rats (Costa  et al., 1995).  Indeed,
several of the risk factors for PM toxicity involve inflammatory response (e.g., asthma,
COPD, and infection).  A similar profile of susceptibility may be shown by the only animal
deaths recorded during the London  Fog  of 1952 linked to the fog.  These were prize show
cattle which suffered from both shipping fever and emphysema.   Thus, the cattle which
shared susceptibility to the London  fog with humans  may also  share some of the same pre-
existing conditions (e.g., COPD and inflammation).    A commonly offered explanation of
the susceptibility of the show cattle was that they were kept in cleaner stalls and thus had
much lower waste ammonia present that might  serve to neutralize the high levels of acid
aerosol portions of the fog and thus decrease  their toxicity.  The original report by the
Ministry of Health (MOH, 1954), however,  also reported cattle death  in previous  fogs with
ordinary stall maintenance and therefore high ambient levels of ammonia that could neutralize
acid particles.
       Seaton et al., (1995) has proposed the hypothesis that the mechanism of PM involves
production of an inflammatory response by ultrafine particles (< 0.02 /xm diameter) in the
urban paniculate cloud.  As a result, mediators may be released capable of causing
exacerbation of lung disease in susceptible individuals and increased coagulability of the
blood.  Thus a rationale is provided for the observed increase  in cardiovascular deaths
associated with urban pollution episodes.  Several hematological factors, including  plasma

-------
                                           D-5
 viscosity, fibrinogen, factor VII, and plasminogen activator inhibitor are not only known to
 be predictive of cardiovascular disease (Lowe, 1993) but to also rise as a consequence of
 inflammatory reactions.  Low grade inflammation has been hypothesized to be particularly
 important in altering the coagulability of blood as a result of activation of mononuclear cells
 in the lung  (MacNee and Selby,  1993).  Activated white cells may initiate and promote
 coagulation (Helin,  1986) via the final clotting pathway (Ottaway et al., 1984).  Alveolar
 inflammation may also cause the release of interleuken - 6 from macrophages and thus
 stimulate hepatocyte to secrete fibrinogen (Akira and Kishimoto,  1992).  Crapo et al., (1992)
 has suggested that activation of lung macrophages in  the absence of recruited neutrophils
 leads to acute damage of capillary endothelial cells as well as alveolar lining cells, resulting
 in intracellular edema, hemorrhage and fibrin deposition.
       In support of Seaton's proposed mechanisms is the observation that ultrafine particles
 cause greater  inflammation (assayed by broncho-alveolar lavage) than larger particles of the
 same substance (Chen  et al., 1992; Oberdorster et al., 1992). Fine particles have been
 shown to be taken up by lung epithelial cells (Stringer et al., 1995) and lung macrophages
 (Godleski et al.,  1995).  They have also been shown  to produce inflammation in vitro (Dye
 et al., 1995) and in  vivo  (Kodavanti et al., 1995).  In addition, metals have been shown to
 increase the toxicity of particles.  Intertracheal instillation of residual oil fly ash into rats also
 produces an inflammatory response (Jaskot et al., 1995) with Dreher et al.,  (1995) linking
 such inflammation to soluble vanadium, iron, and nickel compounds on the  particles.  Ferric
 sulfate has been shown to alter pulmonary macrophage function (Skornik and Brain,  1983).
 In support of an inflammatory component to  PM toxicity are several recent  reports involving
 diesel particles which have ascribed observed inflammatory/tumor promoting effects to
 carbon cores rather than  adsorbed organic (CD, Chapter 11, Section 11.5.5). Thus, under
 this proposed  mechanism of PM effect, toxicity may involve a response to PM which
 involves inflammation.
       Aggravation of underlying conditions  (chronic cardiopulmonary disease in particular)
 has been observed in epidemiologic studies as increased hospital admissions for such
conditions and decreases  in pulmonary function.  Aggravation of severity  of these conditions
has also been hypothesized to explain increases in daily mortality and longitudinal increases

-------
                                          D-6
in mortality.  Under such a scenario individuals experience more frequent and severe
symptoms of their preexisting disease or a more rapid loss of function.
       Airflow obstruction could result from laryngeal constriction or broncho-constriction
secondary to stimulation of receptors by PM in the extrathoracic or intrathoracic airways.  In
addition, stimulation of mucous  secretion could contribute to mucous plugging in small
airways. In pre-existing airway diseases, which feature localized airway narrowing or
obstruction, the increased accumulation of PM  may lead to hypoxia in the respiratory regions
of the lung served by the obstructed airways.  In tandem under such condition, there also
may be an  increased particle deposition and adverse effects on the non-obstructed areas of the
lung (CD,  p. 11-184).  Finally,  effects on the surfactant layer in the alveoli by PM may
cause increased leakiness in the  pulmonary capillaries leading to interstitial edema.
Experimentally, acid aerosols have been shown to cause acute effects on pulmonary  function
among some sensitive individuals.  They may induce hyper-reactive airways after 75 /xg/m3
H2SO4  for 3 hours (El  Fawal and Schlesenger,  1994).  Therefore, the elderly with
debilitating disease such as asthma may be stressed by the fine acid aerosols.
       In regard  to particle size, Thurston  et al., (1994b) have reported that hospital
admissions for asthma were more strongly  associated  with fine rather than coarse fraction
particles.  Aggravation  of  asthma symptoms has also been reported for fine particles (Ostro
et al., 1991; Perry et al.,  1983).  In studies of cellular and immunological injury with PM
inhalation, Kleinman et al. (1995) reports that in eliciting responses 0.2 ^m diameter SO42 is
greater than 0.6 /xm diameter NO3", which  in turn is greater than 4^cm diameter resuspended
road  dust.  Measures of alveolar cord length and cross sectional area were most reduced with
the fine sulfate particles which could result in a decrease  in compliance or "stiffening"  of  the
lung  and smaller inflation  volume.
        Related to the potential for aggravation of underlying disease by PM is the issues of
whether increases in mortality reported to be associated with PM are a result of hastening of
imminent death.   While this is  a plausible and reasonable suggestion, other evidence
suggests that it may not explain the full effects of PM on mortality.  For example, in
interviews with the family members of victims of the London pollution episode of 1952,
while some of those victims were reported to having  chronic pre-existing conditions and

-------
                                          D-7
some having infections, several were reported to have no indication of a life threatening
disease process (Ministry of Health, 1954).  As reported by the CD (Chapter 13), it appears
likely that life shortening from PM exposure is highly variable and could range from days to
years.  The CD concludes that duration life  shortening, lag times, and latent periods of PM-
mediated ,nortality are almost certainly distributed over long time periods.   However,
confident quantitative determination of specific estimates of years lost to ambient PM
exposure is not possible at this time.
       There are  several potential  targets for PM  throughout the respiratory tract which  may
involve stimulation of airway neurological receptors to elicit observed health effects (e.g.,
bronchoconstriction and mucous secretion).  The tracheal bronchial tree has been described
as the dominating site for vagal reflexes affecting the airways and most definitely associated
with common conditions such as asthma and chronic bronchitis (Widdicombe,  1988).
However, respiratory receptors which can effect cardiac as well as other pulmonary effects
are distributed through the respiratory tract.  For  example, "irritant"  receptors reside in  the
epithelium  from trachea to respiratory bronchiole, that produce bronchoconstriction and
reflex contraction of constrictor muscles of the larynx as well as secretion of tracheal mucous
(Widdecombe,  1988).  "C" receptors are distributed throughout the tracheobronchial tree and
in the alveolar wall,  and probably  also in the laryngeal mucosa (Sant1 Ambrogio, 1982;
Coleridge and Coleridge, 1986). They have some of the same actions as "irritant"  receptors
and are activated  by  the same group of stimuli (Widdicombe 1988).  Most of the lung
inflammatory and immunologic conditions such as asthma  and chronic bronchitis would
probably activate  C and irritant receptors, which would interact to cause augmented airway
responses (Widdecombe 1988).  "J" receptors, which reside in the alveolar wall, can elicit a
powerful constriction of the larynx as  well as bronchoconstriction. The main activation  of
these receptors  occurs in pathological  changes in pulmonary circulation and the alveolar  wall
rather physiological conditions (Widdcombe, 1974, 1988). Lung  pathologic conditions (e.g.,
edema, pulmonary congestion, pneumothorax, microembolisms and anaphylaxis) as well as
various irritant  gases (e.g., cigarette smoke, sulfur dioxide, and ammonia) and a wide range

-------
                                          D-8
of mediators (e.g., prostaglandins and histamine) have been shown to stimulate lung
"irritant" receptors.  Irritant gases have been shown to stimulate both lung "irritant" and "J"
receptors (Widdecombe 1974, 1988).
       Cessation of cardiac activity is often the terminal event in life.  Pulmonary responses
to PM exposure may include hypoxemia, broncho-constriction, apnea, impaired diffusion,
and production of inflammatory mediators that can contribute to cardiovascular perturbation
(CD, p. 13-71). For example, hypoxia can precipitate cardiac arrhythmias and other cardiac
electrophysiologic responses that may lead to ventricular fibrillation and ultimately cardiac
arrest. In addition stimulation of many respiratory receptors have direct cardiovascular
effects such as bradycardia and hypertension (C-fibers,  nasal receptor or pulmonary J-
receptor, and laryngeal receptors) and arrythmia, apnea and cardiac arrest (laryngeal
receptors) (CD, p. 13-72).
       Particles that may deposit in the lung over time  may induce an inflammatory response
that  could lead to pulmonary fibrosis and impaired pulmonary function.  With repeated cycles
of acute lung injury by PM and subsequent repair, fibrosis may develop.  Persistence of toxic
particles  may also promote a fibrotic response (CD, p.  13-72).  Large lung burdens of
particles  of even relatively low inherent toxicity have been shown to cause lung cancer in rats
(Mauderly et al., 1994).  While there is difficulty in elucidating how long-term particle
accumulation can induce acute mortality, it may be a factor for the elderly who have been
chronically  exposed to PM in the work place,  those who have resided in heavily
industrialized  cities before effective control of PM, or smokers.  As reported in the previous
section, sensitive subpopulations with obstructive pulmonary diseases may have focalized
particle accumulation in their lungs due to ventilation abnormalities.  However, the
mechanism by which prior exposure to paniculate  could predispose an individual to acute
PM  effects  is  unknown.
       Impaired respiratory defense has also been  proposed as a contributing factor to PM
toxicity.   Patients with pneumonia have increased risk of mortality and morbidity from PM
exposure.  Cough, bronchitis, and lower respiratory illness have been reported to be
associated with increased ambient particle concentrations (CD, Chapter  12, see below).

-------
                                           D-9
       Both mucociliary transport and macrophage function are critical to host defense
against inhaled pathogens.  Potentiation of inflammation and infection  from biologically
active particles (e.g., spores, fungi, and bacteria) may result from effects on clearance and
macrophage function by the acid aerosol component of PM (CD, p.  13-75).  Increased risk
of infection has been associated with changes in mucociliary clearance (e.g., excessive mucus
secretion into the airways can cause airway blockage and reduced clearance).  Alveolar
macrophages are the primary defense cells of lungs and impairment of their function would
also be expected to increase risk of infection. Clearance and macrophage function have been
shown experimentally to be affected by constituents of PM, notably fine acid aerosols.
       H2SO4 and  trace metals have been  shown to have direct effects  on alveolar
macrophages in animal  experiments (CD, p. 13-75). Kleinman et al. (1995) also reported  in
their study of cellular and immunological  injury by PM that antigen binding to receptors in
and respiratory burst activity by macrophages was depressed by  exposure to fine (0.2/im
diameter)  SO4~2 particles.  H2SO4 has also been  shown  to affect mucociliary
transport and, in combination  with ozone, resistance to bacterial infection.  However, these
effects have been shown at concentrations which are much  higher than those reported in the
recent epidemiological studies for which PM effects have been reported. Effects mediated
through clearance, in particular, would be expected to  be manifested over an extended period
of exposure rather than  a few  days.  While  impaired host defense may not be plausible as a
mechanism for mortality associated with short-term fluctuations of PM level, it may
contribute to the long-term exposure mortality.  In addition, the lag-time reported between
PM concentration elevations and general  indicators of morbidity  (e.g.,  missed school and
work loss days) is consistent with an increased  susceptibility to infection which may
precipitate respiratory symptoms (see discussion in section  V.C).

-------
                                        D-10
 II.  EXTRAPOLATION OF RESULTS FROM LABORATORY STUDIES TO THOSE
       OF EPIDEMIOLOGIC STUDIES: STRENGTH AND LIMITATIONS OF
                  CONTROLLED HUMAN AND ANIMAL STUDIES
       As discussed above, the adverse effects of paniculate matter exposure have been
shown to be consistent between historical and more recent studies.  The effects can be severe
and tend to be concentrated in sensitive sub-populations who have pre-existing conditions or
characteristics that tend to make them vulnerable to respiratory insult (the very young and
old, asthmatics, COPD patients, patients with pneumonia etc). The additional risk of
reported mortality and morbidity from paniculate matter exposure is relatively small in terms
of the whole population.  Therefore, large numbers of people must be exposed before effects
can be discerned in studies.  The question arises as to how to elucidate the mechanism of
action of paniculate matter in  humans.  What are the considerations that must be taken into
account when an analysis of the body of human clinical data and experimental animal  work is
done in order to infer a plausible mechanism for paniculate matter effects?
1.     Numbers of Individuals Affected
       An issue of primary concern is that of statistical power.  The nature of the effect
described in epidemiological work is consistent, and serious, but occurring in a relatively
small fraction of the total population (1 in a million increased risk for daily mortality).
Therefore, theoretically a relatively  large number of animals would be needed to  mimic the
frequency  of response at similar doses. The use of a similar number of animals to mimic the
frequency  of response to ambient air concentrations of particles which have been  associated
with effect in humans is impractical. Therefore, in many  experimental paradigms, relatively
large concentrations are often  given investigate the response from a limited number of
animals.  However, the questionable relevancy and  sensitivity of such  paradigms  limits their
use in the  determination of the mechanism of action of relatively low changes in
concentrations of inhaled paniculate matter.

-------
                                          D-ll
2.     Heterogeneity of Human Population
       The human population for which the effects are most demonstrable are a sub-
population from a genetically heterogenous group. Furthermore, consistency of response is
highly variable among  the population at risk (e.g., a relatively small group of asthmatics
have aggravation of symptoms and not all patients with pneumonia or COPD die as  a result
of an increase in inhaled particle concentration).  The CD suggests that for clinical studies
involving asthmatics, differences among subjects may explain in part the differing results
between  laboratories who study effects of acid aerosols.  As an example of differential
susceptibility to a respiratory insult,  a minority of individuals (3-5%) who are exposed to
etiologic agents responsible for hypersensitivity pneumonitis (allergic alveolitis) will  develop
disease.  Determinants of susceptibility for that disease have been described as both  the
genetic constitution of  the individual and the presence of preexisting lung disease. Similar
factors probably play a role in susceptibility to inhaled particulate matter effects.
       By contrast experimental animals are bred as much as possible to be homogenous
genetically so as to give great consistency in response.  They are also usually studied in their
prime in regard to age and general health.  Presence of disease is generally  considered  to be
a confounding factor to be stringently controlled in most animal paradigms.   As stated above,
those segments of the general population most affected from PM10 exposure are the  sick,  the
very young, and the old.  Therefore the sensitivity of studies using relatively small numbers
of healthy, genetically  homogenous, laboratory animals who are in their prime is diminished
in exploring mechanism of particulate matter effects.
3.     Heterogeneity of PM)P Composition
       Another key element helps to frame the discussion of the  relevance of human clinical
studies and experimental animal work to establish a mechanism of action of particulate matter
in humans.  That is the issue of heterogeneity of both the composition of and exposure to
particulate matter.  Particulate matter is a broad class of physically and chemically diverse
substances (as described  in Chapter IV).  The PM10 fraction is composed of two distinct sub-
fraction  of particle:  fine and coarse particles.   PM,0 samplers collect all of the fine particles

-------
                                          D-12
and a portion of the coarse ones.  There is a fundamental uncertainty regarding which
components or properties of paniculate matter is essential to the observed effects in humans.
       Coarse particles are typically composed of re-suspended dusts from fields and streets
and may contain metal oxides of silica, aluminum, magnesium, titanium, and iron.  Coal and
oil fly ash, calcium carbonate, sodium chloride, sea salt, small pollen, mold spores, and
plant parts may also be present.  Fine particles are generally composed of sulfate, nitrate,
hydrogen ion, elemental carbon,  organic compounds, biogenic organic compounds such as
terpenes,  and metals such as iron, lead, cadmium, vanadium, nickel, copper, and zinc.
Some materials  which are more typically  found  in the coarse fraction, may be also found the
fine fraction.  Similarly, some materials typically found in the fine fraction may also be in
the coarse fraction due to particle growth  in conditions of high relative humidity (e.g.,
sulfates).  Additionally, the properties of  PM10 vary greatly from place to place because of
differences in source mixes and atmospheric conditions.
       Thus unlike a typical experimental paradigm, where the agent to be studied is isolated
and the effects  of exposure described in well controlled studies, the heterogeneity of the PMU,
entity forces a different experimental approach.  Typically constituents of the fraction  are
tested individually to see if effects similar to those observed in humans are reproduced.
Consequently, animal studies are further weakened in regard to ability to establish a
mechanism of action of particulate matter and to either refute or validate epidemiological
observation of effect in humans.
4.     Dosimetric Heterogeneity
       Finally, dosimetric comparisons between laboratory animals and humans, show that
there are  significant differences in the respiratory architecture and ventilation of the two
which adds additional complication to comparisons of experimental and observed data.
Ventilation differences coupled with differences in upper airway respiratory tract structure
and size,  branching pattern, and structure of the lower respiratory tract occur between
species as well as between healthy versus diseased states. These differences may result in
significantly different patterns of airflow affecting particle deposition patterns in the
respiratory tract (CD, Chapter 13).  Additionally, inter-species variability in regard to cell
morphology, numbers, types,  distribution, and functional capabilities between animal and

-------
                                          D-13
human respiratory tracks, leads to differences in clearance of deposited particles which may
in turn affect the potential for toxicity. (CD, Chapter 13).  Consequently the difficulty of
using experimental animal data to investigate paniculate matter effects is further defined.
5.     Lack of Distinct Disease Pathology
       The background levels of cardiopulmonary disease as the cause of death for the
general population is very high.  Given that COPD and heart diseases are frequent causes of
death, it is difficult to  discern those who die from the additional effects of paniculate matter
from  those already dying from such diseases and to do autopsy to identify a specific
pathology associated with paniculate matter caused mortality. Even in historical  studies
involving higher levels resulting in more pronounced  effect it is hard to get an adequate
characterization of pathology related to paniculate matter  effects.  Thus without such a
characterization of the pathology of paniculate matter induced mortality, development and
validation of appropriate models to study such effects are  more difficult.
6.     Lack of Appropriate Equivalents to Epidemiological Endpoints
       Animal toxicological equivalents of such epidemiological endpoints as hospital
admissions and emergency room  visits as an indication of morbidity cannot be obtained.
Although mortality can be recreated in a laboratory setting, the relevance of mechanism is
currently an issue.  In  addition, there is question as to what the most appropriate measure of
paniculate matter is in regard to its toxicity. Specifically is it the inhalable  mass which is
the most relevant metric of the toxic quantity of paniculate matter or is it the number of
particles which reaches specific targets?  Particles may have  low inherent toxicity at one size,
yet greater potency at  another (CD, Chapter 11).  A recent study by Chen et al.  (1995)
confirmed that the number of particles in the exposure atmosphere not just total mass
concentration  is an important factor in biological responses following acidic sulfate inhalation
(CD, Chapter 11).  Specifically,  ultrafine particles with a diameter of 20 ^m have an
approximately 6 order of magnitude increased number than a 2.5 ^m diameter particle  of the
same mass concentration (CD, Section 11).  Comparisons  of particle number and size are
shown in Table  11-1 of the CD.
       In addition to considerations of dose (inhalability and appropriate metric), the nature
of the response to particles and correlations of the appropriate response to susceptible

-------
                                          D-14
population are yet to be resolved. Thus,  identification of the dosimeter which induces
mortality and morbidity has not been elucidated with consequent difficulty interpretation and
design of controlled animal and human studies.

-------
                                          E-l
                                       Appendix E
               CONCENTRATION-RESPONSE RELATIONSHIPS FOR
              MODEL SENSITIVITY ANALYSIS IN RISK ASSESSMENT

       The interpretation of specific concentration-response relationships is understood to be one
of the most problematic issues at this time for the assessment of health risks associated with
exposure to ambient PM.  The approach to addressing this issue taken in the risk assessment
discussed in Chapter VI and in the technical support documents (Abt Associates, 1996a,b) is to
consider alternative concentration-response models through a sensitivity analysis. The sensitivity
analysis is intended to develop ranges of estimated risks, without attempting to develop any single
best estimate of health risks One of the elements needed to  frame such a sensitivity analysis is the
development of alternative PM concentration ranges over which reported concentration-response
functions would be applied. Alternative approaches to identifying appropriate PM concentration
cut-points which define the lower end of such ranges are discussed below.  The application of
these approaches to a number of epidemiological studies using PM10 and PM2 5 indices of
exposure for mortality, hospital admissions, and respiratory effects in children is also presented.
A.     Alternative Approaches to Defining Concentration Cutpoints
       The characterization and interpretation of observed PM concentration-response
relationships are of particular importance in adequately assessing risks from ambient PM
Varying degrees of uncertainty exist concerning the PM concentration-response relationship.
Such uncertainties may limit the ability to discriminate between a range of plausible alternative
concentration-response relationships, and this in turn weakens the ability to estimate potential
risks associated with exposure to PM, especially at low ambient concentrations1.  Key issues for
consideration include: 1) what tests and procedures have been done to examine the possibility of
linear versus nonlinear dose-response relationships, 2) to what degree do statistical uncertainty
and inadequate power preclude exclusion of different alternative concentration-response

   1 The terminology of "low" or "lower" concentrations is used  to simply refer to observed PM
concentrations generally within the lower half to twenty-five  percentile of the reported
observations,  rather than any concentrations "lower" than those observed

-------
                                           E-2
functions; and 3) how factors such as measurement error or copollutants may potentially obscure
an underlying concentration-response relationship substantially different and possibly less linear
than the reported apparently linear relationship.
       Epidemiological investigations of PM generally have taken several approaches to
addressing the shape of the concentration-response relationship.  A number of investigators have
addressed possible non-linearity in this relationship by the use of categorical variables (CD, p.  12-
18). Using categorical variables (e.g., quintiles, quartiles) disaggregates the PM concentration
spectrum into discrete ranges, and allows risk estimates to be generated independently for each
interval.  This may increase the likelihood for detecting those ranges of PM concentrations that
may be associated with little risk from those associated with substantially higher risk  However,
by partitioning the PM data into smaller groups, this procedure may increase the impact of
measurement error and reduce the statistical power of the analyses.  (CD,  p. 12-18). More recent
studies (1993-on) have used various nonparametric approaches—locally estimated smoothing,
cubic splines, etc.— applicable in Generalized Additive Models to  allow better assessment of
nonlinearities in the PM concentration-response relationships, as well as control for confounders
such as weather, season, and time trends (CD, p. 12-19). In addition, potential nonlinearity in
these nonparametric concentration-response models are often assessed through statistical tests as
well.
       In the base case risk analyses described in Chapter VI, reported linear concentration-
response functions have been applied across the range of reported PM concentrations, when
available, with estimated risk never being quantified below estimate of PM background
concentrations.  However, given the uncertainty concerning PM concentration-response
relationships, especially at lower concentrations, alternatives to the base case assumptions are
examined through a sensitivity analysis.  Of particular interest is the possibility of substantial
nonlinearity - i.e., a less steep or zero slope in PM concentration-response relationships at lower
concentrations.  To address such possibilities, concentration-response information from key
studies can be assessed to determine for which concentrations it may be most reasonable to posit
a reduced or zero slope in the concentration-response relationship.

-------
                                            E-3
       Several approaches to determine possible outpoint PM concentrations of particular interest
for use in modeling alternative concentration-response relationships are discussed below. Staff
recognizes that  no consensus exists on the best approach to identify, test, or interpret the effect
cf such cutpoints on concentration-response information.  Detailed evaluation of concentration-
response relationships is made more difficult by a lack of information on data densities and
confidence intervals (CD, 12-310-311).  Given these circumstances, alternative approaches are
used to  generate a range of potential cutpoints, with no attempt to identify the best or most
appropriate cutpoint for risk assessment purposes.
       The overall approach taken here is to evaluate the extent to which detailed concentration-
response information from key studies suggests statistical limitations or nonlinearities in PM
concentration-response relationships over the range of PM concentrations observed in the studies.
This evaluation focuses on  lower concentrations ranges, given that several concerns raised about
PM concentration-response relationships center on whether reported linear functions may be
disguising flat or essentially flat relationships (i.e., show no increase in risk) in the lower portions
of the concentration-response relationship  Three approaches, identified as "lower limit of
detection," "minimum mean concentration," and "visual interpretation" are defined below. These
approaches have been used to identify reasonable cutpoint concentrations for the concentration-
response model sensitivity analysis.
•      Lower Limit of Detection: A number of studies present concentration-response
       information which suggests a generally monotonic increase in response as PM increases
       (CD, p. 12-23, 12-309). Even if such studies for which the concentration-response
       information does not suggest a substantially nonlinear relationships across the range of
       data, the ability to detect any potential effects thresholds or other nonlinearities is limited
       by the data (CD, p. 12-309-311).  For example, plots of RR as a function of the quantile
       PM concentrations are inherently not able to detect any nonlinearities that may be present
       within the lowest quantile (CD, p. 12-309-310).  Thus, for studies that only present
       concentration-response information  in quantile plots and do not show apparent
       nonlinearities, the maximum  concentration (the 20th or 25th percentile value for quintile

-------
                                     E-4
and quartile plots, respectively) of the lowest quantile can be considered to be the lower
limit of detection of possible nonlinearities.
       Reported concentration-response relationships using nonparametric smoothed
curves allow a much better assessment of nonlinearities in the concentration-response
model (CD p. 12-19).  Statistical tests can be performed to indicate whether any
fluctuations seen in these smoothed curves reflect a substantially nonlinear overall
relationship that is statistically discriminable from a linear relationship . Limited numbers
of air quality observations can reduce the power of this test, however, and even the visual
presentations of smoothed curves are not able to discriminate nonlinearities in regions
where there are not enough data points to obtain a stable curve shape (CD, p. 12-310).
For studies in which an overall linear relationship cannot be statistically rejected and
substantial nonlinearities are not evident, the lower limit for detection of nonlinearity may
be considered to be around the 10th percentile.  Use of the 10th percentile reflects the
greater sensitivity of these smoothing methods compared to quantile analyses to examine
whether an observed linear relationship appears to hold toward the lower end of the range
of observed concentrations.
Minimum Mean Concentration:  The second approach considered is to use a central
tendency concentration as the cutpoint of interest,  which is generally available for all
studies  The mean (or median) concentration may serve as a reasonable cutpoint of
increased PM health risk since at this point there is generally the greatest confidence (i e ,
the smallest confidence intervals) in the association and the reported RR estimates. The
mean concentration  considered by staff as most informative to test implications of
potential alternative concentration-response functions is the minimum mean concentration
associated with a study or studies reporting statistically significant increases in risk across
a number of study locations, provided that  the monitoring data is sufficient and
representative of the area to which the RR estimate is applied.  Alternatively,  averages of
mean concentrations across a group of locations or studies may be more appropriate if
location-specific data are inadequate.

-------
                                           E-5
•      Visual Interpretation Concentration-response relationships reported by some studies
       sometimes visually suggest that nonlinearities may exist within the range of the data, even
       when PM concentrations are significantly associated with health effects in a linear model.
       Caution is warranted in any visual interpretation of available PM concentration-response
       information, given the limited information provided  and the amount of measurement error
       that often is involved (CD, p. 12-309-311). Use of quantiles can exacerbate this problem
       as it might increase the likelihood of identifying an apparent nonlinearity in the effect
       estimate entirely due to increased uncertainty in each quantiles' smaller sample size.
        In conjunction with the use of these methods to identify cutpoints for estimating adjusted
concentration-response functions, consideration  is given to  adjustments to the slope of the
reported concentration-response relationship.  If an underlying nonlinearity is present, the
reported slope of a linear concentration-response relationship would change both below the
cutpoint concentration (where the reported slope would be  too high) and above the cutpoint
concentration (where the reported slope would be too low). Adjustments to the slopes of such
segments in concentration-response relationships used in this sensitivity analysis are described in
the technical support documents (Abt Associates, 1996a,b).
B      Concentration Cutpoints from Key Studies
       The three methods described above were applied where appropriate to the studies used in
the risk assessment (Table VI-2 in section VLB of this Staff Paper), including both PM10 and
PM2 5 studies where applicable, for mortality, hospital admissions, and respiratory symptoms
effects. As outlined below, judgments are necessary to apply such methods, and staff recognizes
that other judgments could reasonably be made.  However,  staff believes that the approach taken
here is reasonable and results in selected cutpoints that are useful for the purpose of defining
sensitivity analyses that help to address uncertainties in the quantitative assessment of risks based
on the available epidemiological evidence. Following the identification of a number of potential
cutpoints from these alternative approaches, summarized in Tables E-l and E-3, the last section
condenses this information into a few selected cutpoints, for use in the sensitivity analyses
presented in section VI C of this Staff Paper.

-------
                                          E-6
1.      Concentration-Response Relationships Associated with Short-Term PM Exposures
       The potential concentration cutpoints identified in the following discussion of short-term
exposure studies are summarized in Table E-l  for both PM,0 and PM2 5 studies.
       a.     PMip Mortality Studies
       The five studies, conducted in ten locations, included in Table IV-2 which reported PM10
mortality relationships were examined.
       Lower Limit of Detection: This method was applied to the two studies (Birmingham,
Schwartz 1993a; Utah Valley, Pope et al., 1992 and Pope and Kalkstein, 1995) which reported
concentration-response relationships between mortality and PM10 concentrations.  Although some
nonlinearity may be evident in the nonparametric smoothed curve reported by Schwartz (1993 a;
1994g) in the central portion of the range, from approximately 40 - 60 ng/m3 (Fig E-l), these are
concentrations at which mortality risk is elevated (Samet et al., 1995).  Tests failed to indicate the
overall PM-mortality relationship could be statistically discriminated from a possible linear
relationship (p value of 0.7 for rejecting linearity).  The 10th percentile concentration in
Birmingham was reported to be 21 ng/m3 (Schwartz, 1993a).  The nonparametric smoothed curve
reported in Pope and Kalkstein's (1995) reanalysis of Utah Valley mortality (Fig. E-2) was also
reported as not significantly different from linear (p>0.5).  In this study, the 10th percentile
concentration was not directly reported but is likely to be approximately 20 ng/m3, the
approximate midpoint of the lowest quintile reported for Utah Valley by Samet et al. (1995),
These concentrations are consistent with the lower limit of detection for nonlinearities of 20
Hg/m3 PM,0 identified in the CD discussion of PM mortality exposure-response functions (CD,
12-310).
       Minimum Mean Concentration: The lowest mean PM10 concentration reported in these
mortality studies was 30 ng/m3, from Schwartz et al. (1996a). This combined mean, averaged
across the cities in the study,  rather than the lowest mean concentration from any one city in this
study, was judged to be appropriate to use for this purpose, since the single monitors used to
characterize air quality for each city were sited in locations that may underestimate the average

-------
Table E-1.  Potential Concentration Cut points of Interest for Assessing the Sensitivity of Risk Estimates Derived from
                                                   Short-Tcrm Exposure Studies
                                       TOTAL MORTALITY
    Alternative Approaches
 Cone.
(,ug/m3)
Reference
    HOSPITAL ADMISSIONS

 Cone.
(//g/m3)   Reference
RESPIRATORY SYMPTOMS

Cone.
     3)   Reference
  PM,,. STUDIES

    Lower Limit of Detection



    Minimum Mean Concentration


    Visual Interpretation
   20        Pope & Kalkstein,  1996
   21        Schwartz, 1994g


   30        Schwartz et al., 1996a
   37        Popeet al., 1992
   42        Samet et al., 1995
   43***     Cifuentes and Lave, 1996
34-57***     Samet etal., 1995
                            19     Schwartz, 1994e
                            30     Schwartz & Morris,
                                   1996

                            36     Schwartz, 1994f
                            37     Schwartz, 1994d
                                13
         Schwartz et al., 1994
                                30*     Schwartz et al., 1994
                                                                                                                                            m
                                                                                                                                            O\
                                                                                                                                            P
  PM,, STUDIES

    Lower Limit of Detection


    Minimum Mean Concentration



    Visual Interpretation
   18
             Schwartz, et al., 1996a
Schwartz, et al., 1996a
  13**   Burned et al., 1995      12


  19     Thurston et al.,  1994     18*
  15**   Burnett, et al., 1995
                                                                  Schwartz et al., 1994
         Schwartz et al., 1994
   29***     Cifuentes and Lave, 1996
22-36***     Samet et al.,  1995
  Footnotes:    *    Median estimate.
               **   Converted from sulfate data.
               *** Converted from TSP data.  Range for Samet et al.,  1995 reflects elderly and all mortality results, respectively.

-------
 FIGURE E-l.
                                      E-6b
 RELATIONSHIP BETWEEN RELATIVE RISK OF DEATH AND
 PM-10 IN BIRMINGHAM (SCHWARTZ, 1994g)
                              20
                                  40
                                      60
                         i	!	1	r

                        80   100  120   140
                                  PM10 (mioograms/meler cubed)
               The smoothed plot of the relative risk of death veons rulO in Binninghatn. Alabama.
               after controlling for smoothed ftmcoom of time, tempemare. and dew-point lanpcmnre
               (and day-of-week dummy variables) in a generalized additive model. Pcrintwise one-
               standard-error confidence intervals are also shown.
FIGURE E-2.
RELATIONSHH* BETWEEN RELATIVE RISK OF DEATH
AND PM-10 IN UTAH VALLEY (POPE AND KALKSTEIN, 1996)
                     0.0     0.5     1.0     1.5     2.0    2.5    3.0

                       PM-10 Cone (5-day lagged moving average, 100^9/m3 units)

-------
                                           E-7
concentrations experienced across the cities as a whole  The mean concentrations in the three
cities in which statistically significant results were reported ranged from 24 - 32 ug/m3.
       Visual Interpretation.  A quintile analysis of a Utah Valley study provided by Pope et al.,
(1992) suggests that any increased risk associated with the second quintile may be less than the
increases associated with the three higher concentration quintiles (Fig. E-3). Alternatively, Samet
et al. (1995), using quintiles in a slightly different approach, reported that mortality appeared to
increase in the two highest quintiles only (Table E-2).  This information would suggest a possible
cutpoint of interest in the range of 37 (midpoint of quintile showing reducing increased risk in Fig.
E-3) to 42 ^ig/m3 (maximum concentration of quintile  showing no increase in risk in Table E-2)
The staff judges that the weight given these observations should take into consideration the more
recent Utah Valley results discussed above, given the greater sensitivity of the nonparametric
methods that have been subsequently been applied to the Utah Valley data.
       Various analyses have been done on data from Philadelphia examining PM-mortality
relationships using TSP as the measure of PM.  Table  E-l also contains converted PM10
"cutpoint equivalents" from the TSP findings of these  studies that examined TSP concentration-
response relationships when associated copollutants were included in the model.  There are
substantial uncertainties both in interpreting this TSP data in relation to smaller particle indicators
(PM10, PM2 5) (CD, p. 243), especially when evaluation between copollutants is attempted, and
inherent in converting TSP findings into estimates of PM2 5. The method and issues involved in
deriving these PM10 "cutpoint equivalents" are discussed in Section C.
       b.      PJvlin Hospital Admissions Studies
       Studies conducted in seven locations included in Table IV-2 reporting respiratory and
cause-specific hospital admissions relationships with PM10 were examined.
       Lower Limit of Detection  Nonparametric smooth curves of the concentration-response
relationships between PMj0 and pneumonia (Fig. E-4) and COPD hospital admissions in the
elderly in Birmingham have been reported by  Schwartz (1994e).  No apparent  nonlinearities are
observed,  and the relationships are not statistically distinguishable from linearity (p z 0.25). The
10th percentile concentration is approximately 19 ug/m3.  A quartile plot of an analysis of cardiac

-------
                                   E-7a
FIGURE E-3.
RELATIONSHIP BETWEEN RELATIVE RISK OF DEATH AND
PM-10 IN THE UTAH VALLEY (POPE ET AL., 1992)
                 V
                 o
    1.1
                 «   1.0
                 V
                     0.9
                             20     40     60     80

                                PM10 Concentration
                                     100
                      Relative risk of death, by qumtile of PM,. concentration.
TABLE E-2.
RELATIONSHIP BETWEEN RELATIVE RISK OF DEATH AND
PM-10 IN UTAH VALLEY (SAMET ET AL., 1995)
                Table 28. Relative Risks and Confidence Intervals by
                Quintile of Five-Day-Lagged Average PMio for Utah
                Valley Total Mortality. April 1985-December 1989.
                Controlling for Weather
PMio
Quintile
i
2
3
4
S
Range
11.2-27
27-34.2
34.2-42.2
42^-56.2
56.2-296
Relative
Risk
1.00
0.98
0.99
1.04
1.08
95% 0*

(0.89. 1.09)
(0.89. 1.09)
(0.94. 1.16)
(0.97. 1.21)
                * Corrected for constant over-dispersion.

-------
                                       E-7b
FIGURE E-4.
RELATIONSHIP BETWEEN RELATIVE RISK OF PNEUMONIA
ADMISSIONS AMONG THE ELDERLY AND PM-10 IN
BIRMINGHAM (SCHWARTZ, 1994e)
                        ! =
                        I
                                Nonparametnc smooth of counts of pneu-
                        monia adrrusKons (persons per day) versus the con-
                        centration of airborne partioulata matter with an aero-
                        diameter of slO in (PMJ after oontroOing by
                        regression for tonJHerm temporal patterns  and
                        weather. The pointwise 95 percent confidence Units of
                        the smooth curve art also shown.
FIGURE E-5.
RELATIONSHIP BETWEEN ISCHEMIC HEART DISEASE
ADMISSIONS AMONG THE ELDERLY AND PM-10
(SCHWARTZ AND MORRIS, 1996)
                           46
                         c
                         o
                         •o

                         i
                           43  r
                           42
                             20   30   40   50    60   70   80   90

                                         PMJO (tig/ru1)

                      FIGURE 5.  The number of ischemic heart disease (IHD) admis-
                      sions of the elderly in Detroit, Michigan, during 1986-1989 by quar-
                      tite of paniculate matter with an aerodiameter of S10 JUTI (PM,0).
                      The plot is after adjusting for all other covanates.

-------
                                          E-8
hospital admissions for the elderly in Detroit (Schwartz and Morris, 1996) displays increased risk
at and above the second quartile (Fig. E-5), with a 25th percentile concentration of 30 ug/m3.
       Minimum Mean Concentration:  The year-long study with the lowest mean PM10
concentration, 36 ug/m3, reporting significant associations was the Schwartz (1994f) study of
COPD and pneumonia hospital admissions among the elderly in Minneapolis.  This compares
closely to the mean concentration was reported by Thurston et al. (1994) in their study of
summertime hospital admissions in Toronto, with a PM10 mean concentration of 33 ug/m3
averaged across three summers.
       Visual Interpretation:  The quartile plot of Schwartz (1994d) for elderly pneumonia
hospital admissions in Detroit (Fig. E-6) indicates that pneumonia risk may not increase as sharply
for the second quartile of PM concentrations as for subsequent quartiles.  The midpoint
concentration of this second quartile is 37 ug/m3.
       c.     PMio Respiratory Symptoms Studies
       The two studies listed in Table VI-2 reporting PM,0 associations with respiratory
symptoms were examined.
       Lower Limit of Detection: The Six City study (Schwartz et al., 1994) provides
nonparametric smoothed plots for PM10 associations with cough (Fig. E-7) and  lower respiratory
symptoms (Fig.  E-8)   Statistical tests of deviations from linearity for these associations are not
significant. However, the ability to detect nonlinearities is not likely to extend below the 10th
percentile  concentration of 13 ug/m3 PMIO
       Minimum Mean Concentration.  The Six City study (Schwartz et al., 1994) reports the
lower mean PM10 concentration of 30 ug/m3.
       d.     PM7 < Mortality Studies
       There is less available information concerning PM2 5 concentration-response relationships
for mortality in comparison to PMJ0.  However, the Harvard Six Cities study (Schwartz et al.,
1996a) reports significant associations between PM25 and mortality in a combined analysis of six
cities, as well as associations in  individual cities, that indicate that PM2 5 mortality associations
were relatively consistent in magnitude and statistically significant for three locations (Boston, St.
Louis, and Knoxville) with mean concentrations ranging from approximately 16 to 21 ug/m3

-------
                                      E-8a
FIGURE E-6.
RELATIONSHIP BETWEEN RELATIVE RISK OF PNEUMONIA
ADMISSION AMONG THE ELDERLY AND PM-10 IN DETROIT
(SCHWARTZ, 1994d)
                         108
                      i  106
                      •a
                         102
                       «i
                       E
                       «      A
                       i  100 A
                       o
                       a:
                         0.98
                            20   30  iO   50  60   70   80   90

                                        PM,. (m/ml

                        The retains* of pneumontt admissions in the elderty in Detroit.
                  Michigan, by quartila of PM,, a shown. Th« p*ot b after adjusong tor all
                  other covanates. A stepped response w«h inciwwng dose « evrtentwith
                  no evidence (or a threshold.

-------
                                   E-8b
FIGURE E-7.
RELATIONSHIP BETWEEN THE ODDS OF COUGH
INCIDENCE VERSUS PM-10 CONCENTRATION FROM THE
SIX CITY STUDY (SCHWARTZ ET AL., 1994)
                    to
                  §>
                  o
                  o
                 8
                 1
                  IK
                  tr
                        i
                        0
               20
      40

PM10 (ug/m3)
                                  60
                        Relative odds of incidence of coughing smoothed against 3-d
                  mean PM,, (woAn1). controlling tor temperature, city, day of the week, and
                  ozone concentration.
FIGURE E-8.
RELATIONSHIP OF THE ODDS OF LOWER RESPIRATORY
SYMPTOMS INCIDENCE VERSUS PM-10 CONCENTRATION
FROM THE SIX CITY STUDY (SCHWARTZ ET AL., 1994)
                   to
                   
-------
                                           E-9
PM2 5  No concentration-response curves were provided, precluding any visual interpretation of
results presented in terms of PM2 5.
       Lower Limit of Detection:  For this Six City study, a potential cutpoint could be chosen at
the 25th percentile concentration, 9 ug/m3, consistent with similar interpretations of studies
reporting results in terms of quartile plots.
       Minimum Mean Concentration: The PM25 mean of the combined results from this Six
Cities study is 18 ug/m3.
       Visual Interpretation: Consistent with the approach used above for PM10 mortality and
discussed more fully in Section C, Table E-l also gives potential PM2 5 "cutpoint equivalents"
based on conversions of recent reanalyses of TSP/copollutant concentration-response
relationships.
       e.     PM2; Hospital Admissions Studies
       Minimum Mean Concentration.  The only study to examine respiratory hospital
admissions directly in terms of PM2 5 (Thurston et al., 1994) reported mean concentrations for
three summers ranging from approximately 16 to 22 ug/m3, with an overall average of
approximately 19 ng/m3. This is roughly consistent with the more uncertain estimate obtained
from the Burnett et al. (1995) study of sulfates and respiratory and cardiac admissions. The mean
sulfate concentration of 4 4 ug/m3 in that study roughly corresponds to an estimated PM2 5
concentration of 15 ng/m3.
       Lower Limits of Detection: The only study to which this approach can be applied is the
Burnett et al.  (1995) sulfate study which reports that the respiratory and cardiac hospital
admissions from the third quartile were statistically significantly higher than those from the first
two quartiles  combined.  The maximum concentration associated with the bottom two quartiles
was approximately 3.0 ug/m3 sulfate, the 50th percentile value for the nine Ontario monitoring
sites used in the study. To express this finding in terms of a potentially relevant PM2 5 cutpoint of
interest, a site-specific conversion between SO4 and PM2 5 was made using conversion factors for
the three largest cities in the study (Toronto, Ottawa, and Windsor),  resulting in a PM2 5
concentration of roughly 13 ug/m3.

-------
                                         E-10
       f.     PM2 5 Respiratory Symptoms Studies
       Lower Limit of Detection:  The Six City respiratory symptoms study (Schwartz et al.,
1994) found significant relationships between PM25 and cough and lower respiratory symptoms in
children, although it did not provide either separate quantile or nonparametric smoothed plots for
PM2 5.  Consistent with the approach taken for PM2 5 mortality, a potential cutpoint could be
chosen at the 25th percentile concentration of 12 ug/m3 for this study.
       Minimum Mean Concentration: The PM2 5 mean concentration for this study (Schwartz et
al., 1994) was 18 ug/m3.
2.     Concentration-Response Relationships Associated with Long-Term PM Exposures
       The potential concentration cutpoints identified in the following discussion of short-term
exposure studies are summarized in Table E-3 for both PM10 and PM2 5 mortality studies.
       Lower Limit of Detection:  The Dockery et al. (1993)  Six City study provides plots of
long-term mean fine particle concentrations versus adjusted mortality risk for PM10 and PM2 5.
For PM10, increased risks from particles may extend as low as 24 ug/m3, the mean concentration
for Watertown, which shows an increase in relative risk compared to Portage (Fig. E-9).  For
PM25, increased risks may extend as low as 12.5 ug/m3, the mean PM25 concentration for
Topeka, which shows a slight increase in relative risk compared to Portage (Fig. E-10).
       Minimum Mean Concentration: The mean PM,0 concentration for the Six  City study
(Dockery et al., 1993) as a whole was 30 ug/m3 The mean PM2 5 concentration for the Six Cities
study (Dockery et al., 1993) and the mean of the median PM25 concentrations for  each city in the
ACS study (Pope et al.,  1995) were both reported as 18 ug/m3.
       Visual Interpretation.  For PMIO, a case might be made from visually inspecting the results
of the Six City study (Dockery et al., 1993) that risk consistently increases only beginning with St.
Louis,  with a long-term PM,0 mean of approximately 32 ug/m3. For PM2 5,  a similar case might
be made that risk consistently increase beginning with Watertown, with a long-term PM2 5 mean
of approximately 15  ug/m3. Such comparisons, however, are  limited by the small number of cities
in the study. The ACS study (Pope et al.,  1995) provides concentration-response  information for
PM2 5 which appears to more consistently increase at concentrations above the median PM2 5
concentration of approximately  15 ug/m3 (Fig E-l 1).

-------
                                            E-lOa
Table E-3.  Potential Concentration Cutpoints of Interest for Assessing the Sensitivity of
                 Risk Estimates Derived from Long-Term Exposure Studies
                 Alternative Approaches
TOTAL MORTALITY
 Cone.
(/ig/m3)     Reference
               PM,. STUDIES
                 Lower Limit of Detection

                 Minimum Mean Concentration

                 Visual Interpretation
  24       Dockeryetal., 1993
  30       Dockeryetal., 1993
  32       Dockeryetal., 1993
               PM,. STUDIES
                 Lower Limit of Detection

                 Minimum Mean Concentration


                 Visual Interpretation
  12.5      Dockeryetal., 1993

  18       Dockeryetal., 1993
  18       Popeetal., 1995

  15       Dockeryetal., 1993
  15	Popeetal., 1995

-------
                              E-lOb
FIGURE E-9.
RELATIONSHIP BETWEEN MORTALITY RISK RATIOS AND
INHALABLE PARTICLES (PM1S,10) IN THE SIX CITY STUDY
(DOCKERY ET AL., 1993)
               i

4
1.2-


1.1-
4
1.0


s

H
L

W
P T

                    1520263035404660
 Source: CD, Figure 12-8
 FIGURE E-10.
RELATIONSHIP BETWEEN MORTALITY RISK RATE RATIOS
AND PM-2.5 IN THE SIX CITY STUDY (DOCKERY ET AL.,
1993)

                  1.4 r
                  1.3
                  12
                  1.1
                  1.0
                                    H
                                W
                                        J	L
                            10   15  20
                            Fine Partides
                        25   30   35

-------
                                    E-lOc
FIGURE E-ll.
RELATIONSHIP BETWEEN ADJUSTED MORTALITY AND PM-
2.5 IN THE AMERICAN CANCER SOCIETY STUDY (POPE ET
AL., 1995)

o 1000
o
o"
o
\
i | i i i
'1
1

1 •
1 •
> ! I • . •
£ 5°°r i./. ^i
^
u
s
^ 800
« ,J *
I • * •»
• * j % »
• • •
• . •• I • .
< 1 ' .J
f- • * "
o: : •
o
• 1
Q 700 1- « f
t: i * !
H
V3

r
                  600
                     5      10      •=       20      25      30      35
                        FINE PARTICLES  (micrograms per cubic meler)

                     Age-, sex-, and race-adjusted population-based mortality rates
              for 1980 plotted against mean fine paniculate air pollution levels for 1979
              to 1983. Data from.metropolitan areas that correspond approximately to
              areas used in prospective cohort analysis.

-------
                                           E-ll
C.     Potential Effects of Copollutants or PM Measurement Error on Concentration- Response
       Relationships
       The approach carried out in the sections above for assessing whether underlying
nonlinearities exist in PM concentration-response relationships (e.g., resulting from the presence
of biological thresholds) uses existing reported concentration-response relationships. The large
majority of these relationships were derived considering ambient PM concentrations alone (e.g.,
without simultaneous inclusion of copollutants) As discussed in Section V.E., several
commentors have raised the issue that if the observed concentration-response relationship reflect
PM-health effects relationships in which PM is serving as a proxy for other non-considered factors
(e g., the effects of coassociated pollutants, or of total personal exposure to particles) that may
causally give rise to health effects, then analyses of observed concentration-response data that do
not fully take into account the potential role of these other factors may fail to reveal a genuine
underlying nonlinear relationship between ambient PM and health effects.  The failure to consider
these factors, if they have a genuine causal role, may potentially serve to "disguise" nonlinear
concentration-response relationships, and might result in an apparently linear PM concentration-
response relationship in cases in which a genuine nonlinear relationship existed.
       The two factors advanced as issues of particular concern to consider in this regard have
been the influence of coassociated pollutants (Samet et al, 1995; Samet et al., 1996b, Moolgavkar
et al, 1995b; Moolgavkar and Luebeck, 1996, Cifuentes and Lave,  1996, Lipfert and Wyzga,
1995b), and the potential influence of different types of measurement error.  Measurement error
in this context includes concerns over the potential implications that measurements of ambient PM
may not accurately reflect total personal exposures to particles, either exposures to all particles or
at a mininum a subset of particles including particles of nonambient origin (e.g., from indoor
combustion sources).  In both the case of potential effects of copollutants and of measurement
error,  concerns have been raised that available concentration-response relationships may create
erroneous estimates of PM-health effects relationships for risk analyses purposes by failing to
consider the possibility that these unacknowledged factors may alter the shape of the estimated
PM concentration-response relationship.

-------
                                          E-12
1.  Potential Effects of Copollutants on Determining Effects Thresholds
       Several authors have evaluated concentration-response relationships for particles while
simultaneously including other combustion source copollutants as variables in the health effects
concentration-response regression.  Samet et al. (1995) reanalyzed information from Philadelphia
for 1973-1980 simultaneously considering S02 in the model.  One form of presentation they give
to their results leads to the question of whether potential TSP effects thresholds exist when
copollutants are considered simultaneously. Figure 11 of their report appears to indicate a linear
response between mortality and TSP only for TSP > 100 ug/m3 (all ages) or TSP > 60 ug/m3 (age
65+) (CD, p.  12-311). However, the CD also acknowledges that other approaches undertaken by
Samet et al. (1995), such as nonparametric smoothed surfaces simultaneously displaying TSP and
S02 relationships (CD, pp. 335-344), differs significantly from the simple threshold model shown
in  their Figure 11 (CD, p. 12-311).
       Cifuentes and Lave (1996) analyzed a later period in Philadelphia simultaneously
considering two copollutants in the model, SO2 and O3. They presented a number of results from
several different approaches investigating potential thresholds. The CD finds that Cifuentes and
Lave (1996) provides no precise estimate of a change point in the TSP mortality relationship, with
the lower portion of a potential cutpoint relationship not showing significance below 60 ug/m3
and showing general significance at 90 ng/m3 and above (CD, p. 301, Figure 12-32). The study's
authors particularly call out the concentration of 78 ug/m3 as a concentration below which "the
effects of TSP decreased significantly," a concentration representing roughly the midpoint of the
range identified by the CD. Although as pointed out by the CD, the methods applied by Cifuentes
and Lave do not necessarily imply a slope of zero below the tested cutpoints (CD, pp. 301-302),
this central value of 78 ng/m3 TSP will be used to summarize the results of their findings in the
cutpoint  sensitivity analyses for the risk analysis, which does presume a slope of zero below the
cutpoint  (Appendix F).
       To enable the general findings of Samet et al (1995) and Cifuentes and Lave (1996) to be
considered in  the risk analysis, conversion of their TSP cutpoint findings to fine particles (PM2 5)
were carried out.  Such an approach involves substantial uncertainties both in determining both an
appropriate conversion factor to express TSP results as PM2 5 as well as the possibility that

-------
                                         E-13
substantially different results may have been obtained in the copollutant models if PM3 5 data had
been available for inclusion in the model rather than the less robust surrogate measure of TSP,
especially when discriminations between the particle measure and an associated copollutant are
attempted simultaneously in the health model.  As indicated by the CD, there is less basis for
assuming that analogous results would be obtained for other PM indices, such as PM10 or PM2 5
(CD, p. 343).
      With these concerns in mind, conversion factors were derived from information in Table
6-13 of the  CD to allow rough estimates of the potential impacts of application of cutpoints based
on the TSP-copollutant analyses of Samet et al. (1995a) and Cifuentes and Lave (1996) to be
considered.   The Samet et al. (1995) findings were represented by converting the all mortality
and elderly  2-D nonparametric smoothed plot findings (reported in Figure 11 of their report) to
PM25 by using the PM25/TSP ratio (for TSP > 80 ng/m3) of 0.36 for the Inhalable Particle
Network (IPN), 1979-1983, which provided a rough central estimate PM25/TSP ratio of 0.36
(CD, Table  6-13)  The Cifuentes and Lave (1996) findings were converted to an estimated PM2 5
concentration by using the PM25/TSP ratio available from a site reported to AIRS,  1987-1990
(CD, Table  6-13). Applying these conversions, the Samet  et al. (1995)  findings could be
interpreted  as suggesting potential cutpoints in the range of 22  -  36 |ag/m3 for elderly and all age
mortality, respectively, and the Cifuentes and Lave (1996)  findings could be interpreted as
suggesting the potential for a cutpoint of roughly 29 ug/m3 for all age mortality.
      Comparable conversions based on Table 6-13 also can be done for PMIO, although some
additional concern exists for deriving a PM10 /TSP conversion factor for Samet et al. (1995) in
that the IPN dataset that overlapped the period of study provided information only in terms of
PM15. Use of a single monitor operating two years after the study (1982-1983),  which was not
used in determining the PM25 conversion factor for Samet et al. (1995) presented previously
because the earlier, more extensive network was available, would provide a PMIO /TSP
conversion  factor of approximately 0.57.  Use of this factor and a PMIO /TSP conversion factor
of 0.53 for the AIRS 1987-1991 site provides possible PM10 cutpoint concentrations of
approximately 34 - 57 ^g/rn3  for the Samet et al. (1995)  findings  and approximately 43
for the Cifuentes and Lave (1996) findings.

-------
                                          E-14
       For the purposes of sensitivity analyses for the risk analyses, the various outpoints
findings from Samet et al. (1995) and Lave and Cifuentes were represented with a outpoint of
30 /xg/m3 PM25. Given the following considerations: (1) that the Lave and Cifuentes, Samet et
al. (1995) findings for the elderly, and the central tendency of the findings for the elderly and
all mortality for the two studies combined suggest PM10 cutpoints at or below the range of 40
- 45 /xg/m3, (2) the increased uncertainty in estimating PM10 outpoint equivalents for the
Samet et al. (1995) study, and (3) the emphasis of the alternative standards portion of the risk
analysis on PM25, it was judged that there was not a sufficient need to add a separate PM10
cutpoint to the sensitivity analyses above 40 /xg/m3, a concentration that also summarizes the
upper end of the analyses of reported concentration-response relationships in Table E-l (see
Summary  Section D).
2.      Potential Effects of Measurement Error on Determining Effects Thresholds
       Another issue to consider in estimating PM concentration-response relationships is the
potential effects of measurement error.   As  discussed in Chapter V, the term measurement
error in the broadest sense refers to errors or mis-estimation of several forms that can arise
from the use of outdoor monitors to indicate exposure. Measurement  error includes both
errors resulting from errors in the direct measurement of ambient concentrations, and
inaccuracies in the ability of central measurements to proxy for individual exposures, either to
ambient pollutant concentrations or potentially the more broad array of paniculate pollution
from both indoor or outdoor sources to which an individual is personally exposed.
       The potential of ambient exposure measurement error (i.e., either error in the direct
measurement of ambient concentrations or in the ability of a central monitor to proxy for an
individual's exposure to ambient pollutants) to give rise to an apparent more linear-seeming
relationship that can disguise an underlying nonlinear relationship  has  been discussed to some
extent in the air pollution and statistics literature (e.g.,Yoshimura,  1990). However, some
evidence exists  suggesting that the extent of such error may not serve to have large practical
significance for current ambient particle concentration-response relationships. As discussed in
Section V.E., Schwartz et al. (1996a) reported that statistical relationships between ambient
PM2 5 concentrations and mortality were observed even when the analysis was restricted to only

-------
                                          E-15
days with PM25 concentrations of 25 /ig/m3 or below.  A number of other studies (Pope, 1991;
Schwartz etal.,1993a; Schwartz, 1994d; Schwartz,  1994e; Schwartz, 1994f) have excluded
higher PM concentrations (e.g., PM10 concentrations above 150 /ig/m3).  The similar or
slightly larger relative risks observed in these studies when days with high concentrations are
excluded from the analysis suggests that it is unlikely that measurement error is serving to
disguise a nonlinear relationship that extends far into the range of observed concentrations.
These studies also suggest that any "personal exposure measurement error" (errors in the ability
of a central monitor to proxy for an individual's total exposure to indoor and outdoor particles,
or some relevant subset of total exposure such as,  exposures to all outdoor and indoor
combustion sources), if present, may be affecting  reported ambient PM2 5 concentration-
response relationships to  only a limited extent.  If ambient particle exposures are associated
with mortality risk at 25 fj.g/m3 PM2 5 or below, it seems unlikely that a nonlinear
concentration-response relationship with little or no risk for ambient particles may be being
"disguised" by the unacknowledged role of other particle exposures, since relationships
between ambient PM25 and health effects,  in general, would not be expected to be influenced
by exposures to nonambient indoor sources, which are largely independent of ambient
exposures (CD, p. 1-10).
       To allow for assessment of the potential effects on the risk analysis if measurement
errors were found to be substantially affecting the shape of reported concentration-response
relationships,  cutpoint concentrations and slope adjustments of the type described in Chapter
VI can be used to remodel ambient concentration-relationships to reflect hypothetical
measurement error.  For this purpose, although they were originally derived using the results
from other lines of investigation, the cutpoint levels effects selected in Section D of this
Appendix, which provide cutpoints across a substantial portion of the lower range of  ambient
concentrations, can be used to also model  the possibility that measurement errors might be
obscuring a nonlinear ambient concentration response function with little or no risk in this
lower range of concentrations. For example, the possibility that exposure error might be
obscuring ambient concentration-response nonlinearities at cutpoints of 10, 18 and 30 /ng/m3
PM25 can be examined.   Although the very issue raised by concerns about measurement error

-------
                                           E-16
is that these reported functions may "disguise" nonlinearity through the operation of errors in
measurement of exposure, the results of the analyses in Sections A - C.I above generated
generate a set of potential cutpoints that include substantial PM concentrations, and thus for
practical purposes can be used to examine of the potential impacts of substantial measurement
error as well.
D.     Summary
       Staff believes that it is most appropriate to combine the potential concentration
cutpoints summarized in Tables E-l and E-3 into a few cutpoints for the purpose of doing
sensitivity analyses.  Combining information across studies, effects, and alternative approaches
avoids giving undue weight to any particular study or approach.   From these efforts, the
following specific cutpoints judged of use for illustrating the sensitivity of risk analyses results
have been identified:
•      Short-term PM10 studies: 20, 30, 40 jig/m3
•      Short-term PM25 studies:  10, 18, 30 ng/m3
•      Long-term PM10 studies: 24, 30, 32 /xg/m3
•      Long-term PM25  studies:  12.5, 15, 18 fig/m3
       These cutpoints were derived for the purposes of obtaining  a reasonable range of
possible  cutpoints for the purposes of investigating the potential  sensitivity of the risk analyses
results to alternative concentration-response relationships reflecting alternative interpretations
of reported relationships, potential changes in the concentration-response relationships from the
consideration of copollutants, and/or potential effects of different types of measurement error.
The material in Appendix E is not intended to be a critical or rigorous assessment of relative.
weight of evidence for any particular cutpoints from the available  literature.

-------
                                          F-l
                                      Appendix F
                SENSITIVITY ANALYSES OF KEY UNCERTAINTIES
                             IN THE RISK ASSESSMENT
       As indicated in Chapter VI, a number of assumptions are involved in conducting a
quantitative risk analysis of the effects of ambient PM, and any such effort involves a number
of significant uncertainties. Sensitivity analyses are one approach that can provide insight into
the potential effects of uncertainties and selection of alternative input assumptions on the risk
analyses results. The results of a number of sensitivity analyses for the risk analyses are
presented below.  A more detailed discussion of the sensitivity analyses conducted for the PM
health risk assessment can be found in the technical support document (Abt Associates,
1996b).
A.     Sensitivity Analyses of Key Air Quality Uncertainties
1.      Sensitivity Analysis of Alternative Background Concentrations
       An important uncertainty concerning the air quality information used in the risk
analysis involves estimates of background concentrations (see Table IV-3 for range of
estimated background PM10 and PM2 5 concentrations based on Chapter 4 of the CD).   For the
base case PM risk estimates,  effects were quantified across the range of observations in the
original study or to background concentrations,  whichever was higher. For the base case risk
analysis results  reported in Chapter VI, the midpoint of the range of estimated annual
background concentrations has been used.   Tables F-l A and F-1B show the sensitivity  of the
risk estimates to using either the low end of the  annual background concentration range
identified in the CD (5 /xg/m3 PM10 and 2 /ig/m3 PM25 in the eastern U.S.) or the high end of
the annual background concentration range identified in the CD (11 ^g/m3 PMi0 and 5 /xg/m3
PM25 in the eastern U.S.) as the estimate for background concentrations rather than the
midpoint of the range.
       One important point from Table F-l A and F-1B is that the estimates of mortality and
bronchitis  risks associated with long-term exposure to PM do not change  as a result of
alternative background concentrations.  Because these long-term studies relate  health effects to

-------
Table F-1a.  Sensitivity Analysis: The Effect of Alternative Background Levels on
Predicted Health Effects Associated With "As-ls"  PM-10
Philadelphia County, September 1992 - August 1993
Health Effects*
Mortality (all ages)
Hospital Admissions
Respiratory
Hospital Admissions
Cardiac
.ower Respiratory
Symptoms in Children
(A) Associated with short-term exposure
(B) Total Respiratory
(>64 years old)

(E) Ischemic Heart Disease
(>64 years old)
(C) COPD
(>64 years old)
(D) Pneumonia
(>64 years old)

(F) Congestive Heart Failure
(>64 years old)
(G) Lower Respiratory Symptoms (# of cases)
(8- 12 year olds)
(H) Lower Respiratory Symptoms (# of days)
(9-1 1 year old asthmatics)
Percent of Total Incidence Associated with PM-10 Above Background**
BASE CASE
Background
= 8 ug/m3
1.1%
(0.8 -1.4)
2.4%
(1.5 -3.3)
3.7%
(2.5 -4.7)
1.9%
(1.3 -2.6)
0.8%
(0.3 - 1.3 )
1.4%
(0.7 - 2.1 )
17.5%
(15.3 -19.6)
6.8%
(2.4 - 10.9 )
Background
= 5 ug/m3
1.3%
(1.0 -1.7)
2.87%
(1.8 -4.0)
4.4%
(3.1 -5.7)
2.3%
(1.6 -3.1)
1.0%
(0.4 -1.5)
1.7%
(0.8 -2.5)
20 8%
(18.2 -23.3)
8.2%
(2.9 -13.0)
Background
= 11 pg/m3
0.9%
(0.6 -1.1 )
1.9%
(1.2 -2.7)
3.0%
(21 -3.8)
1.6%
(1.1 -2.1)
0.6%
(0.2 -1.0)
1.1%
(0.5 -1.7)
14.2%
(12.4 -15.9)
5.5%
(2.0 -8.8)
                                                                                                                                       I
                                                                                                                                      f—'
                                                                                                                                      0)
*  Health effects associated with short-term exposure to PM
** Health effects incidence was quantified across the range of PM concentrations observed in each study,
    when possible, but not below background level  Background PM-10 is assumed to be 8 ug/m3 .
The numbers in parentheses for pooled functions are NOT standard confidence intervals.
All numbers in parentheses are interpreted as 90% credible intervals based on
uncertainty analysis that takes into account both statistical uncertainty and
possible geographic variability.  See text in Chapter VI for details.
Sources of Concentration-Response (C-R) Functions:
(A) PM-10 C-R function based on pooled results from
   studies in 10 locations.
(B) PM-10 C-R based on pooled results from 4 functions
(C) PM-10 C-R based on pooled results from 4 functions
(D) PM-10 C-R based on pooled results from 4 functions
(E) Schwartz & Morris, 1995
(F) Schwartz & Morris, 1995
(G) Schwartz, etal., 1994
(H) Pope et al., 1991

-------
Table F-1b.  Sensitivity Analysis:  The Effect of Alternative Background Levels on
Predicted Health Effects Associated With "As-ls"  PM-2.5
Philadelphia County, September 1992 - August 1993
Health Effects*
Mortality (all ages)
Hospital Admissions
Respiratory
Hospital Admissions
Cardiac (>64 years old)
uOwer Respiratory
Symptoms in Children
(A) Associated with short-term exposure
(B) Total Respiratory
(all ages)
(C) Ischemic Heart
Disease***
(D) Congestive
Heart Failure***
(E) Lower Respiratory Symptoms
(# of cases) (8-1 2 years old)
Percent of Total Incidence Associated with PM-2.5 Above Background**
BASE CASE Background
- 3.5 ug/m3
1.8%
(1.1 -2.5)
2.0%
(0.5 -3.5)
0.7%
(0.3 -1.2)
1.3%
(0.6 -2.0)
20.1%
(10.3 -28.3)
Background
= 2.0 ug/m3
2.0%
(12 -2.8)
23%
(0.6 -3.9)
0.8%
(0.3 -1.3)
1.5%
(0.7 -2.2)
22.2%
(11.5 -31.3)
Background
= 5.0ug/m3
1.6%
(1.0 -2.2)
1.8%
(0.5 - 3.1 )
0.7%
(0.3 -1.1)
1.2%
(0.6 -1.7)
17.8%
(9.2 -25.2)
 * Health effects associated with short-term exposure to PM
 ** Health effects incidence was quantified across the range of PM concentrations observed in each study,
    when possible, but not below background level. Background PM-2.5 is assumed to be 3.5 ug/m3.
*** PM-2 5 results based on using PM-2 5 mass as PM-10 mass in the PM-10 functions.

The numbers in parentheses for pooled functions are NOT standard confidence intervals.
All the numbers in parentheses are interpreted as 90% credible intervals based on uncertainty
analysis that takes into account both statistical uncertainty and possible geographic variability.
See text in Chapter VI for details.
Sources of Concentration-
Response (C-R) Functions'
(A) PM-2.5 C-R function based on
    pooled results from 6
    locations.
(B)Thurston, etal., 1994
(C) Schwartz & Morris, 1995
(D) Schwartz & Morris, 1995
(E) Schwartz, et al, 1994

-------
annual mean concentrations, and the lowest observed annual mean concentration (the limit
used for quantification of risk) is well in excess of current estimates of background (e.g., the
range of concentrations observed for the cities in the ACS study (Pope et al., 1995) was 9.0 -
33.4 /ig/m3 PM2 5), the estimates of health risks associated with these endpoints do not change
in relation to estimates of background concentrations in the ranges used here (e.g., 2 -5 /xg/m3
PM2.5).
2.     Sensitivity of Health Risks Estimates to Alternative Rollback Methods for Simulating
       Attainment of Alternative Standards
       In addition to uncertainties concerning "as is" air quality, there is inherent uncertainty
concerning any effort to estimate air quality distributions that would occur upon attaining
standards  at some future date. In the risk analysis, such uncertainties are introduced both in
efforts to  model health risks upon attainment of the current standard (Chapter VI, Table VI-8)
and upon  attainment of alternative PM2 5 standards (Chapter VI, Tables VI-12a -13b).  The
base case  analysis assumes that proportional reductions would be observed in air quality
concentrations as an area attained either a controlling annual mean or 24-hr standard.   A
sensitivity analysis was conducted to examine the sensitivity of risk reduction estimates
associated with alternative PM2 5 standards to an alternative assumption concerning the pattern
of air quality rollbacks and the resulting air quality distribution that might be observed in
reaching attainment of PM2 5 standards (Table F-2). Because PM2 5 standards do not currently
exist, information on past air quality rollbacks in response to PM2 5 standards is not available.
However, monitoring information for PM2 5 can be examined, although it is uncertain how
much of the variation observed between years in the air quality distribution at a location
reflects actual control strategies versus more general year-to-year variability.  In a preliminary
examination of changes in the distribution of PM2 5 concentrations from sites with multiple
years of data (from AIRS and CARS data sets),  Abt Associates found that proportional
rollback reasonably approximated the central tendency of variations in PM2 5 air quality
distributions, however, considerable variation could be observed in this relationship across
time and  location (see Abt Associates, 1996b for more information).

-------
Table F-2.  Sensitivity Analysis:  Effect of Alternative Rollback Methods
on Predicted Health Effects of PM-2.5

Philadelphia County, September 1992 - August 1993
Health Effects
(A) Mortality associated with
short-term exposure
(B) Mortality associated with
ong-term exposure
Total
PM-related
Incidence
370
(220 -510)
Alternative
Standard
15 ug/m3 annual
Entire AQ distribution reduced equally
Rollback
Required
10.5%
Resulting Air
Quality (Annual
Mean/ 2nd
Daily Max)
15.0/ 64.4
Percent reduction in PM-related incidence:
370
(220 -510)
50 ug/m3 daily
29.4%
12.3/50.0
Percent reduction in PM-related incidence:
900
J560 -1230)
15 ug/m3 annual
10.5%
Percent reduction in PM-related incidence:
900
(560 -1230)
50 ug/m3 daily
29.4%
15.0/64.4

12.3/50.0
Percent reduction in PM-related incidence:
Change in Total
Incidence*
40
(20 -60)
10.6%
110
(70 -170)
29.7%
170
(130 -280)
19.4%
490
(350 -770)
54.1%
Upper 10% of AQ distribution reduced more
Base Rollback
Required**
14.5%

18.4%

9.0%

18.4%
Resulting Air
Quality (Annual
Mean/ 2nd
Daily Max)
15.0/62.6

13.3/50.0

15.0/62.6

13.3/50.0
Change in Total
Incidence*
30
(20 -50)
9.2%
70
(40 -110)
18.6%
170
(110 -240)
19.4%
350
(220 -480)
39.3%
* Health effects incidence was quantified across the range
of PM concentrations observed in each study, but now below
background PM-2.5 levels, assumed to be 3.5 ug/m3.
(A) C-R function based on studies in 6 locations.
(B) Pope et al., 1995
** The base rollback is the rollback on the lower 90% of the air quality distribution. The upper 10% is reduced by more.

The numbers in parentheses are NOT standard confidence intervals. They are 90%
credible intervals based on Monte Carlo analysis that takes into account both statistical
uncertainty and possible geographic variability. See text for details.

-------
                                          F-3
       An attempt to bound the potential effects of alternative PM air quality reduction
patterns has been examined in a sensitivity analysis of PM-associated risks by choosing
alternative assumptions  for modeling PM2 5 rollbacks.  Table F-2 shows the sensitivity of risks
reduction estimates associated with alternative PM2 5 standards to the rollback assumption in
which the upper 10%  of the PM2 5 24-hr air quality concentrations are reduced by a larger
amount (a ratio of 1.6) than in the remaining 90% of the distribution of PM air quality
concentrations.  This alternative rollback case is intended to model a control strategy that
preferentially targets peak PM2 5 levels.  The proportion of preferential reduction in peak
concentrations (a 1.6 ratio in reduction for the upper 10% of concentrations) is based on
empirical observation of the 99th  percentile of observed year-to-year variation in PM2 5 air
quality among site-years for all available PM2 5 monitoring sites with multiyear data from the
AIRS or CARB PM25 datasets.
       Table  F-2  shows for both a proportional rollback and the preferential peak reduction
rollback the amount of reduction in PM2 5 concentrations necessary to reach alternative
standards (for simplicity, the annual and daily standards are considered alone) and the air
quality distribution (summarized as the annual mean and 2nd daily max concentration) that is
projected to occur upon attainment.  In this example, the  annual standard provides less of a
change in total incidence of health effects, but this is simply a consequence of the annual
standard chosen (15 ^g/m3) being less controlling than  the daily standard chosen (50 jug/m3)
for Philadelphia County (Chapter VI, Table  lib).
       More important to consider are  the PM-associated risk reductions and resulting air
quality observed when the operation  of the same standard (annual or daily) is modeled under
the two rollback cases rather than any comparison of total incidence reduction between the two
standards.  The important observation is that estimated changes in incidence of health effects
provided by attainment of annual standards are less sensitive to deviation from the base case
assumption on rollback than estimated reductions in health effects incidence risk resulting from
attainment of a daily standard.  For instance, the results  in Table F-2 indicate that for a
controlling annual standard, past  patterns of air quality change would suggest the reduction in
health effects from short-term exposures, as represented by mortality from short-term

-------
                                           F-4
exposures, could potentially vary more than 35% with a controlling 24-hr standard (mean
change in total incidence of 70 versus 110), compared to approximately 25% with a
controlling annual standard.  For mortality from long-term exposures, this contrast is greater.
For example, under a controlling short-term standard estimated risk reduction could potentially
vary 30%, while under an annual standard there would be no change in estimated risk
reduction. This is a result of the fact that mortality from long-term exposures are related to
central estimate air quality measures such as annual mean concentration in the reported
concentration-response relationships, thus the distribution of 24-hr concentrations associated
with this annual mean concentration does not influence the estimated health risk reduction as
long as the same annual mean (in this case, 15 ng/m3) is achieved under both rollback
conditions.
       Figure F-l illustrates some of the characteristics of the integration of current air quality
distributions and reported concentration-response relationships as used to predict the total risk
from ambient particle exposures across a year.  Figure F-l shows the relative contribution of
different portions of the ambient PM2 5 concentration distribution for Philadelphia County to
the "as is" mortality health risk from short-term exposures.  The Figure shows in bar graph
form the proportion of total observed PM-2.5 concentrations across the year (in groups of 4
Mg/m3 per bar), with the number of days out of the whole year (361 observations) that
concentrations fell within each concentration range shown on the left-hand Y axis.  On top of
this frequency distribution has been overlaid the proportion of "as is" mortality  risk under base
case assumptions associated with each 4  pig/m3 concentration range (Since "as is" mortality risk
from short-term exposures was calculated using a two-day  mean averaging time, the averaging
time used at the largest number of mortality study locations, the proportion of "as is" mortality
risk is calculated for each two-day mean interval of 4 /*g/m3).   This Figure shows that for
base case assumptions, concentrations in the range of 16-20 jig/m3 contribute the largest
amount to the estimated mortality risk on an annualized basis for Philadelphia County.  Even
though concentrations in the range of 44 /ig/m3 PM2 5 and above clearly contribute more
mortality per day for these concentrations, the much  larger number of days within the  16-20
/ig/m3 range results in this interval  being associated with the largest total risk.  Standards with

-------
Oi
T3
CD



I
ch

oT
§
l-t-

3

CD




SI
CD
CD

01
0


B

CD
cb~
                      Frequency of Mean PM-2.5 Concentration
N)
o
                                     en
                                     o
CO
o
o
o
                                                                 3
                                                                 do'


                                                                 a
                                                                                  BT
                                                                                  e
                                                                 3.
                                                                 a-

                                                                 a
                                                                 O
                                                                 9
                                                                                  rV  »-^


                                                                                  i  Is
                                                                                     3
                                                                                     5s.
                                                                                     5'
                                                                                  "^d  s
                                                                                  65  9
                                                                                  a  a
                                                                                  t  s,

                                                                                  I
                                                                                  e
B
                        Associated Mortality (5%, mean, 95%)
                                                                 M
                                                                 <^

                                                                 3
                                                                 »


                                                                 a

                                                                 2
                                                                 5
                                                                 w
                                                                 er
                                                                     71

-------
                                           F-6
a 24-hr averaging time are traditionally based on peak air quality statistics, concentrations for
which the risk on an individual day is highest, but, as a result of the ambient air quality
distribution and the PM2 5 concentration-response functions that have been observed, appear to
contribute a relatively small amount of the total health risk compared to the distribution as a
whole.  The annual mean statistic contains information about the aggregate total of all the air
quality concentrations, a quantity similar to the quantity of all air quality concentrations minus
estimated background that contributes to estimates of annualized mortality risk in the base case
risk analysis.
       The difference between the air quality distribution as a whole and that estimated to
contribute to aggregate annualized health risk will be more pronounced if assumptions about a
substantial  cutpoint concentration are made.  However,  even in these cases, the aggregate
annualized risk will be a function of the concentrations across a wide portion of the upper end
of the PM2 5 air quality distribution. Since reducing high concentration days can provide  a
greater microgram reduction in PM2 5 annual average mass for a lesser percentage reduction in
air quality, an annual standard will still favor reducing high concentration values.  In contrast
to the 24-hr standard, however, an annual standard is less likely to allow areas whose  air
quality concentrations are substantially above those necessary for attainment to reduce
concentrations in a fashion that might not result  in meaningful risk reduction (e.g., by
reducing just a few high peak values).  In so doing, an annual controlling standard might  be
expected to lead to less variation in the risk reduced in different geographic areas having
similar initial air quality that reduce PM concentrations to attain a set of PM2 5 alternative
standards.
       Table F-2 conveys this point in a related  fashion.  Table F-2 shows that under the
preferential peak reduction rollback considered,  the lower 90% of air quality concentrations
are reduced only 18% versus the 30%  reduction observed if the entire distribution is reduced
evenly.   Because the lower 90 percent of the air quality values contribute so substantially to
the aggregate annualized risk (Figure F-l), a lesser reduction across this wide range of
concentration values leads to less total  PM2 5 reduction [as reflected by the higher annual  mean
upon  attainment of a daily standard of  50 ^g/m3 in which lower concentrations have been less

-------
                                           F-7
substantially reduced (13.6 /ig/m3) than when concentrations have been reduced evenly (12.6
jig/m3)], and thus less total annual health risk being reduced.
       Absent information that allows the possibility to be excluded that PM concentrations
through a wide portion of the air quality distribution may contribute to risk, an annual
controlling standard is likely to be less sensitive to alternative rollback assumptions.  This is in
large part because the standard employs an air quality measure (the annual mean) that
inherently captures  more information reflective of the concentrations across the bulk of the air
quality distribution.  In general, annual standards would be expected to decrease uncertainty in
risk reductions observed for areas that might undergo different air quality rollbacks to reach
attainment of PM2 5 alternative standards relative to comparably stringent controlling 24-hr
standards.
       For the special case of modeling the "attainment of current PM10 standards" case for
Los Angeles County, since the current daily PMi0 standard is controlling in Los Angeles, it is
relevant to consider the potential effects of variations from a proportional rollback for PM10
on the risk estimates for alternative PM2 5  standards.  Variations in the PM10 rollback that
would result in attainment of the current standards from the proportional rollback assumed
could either increase or decrease the amount of risk associated with PM remaining to be
affected by alternative PM2 5 standards. In addition, the risk estimate for the "attainment of the
current standards" case in Los Angeles has an important additional source of uncertainty
relating to patterns of reductions.  If control strategies to meet the current PM10  standards
preferentially reduce the coarse fraction of PM10 in relation to the fine fraction of PM10, risks
associated with PM2 5 as an indicator of PM under the "attain current standards" case could be
higher and, thus, proportions of estimated risk reduced under the alternative PM2 5 standards
also would be greater.  Alternatively, if control strategies to meet the current standards
preferentially reduce the fine fraction, then risks associated with PM2 5 as an indicator of PM
would be less under the "attain current standards" and the proportion of estimated risks reduced
under the alternative PM2 5 standards would be less.

-------
                                            F-8
B.     Sensitivity Analyses nf Key Concentration-Response Uncertainties
       The area of the risk analysis with the largest number of uncertainties amenable to
sensitivity analyses involves the application of PM concentration-response relationships in the
risk analysis.  The sensitivity of risk estimates for "as is" air quality in Philadelphia has been
analyzed to determine the potential impact of alternative analytic approaches to addressing
uncertainty in the concentration-response relationships.  The following sensitivity analyses
about concentration-response relationships are summarized in this Section:
•     The effect of alternative assumptions concerning the shape of the concentration-
       response relationships, especially concerning the effect of cutpoint concentrations below
       which variations in PM concentration are not associated with increases in risk, is
       analyzed.  Alternative assumptions about the slope of the concentration-response
       relationship above any presumed cutpoints also is addressed.
•     The effect of pooling studies to combine information from a number of studies to apply
       to the two risk analysis locations is examined.  The sensitivity of short-term mortality
       risk estimates is analyzed, especially with respect to the effects of combining studies
       that are heterogenous in averaging time.
•     The effect of using coefficients for PM obtained simultaneously with other copollutants
       in the regression model is addressed.
•     The effect of alternative assumptions concerning the potential role of air quality
       previous to that monitored in studies of the effects on mortality associated with long-
       term exposure is examined.
       All of these sensitivity analyses are conducted using "as-is" air quality in Philadelphia
County.  Further sensitivity analyses are provided in the technical support document (Abt
Associates, 1996b).
1.     Sensitivity Analyses of Alternative Cutpoint Concentrations
       Tables F-3A-E present the results from sensitivity analyses of different alternative
cutpoint concentrations for short-term and long-term exposures to PM. The concentrations
chosen as cutpoints for these sensitivity analyses were selected from the analysis of potential
cutpoints of interest described in Appendix E and summarized in Chapter VI.  For the base
case analysis, no cutpoint has been  assumed.  In the sensitivity analyses, various cutpoint
concentrations have been examined, and no health risks associated with PM are estimated for

-------
Table  F-3a.  Sensitivity Analysis:  The Effect of Alternative Outpoint Models on
Predicted Health Effects Associated With "As-ls" PM-10
Slope Adjustment Method 1"
Philadelphia County, September 1992 - August 1993
Health Effects"
Mortality (all ages)
Hospital Admissions
Respiratory
Hospital Admissions
Cardiac
Lower Respiratory
Symptoms in Children
(A) Associated with short-term exposure
(B) Total Respiratory
(>64 years old)
(C) Ischemic Heart Disease
(>64 years old)
(D) Congestive Heart Failure
(>64 years old)
(E) Lower Respiratory Symptoms (# of cases)
(8-1 2 year olds)
Percent of Total Incidence Associated with PM-10 Above Cutpoint
BASE CASE Background
= 8 pg/m3
1.1%
(0.8 -1.4)
2.4%
(1.5 -3.3)
0.8%
(0.3 -1.3)
1.4%
(0.7 -2.1)
17.5%
(15.3 -19.6)
Cutpoint
= 20 pg/m3
0.4%
(0.3 - 0 6)
1.3%
(08-17)
03%
(0.1 - 0.4)
05%
(0.2 - 0.2)
9.3%
(5.4 - 12.7)
Cutpoint
= 30 ug/m3
0.2%
(0.1 -0.2)
0.7%
(0.4 - 0 9)
0.1%
(0.1 - 0.2
0.2%
(0.1 -0.1)
63%
(3.9-8.1)
Cutpoint
= 40 ug/m3
0.1%
(0.0-0.1)
0.4%
(0.2 - 0.5)
0.1%
(0.0-0.1)
0.1%
(0.1-0.2)
4.7%
(3.4 - 5 5)
                                                                                                                                                      CO
* Two methods examine the potential impact of a concentration-response function having a steeper slope (i.e., larger coefficient) above
   specified cutpoints.  In both methods the slope below the cutpoint is set = 0, while the slope above the cutpoint is set to be greater
   than the slope in the original study. In Adjustment Method 1, the cutpoint C-R relationship is modeled to intersect with the original
   relationship, exceeding the RRs predicted for the original study at higher concentrations. The relationship was modeled to match the reduction in
   the range of PM concentrations upon application of the cutpoint with an identical percentage increase in the risk observed
   at the highest concentration.  Method 2 estimates a smaller increase in the slope See text for further information.
"Health effects associated with short-term exposure to PM

The numbers in parentheses for pooled functions are NOT standard confidence intervals.
All numbers in parentheses are interpreted as 90% credible intervals based on
uncertainty analysis that takes into account both statistical uncertainty and
possible geographic variability.  See text in Chapter VI for details.
Sources of Concentration-
  Response (C-R) functions:
(A) C-R function based on pooled
    results from 10 locations
(B) C-R function based on pooled
    results from 4 locations.
(C) Schwartz & Morris, 1995
(D) Schwartz & Morris, 1995
(E) Schwartz, et al., 1994

-------
Table F-3b.  Sensitivity Analysis:  The Effect of Alternative Outpoint Models on
Predicted Health Effects Associated With "As-ls" PM-10
Slope Adjustment Method 2*
Philadelphia County, September 1992 - August 1993
Health Effects"
Mortality (all ages)
Hospital Admissions
Respiratory
Hospital Admissions
Cardiac
Lower Respiratory
Symptoms in Children
(A) Associated with short-term exposure
(B) Total Respiratory
(>64 years old)
(C) Ischemic Heart Disease
(>64 years old)
(D) Congestive Heart Failure
(>64 years old)
(E) Lower Respiratory Symptoms (# of cases)
(8-12 year olds)
Percent of Total Incidence Associated with PM-10 Above Cutpolnt
BASE CASE Background
= 8 ug/m3
i.1%
(0.8 - 1.4 )
2.4%
(1.5 -3.3)
0.8%
(0.3 -1.3)
1.4%
(0.7 - 2.1 )
17.5%
(15.3 -19.6)
Cutpoint
= 20 pg/m3
0.4%
(0 3 - 0.5)
1 .0%
(06-1.3)
03%
(0.1 -0.4)
0.5%
(0 2 - 0.7)
79%
(4.5- 11.0)
Cutpoint
= 30ug/m3
0.1%
(0.1-02)
0.4%
(0.3 - 0.6)
01%
(0.0 - 0.2)
0.2%
(0.1 -0.3)
4.1%
(2.4 - 5.6)
Cutpoint
= 40 (jg/m3
0.1%
(0.0-0.1)
0.2%
(0.1-0.3)
0.0%
(0.0-0.1)
0.1%
(0.0-0.1)
25%
(1.5-3.2)
                                                                                                                                                 00
                                                                                                                                                 cr
* Two methods examine the potential impact of a concentration-response function having a steeper slope (i e., larger coefficient)
   above specified outpoints  In both methods the slope below the cutpoint is set = 0, while the slope above the cutpoint is set to
   be greater than the slope in the original study  In Adjustment Method 2, the slope is increased so that the new C-R function estimates
   the same health risk at the highest observed PM value as the original function Method 1 estimates a larger increase in the slope

"Health effects associated with short-term exposure to PM.

The numbers in parentheses for pooled functions are NOT standard confidence intervals.
All numbers in parentheses are interpreted as 90% credible intervals based on
uncertainty analysis that takes into account both statistical uncertainty and
possible geographic variability. See text in Chapter VI for details.
Sources of Concentration-
 Response (C-R) functions'
(A) C-R function based on pooled
    results from 10 locations.
(B) C-R function based on pooled
    results from 4 locations.
(C) Schwartz & Morris, 1995
(D) Schwartz & Morris, 1995
(E) Schwartz, et al., 1994

-------
Table F-3c.  Sensitivity Analysis:  The Effect of Alternative Cutpoint Models on
Predicted Health Effects Associated With "As Is"  PM-2.5
Slope Adjustment Method 1*
Philadelphia County, September 1992 - August 1993
Health Effects**
Mortality (all ages)
Hospital Admissions
Respiratory
Hospital Admissions
Cardiac
Lower Respiratory
Symptoms in Children
(A) Associated with
short-term exposure
(B) Total Respiratory
(all ages)
(C) Ischemic Heart Disease
(>64 years old)
(D) Congestive Heart Failure
(>64 years old)
(E) Lower Respiratory Symptoms
(8 -12 years old)
Percent of Total Incidence Associated with PM-2.5 Above Cutpoint
BASE CASE:
Background
= 3.5 pg/m3
1.8%
(1.1 -2.5)
2.0%
(0.5 - 3.5 )
0.7%
(0.3 -1.2)
1.3%
(0.6 -2.0)
20.1%
(10.3 -28.3)
Cutpoint
= 1 0 ug/m3
1.1%
(0.6- 1.5)
1 .4%
(0.4 - 2 4)
0.4%
(0.1 -06)
0.6%
(0.3 - 1 0)
13.1%
(7.1 - 18.5)
Cutpoint
= 18 ug/m3
0.5%
(0.3 - 0.6)
0.8%
(0.2 - 1 .4)
0.2%
(0.1 -0.3)
0.4%
(0.2 - 0 5)
9.7%
(5.6-13.0)
Cutpoint
= 30 ug/m3
0.1%
(0.1 -02)
0.4%
(0.1 - 0.7)
0.1%
(0.0-0.1)
0.1%
(0.1 -0.2)
6.5%
(5.2-7.1)
                                                                                                                                                    00
                                                                                                                                                    o
* Two methods examine the potential impact of a concentration-response function having a steeper slope (i e., larger coefficient) above
   specified cutpoints.  In both methods the slope below the cutpoint is set = 0, while the slope above the cutpoint is set to be greater
   than the slope in the original study. In Adjustment Method 1, the cutpoint C-R relationship is modeled to intersect with the original
   relationship, exceeding the RRs predicted for the original study at higher concentrations. The relationship was modeled to match the redu
   the range of PM concentrations upon application of the cutpoint with an identical percentage increase in the risk observed
   at the highest concentration.  Method 2 estimates a smaller increase in the slope. See text for further information.

** Health effects associated with  short-term exposure to PM.

The numbers in parentheses for pooled functions are  NOT standard confidence intervals.
All the numbers in parentheses are interpreted as 90% credible intervals based on uncertainty analysis
that takes into account both statistical uncertainty and possible geographic variability.
See text in Chapter VI for details.
Sources of Concentration-
Response (C-R) functions:
(A) C-R function based on pooled
  results from six locations.
(B) Thurston, et al, 1994
(C) Schwartz & Morris, 1995
(D) Schwartz & Morris, 1995
(E) Schwartz etal., 1994

-------
Table F-3d. Sensitivity Analysis:  The Effect of Alternative Outpoint Models on
Predicted Health Effects Associated With "As Is" PM-2.5
Slope Adjustment Method 2*
Philadelphia County,  September 1992 -August 1993
Health Effects**
Mortality (all ages)
Hospital Admissions
Respiratory
Hospital Admissions
Cardiac
Lower Respiratory
Symptoms
(A) Associated with short-term exposure
(B) Total Respiratory
(all ages)
(C) Ischemic Heart Disease
(>64 years old)
(D) Congestive Heart Failure
(>64 years old)
(E) Lower Respiratory Symptoms
(8 -12 years old)
Percent of Total Incidence Associated with PM-2.5 Above Outpoint
BASE CASE: Background
= 3.5 ug/m3
1.8%
(1.1 -2.5)
2.0%
(0.5 -3.5)
0.7%
(0.3 -1.2)
1.3%
(0.6 -2.0)
20.1%
(10.3 - 28.3 )
Cutpoint
= 10pg/m3
1 .0%
(06-14)
1 2%
(03-2.1)
0.4%
(0.2 - 0.6)
0.7%
(03-1 .0)
12.1%
(65-172)
Cutpoint
= 18pg/m3
0.4%
(0 2 - 0.6)
0.6%
(0.2-1.1)
0.2%
(0.1-0.3)
0.3%
(0.1 -0.5)
6.9%
(3.8 - 9 6)
Cutpoint
= 30 pg/m3
0.1%
(0.1 - 0 2)
02%
(0.1 -04)
0.1%
(0.0-01)
0.1%
(0.0-0.1)
3.6%
(2.3 - 4 5)
                                                                                                                                               Co
* Two methods examine the potential impact of a concentration-response function having a steeper slope (i.e., larger coefficient)
    above specified cutpoints. In both methods the slope below the cutpomt is set = 0, while the slope above the cutpoint is set to
    be greater than the slope in the original study In Adjustment Method 2, the slope is increased so that the new C-R function estimates
    the same health risk at the highest observed PM value as the original function  Method 1 estimates a larger increase in the slope.
"Health effects associated with short-term exposure to PM

The numbers in parentheses for pooled functions are NOT standard confidence intervals.
All the numbers in parentheses are interpreted as 90% credible intervals based on uncertainty analysis
that takes into account both statistical uncertainty and possible geographic variability.
See text in Chapter VI for details.
Sources of Concentration-
Response (C-R) functions:
(A) C-R function based on pooled
  results from six locations
(B)Thurston, etal., 1994
(C) Schwartz & Morris, 1995
(D) Schwartz & Morris, 1995
(E) Schwartz etal., 1994

-------
Table F-3e. Sensitivity Analysis:  The Effect of Differing Cutpoints on Estimated
Mortality Associated with Long-term Exposure to PM-2.5
Philadelphia County, September 1992 - August 1993

(A) Mortality associated with
long-term exposure
BASE CASE
Lowest Observed =
9 ug/m3
4.6%
(2.8 - 6.2)
Outpoint = 12.5
ug/m3
2.4%
(1.5-3.3)
Cutpoint =15 ug/m3
0.8%
(0.5-1.1)
Cutpoint = 18 ug/m3
0.0%
(0.0 - 0.0)
                                                                    (A)Popeetal., 1995
Health effects incidence was calculated down to the lowest level observed in the study (9 ug/m3).
No adjustments to the slope were performed.

The numbers in parentheses for pooled functions are NOT standard confidence intervals.
All the numbers in parentheses are interpreted as 90% credible intervals based on uncertainty analysis
that takes into account both statistical uncertainty and possible geographic variability.
See text in Chapter VI for details.
00
n>

-------
                                           F-9
any days whose 24-hr concentrations are below the specified cutpoint concentration.  In
addition, the slope of the relationship above the cutpoint has been remodeled using one of two
approaches.  For both approaches, the relationship is assumed to begin at zero increased risk at
the cutpoint concentration, and to extend upward with an increased slope compared to the
original reported relationship (see Fig. VI-6).  In Approach 1 it is  assumed that the new slope
would increase to an extent where the increased health risk predicted at the highest
concentration is increased proportional to the proportion of the range of original concentrations
that fall below the cutpoint. While this adjustment produces a slope resembling those
generally posited to result in a model incorporating a cutpoint  (e.g., Fig VI-6), there is no
clear guidance on how to most appropriately model changes in slope for purposes such as the
PM risk analysis (where, for instance,  primary datasets are not readily available).
       In light of this uncertainty, a second approach, involving a  more minimal adjustment to
slope (labeled "Approach 2" on Figure VI-6) also  has been carried out as a potential  lower
bound for an adjusted slope.  In Approach 2, the concentration-response  relationship has been
remodeled to begin at zero at the cutpoint and intersect with the same health risk estimated at
the highest concentrations observed in the original  relationship. As cutpoints are chosen that
exclude successively larger number of observations,  it is expected that the milder degree of
increased slope represented by Approach 2  would be less likely to be observed.
       Figure F-2 suggests that relatively mild increases in slope may be observed for some
TSP concentration-response relationships compared to a linear model meta analysis from the
CD.  However,  other TSP concentration-response  relationships which examined cutpoints well
within the range of data observed a pattern  of increased slope more like that modelled in
Approach 1 (Philadelphia  1983-88,  which included SO2 and O3 in the analysis, compared with
a meta analysis of PM coefficients from models including copollutants).
       As might be expected, Tables F-3A - D indicate that the two slope adjustment
approaches agree mostly closely at the lowest cutpoint concentration. In addition, these tables
suggest that the method of adjusting the slope of the remaining relationship is less important to
the estimates of health risk than the choice of cutpoint concentration itself.  The higher the
cutpoint, the greater the proportion  of observations for each city that is associated with no
increase in risk.  Depending on judgments concerning the weight to be given the estimates at

-------
                                        F-10
          _
          fi
1.16-

1.14-

1.12-

1.10

1.08

1.06-

1.04-

1.02-

 1.0-
                    EPA Metaanalysis

         Philadelphia (1973-80).

Cincinnati (1977-82)-
                                                                \
                                                        EPA Metaanalysis
                                                        with Copollutants
                                          Phila. (1983-88) w/Copoll
                               l
                               50
  -1.16

  -1.14

  -1.12



  •1.08

  •1.06

  •1.04

  •1.02

  •1.0
                              100
                            TSP ug/m3
                         150
200
Figure F-2.  Comparison of Smoothed Nonlinear and Linear Mathematical Models for
Relative Risk of Total Mortality Associated with Short-Term TSP Exposure (CD, Figure
13-6).  Curves show smoothed nonparametric models for Philadelphia (based on Schwartz
19945) and for Cincinnati (based on Schwartz, 1994a), and piecewise linear models for
Philadelphia (based on Cifuentes and Lave, 1996).  Solid curve shows linear model from EPA
metaanalysis using studies with no copollutants, dash-dot curve shows linear model from EPA
metaanalysis using studies with SO2 as a copollutant (described in CD Chapter 12).

-------
                                          F-ll
higher cutpoint concentrations, assumptions concerning cutpoint concentrations can make a
substantial difference in the estimates of risks associated with PM.
       For the concentration-response relationship of mortality from long-term exposures
(Table F-3E), the upper cutpoint eliminates estimated risk for Philadelphia County because
Philadelphia County's annual mean concentrations are below 18 //g/m3.  For health risks both
from short-term and long-term exposures, the sensitivity of estimates of risks would be
expected to vary with location, especially for locations with  substantially different overall PM
air quality (e.g., Los Angeles  County).
2.     Effect on Pooled Concentration-Response Analyses Using  Studies with Different
       Averaging Times
       In their review of the PM mortality literature, the CD pointed out that heterogeneity in
averaging time is an important factor to consider in assessing results (CD, p. 12-72).  In the
PM risk analysis estimates  from a number of studies have been pooled for several endpoints.
For the mortality pooled analysis, studies that used averaging times ranging from 1 to 5 day
mean PM concentrations have been included. Table F-4 disaggregates the pooled analysis to
examine the effect of restricting the estimates of mortality risk to  those studies using only the
same averaging time (with  the exception  of the three-day and five-day mean studies, which
were combined). Results vary considerably over averaging times. In the base case analysis,
two-day mean air quality concentrations were used to estimate mortality, since the largest
number of functions used that  averaging time.  Table F-4 indicates that using two-day mean
concentrations to represent  Philadelphia County PM10 concentrations results in an increase in
the risk estimates predicted by the single  study that reported  results related to a one-day mean
concentration (Kinney et al., 1995), and a slight increase in the risk predicted for the set of
two studies using three- to five-day  mean concentrations (Schwartz, 1993 and Pope et al.,
1992).   However, the Table also indicates that applying an alternative averaging time, such
as one-day or five-day mean concentrations, results in no apparent difference  in estimated risk
from the base case two-day mean assumption.

-------
Table F-4.  Sensitivity Analysis:  Effect of Combining Different Averaging Times
In Pooled Short-Term Exposure Mortality Functions on
Predicted Health Effects Associated With "As-ls" PM-10
Philadelphia County, September 1992 - August 1993

Matching study and data
averaging times
Using 2-day average PM data
Using 1-day average PM data
Using 5-day average PM data
Percent of Total Incidence Associated with PM-10 Above Background*
BASE CASE"
Studies Using All
Averaging Times
(10 studies)
2-day average PM
1.1%
(0.8 -1.4)
2-day average PM
same
1-day average PM
1.1%
(0.8 -1.4)
5-day average PM
1.1%
(0.8 -1.4)
Studies using 1-day
average PM
(1 study)
1 -day average PM
0.4%
(0.0 - 0.8 )
2-day average PM
0.4%
(0.0 - 0.8 )
Studies using 2-day
average PM
(7 studies)
2-day average PM
1.0%
(0.5 -1.5)
2-day average PM
same
Studies using 3-5 day
average PM
(2 studies)
5-day average PM
1.8%
(1.3 -2.4)
2-day average PM
1.9%
(1.3 -2.4)

                                                                                                                                           I
                                                                                                                                          h-1
                                                                                                                                          I—*
                                                                                                                                          (U
•Health effects incidence was quantified across the range of PM concentrations observed in each study,
     when possible, but not below background level Background PM-10 is assumed to be 8 ug/m3

** The base case is a random-effects pooled function used with 2-day average PM data
    All other pooled functions are also random effects, except the pooled function derived from
    studies using 3-5 day average PM data, for which a fixed effects model was used, since
    it is not possible to calculate a random effects model for those two functions

The numbers in parentheses for pooled functions are NOT standard confidence inter
All numbers in parentheses are interpreted as 90% credible intervals based on
uncertainty analysis that takes into account both statistical uncertainty and
possible geographic variability. See text in Chapter VI for details.
      The studies that contribute to the pooled
      function are:
1-day: Kinney et al., 1995 (Los Angeles)
2-day: Ito and Thurston, 1996 (Chicago)
      Schwartz et al. 1996 (Boston, MA;
         Knoxville, TN; St. Louis, MO;
         Steubenville, OH; Portage, Wl;
         Topeka, KS)
3-day: Schwartz 1993 (Birmingham, AL)
5-day: Pope et al., 1992 (Utah Valley)

-------
                                            F-12
 3.     Effect of Using Concentration-Response Relationships Simultaneously Considering
        Copollutants
        PM is part of a mix of combustion source pollutants originating from a variety of
 stationary and mobile sources and, thus generally occurs along with other pollutants generated
 by combustion sources (e.g., sulfur oxides, nitrogen oxides, volatile organic compounds) or
 produced through the transformation of these pollutants (e.g., O3).   Such copollutants could
 either serve as potential confounders of the observed PM-health associations or as effect
 modifiers that influence the magnitude of PM associated effects.  The studies used in the risk
 analysis provide PM coefficients from areas with widely varying levels of copollutants.  One
 approach to controlling for the potential effects of copollutants is to include copollutants
 simultaneously in the model with PM when estimating the PM coefficient for a health
 endpoint. However, this method may be limited by collinearity in  the pollutants of interest
 (Samet et al.,  1996b).  (For a fuller treatment of copollutants, potential confounding, and the
^significance of observed variations across study locations, see Chapter V and CD, Chapters 12
 and 13).
        The base case analysis used concentration-response relationships estimated without
 inclusion of copollutants, and it is not possible to directly estimate the sensitivity of the base
 case results taking into account the effect of simultaneous inclusion of copollutants, since not
 all the studies used for the base case examined copollutants  in this manner.   As an alternative,
 the sensitivity of individual study estimates in relationship to inclusion of copollutants is
 examined in Tables F-5A and F-5B.  Table F-5A provides a comparison of the coefficients for
 studies that reported PM coefficients both with and without inclusion of  copollutants, and
 Table F-5B provides the risk estimates obtained from applying those coefficients to
 Philadelphia County in the risk analysis.  The results in these two tables provide a more
 general sense of  how much of an effect inclusion  of copollutants typically has on the
 magnitude of the health risk estimates and, thus, potentially on the  base case results. The
 results for many, but not necessarily all, of the studies are consistent with the assessment in the
 CD that PM effect sizes and their statistical uncertainty in most studies showed little sensitivity
 to the adjustment for copollutants (CD, p. 13-55).

-------
Table F-5a.  Sensitivity Analysis:  Effect of Copollutants
Relative Risks for Change of 50 ug/m3 PM-10 or 25 ug/m3 PM-2.5
Health Effects Associated with
Short-Term Exposure
Mortality
Hospital
Admissions
All respiratory
(all ages)
All respiratory
(ages >64)
Pneumonia
(ages >64)
COPD
(ages >64)
Ischemic Heart Disease
Congestive Heart Failure
Study, Pollutant, & Location
Ito & Thurston 1995, PM-10
Chicago
Kinneyetal., 1995, PM-10
Los Angeles
Pope 1994, PM-10
Utah Valley, summer only
Thurston et al., 1994, PM-2.5
Ontario, Canada
Schwartz 1995, PM-10
New Haven
Schwartz 1995, PM-10
Tacoma
Schwartz 1994, PM-10
Minneapolis/St. Paul
Schwartz 1994, PM-10
Detroit
Schwartz 1994, PM-10
Detroit
Schwartz & Morris 1995, PM-10
Detroit
Schwartz & Morris 1995, PM-10
Detroit
Relative Risk
No Copollutant
1.02
(1.02-1.04)
1.02
(1.00-1.05)
1.11
(0.95-1.31)
0.086*
(0.024 - 0.15)
1.06
(1.01 -1.12)
1.10
(1.04-1.16)



1.028
(1.011 -1.047)
1.050
(1.024-1.077)
Relative Risk with
Daily Average
SO2




1.07
(1.02-1.13)
1.11
(1.03-1.19)



1.024"
(1.005-1.043)

Relative Risk with
Daily 1-hour Maximu
CO

1.02
(0.99-1.04)







1.025
(1.007-1.044)
1.038
(1.011-1.064)
Relative Risk with
Daily Average
O3
1.02
(1.01 -1.03)

1.14
(0.96-1.37)

1.09
(1.01 -1.181
1.12
(0.99-1.26)
1.08
(1.02-1.14)
1.06
(1.03-1.09)
1.10
(1.06-1.16)


Relative Risk with
Daily 1-hour Maximum
03

1.02
(1.00-1.05)
1.19
(1.00-1.43J
0.045*
(-0.028 - 0.12)







Results presented in bold come from functions used in the base case analysis.
The number of significant digits given for each relative risk is the same as the number reported in the original study.

* Thurston et al. 1994 provides a function relating changes in PM to changes in the number of cases.
    The relative risk calculated from this coefficient may vary widely from location to location, depending on baseline incidences.
    Therefore, the coefficient, adjusted to a rate per 100,000 people, is reported, instead of a relative risk.
" Based on 1-hour maximum S02.

-------
Table F-5b. Sensitivity Analysis:  Effect of Copollutants on
Predicted Health Effects Associated With "As-ls" PM*
Philadelphia County, September 1992 - August 1993

Health Effects
Mortality
Hospital
Admissions
All respiratory
(ages >64)
All respiratory
(ages >64)
Pneumonia
(ages >64)
COPD
(ages >64)
Ischemic Heart Disease
Congestive Heart Failure
Study & Location
Ito & Thurston 1996, PM-10
Chicago
Kinneyetal., 1995, PM-10
Los Angeles
Pope 1994, PM-10
Utah Valley, summer only
Thurston et al., 1994, PM-2.5
Ontario, Canada
Schwartz 1995, PM-10
New Haven
Schwartz 1995, PM-10
Tacoma
Schwartz 1994, PM-10
Minneapolis/St. Paul
Schwartz 1994, PM-10
Detroit
Schwartz 1994, PM-10
Detroit
Schwartz & Morris 1995, PM-10
Detroit
Schwartz & Morris 1995, PM-10
Detroit
Percent of total incidence associated with PM above background
with
no copollutant
0.8%
(0.3-1.3)
0.4%
(0.0 - 0.8)
3.0%
(-1.5-7.2)
NA
NA
2.4%
(0.3 - 4.5)
3.2%
(-0.2 - 6.4)



0.8%
(0.3-1.3)
1.4%
(0.7 - 2.1)
with
daily average
SO2




1.9%
(0.6 - 3.4)
2.9%
(1.0-4.7)



0.7%"
(0.1 -1.2)

with daily
1 -hour maximum
CO

0.3%
(-0.0 - 0.7)







0.7%
(0.2-1.2)
1.1%
(0.3-1.8)
with
daily average
O3
0.6%
(0.2 - 0.9)

3.7%
(-1.3-8.3)

2.4%
(0.4 - 4.6)
3.2%
(-0.2 - 6.4)
2.2%
(0.6-3.8)
1.6%
(0.7 - 2.5)
2.8%
(1.5-4.2)


with daily
1-hour maximum
03

0.4%
(0.0 - 0.8)
4.8%
(-0.2 - 9.4)
NA
NA







                                                                                                                                   S3
Results presented in bold come from functions used in the base case analysis.
* Health effects associated with short-term exposure to PM. Incidence was quantified across the range of PM concentrations observed in each study,
    but not below background PM levels, assumed to be 8 ug/m3 for PM-10 and 3.5 ug/m3 for PM-2.5.
** Based on 1-hour maximum SO2.

The numbers in parentheses for pooled functions are NOT standard confidence intervals. All numbers in parentheses are interpreted
as 90% credible intervals based on uncertainty analysis that takes into account both statistical uncertainty and possible geographic
variability. See text  in Chaptet VI for details.

-------
                                          F-13
       Two substantial uncertainties remain concerning copollutants and the method of
controlling for their effects through simultaneous inclusion in the health risk model.   First, to
what degree is it possible that the associated copollutant does not have a bona fide independent
effect on mortality separate from PM? If the copollutant does not have an independent effect
on mortality, then changes in the PM  coefficient resulting from inclusion of the second
pollutant may just be the results of collinearity between the pollutants and may not accurately
reflect the underlying PM coefficient.  Second, if the changes  seen with inclusion of
copollutants actually do reflect a bona fide  improvement in the estimate of the PM effect, then
is it possible simultaneous inclusion of additional copollutants  would further reduce the
coefficient?  As pointed out by  Samet et al. (1996b) and in Chapter V, examination of effects
within a single location may often be  limited by collinearity between pollutants and
comparison across geographic areas may be required for a fuller assessment of the potential
effects of copollutants on reported PM concentration-response relationships.
 4.     Sensitivity Analysis Concerning Reduction in the Slope of Concentration-Resposer
       Relationships for Risks from Long-Term Exposures
       Two major concerns have been raised concerning whether the slope of the
concentration-response relationships from recent studies of mortality from long-term exposures
(Dockery et al., 1993, Pope et al., 1995) may be misestimated.  One major uncertainty
concerning the studies of health risks  associated with long-term exposures to PM for adults is
the potential relevance of air  quality concentrations previous to the period of monitoring in the
study. If long-term air quality concentrations previous to the period being monitored: 1) are
relevant for a substantial portion of the population for the endpoint being studied, and 2) are
substantially different than concentrations monitored during the study, then the actual long-
term concentration-response relationship may be substantially different than that observed in
the reported study (CD, p. 13-34). The second major uncertainty relates to whether inadequate
control of potential confounders may  substantial alter the reported concentration-response
relationships (CD, pp. 12-140-43, 12-165,  12-176-178).

-------
                                               F-14

        The question of the degree to which previous (from years to decades) air quality
exposures might have affected mortality risk is complex.1 In addition,  quantitative
information on the levels of previous air quality  concentrations is difficult to ascertain,
especially for PM2 5.  The CD reports that for  the monitoring data reported in the Six City
mortality study,  downward trends in PM2 5 mass are evident for four of the six cities (CD, p.
13-14).
        Given these uncertainties in developing a quantitative basis for a sensitivity  analyses
concerning  historical air quality, Table F-6 simply shows the potential impact of mortality risk
estimates associated with long-term exposures  if one assumes that previous  air quality
concentrations reduce the observed slope of the PM  concentration-response relationship by
33% (modeling the case if relevant previous PM25 concentrations averaged approximately 50%
higher than that monitored  in the study period  ) and  by 50% (modeling the  case  if relevant
previous PM2 5 concentrations were twice as high).   As expected, positing that the most
important PM2 5  concentrations in regards to effects  on mortality risk occurred before the  study
monitoring  period leads directly to similarly proportional reductions (approximately 33%  and
    1 Judging the extent to which previous air quality may be a significant concern for the estimates of risk from long-
term exposures requires consideration of both of past air quality variability and of the relevant exposure period that
might be expected to affect mortality nsk for a substantial portion of the cohort population  The CD notes that a
detailed investigation of temporal relationships has not been attempted in the cohort studies, but also notes that if
responses reflect primarily the last few years of integrated exposure then the concurrent average monitoring data would
be reasonably predictive (CD, p. 12-171, 12-181).  Some findings from air pollution epidemiology suggest recent
exposures may be of primary importance. The reduction in mortality incidence observed with a reduction in PM
concentrations for 14 months in Utah Valley suggests that a significant amount of the mortality of substantial
prematurity associated with particles in that location did not appear dependent on exposures over the span of years, since
changes in mortality rates could be observed with a relatively brief temporal change (a 14 month period of reduced
concentrations) in long-term average PM pollution
        Observations of the temporal relationship of exposure to mortality risk for a large portion of cardiovascular
mortality (deaths from myocardial infarction) and for lung cancer from cohort studies on active cigarette smoke exposure
suggest that elevated risks for myocardial infarction generally return to close to baseline nonsmoking relative risks
within three to ten years (Rosenberg et al., 1985; 1990) and that much of the lung cancer risk is reduced close to the risk
for never smokers (compared to the marked elevation in relative risk for lung cancer among current smokers) within 10-
15 years after cessation of smoking (USEPA, 1992, Table 4-6 and 4-7). The significance of these findings to air
pollution effects cannot be assumed, since quite distinct mechanisms for cigarette smoking and particular matter
exposure and mortality from cardiovascular and lung cancer causes may be likely.  However, the smoking cohort studies
show that in one area in which the temporal relationship of exposure to mortality risk from cardiovascular and lung
cancer causes has been examined, evidence suggests recent exposures may be substantially more important than less
recent exposures

-------
Table F-6.  Sensitivity Analysis:  The Effect of Concentration-Response
Function Slope on Estimated Mortality Associated with Long-term
Exposure to PM-2.5*
Philadelphia County, September 1992 - August 1993
Health Effect"
(A) Mortality associated with
long-term exposure
BASE CASE
Assuming AQ as
reported
4.6%
(2.8 - 6.2)
Assuming relevant
AQ 50% higher*
3.4%
(2.1-4.7)
Assuming relevant
AQ twice as high***
2.3%
(1.4-3.2)
*This Table illustrates the sensitivity of mortality risk associated with long-term exposure (A) Pope et al., 1995
 if concentration-response function slope were adjusted to reflect possible effects of previous
 air quality or potential confounders not addressed in the original PM health effects model.
"Health effects incidence was calculated down to the lowest level observed in the study.
*** Adjusted function from Pope et al., 1995.  Had historical air quality (AQ) been 50% higher, the
   relative risk calculated by the study would have been two thirds of that reported. Had historical
   air quality been twice as high, the relative risk calculated would have been half that reported.
The numbers in parentheses for pooled functions are NOT standard confidence intervals.
All the numbers in parentheses are interpreted as 90% credible intervals based on uncertainty analysis
that takes into account both statistical uncertainty and possible geographic variability.
See text in Chapter VI for details.

-------
                                          F-15
50%) in the estimates of long-term mortality risk. To the extent that the estimates of mortality
risks from long-term exposure reflect the net sum of acute events that take place over that year
(which will occur when increases in daily death rates associated with acute events are not
subsequently canceled by decreases ("harvesting") (CD p. 12-139), this component of mortality
risk from long-term exposures risk is not sensitive to assumptions about previous air quality.
       Similar slope reductions can also serve to model concerns about uncontrolled
confounding.  The CD provides as an example how inclusion of additional ecological variables
can attentuate the PM2.5-mortality relationship observed in a initially simply age- and race-
adjusted dataset.  The direction and extent of change in slope that might be observed by
control of such confounders in a prospective cohort design, which features individual data for
some risk factors is not certain (CD, pp.  12-176-77), however for the purposes of sensitivity
analyses reductions in slope of 33-50% for the long-term studies will be assumed appropriate
appropriate to reflect the viewpoint that exhibits substantial concerns about residual
uncontrolled confounding in these studies. These would result in the same proportional
reductions of approximately 33-50% in the estimates of long-term mortality risk (relative to
base case assumptions) as when this slope reduction was considered as a sensitivity analysis for
the potential effects of previous air quality.

-------
                                          G-l
                                      Appendix G
       MEASURES OF VISIBILITY IMPAIRMENT AND LIGHT EXTINCTION
       Several atmospheric optical indices and approaches can be used for characterizing
visibility impairment and light extinction.  The CD discusses several indicators that could be
used in regulating air quality for visibility protection, including:  1) light extinction (and
related parameters of visual range and deciview) calculated from measurements of fine
particle constituents and their associated scattering and absorption; 2) light extinction
measured directly by transmissometer; 3) light scattering by particles, measured by
nephelometer; 4) fine particle mass concentration; 5) contrast transmittance (CD,  8-125).
       In conjunction with the National Park Service, other Federal land managers, and State
organizations, EPA has supported since 1986 a monitoring protocol utilizing a combination
of the first four measurements.  This long-term visibility monitoring network is known as
IMPROVE (Interagency Monitoring of PROtected Visual Environments.  The following
discussion briefly describes the IMPROVE protocol  and provides rationale supporting use of
the light extinction coefficient, derived from both direct optical measurements and
measurements of aerosol constituents, for purposes of implementing air quality management
programs to improve visibility.
       IMPROVE provides direct measurement of fine  particles and precursors that
contribute to visibility impairment at more than 40 mandatory Federal Class I areas across
the country.  The IMPROVE network employs aerosol, optical, and scene measurements.
Aerosol measurements are taken for PM10 and PM25 mass, and for key  constituents of PM2.5,
such as sulfate, nitrate, organic and elemental carbon, soil dust, and several other elements.
Measurements for specific aerosol constituents  are used to calculate "reconstructed" aerosol
light extinction by multiplying the mass for each constituent by its empirically-derived
scattering and/or absorption efficiency. Knowledge  of the main constituents of a  site's light
extinction "budget"  is critical for source apportionment  and control strategy development.
Optical measurements are used to directly measure light extinction or its components. Such
measurements are taken principally with either a transmissometer,  which measures total  light
extinction, or a nephelometer, which measures particle scattering (the largest human-caused
component of total extinction).  Scene characteristics are recorded  3 times daily with 35

-------
                                           G-2
millimeter photography and are used to determine the quality of visibility conditions (such as
effects on color and contrast) associated with specific levels of light extinction as measured
under both direct and aerosol-related methods.   Because light extinction levels are derived hi
two ways under the IMPROVE protocol,  this overall approach provides a cross-check in
establishing current visibility conditions and  trends and in determining how proposed changes
hi atmospheric constituents would affect future visibility conditions.
       The light extinction coefficient has been widely used in the U.S. for many years to
describe visibility conditions and the change  in visibility experienced due to changes in
concentrations of air pollutants.  As noted earlier, the extinction coefficient can be defined as
the fraction of light lost or redirected per unit distance through interactions with gases and
suspended particles in the atmosphere.  Direct relationships exist between measured ambient
pollutant concentrations and their contributions to the extinction  coefficient.   The contribution
of each aerosol constituent to total light extinction is derived by  multiplying the aerosol
concentration by the extinction efficiency  for that aerosol constituent. Extinction efficiencies
vary by type of aerosol constituent and have been obtained  through empirical studies.  For
certain aerosol constituents, extinction efficiencies increase  significantly with increases in
relative humidity.
       In addition to the optical effects of atmospheric constituents as characterized by the
extinction coefficient, lighting conditions  and scene characteristics play an important role in
determining how well we see objects at a distance.  Some of the conditions that influence
visibility include whether a scene is viewed towards the sun or away from it, whether the
scene is shaded or not,  and the color and  reflectance of the scene (NAPAP, 1991).  For
example,  a mountain peak in bright sun can  be  seen from a much greater distance when
covered with snow than when it is not.
       One's ability to  see an object is degraded both by the reduction of image forming
light from the object caused by scattering and absorption, and by the addition of non-image
forming light that  is scattered into the  viewer's  sight path.  This non-image forming light is
called path radiance (CD, 8-23).  A common example  of this effect is our inability to see
stars in the daytime due to the brightness  of  the sky caused by Rayleigh scattering.  At night,
when the sunlight  is not being scattered, the  stars are readily seen.  This same effect causes a

-------
                                           G-3
haze to appear bright when looking at scenes that are generally towards the direction of the
sun and dark when looking away from the sun.
       Though these non-air quality related influences on visibility can sometimes be
significant, they cannot be accounted for in any practical sense in formulation of national or
regional measures to minimize haze.  Lighting conditions change continuously as the sun
moves across the sky and as cloud conditions vary.  Non-air quality influences on visibility
also change when a viewer of a scene simply turns his head. Regardless of the lighting and
scene conditions, however, sufficient changes in ambient concentrations of PM will lead to
changes in visibility (and the extinction coefficient).  The extinction coefficient integrates the
effects of aerosols on visibility, yet is not dependent on scene-specific characteristics.  It
measures the changes in visibility linked to emissions of gases and particles that are subject
to some form of human control and potential regulation, and therefore can be useful in
comparing visibility impact potential of various air quality management strategies over time
and space (NAPAP, 1991).
       By apportioning the extinction coefficient to different aerosol constituents, one can
estimate changes in visibility due to changes in constituent concentrations (Pitchford and
Malm,  1994).  The National Research Council's 1993 report Protecting Visibility in National
Parks and Wilderness Areas states that  "[PJrogress toward the  visibility goal  should be
measured  in terms of the extinction coefficient, and extinction measurements should be
routine and systematic."  Thus, it is reasonable to  use the change in the light extinction
coefficient, determined in multiple ways, as the primary indicator of changes in visibility for
regulatory purposes.
       Visual range is a  measure of visibility that  is inversely  related to the extinction
coefficient.  Visual range can be defined as the maximum distance at which one can identify
a black object against the horizon sky.  The colors and fine  detail of many objects  will be
lost at a distance much less than the visual range, however.  Visual range has been widely
used in air transportation and military operations in addition to its  use in characterizing air
quality. Because it is expressed in familiar units and has a straightforward definition, visual
range is likely to continue  as a popular measure of atmospheric visibility (Pitchford and
Malm, 1994).  Conversion from the extinction coefficient to visual range can be made with

-------
                                           G-4
the following equation (NAPAP, 1991):

                                Visual Range = 3.91/aext

       Another important visibility metric is the deciview, which describes changes in
uniform atmospheric extinction that can be perceived by a human observer.  It is designed to
be linear with respect to perceived  visual changes over its entire range in a way that is
analogous to the decibel scale for sound (Pitchford and Malm, 1994). Neither visual range
nor the extinction coefficient has this property.  For example, a 5 km change in visual range
or 0.01 km'1 change in extinction coefficient can result in a change that is either
imperceptible or very apparent depending on baseline visibility conditions.  Deciview allows
one to more effectively express perceptible changes in visibility, regardless of baseline
conditions.  A one deciview change is a small but perceptible scenic  change under many
conditions, approximately equal to  a 10% change in the extinction coefficient.  The deciview
metric also may be useful in defining goals for perceptible changes in visibility conditions
under future regulatory  programs.   Deciview can be calculated from  the light extinction
coefficient by the equation:

                               dv = 101og]0(aext/10 Mm-1)

Figure G-1 graphically illustrates the  relationships among light extinction, visual range, and
deciview.

-------
                                  G-4a
  1.6 T.
                                Extinction Coefficient

                           • • • • Visual Range
                              20    25    30
                               Haziness (dv)
FIGURE G-l.   VISUAL  RANGE AND  EXTINCTION COEFFICIENT  AS A
               FUNCTION OF HAZINESS EXPRESSED  IN DECIVIEW
                        Source:  Pitchford  and Malm, 1994

-------
              APPENDIX H
CLEAN AIR SCIENTIFIC ADVISORY COMMITTEE
            CLOSURE LETTERS

-------
               UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
                           WASHINGTON, D.C. 20460
                               March 15,1996
                                                            Office OF THE ADMNWnUTOa
                                                             SCtNCt ABVKOKY BOAHIJ
EPA-SAB-CASAC-LTR-96-005

Honorable Carol M. Browner
Administrator
U.S. Environmental Protection Agency
401 M. Street SW
Washington. DC 20460

            Re    Closure by the Clean Air Scientific Advisory Committee (CASAC)
                  on the draft Air duality Criteria for Paniculate Matter

Dear Ms. Browner:

      The Clean Air Scientific Advisory Committee (CASAC) of EPA's Science
Advisory Board (SAB) has held a series of public meetings during its peer review of the
Agency's draft documents which will form part of the basis for your decision regarding
the National Ambient Air Quality Standards (NAAQS) Tor Paniculate Matter (PM). The
Commrttfie has held public meetings on December 12-13.1994 (planning and
introductory issues): August 3-4. 1995 (review of the initial draft Criteria Document);
Decembei 14-15,1995 (review of the revised draft Criteria Document and the first draft
of the Staff Paper); and February 29,1996 (review of the revised draft Criteria
Document - specified chapters only). A review of the revised draft Staff Paper is
planned for May 16-17,1996. The primary Agency draft documents that we have
reviewed are the: a) Air Quafty Criteria for Particulate Mattfir (the Criteria  Document
prepared by the National Center for Environmental Assessment - Reseaidi Triangle
Park. NC - ORD). and b) Review of the National Ambient Air Quality Standards for
Particulate Matter: Policy Assessment of Scientific and Technical Information (the Staff
Papei prepared by the Office of Air Quality Planning and Standards - Research Triangle
Park. NC - OAR).

      As part of our review process, we have kept you informed of our findings through
two letter reports: a) Clean Air Scientific Advistvy Committee (CASAC) Comments on
the April 1995 draft Air Quality Criteria for Paniculate Matter (EPA-SAB-CASAC-
LTR-95-005; August 30.1995): and b) Clean Air Scientific Advisory Committee
(CASAC) Comments on the November. 1995 Drafts of the Air Quality Criteria for
Particulate Matter and the Review of the National Ambient Air Qua/Try Standards for
Particulate Matter Poficy Assessment of Scientific and Technical Information (OAQPS
Staff Paper), (EPA-SAB-CASAC-LTR-96-003. January b. 1996)
                                   OPTIONAL FORM 99 (7 9
-------
      The Clean Air Scientific Advisory Committee, supplemented by a number of
expert Consultants (hereinafter referred to as the Panel) reviewed a revised draft cf  the
PM Criteria Document and a first draft of the Staff Paper for Particulate Matter at a
meeting on December 14-15,1995 in Chapel Hill, NC. At that meeting and in
subsequent written comments by individual members which were provided to
EPA Staff, the Panel made numerous recommendations for improving the draft
document.  The Panel was impressed with the breadth and scope of the latest revision
of the draft Criteria Document and agreed that, except for Chapters 1 (Executive
Summary), 5 (Sources and Emissions), 6 (Air Quality), and 13 (Integrative Synthesis),
only minor revisions would be necessary to make the remainder of the draft Criteria
Document satisfactory for providing an adequate scientific basis for regulatory decisions
on PM based on available information. However, the Panel felt that Chapters 1, 5,6,
and 13 required major revisions which the Panel would need to review again.

      On February 29,1996, the Panel again met in Chapel Hill.  NC to review revised
drafts of Chapters 1, 5, 6, and 13 of the Criteria Document While Chapter.! 3 can be
improved, as suggested below. I want to take this opportunity, on  behalf of the entire
Panel, to commend Dr. Lester Grant and his staff in the National Center for
Environmental Assessment (NCEA) for producing its best ever example of a true
•integrative summary of the state of knowledge about the health effects of airborne PM
and the associations between the effects and the various available indices of PM
exposure.  NCEA has outlined some of the options for your subsequent choice of
available PM indicators for a NAAQS by examining the degrees of association between
various health indices and PM indicators including total suspended paniculate (TSP),
thoracic particulate (PM10). fine particulate (PM^), sulfate particulate (SO/), acid
paniculate (H*) and carbonaceous particulate (BS and  CoH), with available knowledge
from dosimetry, results of controlled human exposure studies in humans and laboratory
animais. and mechanistic understandings. This thorough review and evaluation also
provides an important starting point for focussing the future PM research program on
studies that can better identify the compositional and particle size characteristics of the
most biologically active agents within the PM10.  We were especially impressed that this
integrative summary could be produced in the short time period since our review of the
initial rough draft in December 1995.

       This letter is a summary of our findings and conclusions from the February 29th
meeting.  Our comments reflect our satisfaction with the improvements made in the
scientific quality and completeness of these chapters.  The changes made in these
chapters are consistent with our earlier recommendations. However, the Panel
 provided additional comments to your staff at the meeting and subsequently in writing.
Although we feel that it is essential to have these additional comments considered for
 incorporation in the Criteria Document, we did not feel that it was essential to review
 another revised version  and, thus, we came to closure on the entire Criteria Document
 anticipating incorporation of our suggested changes. It was our consensus that
 although our understanding of the health effects of PM is far from complete, a revised
 Criteria Document which incorporates the Panel's latest comments will provide  an

-------
adequate review of the available scientific data and relevant studies of PM.  With the
incorporation of our suggested changes, the revised Criteria Document will be very
comprehensive and will provide an adequate scientific basis for regulatory decisions on
paniculate matter based on available information. However, a number of members
have expressed concern that since we are closing on the Criteria Document before we
will be able to see the revised version, we have no assurance that our comments will be
incorporated. I will return to this concern later.

      I would like to summarize for you the Panel's major comments on Chapters 1,6,
and 13.  There were no major comments on Chapter 5. In Chapter 6, Panel members
raised issues concerning the definition and level  of background PM concentrations. The
Panel has provided the Agency with guidance in  the written comments to resolve these
concerns.  This is an important issue because some studies suggest effects at levels
which approach background concentrations.

      Of the 17 members of the Panel present, five were satisfied with Chapter 13 as
is, four had no substantive comments because their expertise was outside of Chapter
13. and eight had some substantive comments on one or more aspects of the chapter
which I summarize below.  The members who were satisfied with the chapter praised
the Agency for making a compelling case for PM25 being the best available surrogate
index for the causal agenL They thought EPA presented a large body of consistent and
coherent studies and that they were appropriately presented as an integrative synthesis.
The issues raised by the other Panel members regarding Chapter 13 fell into three
categories. First, several Panel members felt that additional discussions of the inherent
errors associated with air  sampling, estimating human exposure from central monitoring
data, and relating these data to excess mortality  and morbidity were necessary so that
the uncertainties of the relative risk estimates would be better appreciated.

      Second, about half of the Panel members  expressed concern that the case made
in the Criteria Document for PM2J5 being the best available surrogate for the principal
causative agent in PM1Q may be overstated, and  that EPA has not adequately justified
its rejection of other alternative explanations discussed next.  In addition, it needs to be
acknowledged that large particles (e.g., dc4.0 ;/m) may be responsible for acute
respiratory effects, especially in susceptible groups such as asthmatics.

      Third, several Panelists pointed out that a number of recently published (or
in-press) studies (including the Health Effects Institute study), which were conducted to
critically evaluate some of the epidemiological studies using alternative models or
including additional gaseous pollutant data, present a different perspective of the
PM/mortality issue than the one presented in this chapter. Collectively, these
reanalyses have confirmed the reprodua'bility of  the earlier studies, but they also
present a more complicated relationship in which causality does not appear to be
unambiguously attributed to any single pollutant  let alone a specific portion of the PM.
EPA on the other hand emphasized a PM causa! conclusion based on the pattern of

-------
associations across multiple sites having different pollutant mixtures. These results
need to be discussed adequately in Chapter 13.

      Our only comments on Chapter 1, the Executive Summary, were that it reflect the
revisions that have been recommended for Chapters 6 and 13.

      As mentioned above, Panel members have expressed concern that the Agency
may not be responsive to some of our comments or may misinterpret them since we will
not have another opportunity to review  the final document. This concern is another
unfortunate consequence of the court- mandated "accelerated" time schedule, but
nevertheless, it is a real concern. We anticipate being advised of text changes made in
response to our concerns prior to or at the May 16-17.1996 meeting, and we can
advise you afterward about whether our concerns have been adequately addressed by
the Agency.

      On behalf of the Panel, I would like to thank EPA staff for their considerable
efforts in preparing the Criteria Document on the accelerated schedule. We-iook
forward to seeing the revised final version once it is completed. The Panel also looks
forward to reviewing the revised Staff Paper during the public meeting presently
scheduled for May 16-17,1996.
                                         Sincerely.
                                         Dr. George T. Wolff, Chair
                                         Clean Air Scientific
                                          Advisory Committee

-------
                                     March 20,1996
Honorable Carol M- Browner
Administrator
U. S. Environmental Protection Agency
401 M Street SW
Washington DC 20460
   re: Supplement to the Closure Letter from the Clean Air Scientific Advisory Committee

Dear Ms. Browner:

 The co-signers of this letter are members of the Particulate Matter Criteria Document Review
Panel and consultants to flic Clean Air Scientific Advisory Committee (CASAC) of the Science
Advisory Board, U.S. EPA. This letter is not being sent as a minority report to the CASAC
closure letter, but as a supplement to address some of the concerns raised in the CASAC letter.
We were selected for the CASAC review of the Particulate Matter Criteria Document because of
our combined expertise in the interpretation of epidemiological studies, our understanding of
the literature on the human health effects of particulate air pollution, and our familiarity with
the use of air monitoring data in analyzing human health effects. As individuals, we have been
extensively involved in conducting studies of population exposure to air pollution and
evaluating the human health effects of this exposure.

As noted in the closure letter to you on the draft Air Quality Criteria for Particulate Matter from
the Chair of CASAC, the Panel members praised the EPA criteria document for its excellent
integrative synthesis of the literature. Overall, most panel members concluded that the
document made a persuasive case that population exposure to particulate matter (PM)  is
causally associated with excess mortality and morbidity in the U. S. even  at concentrations at
and below the existing primary air quality standard.  WMe the cosigners of this letter are in
agreement with this judgment, we are aware that some of our Panel colleagues have
reservations about this important conclusion. Our purpose in this supplementary letter is to
make explicit our reasons for reaching our conclusion, in order to assist the staff of the National
Center for Environmental  Assessment in addressing the reservations of our colleagues. We also

-------
March 20, 1996                              2                     PM Supplement

intend our comments to aid the staff of the Office or Air Quality Planning and Standards in
preparing its staff paper in support of a revised particulate air quality standard.

The closure letter from the Chair of CASAC notes that the concerns of Panel members who are
not in full agreement with the above conclusion fall into three categories:

        1. Uncertainties in the human health risks of particalate air pollution, arising from
        errors in air monitoring, from estimating human exposure from central monitoring ^"ky
        and from relating these data to excess mortality and morbidity.

        2. Concern that the case for PM23 being the best available surrogate for the principal
        causative agent in particulate air pollution may be overstated, and that EPA has not
        adequately justified its rejection of other alternative explanations.

        3. Recently published studies that appear to contradict, or at least to present a different
        perspective on. the conclusions reached by EPA in its integrative synthesis of the
        literature.

Regarding these concerns overall, the writers of this letter wish to make it dear that we are not
arguing that PM^S fc the causal agent of the observed excess mortality and morbidity
associated with particulate air pollution. In our judgment the studies reviewed in the criteria
document, specifically those considered in Chapter 12 (Epidemiological Studies), are persuasive
in demonstrating a causal relationship between particulate air pollution, as measured by
different methods in the various studies, and excess mortality and morbidity. However, the
evidence does not allow us to conclude that a specific physical or chemical component of the
particulate mass is clearly the responsible causal agent Our conclusion is analogous to making
the assertion that cigarette smoke is a cause of hing cancer and nonmalignant respiratory
disease,evcn though the specific causal agent in cigarette smoke has not been identified among
the many chemicals known to be present in cigarette smoke.

The reasons for concluding that particulate air pollution is causally related to excess mortality
and morbidity have been well stated in the integrative synthesis (Chapter 13) of the criteria
document. For heuristic purposes, we will summarize these reasons here, and cite locations in
Chapter 13 where supporting sentences and paragraphs are presented:

-------
March 20,19%                              3                      PM Supplement

        A large number (20) of epidemioiogical time-series studies have consistently found a
        statistically significant association between daily variation in particulates and total
        mortality in cities of the U.5., Canada, Latin America, the U.K. and continental Europe.
        These findings argue against the associations being attributable to statistical sampling
        variation, ie. the role of chance (Section 13.4.1.1).

•       The results of these time-series studies cannot be attributed to the vagaries of statistical
        modeling (Section 13.4.3.2), nor to confounding by season or weather (Section 13.4.33).

•       The results of the time-aeries studies cannot be attributed to other criteria air pollutants.
        The mortality effect of particulates is found whether or not other pollutants are present
        at elevated concentrations, though it is difficult to separate the effects of particulates
        from other pollutants when the latter covary with particulates. The most persuasive
        evidence that the causal agent is some component of the airborne particulate mass is in
        studies of cities or seasons where other poDutants are present at very low
        concentrations. Across the range of the 20 studies mentioned above, particulate air
        pollution is the only pollutant that is consistently associated, with excess daily mortality,
        and the estimate of its effect is relatively stable when adjusted for the presence of co-
        pollutants. There are exceptions to this stability, particularly in those cities where
        particulate and gaseous air pollutants are highly intercorrelated. But no monitored air
        pollutant, other than particulate matter, can account for tha consistently observed excess
        mortality in these studies (Section 13.43.4). Excess morbidity from cardiopulmonary
        diseases has also been observed in a considerable number of studies (Section 13.4.1.2),
        and the morbidity relationship with ambient particulate concentrations is stronger
        overall and more consistent than for any other air pollutant,

 •       There is considerable coherence between the observed mortality and morbidity effects
        of particulate air pollution. Not only is excess mortality from cardiovascular and
        respiratory diseases observed, but on days of higher particulalea excess hospiUlizatk>ns
        for cardiovascular and respiratory diseases are reported. These mortality and
        morbidity excesses are strongest in populations that would be expected to be more
        susceptible to the effects of air pollution, particularly the elderly. The relation of
        particulates with mortality is strongest also for cardiopulmonary diseases rather than
        for other disease categories. On days of high particulates, there is an increased
        proportion of deaths from chronic obstructive pulmonary disease, pneumonia, heart
        disease and deaths among the elderly  than on days of low particulates. These findings

-------
March 20,1996                             4                      PM Supplement

        are supportive of a causal role for participate air pollution, since they are health
        endpoints one would most anticipate from exposure by the inhalation route (Section
        13.435 and Section 13 J.I).

Given the striking consistency of the above studies, their robustness to variations in statistical
modeling,, the coherence among different but closely related health endpoints, and the  empirical
elimination of any alternative explanation for the findings, we conclude that a causal
interpretation for particulate air pollution exposure is reasonable and defensible. This
conclusion is farther supported by longitudinal cohort studies of populations in which a
geographical gradient in particulate air pollution was associated with « corresponding gradient
in total mortality, in cardiopulmonary mortality and in lung cancer.  These studies carefully
controlled for other individual risk factors for these health endpoints (Section 13.4.1.1).

With specific reference to the first category of concern expressed by our Panel colleagues,
although population exposure to air pollution cannot be perfectly estimated based on central
monitoring, these inherent errors in exposure estimation are more likely to cause an
underestimation of the adverse health effects associated with pollution exposure, particularly in
longitudinal cohort studies where individual risk factor* and exposures are directly related to
health effects. Thus the consistent positive findings cannot be attributed to exposure
measurement error. Furthermore, there is growing evidence that fine particles are more
uniformly distributed over large geographic areas than are coarse particles (Section 13.2.4), that
measurements at one site give a reasonable estimate of the fine particulate concentrations across
a city (Section 13.2.6), and that fine particles penetrate and have longer lifetimes indoors than
coarse particles (Section 13.2.6). This evidence supports using ambient measures of fine
particuLttes at a central site as an acceptable estimate of the average exposure of people in the
community (Section 13.2.6). For these reasons, we judge that uncertainties arising from air
monitoring and human exposure estimation do not negate the consistent excess mortality and
morbidity associations discussed above.

 With regard to the second concern of our Panel colleagues, we believe that the case has been
 made thai fine particulates, as measured by PM25, «re the best surrogate currently available for
 the component of particulate air pollution that is associated with excess mortality and
 morbidity. We emphasize once again that we are not claiming that PM25 is the causal agent,
 but rather that PM23 i» a better measure, than any alternative metric,  of the complex in the
 particulate mass that is causing excess mortality and morbidity. Distinguishing between PMio
 and PM2.5 is difficult, given the high correlation between these two pollutants in both time and

-------
March 20,19%                              S                      PM Supplement

space. In many studies, either metric will provide nearly the same estimate of the exposure-
response relationship.  However, a number of recert re-analyses of mortality and morbidity
have been performed to address the issue of whether fine or coarse particulates (the latter
indexed by subtracting PM25 from PMjo) more consistently predicts a relationship with
adverse health effects.  These studies, as reviewed Section 13.4.1.1 and Tables 13-3,13-4  and 13-
5 of the Criteria Document, conclude that excess mortality, hospital admissions for respiratory
diseases and decreased lung function are more strongly and consistently associated with fine
rather than with coarse mode particulates. These finding!* are also supported by earlier studies
in the U.K. in which British Smoke measurements, which primarily reflect the contribution of
the- fine particle mode, were consistently associated with excess mortality. Finally, several
characteristics of fine mode particles, as opposed to the coarse mode, arc more consistent with
the observed excess mortality and morbidity observed in epidemiological studies. As noted
earlier, these characteristics are: (1) fine particulates are more uniform in distribution than the
coarse mode across urban areas, (2) fine particulates penetrate into indoor environments more
completely than coarse particles, and (3) fine particulittea have a more prolonged residence time
in indoor air than coarse particles.  These points are discussed in Section 13.7, Summary and
Conclusions. Given that a causal association of excess mortality and morbidity with particulate
air pollution has been established, we concur with staffs judgments that fine particulates are the
best available surrogate for the population exposures associated with these health effects.

With regard to the third concern of our Panel colleagues, some studies have recently been
published that arc interpreted as contradicting the conclusion that particulate air pollution is
causally associated with excess  mortality «nd morbidity. We agree that, in its revision of the
criteria document, EPA needs to address these apparent discrepancies more explicitly, and we
offer the following comments to assist staff in that task.

First, the Health Effects Institute (HEI) reanalysis does not contradict any of the above
conclusions. The HEI analysis conclusively demonstrated that the positive findings from the
original studies selected for reanalysis were replicable, were not an artifact of statistical
modeling. and were not confounded by idiosyncrasies in the method to control for season or
weather. The HEI investigators then proceeded to apply their statistical modeling procedure to
data from Philadelphia. They reported moderately high inU>icoi relations between parttculatts.
as measured by total suspended particulate (TSP) measurements, and several of the pollutant
gawes,  and, as expected, found that under these conditions, they could not attribute the observed
exposure-response mortality relationships to TSP alone. They further observed that the TSP and
SO2 effects were not independent of one another, and that the TSP effect was stronger in some

-------
March 20,19%                              6                      PM Supplement

seasons of the year and at some concentrations of SC>2/ while the SC>2 effect was stronger in
other seasons and at some concentrations of ISP. The HEI investigators appropriately
concluded that, because of the high intercorrelations between pollutants in Philadelphia,
mortality effects could not be attributed solely to particulates. More importantly, in their
further report on this phase of their study, they concluded that "insights into the effects of
individual criteria pollutants can be best gained by assessing rffects across locations having
different pollutant mixes and not from regression modeling of data from single locations" ("Air
Pollution and Mortality in Philadelphia, 1974-1988", interim report dated February 9,1996). The
EPA Criteria Document undertakes this assessment of effects across locations having different
pollutant mixes, and this assessment was discussed above (in the third bulletted paragraph)

One published reanalysis (Moolgavkar S: Epidemiology 1995; 6: 476-4S4) of the Philadelphia
mortality data set has been interpreted as contradicting the findings of the original study
(Schwartz J & Doctcry DW: Am Rev Resp Dis 1992:145:60O404), which concluded that
particulates were positively associated with variations in daily mortality.  However, the HEI re-
analysis, reported above, confirmed the findings of the original study, but, more importantly,
noted that it was not possible in Philadelphia to attribute the mortality effect exclusively to
particulates or individual gaseous pollutants, due to their high intercorrelations, as previously
discussed. Separation of the effects of these pollutants requires analyses in a variety of locations
with different pollutant mixes.

 Presentatiuns and papers by Upfert and Wyzga (Inhalation Toxicology 1995; 7:671-689) discuss
 uncertainties in identifying responsible pollutants in epidemiological studies. The latter article
 raises the important issue of measurement error, but in applying its analysis to the Philadelphia
 data set, it encounters the same problem of intercorrelated pollutants and the inability to
 partition health effects exclusively or primarily to one of the pollutants. Similarly, the analysis
 of the Philadelphia data set by U and Roth (Inhalation Toxicology 1995; 7:45-58) purports to
 show that a panoply of seemingly conflicting findings is produced with different modeling
 strategies, but this paper is superseded by the HEI report, which shows conclusively that the
 confounding effect of weather was appropriately controlled in the original analysis, and that the
 original results arc not an artifact of the modeling strategy.

 Finally, among papers considered as not supporting the main conclusion of the EPA criteria
 document, that of Styer et al. (Environ Health Perspec 1995; 103:49CM97) fitted separate
 regressions to each month of the year and found significant part'culate effects only in a few of
 the months.  But such partitioning of data in small time segments is considered to be

-------
March 20,1996                              7                          PM Supplement

inappropriate because it results in a significant loss of statistical power and thus a loss of sensitivity to
the moderate relative risk associated with ambient air pollution and a loss of ability to separate the
effects of one pollutant as opposed to another.
There arc several reasons why the mortality and morbidity effects of particulate air pollution will not be
the same in all cities and at aH seasons of the year. Therefore, there will not be total agreement among
all published studies in the magnitude of the adverse effect per unit of particulate exposure.  The
reasons for these variations in estimates of the exposure-response relationship are several (as discussed
in Section 13.4.1.1): (1) the toxicity of particulates likely depends on size distribution and chemical
composition, and these characteristics vary among geographic areas. (2) local populations differ in
demographic and soaoeconoxnic characteristics, and these differences will be likely to modify the health
effects of particulate exposures. (3) the health status of communities differs among geographic areas,
and thus the susceptibility of populations to the same level of particulate air pollution will vary.  (4)
average levels of copollutants will vary across geographic areas, and these may cause small or moderate
variations in the particulate effect. In spite of these considerations, there is a remarkable consistency in
the body of epidemiological studies/ showing a positive exposure-response association between
particulars and mortality and morbidity.  In our judgment, EPA has appropriately synthesized this
evidence and drawn a responsible public health conclusion, namely, that particnlate concentrations at
current levels are causally associated with excess mortality and morbidity. Furthermore, we agree that
fine participates, as currently indexed by PM2JJ, are the most appropriate indicator for the component
of the particulate air mass to which these adverse effects are attributed. We also agree that some
adverse health effects may be related to the coarse particulate mode, and that therefore it is desirable to
consider fine and coarse mode particulates as separate candidates for air quality standards.  This is the
final conclusion of Chapter 13 of the Criteria Document, and we hope that our discussion -will assist the
EPA staff in presenting firmer support for their conclusion.

                                            Sincerely,
    Morton Lippmonn, Professor                             Jan Stolwijk, Professor
    Nelson Institute of Environmental Medicine                Department of Epidemiology and
    New York University                                    Public Health
                                                            Yale University
    Carl Shy, Professor and Chair                             Frank Speizer, Professor
    Department of Epidemiology                             Charming Laboratory
    University of North Carolina at Chapel Hill                Harvard Medical School
c: Members of the Particulate Matter Criteria Document Review Panel

-------
               UNITED STATES ENVIRONMENTAL PROTECTION AGENCY

                            WASHINGTON, D.C. 20460
                                                              OFFICE OF THE ADMINISTRATOR
                                June 13, 1996                   SCIENCE ADVISORY BOARD
EPA-SAB-CASAC-LTR-96-008
Honorable Carol M. Browner
Administrator
U.S. Environmental Protection Agency
401 M. Street SW
Washington, DC 20460

Subject:     Closure by the Clean Air Scientific Advisory Committee (CASAC) on the Staff
            Paper for Paniculate Matter

Dear Ms. Browner:

      The Clean Air Scientific Advisory Committee (CASAC) of EPA's Science
Advisory Board (SAB) has held a series of public meetings during its peer review of the
Agency's draft documents which will form part of the basis for your decision regarding
the National Ambient Air Quality Standards (NAAQS) for Particulate Matter (PM). The
Committee has held public meetings on December 12-13, 1994 (planning and
introductory issues); August 3-4,  1995 (review of the initial draft Criteria Document);
December 14-15, 1995 (review of the revised draft Criteria Document and the first draft
of the Staff Paper); February 29,  1996 (review of the revised draft Criteria Document -
specified chapters only, and the Office of Air Quality Planning and Standards (OAQPS)
Risk Assessment Plan); and May 16-17, 1996 (review of the revised draft Staff Paper).
The primary Agency draft documents that we have reviewed are the: a) Air Quality
Criteria for Particulate Matter (the "Criteria Document" prepared by the National Center
for Environmental Assessment - Research Triangle Park, NC - ORD), b) Review of the
National Ambient Air Quality Standards for Particulate Matter: Policy Assessment of
Scientific and Technical Information (the "Staff Paper" prepared by the Office of Air
Quality Planning and Standards - Research Triangle Park, NC - OAR), and c) A
Particulate Matter Risk Analysis for Philadelphia and Los Angeles (draft), 1996,
Prepared by Abt Associates for US EPA.

       As part of our review process, we have kept you informed of our findings through
three letter reports: a) Clean Air Scientific Advisory Committee (CASAC) Comments on
the April 1995 draft Air Quality Criteria for Particulate Matter (EPA-SAB-CASAC-LTR-
95-005; August 30. 1995); b) Clean Air Scientific Advisory Committee (CASAC)
Comments on the November. 1995 Drafts of the Air Quality Criteria for Particulate
Matter and the Review of the National Ambient Air Quality Standards for Particulate

-------
Matter Policy Assessment of Scientific and Technical Information (OAQPS Staff
Paper), (EPA-SAB-CASAC-LTR-96-003, January 5, 1996), and c) Closure by the Clean
Air Scientific Advisory Committee (CASAC) on the draft Air Quality Criteria for
Particulate Matter (EPA-SAB-CASAC-LTR-96-005, March 15, 1996).

      The Clean Air Scientific Advisory Committee, supplemented by a number of
expert Consultants (hereinafter referred to as the "Panel"), reviewed a first draft of the
Staff Paper for Particulate Matter at the December 14 and 15, 1995 meeting in Chapel
Hill, NC. At that meeting and in subsequent written comments by individual members
which were provided to EPA Staff, the Panel made numerous recommendations for
improving the draft document. The Panel met again on May 16, 1996 in Chapel Hill, NC
and on May 17,1996 in Research Triangle Park, NC to review a revised draft of the
Staff Paper and the recommendations contained within the Staff Paper for the level and
form of the proposed PM NAAQS, This letter is a summary of our findings and
conclusions from that meeting.

      It was the consensus of the Panel that although our understanding of the health
effects of PM is far from complete, the Staff Paper, when revised, will provide an
adequate summary of our present understanding of the scientific basis for making
regulatory decisions concerning PM standards. Seventeen of the twenty-one Panel
members voted for closure. There were two no votes, one abstention,  and one
absence. However, most of the members who voted for closure did so under the
assumption that the Agency would make significant changes to the next version of the
Staff Paper which is due by July 15, 1996 (a court ordered mandate).  The desired
changes have been articulated to your staff at the meeting and subsequently in writing.

      The Panel endorses the EPA Staffs recommendation not to establish  a separate
secondary PM NAAQS for regulating regional haze and agrees that there is an
inadequate basis for establishing a secondary NAAQS to reduce soiling and  material
damage effects.

      The attached table (Table I) summarizes the Panel members' recommendations
concerning the form and levels of the primary standards. Although some Panel
members prefer to have a direct measurement of coarse mode PM (PM1ft.zs) rather than
using PM10 as a surrogate for it, there is a consensus that retaining an annual PM10
NAAQS at the current level is reasonable at this time. A majority of the members
recommend keeping the present 24-hour PM10 NAAQS, at least as an option for the
Administrator to consider, although those commenting on the form of the standard
strongly recommended that the form be changed to one that is more robust than the
current standard. There was also a consensus that a new PM2i5 NAAQS be
established, with nineteen Panel members endorsing the concept of a 24-hour  and/or
an annual PMj3 NAAQS. The remaining two Panel members did not think any  PM25
NAAQS was justified.  However, as indicated in Table I, there was no consensus on the
level, averaging time, or form of a PM2S NAAQS. At first examination of Table I, the
diversity of opinion is obvious and appears to defy further characterization.  However,

-------
the opinions expressed by those endorsing new PM2S NAAQS can be classified into
three broad categories. Four Panel members supported specific ranges or levels within
or toward the lower end of the staffs recommended ranges.  Seven Panel members
supported specific ranges or levels near, at. or above the upper end of staffs
recommended ranges. Eight other Panel members declined to select a specific range
or level, but most had comments which appear as footnotes in Table I.

      A number of Panel members based their support for a PM25 NAACS on the
following reasoning: there is strong consistency and coherence of information indicating
that high concentrations of urban air pollution adversely affect human health, there are
already NAAQS that deal with all the major components of that pollution except PM2 s>
and there are strong reasons to believe that PM2.5 is at least as important as PM10>23 in
producing adverse health effects.

      Part of this diversity of opinion can be attributed to the accelerated review
schedule.  While your staff is to be highly commended for producing such quality
documents in such a short period of time,  the deadlines did not allow adequate time to
analyze, integrate, interpret, and debate the available data on this very complex issue.
Nor does a court-ordered schedule recognize that achieving the goal of a scientifically
defensible NAAQS for PM may require iterative steps to be taken in which new data are
acquired to fill obvious and critical voids in our knowledge.  The previous PM NAAQS
review took eight years to complete.

      The diversity of opinion also reflects the many unanswered questions and
uncertainties associated with establishing causality of the association between PM25
and mortality. The Panel members who recommended the most stringent PM2 5
NAAQS, similar to the lower part of the ranges recommended by the Staff, did so
because they concluded that the consistency and coherence of the epidemiology
studies made a compelling case for causality of this association.  However, the
remaining Panel members were influenced, to varying degrees by the many
unanswered questions and uncertainties regarding the issue of causality.  The concerns
include: exposure misclassification, measurement error, the influence of confounders,
the shape of the dose-response function, the use of a national PMiS/PM,0 ratio to
estimate local PM2 5 concentrations, the fraction of the daily mortality that is advanced
by a few days because of pollution, the lack of an understanding of lexicological
mechanisms, and the existence of possible alternative explanations.

      In recommending that the staff carry out a risk assessment, it was the
expectation of CASAC that the risk assessments would narrow the diversity of opinion
by evaluating how all of the uncertainties propagate throughout the entire model.
However, not all of the uncertainties could be included and the combined effect of all of
them could not be examined.  The Panel recommended that additional analyses be
conducted to present combined uncertainties.  However, currently the risk assessments
are of limited value in narrowing the diversity of opinion within the Panel.

-------
      The Panel is unanimous, however, in its desire to avoid being in a similar
situation when the next PM NAAQS review cycle is under way by a future CASAC
Panel.  The Agency must immediately implement a targeted research program to
address these unanswered questions and uncertainties. It is also essential that we
obtain long-term PM^ measurements. CASAC is ready to assist the Agency in the
development of a comprehensive research plan that will address the questions which
need answers before the next PM review cycle is completed.  We understand that your
staff is preparing a PM research plan for our review later this summer. We look forward
to providing our comments on this important matter.

      CASAC recognizes that your statutory responsibility to set standards requires
public health policy judgments in addition to determinations of a strictly scientific nature.
While the Panel is willing to advise you further on the PM standard, we see no need, in
view of the already extensive comments provided,  to review any proposed PM
standards prior to their publication in the Federal Register.  In this instance, the public
comment period will provide sufficient opportunity for the Panel to provide any additional
comment or review that may be necessary.

      Thank you for the opportunity to present the Panel's views on this important
public health issue. We look forward to your response to the advice contained in this
letter.
                                          Sincerely,
                                          Dr. George T. Wolff, Chain
                                          Clean Air Scientific Advisory Committee

-------
                               TABLE!
      Summary of CASAC Panel Members Recommendations
                             (all units ug/m3)

Current NAAQS
EPA Staff Recommendation
PNLs
24-hr
N/A
18-65
PM2.B
Annual
N/A
12.5-20
PM10
24-hr
150
15013
PM,o
Annual
50
40-50

Name
Ayres
Hopke
Jacobson
Koutrakis
Larntz
Legge
Lippmann
Mauderly
McClellan
Menzel
Middleton
Pierson
Price
Shy
Samet1
Seigneur
Speizer1
Stolwijk
Utell
White
Wolff
Discipline
M.D.
Atmos. Sci.
Plant Biologist
Atmos. Sci.
Statistician
Plant Biologist
Health Expert
Toxicolooist
Toxicologist
Toxicologist
Atmos. Sci.
Atmos. Set.
Atmos. Sci./
State Official
Epidemiologist
Epidemiologist
Atmos. Sci.
Epidemiologist
Epidemiologist
M.D.
Atmos. Sci.
Atmos. Sci.

yes2
20 - 503
yes2
yes2'5-8
no
a 75
20 - SO3
50
noe
no
yes2'3*'2
yes2-9
yes3'10
20-30
yes2-'1
^yes3-5
20-50
757
:>65
no
*753-7

yes2
20-30
yes2
yes2-5'6
25-307
no
15-20
20
nofl
no
yes25
yes2-9
yes10
15-20
no
no
no
25-307
no
• 20
no

150
no
150
no
nc
150
no
150
150
150
1503'13
yes4
no3-'
no
150
15013
no
150
150
150
1503

50
40 -50"
50
yes4
yes2
40-50
40-50
50
50
50
50
yes4
yes4
50 I
yes2
50
40-50
50
50
50
50
  not present at meeting; recommendations based on written comments
2 declined to select a value or range
3 recommends a more robust 24-hr, form
* perfers a PM10.2 5 standard rather than a PM10 standard
5 concerned upper range is too low based on national PM2-5/PM10 ratio
6 leans towards high end of Staff recommended range
7 desires equivalent stringency as present PM10 standards

-------
8 if EPA decides a PM2.5 NAAQS is required, the 24-hr, and annual standards
  should be 75 and 25 ug/m3, respectively with a rooust form
9 yes, but decision not based on epidemiological studies
10  low end of EPA's proposed range is inappropriate; desires levels selected to
   include areas for which there is broad public and technical  agreement that
   they have PM7S pollution problems
11  only if EPA has confidence that reducing PM25 will indeed reduce the components
   of particles responsible for their adverse effects
12 concerned lower end of range is oo close to background
13  the annual standard may be sufficient; 24-hr level recommended if 24-hour
   standard retained

-------
                                   NOTICE

   This report has been written as part of the activities of the Science Advisory Board, a
public  advisory group  providing extramural  scientific  information  and advice to the
Administrator and other officials of the Environmental Protection Agency. The Beard is
structured to provide balanced, expert assessment of scientific matters related to problems
facing  the Agency. This report has not been reviewed for approval by the Agency and,
hence,  the contents of this report do not necessarily represent the views and policies of the
Environmental Protection Agency, nor of other agencies in the Executive Branch of the
Federal government, nor does mention of trade names or commercial products constitute
a recommendation for use

-------
                     U.S. Environmental Protection Agency
                            Science Advisory Board
                    Clean Air Scientific Advisory Committee
                        Particulate Matter Review Panel
Chairman
Dr. George T. Wolff
General Motors
Environmental & Energy Staff
Detroit, Mi

Members
Dr. Stephen M. Ayres
Office of international Programs
Virginia Commonwealth University
 /Medical College of Virginia
Richmond, VA

Dr. Phil Hopke
Department of Chemistry
Clarkson University
Pottsdam. NY

Dr. Jay S. Jacobson
Boyce Thompson Institute
Cornell University
Ithaca. NY

Dr. Joe L. Mauderly
Inhalation Toxicology Research
 Institute
Lovelace Biomedical  &  Environmental
 Research Institute
Albuquerque. NM

Dr. James H. Price, Jr.
Texas Natural Resource Conservation
 Commission
Austin. TX

Consultants
Dr. Petros Koutrakis
Harvard School of Public Health
Boston, MA
Dr. Morten Lippmann
Institute of Environmental Medicine
New York University
Tuxedo, NY

Dr. Kinley Larntz
Department of Applied Statistics
University of Minnesota
St. Paul. MN

Dr. Allan Legge
Biosphere Solutions
Calgary, Alberta, Canada

Dr. Roger  Q. McClellan
Chemical Industry Institute of
 Toxicology
Research Triangle Park, NC

Dr. Daniel Menzel
Department of Community
 and Environmental Medicine
University of California, Irvine
Irvine, CA

Dr. Paulette Middleton
Science and Policy Associates
Boulder, CO

Dr. William R. Pierson
Energy & Environmental  Engineering
 Center
Desert Research Institute
Reno, NV

-------
Dr. Carl M. Shy
Department of Epidemiology
School of Public Health
University of North Carolina
Chapel Hill, NC

Dr. John Samet
School of Hygiene & Public Health
Johns Hopkins University
Baltimore. MD

Dr. Christian Siegneur
AER, Inc
San Ramon, CA

Dr. Frank Speizer
Harvard Medical School
Channing Lab
Boston, MA

Dr. Jan Stolwijk
Yale University
New Haven, CT
Dr. Mark Utell
Pulmonary Disease Unit
University of Rochester Medical Center
Rochester, NY

Dr. Warren White
Washington  University
St. Louis, MO

Science Advisory Board Staff
Mr. A. Robert Flaak
Designated Federal Official
U.S. EPA
Science Advisory Board
Washington. DC

Ms. Dorothy Clark
Staff Secretary
U.S. EPA
Science Advisory Board
Washington, DC

-------
                         DISTRIBUTION LIST
Administrator
Deputy Administrator
Assistant Administrators
Deputy Assistant Administrator for Science, ORD
Director, Office of Science Policy, OPD
Director, Office of Air Quality Planning and Standards, OAR
Director, National Center for Environmental Assessment, ORD, RTP. NC
EPA Regional Administrators
EPA Laboratory Directors
EPA Headquarters Library
EPA Regional Libraries
EPA Laboratory Libraries
Library of Congress
National Technical Information Service

-------
                                    REFERENCES
Abbey, D. E.; Lebowitz, M. D.; Mills, P. K.; Petersen, F. F.; Beeson, W.  L.;
       Burchette, R. J. (1995a) Long-term ambient concentrations of particulates and
       oxidants and development of chronic disease in a cohort of nonsmoking California
       residents. Inhalation Toxicol. 7: 19-34.

Abbey, D.E.; Ostro, B.E.; Petersen, F.; Burchette, R.J. (1995b)  Chronic respiratory
       symptoms associated with estimated long-term ambient concentrations of fine
       particulates less than 2.5 microns in aerodynamic diameter (PM2.5) and other air
       pollutants.  J. of Exp. & Environ.  Epidem. 5: 137-159.

Abbey, D.E.; Hwang, B.L.; Burchette, R.J.; Vancuren, T.; Mills, P.K. (1995c)  Estimated
       long-term ambient concentrations of PMIO and development of respiratory symptoms
       in a non-smoking population.  Arch,  of Environ. Hlth. 50: 139-152.

Abbey, D.E.; Petersen, M.P.H.; Mills, P.K.; Beeson, W.L. (1993)  Long-term
       concentrations of total suspended particulates, ozone, and  sulfur dioxide and
       respiratory symptoms in a nonsmoking populations.  Arch, of Environ. Hlth:48: 33-
       46.

Abbey, D.  E.; Mills, P. K.; Petersen, F.  F.; Beeson, W. L. (1991) Long-term ambient
       concentrations of total suspended particulates and oxidants as related to incidence of
       chronic disease in California Seventh-Day Adventists. Environ. Health Perspect. 94:
       43-50.

Abt Associates (1996a).  Proposed Methodology for PM Risk Analyses in Selected Cities
       (Draft).  Prepared by Abt Associates, Inc., Hampden Square,  Suite 500, 4800
       Montgomery Lane, Bethesda, MD  20814-5341  12 February  1996.

Abt Associates (1996b). A Paniculate Matter Risk Analysis for Philadelphia and Los
       Angeles.  Prepared by Abt Associates for US EPA, OAQPS, Hampden  Square, Suite
       500, 4800 Montgomery  Lane,  Bethesda, MD  20814-5341 3 July 1996.

Ackermann-Liebrich, U.; Leuenberger, P.; Schwartz, J.; Schindler,  C.; Monn, C.;
       Bolognini, B.; Bongard, J.P.; Brandi, O.; Domemighetti,  G.;  Elsasses, S.; Grize, L.;
       Karrer, W.; Keller, R.;  Keller-Wossidlo, H.; Kijnzli, N.;  Martin, B.W.; Medicii,
       A.P.;  Perruchoud, A.P.; Schoni, M.H.; Tschopp, J.M.; Villiger, B.;  Wuthrich,
       Zellwege, Zemp, E. Lung Function  and Long-term Exposure to Air Pollutants in
       Switzerland. Am. J. Respir. Crit. Care Med.: submitted.

Akira, S.; Kishimoro,  T.; (1992) Ih-6 and Nf-IL6 in acute phase  response and viral
       infection.  Immunol. Rev. 127:25-50.

-------
                                         R-2

Altshuller, A.P.  (1982)  Relationships involving particle mass and sulfur content at sites in
       and around St. Louis, MO.  Atmos. Environ. 16:837-843.

Amdur, M. O.; Chen, L. C. (1989) Furnace-generated acid aerosols: speciation and
       pulmonary effects. In: Symposium on the health effects of acid aerosols; October
       1987; Research Triangle Park, NC. Environ. Health Perspect. 79: 147-150.

Anderson, P. J.; Wilson, J. D.; Killer, F. C.  (1990) Respiratory tract deposition of ultrafme
       particles in subjects with obstructive or restrictive lung disease. Chest 97: 1115-1120.

Anderson, I., Lundqvist, G.R.; Proctor, D.F.; Swift, D.L.(1979) human responses to
       controlled levels of inert dust. Am. Rev. Respir. Dis.  119: 619-627.

Anuszewski, J.;  Larson, T.V.; Koenig, J.Q.  (1992)  Simultaneous indoor and outdoor
       particle light scattering  measurements at nine homes using a portable nephelometer.
       Presented at:  meeting of the American Association  for Aerosol Research; paper no.
       3A.5.

Askey, A.; Lyon, S.B.; Thompson, G.E.; Johnson, J.B.; Wood, G.C.; Sage, P.W.; Cooke,
       M.J. (1993) Effect of fly-ash particulates on the atmospheric corrosion of zinc and
       mild steel.  Corros. Sci. 34:  1055-1081.

Avol,  E.L.; Linn, W.S.; Whynot, J.D.; Anderson, K.R.; Shamoo, D.A.; Valencia, L.M.;
       Little, D.E.; Hackney,  J.D. (1988a)  Respiratory dose-response study of normal and
       asthmatic volunteers exposed to sulfuric acid aerosol in the sub-micrometer size
       range.  Toxicol.  Ind. Health 4:173-184.

Avol,  E.L.; Linn, W.S.; Wightman, L.H.; Whynot, J.D.;  Anderson, K.R.; Hackney, J.D.
       (1988b) Short-term respiratory effects of sulfuric acid in fog: a laboratory study of
       healthy and asthmatic volunteers.  JAPCA 38:258-263.

Bachmann, J.D. (1996)  Attachment to letter to Dr. George T. Wolff from  Dr. Karen M.
       Martin,  dated July 8, 1996.  Subject of attachment:  Basis for Estimated Maximum
       Daily Background PM2 5 Concentration.

Baedecker, P.A.; Edney, E.G.; Moran, P.J.;  Simpson, T.C.; Williams,  R.S. (1991) Effects
       of acidic deposition on materials.  In: Irving, P.M., ed. Acidic deposition: state of
       science and technology, volume III: terrestrial, materials, health and visibility effects.
       Washington, DC: The U.S. National Acid Precipitation Assessment  Program. (State
       of science and technology report no. 19).

Baily, D.L.R.; Clayton, P. (1980)  The measurement of suspended paniculate and carbon
       concentrations in the atmosphere using standard smoke shade methods. Stevenage,
       Hertfordshire, United Kingdom: Warren Spring Laboratory;  report no. LR 325 (AP).

-------
                                         R-3

Barton, K. (1958) The influence of dust on atmospheric corrosion of metals.  Werkst.
      Korros. 8/9: 547-549.

Bar-Ziv and Goldberg (1974).  Simple silicieous pneumoconiosis in Negev Bedouins.  Arch.
      Envion. Health 29:121-126.

Bates, D.V. (1992) Health indices of the adverse effects of air pollution:  the question of
      coherence.  Environ. Res. 59:336-349.

Blais, G. (1996) Memorandum to File.  PM10 NAAQS Implementation Case Studies.  U.S.
      EPA, Office of Air Quality Planning and Standards. April 24,  1996.

Bobak, M.; Leon,  D. A. (1992) Air pollution and infant mortality in the Czech Republic,
       1986-1988.  Lancet (8826):  1010-1014.

Booz, Allen and Hamilton,  Inc. (1970) Study to determine residential soiling costs of
      particulate air pollution. Washington, DC:  U.S. Department of Health, Education,
      and Welfare, National Air Pollution Control Administration; Available from: NTIS,
      Springfield, VA; PB-250807.

Bowden, D.H. (1987) Macrophages, dust, and pulmonary diseases. Exp.  Lung Res. 12: 89-
       107.

Bouhuys, A.; Beck, G. J.; Schoenberg, J. B. (1978) Do present levels of air pollution
      outdoors affect respiratory health? Nature (London) 276: 466-471.

Brambilla, C., J. Abraham, E. Brambilla, K. Benirschke, and C. Bloor (1979).
      Comparative pathology of silicate pneumoconiosis.  Am. J. Pathol.  96:149-170.

Brauer,  M., Koutrakis, P.,  Spengler, J.D.  Personal exposures to acidic aerosols and gases.
      Environ. Sci. Tech.  1989; 23: 1408-12.

Braun-Fahrlander,  C.; Ackermann-Liebrich, U.; Schwartz, J.; Gnehm, H. P.;
      Rutishauser, M.; Wanner, H. U. (1992) Air pollution and respiratory symptoms
      in preschool children. Am. Rev. Respir. Dis.  145: 42-47.

Buechley, R.W.  (1975).  SO2 Levels, 1967-1972 and Perturbations  in Mortality.  Contract
      No.  ES-5-2101. Report available from National Institute of Environmental Health
      Sciences, Research Triangle Park, NC.

Burnett, R.T., Dales, R.; Krewski, D.; Vincent, R.; Dann, T.; Brook, J.F. (1995)
      Associations between ambient particulate sulfate and admissions to  Ontario hospitals
      for cardiac and respiratory diseases.  Am. J.  Epidemiol. 142: 15-22.

-------
                                         R-4

Burnett, R. T.; Dales, R. E.; Raizenne, M. E.; Krewski, D.; Summers, P. W.;
       Roberts, G. R.; Raad- Young, M.; Dann, T.; Brook, J.  (1994) Effects of low ambient
       levels of ozone and sulfates on the frequency of respiratory admissions to Ontario
       hospitals. Environ. Res. 65: 172-194.

Burton, R. M.; Suh, H. H.; Koutrakis,  P.  (1996) Spatial variation in paniculate
       concentrations within metropolitan Philadelphia. Environ.  Sci. Technol. 30: 400-407.

Butlin, R.N.; Coote, A.T.;  Devenish, M.;  Hughes, I.S.C.; Hutchens, C.M.; Irwin, J.G.;
       Lloyd, G.O.; Massey, S.W.; Webb, A.H.; Yates, T.J.S.  (1992a)  Preliminary results
       from the analysis of stone tables from the National Materials Exposure Programme
       (NMEP). Atmos. Environ. Part B 26: 189-198.

Butlin, R.N.; Coote, A.T.;  Devenish, M.;  Hughes, I.S.C.; Hutchens, C.M.; Irwin, J.G.;
       Lloyd, G.O.; Massey, S.W.; Webb, A.H.; Yates, T.J.S.  (1992b)  Preliminary results
       from the analysis of metal  samples from the National Materials Exposure Programme
       (NMEP). Atmos. Environ. Part B 26: 199-206.

Campbell, G.G.; Schurr, G.G.; Slawikowski, D.E.; Spence, J.W. (1974) Assessing air
       pollution damage to  coatings. J. Paint Technol. 46: 59-71.

Canada Environmental  Protection  Agency (CEPA)/FPAC Working Group on Air Quality
       Objectives and Guidelines (1995).  National ambient air quality objectives for
       paniculate matter less than 10 ptm (PM10).
Carey, W.F. (1959)  Atmospheric deposits in Britain: a study of dinginess.  Int. J. Air
       Pollut. 2: 1-26.

Chapman, R. S.; Calafiore, D. C.; Hasselblad, V. (1985) Prevalence of persistent cough and
       phlegm in young adults in relation to long-term ambient sulfur oxide exposure.
       Am. Rev. Respir. Dis.  132: 261-267.

Chappie, M.; Lave, L. (1982) The health effects of air pollution: a reanalysis. J. Urban
       Econ. 12: 346-376.

Chen, L.C.; Wu, C.Y.; Qu, Q.S.;Schel singer, R.B. (1995) Number concentration and mass
       concentration  as determinants of biological responses to inhaled particles. In. Phalen
       R.F.; Bates, D.V. eds Proceedings of the colloquium on paniculate air pollution and
       human mortality and morbidity, Part II; January 1994; Irvine , CA. Inhalation
       Toxicol. 7:577-588.

Chen, L. C.; Fine, J. M.; Qu, Q.-S.;  Amdur, M.  O.; Gordon,  T. (1992a) Effects of fine
       and ultrafine sulfuric acid aerosols in guinea pigs: alterations in alveolar  macrophage
       function and intracellular pH. Toxicol. Appl. Pharmacol.  113: 109-117.

-------
                                         R-5

Chen, L. C.; Miller, P. D.; Amdur, M. O.; Gordon, T. (1992b) Airway
       hyperresponsiveness in guinea pigs exposed to acid-coated ultrafine particles. J.
       Toxicol. Environ. Health 35: 165-174.

Chen, L.C.; Lam, H.F.; Kim, E.J; Guty, J.; Amdur, M.O. (1990) Pulmonary effects of
       ultrafine coal fly ash inhaled by guinea pigs. J. Toxicol. Envrion. Helath 29:169-184.

Chestnut, L.G.; Dennis, R.L.; Latimer, D. A. (1994) Economic Benefits of Improvements in
       Visibility:  Acid Rain Provisions of the 1990 Clean Air Act Amendments.
       Proceedings of Aerosols and Atmospheric Optics:  Radiative Balance and Visual Air
       Quality. Air & Waste Management Association International Specialty Conference,
       pp. 791-802.

Chestnut, L.G.; Rowe, R.D. (1990b) Preservation values for visibility in the national parks.
       Washington,  DC: U.S. Environmental Protection Agency.

Chow, J.C. (1995)  Measurement Methods to Determine Compliance with Ambient Air
       Quality Standards for Suspended Particles. J.  of the Air and Waste  Management
       Association. May 1995. 45:320-382.

Chow, J.C., J.G. Watson, L.W. Richards, D.L. Hasse, C. McDade, D.L.  Dietrich, D.
       Moon and C. Sloane (1991) The 1989-90 Phoenix PM10  study Volume II:  Source
       Apportionment Final Report.  DRI Document No.  8931.6F1, prepared for Arizona
       Department of Environmental Quality, Phoenix,  AZ, by the Desert Research Institute,
       Reno, NV, April  12,  1991.

Cifuentes, L.; Lave, L. B.  (1996)  Association of daily mortality  and air pollution in
       Philadelphia, 1983-1988. J. Air Waste Manage. Assoc.: in press.

Ciocco, A. and D.J. Thompson (1961). A follow-up of Donora  ten years after:
       methodology and findings. J. Pub.  Health 51:  155-164.

Clayton, C. A.; Perritt, R. L.; Peilizzari,  E. D.; Thomas, K. W.; Whitmore, R. W.;
       Wallace, L. A.; Ozkaynak, H.;  Spengler, J. D. (1993) Particle total exposure
       assessment methodology (PTEAM) study: distributions of aerosol and elemental
       concentrations in personal, indoor,  and outdoor air samples in a southern California
       community. J. Exposure Anal. Environ.  Epidemiol. 3: 227-250.

Cohen, S.; G.W. Evans; D. Stokols; D.S. Krantz (1986) Behavior, Health and
       Environmental Stress.  Plenum Press, New York, NY.

-------
                                         R-6

Coleridge, H.; Coleridge, J. (1986) Reflexes evoked from the tracheolbronchial tree and
       lungs.  In Handbook of physiology, Section 3, The respiratory system.  Vol. II.
       Cherniack, N.S.; Widdicombe, J.G. eds. Washington DC. American Physiological
       Society: 395-430.

Constantine, H.; Dautrebande, L.;  Kaltreider, N.; Lovejoy, F.W., Jr.; Morrow, P.; Perkins,
       P. (1959). Influence of carbachol and  of fine dust aerosols upon the breathing
       mechanics and the lung volumes of normal subjects and of patients with chronic
       respiratory disease before and after administering sympathomimetric aerosols.   Arch.
       Int. Pharmacodyn. 123:239-252.

Cooper, R.L.; Goldman, J.M.; Harbin, J.J., eds. (1991) Aging and environmental
       toxicology: biological and behavioral perspective. Baltimore, MD: John Hopkins
       University Press (series in environmental toxicology).

Costa, D.L.; Lehmann, J.R.;  Smith,  S.; Dreher, K.L. (1995)  Ampliation of particle
       toxicity to the lung by pre-existing inflammation. Am. J.  Respir. Crit. Care Med.
       151:  A265.

Costa, D. L.; Tepper, J. S.; Lehmann, J. R.; Winsett,  D. W.; Dreher,  K.;  Ghio. A. J.
       (1994a)  Surface complexed iron  (Fe+3) on particles: its role in the induction of lung
       inflammation and hyperreactivity. Presented at: Colloquium on paniculate air
       pollution and human mortality and morbidity: program and abstracts; January;  Irvine,
       CA.  Irvine, CA: University of California Irvine, Air Pollution Health Effects
       Laboratory; p. S3.4; report no. 94-02.

Costa, D. L.; Lehmann, J.  R.; Frazier, L. T.; Doerfler, D.;  Ghio, A. (1994b) Pulmonary
       hypertension: a possible risk factor in particulate toxicity. Am. Rev.  Respir. Dis.  149
       (4, pt. 2):  A840.

Council on Environmental Quality  (1978) Visibility Protection for Class I  Areas, the
       Technical Basis. Washington, DC.

Cowling, J.E.;  Roberts, M.E. (1954) Paints, varnishes, enamels, and lacquers. In:
       Deterioration of materials:  causes and preventive techniques.  New York, NY:
       Reinhold Publishing Corp.; pp. 596-645.

Crapo, J.; Miller, F.J.; Mossman, B.; Pryor, W.A.;  Kiley, J.P. (1992) Relationship
       between acute inflammatory responses  to air pollutants and chronic lung disease.
       Am.  Rev.  Respir. Dis. 145:1506-1512.

Creighton, P J.; Lioy, P. J.;  Haynie, F. H.; Lemmons, T J.;  Miller, J. L.; Gerhart, J.
       (1990) Soiling by atmospheric aerosols in an urban industrial area.   J. Air Waste
       Manage. Assoc. 40:  1285-1289.

-------
                                         R-7

Gulp, D.J.; Latchney, L.R.; Frampton, M.W.; Jahnke, M.R.; Morrow, P.E.; Utell, M.J.
       (1995).  Composition of human airway mucins and effects after inhalation of acid
       aerosols. Am. J. Physiol.: Sept. 1995,  Vol. 269 / No. 3 / Part I, pp. L358-L370.

Cupitt, L.T.; Glen, W.G.; Lewtas, J. (1994) Exposure and risk from ambient particle-bound
       pollution in an airshed dominated by residential wood combustion and mobile sources.
       In: Symposium of risk assessment of urban air: emissions, exposure, risk
       identification, and risk quantitation; May-June 1992; Stockholm, Sweden. Environ.
       Health Perspect.  102 (suppl. 4): 75-84.

Dahl, A.R.; Snipes, M.B.;  Muggenburg, B.A.;  Young, T.C. (1983) Deposition of sulfuric
       acid mists  in the respiratory tract of beagle dogs.  J. Toxicol. Environ. Health
       11:141-149.

Dassen, W.;  Brunekreef, B.; Hoek,  G.; Hofschreuder, P.; Staatsen, B.; De Groot, H.;
       Schouten,  E.; Biersteker, K.  (1986) Decline in children's pulmonary function during
       an air pollution episode. J. Air Pollut. Control Assoc. 36: 1223-1227.

Delfino, R.J.; Becklake, M.R.; Hanley, J.A. (1993). Reliability  of Hospital Data for
       Population-based Studies of Air  Pollution.  Arch,  of Environ. Health 48:(No. 3): 140-
       145.

Dockery, D.  W.;  Cunningham, J.; Damokosh, A. I.; Neas,  L. M.; Spengler, J. D.;
       Koutrakis, P.; Ware, J. H.; Raizenne,  M.; Speizer, F. E.  (1996) Health effects of
       acid aerosols on North American children: respiratory symptoms. Environ. Health
       Perspect: in press.

Dockery, D.W.; Schwartz,  J.; Pope, C.A., III.  (1995).  Comment from original
       Investigators, in: Particulate Air Pollution and Daily Mortality:  Replication and
       Validation of Selected Studies.  Prepared by:  Health Effects Institute, 141 Portland
       Street, Suite 7300, Cambridge, MA 02139.  August 1995.

Dockery, D.  W.;  Pope, C.  A., III. (1994) Acute respiratory effects of paniculate air
       pollution.  Annu. Rev. Public Health 15:  107-132.

Dockery, D.  W.;  Pope, C.  A., III. (1994) Air pollution and mortality: the authors reply
       [letter].  N. Engl. J.  Med. 330: 1238.

Dockery, D.  W.;  Pope, C.  A., Ill; Xu,  X.; Spengler, J.  D.; Ware, J.  H.; Fay, M. E.;
       Ferris, B.  G., Jr.; Speizer, F. E. (1993) An association between air pollution and
       mortality in six U.S. cities. N. Engl. J. Med. 329: 1753-1759.

Dockery, D.  W.;  Schwartz, J. (1992a) The authors'  response to  Waller and Swan.
       Am. J. Epidemiol. 135: 23-25.

-------
                                         R-8

Dockery, D.W.; Schwartz, J.; Spengler, J.D.  (1992b) Air pollution and daily mortality:
       associations with participates and acid aerosols. Environ. Res. 59:  362-373.

Dockery, D. W.; Speizer, F. E.; Stram, D. O.; Ware, J. H.; Spengler, J. D.; Ferris, B.  G.,
       Jr. (1989) Effects of inhalable particles on respiratory health of children. Am. Rev.
       Respir. Dis. 139: 587-594.

Dockery, D. W.; Ware, J. H.; Ferris, B. G.,  Jr.; Speizer,  F. E.; Cook, N.  R.;
       Herman,  S. M. (1982) Change in pulmonary function in children associated with air
       pollution  episodes. J. Air Pollut. Control Assoc. 32:  937-942.

Dosman, J.A., D.J. Cotton, B1L. Graham, K.Y.R.  Li, F. Froh and G.D.  Barnett ( 1980).
       Chronic bronchitis and decreased forced expiratory flow rates in lifetime nonsmoking
       grain workers.  Am. Rev. Resp. Dis. 121:11-16.

Dreher, K.; Jaskot, R.; Koduvanti, J.; Lehmann, J.; Winsett, D.; Costa, D. (1995)  Role of
       soluble metals in acute pulmonary toxicity of residual oil fly particles. Am. J. Respir.
       Crit. CareMed. 151:A265.

DRI (1995) PM10 and PM2.5 Variations in Time and Space.  DRI Document No. 4204. IF.
       prepared  by Desert Research Institute,  Reno, NV.  October 24, 1995.

Dusseldorf, A.;  Kruize, H.; Brunekreef, B.; Hofschreuder,  P.; de Meer, G.;
       van Oudvorst, A. B. (1994) Associations of PM10 and airborne iron  with respiratory
       health of adults living near a steel factory. Am. J. Respir.  Crit. Care Med. 152:
       1932-1939.

Dye, J.A.; Richards,  J.R.; Dreher, K.L.  (1995)  Injury of rat trachea! epithelial  cultures  by
       exposure to ozone and/or residual oil fly ash.  Am.  J. Respir.  Crit. Care  Med.
       151:A265.

Edney, E.G.; cheek, S.F.; Corse, E.W.; Spence, J.W.; Haynie, F.H. (1989) Atmospheric
       weathering caused by dry deposition of acidic  species.  J. Environ. Sci. Health Part A
       24: 439-457.

El-Fawal, H.A.N.; Schlesinger, R.B.  (1994)  Nonspecific airway hyperresponsiveness
       induced by inhalation exposure to sulfuric acid aerosol: an in vitro assessment.
       Toxicol.  Appl. Pharmacol.  125:70-76.

Eldred, R.A. and T.A. Cahill (1994). Trends in elemental concentrations of fine particles at
       remote sites in the United States of America.  Atmos. Environ. 28:1009-1019.

-------
                                         R-9

Euler, G.L.; Abbey, D.E.; Hodgkin, J.E.; Magie, A.R. (1988)  Chronic obstructive
       pulmonary disease symptom effects of long-term cumulative exposure to ambient
       levels of total oxidants and nitrogen dioxide in California Seventh-Day Adventist
       residents. Arch, of Environ. Hlth 43:279-285.

Fairley, D. (1990) The relationship of daily mortality to suspended particulates in Santa
       Clara county, 1980-86.  Environ. Health Perspect.  89:  159-168.

Ferin, J.;  Oberdorster, G.; Penney, D.P.; Soderholm, S.C.; Gelein, R.; Piper, H.C.  (1990)
       Increased pulmonary toxicity of ultrafine particles? I. Particle clearance, translocation,
       morphology. J. Aerosol. Sci. 21:  381-384.

Ferris, E.G., Jr.; Ware, J.H.;  Spengler,  J.D.; Dockery, D.W.; Speizer, F.E. (1986)  The
       Harvard six-cities study.  In: Lee, S.D.; Schneider, T.; Grant,  L.D.; Verkerk, P.J.,
       eds. Aerosols: research, risk assessment and control strategies:  proceedings of the
       second  U.S.-Dutch international symposium; May 1985; Williamsburg, VA. Chelsea,
       MI: Lewis Publishers, Inc. pp. 721-730.

Fine, J. M.; Gordon, T.; Thompson, J. E.; Sheppard, D. (1987b) The role of titratable
       acidity  in acid aerosol-induced bronchoconstriction. Am. Rev. Respir. Dis. 135:
       826-830.

Firket, M. (1931) Sur les causes  des accidents survenus dans  la vallee de la Meuse, lors des
       brouillards de decembre 1930  [The causes of accidents which occurred in the Meuse
       Valley  during the fogs of  December  1930]. Bull. Acad. R. Med. Belg.
       ll[ser.  5]: 683-741.

Fishman,  A.P. (1976)  Chronic cor-pulmonale. Am. Rev. Resp. Dis.  114:775-794.

Fitz-Simons, T.; Mintz, D.; Wayland, M. (1996)  Proposed methodology for predicting
       PM25 from PM10 values to assess  the impact of alternative forms and levels of the  PM
       NAAQS.  Document transmitted to members of the Clean Air Scientific Advisory
       Committee on June 26,  1996.

Fochtman, E.G.; Langer, G. (1957) Automobile paint damaged by airborne iron particles. J.
       Air Pollut. Control Assoc. 6: 243-247.

Frampton, M.  W.; Morrow, P. E.; Cox, C.; Levy, P. C.; Condemi, J.  J.;  Speers, D.;
       Gibb, F. R.; Utell, M.  J.  (1995) Sulfuric acid aerosol  followed by ozone exposure in
       healthy and asthmatic subjects.  Environ. Res. 69:  1-14.

Frampton, M.W.; Voter, K.Z.; Morrow,  P.E.; Roberts, N.J., Jr.;  Gulp, D.J.; Cox, C.;
       Utell, M.J. (1992)  Sulfuric acid aerosol exposure in humans assessed by
       bronchoalveolar lavage. Am.  Rev. Respir. Dis. 146:626-632.

-------
                                         R-10

Frankenthal, R.P.; Lobnig, R.; Siconolfi, D.J.; Sinclair, J. D. (1993) Role of particle
       contamination in the corrosion of electronic materials and devices. J. Vac. Sci.
       Technol. All:  2274-2279.

Freeman, A.M., III. (1979)  The benefits of environmental improvement: theory and
       practice.  Baltimore, MD: The Johns Hopkins University Press.

Fujimaki, H.; Katayama, N.; Wakamori, K. (1992) Enhanced histamine release from lung
       mast cells of guinea pigs exposed to sulfuric acid aerosols. Environ. Res. 58:
       117-123.

Friedlander, S.K. (1982).  Letter from Sheldon K. Friedlander, Chair, Clean Air Science
       Advisory Committee (CAS AC) to Administrator Anne M. Gorsuch.   CAS AC  Review
       and  Closure of the OAQPS Staff Paper for Paniculate Matter.  January 29, 1982.

Gardner, D.E.; Miller, F.J.; Illing, J.W.; Kirtz, J.M. (1977). Increased Infectivity with
       Exposure to Ozone and Sulfuric Acid. Tox. Lett.  1:59-64.

Gearhart, J. M.; Schlesinger, R. B.  (1986) Sulfuric acid-induced airway
       hyperresponsiveness.  Fundam. Appl. Toxicol. 7: 681-689.

Gearhart, J. M.; Schlesinger, R. B.  (1988) Response of the tracheobronchial mucociliary
       clearance system to repeated  irritant exposure: effect of sulfuric acid mist on function
       and  structure. Exp. Lung Res. 14: 587-605.

Gergen, P.J.; Weiss, K.B. (1992) The increasing problem of asthma  in the United States.
       Am. Rev. Respir. Disease. 146:823-824.

Ghio, A.J.; Hatch,  G.E.  (1993) Lavage phospholipid concentration after silica installation in
       the  rat is associated with complexed [fe3+] on the dust surface.  Am. J. Respir.  Cell
       Mol. Biol. 8:403-407.

Ghio, A.J.; Kennedy, T.P.; Whorton, A.R.; Crambliss, A.L.; Hatch, G.E.; Hoidal,  J.R.
       (1992)  Role of surface complexed iron for oxidant generation  and lung inflammation
       induced by  silicates.  Am. J.  Physiol. 263:L511-L518.

Ghio, A.J.; Jaskot,  R.H., Hatch, G.E. (1994) Lung  injury after silica instillation is
       associated with an accumulation of iron in rats. Am. J. Physiol. 267: L686-L692.

Gilbert, C.C.S. (1985) Household adjustment and the measurement of benefits from
       environmental quality improvements [Ph.D. dissertation].  Chapel Hill, NC:
       University of North Carolina at Chapel Hill.

-------
                                         R-ll

Gillespe, J. R.  (1980).  Review of the cardiovascular and pulmonary function studies on
       beagles exposed for 68 months to auto exhaust and other air pollutants.  IN: Long-
       term Effects of Air Pollutants in Canine Species.  (J.F. Stara, D.L. Dungworth, J.C.
       Orthoefer, and W.S. Tyler eds.).  EPA Report #6007-80-014, pp. 115-153.

Glasser, M., and L. Greenburg (1971).  Air pollution and mortality and weather, New York
       City, 1960-64. Arch.  Environ. Health 22:334-343.

Glezen, W. P. (1989) Antecedents  of chronic  and recurrent lung disease: childhood
       respiratory trouble. Am. Rev. Respir.  Dis. 140:  873-874.

Godfrey, S. (1993)  Airway inflammation, bronchial reactivity and asthma.  Agents Actions
       Suppl. 40: 109-143.

Godleski, J.J.; Hatch, V.; Hauser.  R.; Christiani, D.; Gasula, G.; Sioutas, C. Ultrafine
       particles in lung macrophages of healthy people.  Am. J. Respir. Crit. Care Med.
       1995;  151: A264.

Gold, D. R.;  Tager, I. B.;  Weiss,  S. T.; Tosteson, T. D.; Speizer, F. E.  (1989) Acute
       lower respiratory illness in childhood as a predictor of lung function and chronic
       respiratory symptoms.  Am.  Rev.  Respir. Dis.  140: 877-884.

Goodwin, I.E.; Sage,  W.; Tilly, G.P. (1969)  Study of erosion by solid particles.
       Proceedings of the Institution  of Mechanical Engineers  184(15): 279-292.

Gordian, M. E.; Ozkaynak, H.; Xue, J.; Morris, S. S.; Spengler,  J. D. (1996) Paniculate
       air pollution  and respiratory disease in  Anchorage, Alaska. Environ. Health Perspect.
       104: 209-297.

Hancock, R.P.; Esmen,  N.A.; Furber, C.P. (1976)  Visual response to dustiness. J. Air
       Pollut. Control Assoc.  26: 54-57.

Hanley, Q.S.; Koenig, J.Q.; Larson, T.V.; Anderson, T.L.; Van Belle, G; Rebolledo. V.;
       Covert, D.S.; Pierson,  W.E. (1992) Response of young asthmatic patients to inhaled
       sulfuric acid.  Am. Rev. Respir. Dis.  145: 326-331.

Hausman, J. A.; Ostro, B.  D.; Wise, D.  A. (1984) Air pollution and lost work. Cambridge,
       MA: National Bureau of Economic Research; NBER working paper no.  1263.

Haynie F. H.; Lemmons, T. J. (1990) Evaluation of an atmospheric corrosion rate monitor
       as a time-of-wetness meter.  Mater. Perform. 31: 48-52.

-------
                                         R-12

Haynie, F. H. (1989) Environmental factors affecting the corrosion of galvanized steel.  In:
       Dean,  S. W.; Lee, T.S., eds. Degradation of metals in the atmosphere:  a
       symposium sponsored by ASTM Committee G-l on corrosion of Metals;  May  1986;
       Philadelphia, PA. Philadelphia, PA:  American Society for Testing and Materials;
       pp. 282-289.  (ASTM Spec.  Tech. Publ. 965).

Haynie, F.H.; Spence, J.W. (1984)  Air pollution damage to exterior household  paints.  J.
       Air Pollut. Control Assoc. 34: 941-944.  Citations: 0082,  12; 0671, 09; 0746,  05.

Haynie, F. H.; Upham, J.B  (1974)  Correlation between corrosion behavior of steel and
       atmospheric pollution data.  In:  Coburn, S. K., ed. Corrosion in natural
       environments:  presented at the 76th annual meeting American Society for Testing and
       Materials;  June 1973; Philadelphia, PA; pp. 33-51.  (ASTM special technical
       publication 558).

Hefflin, B. J.; Jalaludin, B.; McClure, E.; Cobb, N.; Johnson, C. A.; Jecha, L.; Etzel, R.
       A. (1994)  Surveillance for dust storms and respiratory diseases in Washington State,
       1991. Arch. Environ. Health 49:  170-174.

Helin,  H. (1986)  Macrophage procoagulant factors - mediators of inflammatory  and
       neoplastic  tissue  lesions.  Med. Biol. 1986; 64:167-176.

Hennekens, C.H.; Buring, J.E.  (1987)  Epidemiology in Medicine.  Little, Brown and
       Company. Boston/Toronto.

Hermance, H.W.  (1966) Combatting the effects of smog on  wire-spring relays.   Bell Lab.
       Rec. (February): 48-52.

Hill, A. B. (1965) The environment and disease: association or causation? Proc.  R. Soc.
       Med. 58: 295-300.

Hoek,  G. (1992) Acute  effects of ambient air pollution episodes on  respiratory health of
       children [thesis]. Wageningen, The Netherlands: Agricultural University of
       Wageningen.

Hoek,  G.; Brunekreef, B.  (1995)  Effect of photochemical air pollution on acute respiratory
       symptoms in children.  Am.  J. Respir. Crit. Care Med. 151: 27-32.

Hoek,  G.; Brunekreef, B.  (1994)  Effects of low-level winter air pollution concentrations on
       respiratory health of Dutch children. Environ. Res. 64: 136-150.

-------
                                         R-13

Hoek, G.; Brunekreef, B.  (1993) Acute effects of a winter air pollution episode on
      pulmonary function and respiratory symptoms of children. Arch.  Environ.  Health 48:
      328-335.

Hogg, JC.; Macklem, PT; Thurlbeck, WM. (1968) Site and nature of airway obstruction in
      chronic obstructive lung disease in New Eng. J. Med 278:1355.

Holland, W.A.; Bennett, A.E.; Cameron, I.R.; du V. Florey, C.; Leeder, S.R.; Schilling,
      R.S.; Swan, A.V.; Wallter, R.E. (1979. Health effects of paniculate pollution:
      reappraising the evidence. Am. J.  Epidemiol. Ill: 525-659.

Hyde, D.; Orthoefer,  J.; Dungworth, D.; Tyler, W.;  Carter, R.; Lum, H. (1978)
      Morphometric  and morphologic evaluation of pulmonary lesions in beagle dogs
      chronically exposed to high  ambient levels of air pollutants. Lab. Invest. 38: 455-469.

Ishikawa, S.; Bowden, D.H.; Fisher, V.; Wyatt, J.P. (1969) The "emphysema profile" in
      two mid-western cities in North America. Arch. Environ. Health 18:  660-666.

Ito, K.;  Kinney, P.; Thurston, G. D. (1995) Variations in PM-10 concentrations within two
      metropolitan areas and their implications for health effects analyses. In: Phalen,  R.F.;
      Bates, D.V.; eds. Proceedings of the colloquium on paniculate air pollution and
      human mortality and morbidity part II; January 1994; Irvine,  CA.  Inhalation
      Toxicol. 7: 735-745..

Ito, K.;  Thurston,  G.  D.; Hayes, C.; Lippmann, M. (1993) Associations of  London,
      England, daily mortality with paniculate matter, sulfur dioxide, and acidic aerosol
      pollution. Arch.  Environ.  Health 48: 213-220.

Ito, K.;  Thurston,  G.  D. (1996) Daily PMlO/mortality associations: an investigation of at-
      risk sub-populations. J. Exposure Anal. Environ.  Epidemiol.: 6:79-95.

Jakab, G. J.; Hemenway, D.  R. (1993) Inhalation coexposure to carbon black and acrolein
      suppresses alveolar macrophage phagocytosis and  TNF-a release and modulates
      peritoneal macrophage phagocytosis. Inhalation Toxicol. 5: 275-289.

Jaskot, R.H.; Costa, D.L.; Kodavanti, P.; Lehmann,  J.R.; Winsett, D.; Dreher, L.
      Comparison of lung inflammation and airway reactivity in three strains of rats
      exposed to residual oil fly ash particles.  Am. J.  Respir.  Crit. Care Med. 1995;
      151:A264.

Jedrychowski, W.; Becher, H.; Wahrendorf, J.; Basa-Cierpialek, Z. (1990)  A case-control
      study of lung cancer with special  reference to the effect of air pollution in  Poland.
      J. Epidemiol. Commun. Health 44: 114-120.

-------
                                         R-14

John, W.; Wall, S.M.; Ondo, J.L.; Winklmayr, W. (1990) Modes in the size distributions of
       atmospheric inorganic aerosol.  Atmos. Environ. Part A 24: 2349-2359.

Johnson, J.G.; Gideon, R.A.; Loftsgararden, D.O. (1990)  Montana air pollution study:
       children's health effects. J. Off. Stat. 5:391-407.

Johnson, J.B.; Elliot, P.; Winterbottom, M.A.; Wood, G.C. (1977) Short-term atmospheric
       corrosion of mild steel at two weather and pollution monitored sites.  Corros. Sci.  17:
       691-700.

Johnson, K. G.; Loftsgaarden, D. O.;  Gideon, R. A.  (1982) The effects of Mount St. Helens
       volcanic ash on  the pulmonary function of 120 elementary school children.  Am. Rev.
       Respir. Dis. 126:  1066-1069.

Kalkstein, L.S.; Barthel, C.D.; Ye, H.; Smoger, K.;  Greene, J.S.; Nichols,  M.C. (1994)
       The differential  empacts of weather and pollution on human mortality. Newark. DE..
       University of Delaware Dept. of Geography, Center for Climatic Research:
       November.

Kalkstein, L.S. (1991)  A new approach to evaluate the impact of climate on human
       mortality. Environ.  Health Perspect. 96: 145-150.

Katsouyanni, K.; Hatzakis,  A.; Kalandidi,  A.; Trichopoulos, D. (1990a) Short-term effects
       of atmospheric pollution on mortality in  Athens. Arch.  Hellen. Med.  7: 126-132.

Katsouyanni, K.; Karakatsani,  A.; Messari, I.; Touloumi, G.;  Hatzakis,  A.;  Kalandidi, A.;
       Trichopoulos, D. (1990b) Air pollution and cause specific mortality in Athens.
       J. Epidemiol. Commun. Health 44:  321-324.

Katsouyanni, K.; Pantazopoulou, A.; Touloumi, G.; Tselepidaki, I.; Moustris, K.;
       Asimakopoulos, D.; Poulopoulou, G.; Trichopoulos, D. (1993) Evidence for
       interaction between air pollution and high temperature in the causation of excess
       mortality. Arch. Environ. Health 48: 235-242.

Kenny, R.A., ed. (1989) Physiology of aging.   St. Louis,  MO: Mosby-Year Book, Inc.

Kim, C.S.; Lewars, G.A.;  Sackner, M.A.  (1988). Measurement of total lung aerosol
       deposition as an index of lung abnormality. J. Appl. Physiol. 64:1527-1536.

Kinney, P. L.; Ito, K.; Thurston, G. D. (1995) A sensitivity analysis of mortality/PM10
       associations in Los Angeles. Inhalation Toxicol. 7: 59-69.

-------
                                         R-15

Kinney, P. L.; Ozkaynak, H. (1991) Associations of daily mortality and air pollution in Los
       Angeles County. Environ. Res. 54: 99-120.

Kleinman, M.T.; Bhalla,  O.K.; Mautz, W.J.; Phalen, R.F. (1995) Cellular and
       Immunologic Injury with PM-10 Inhalation.  Inhalation Toxicol. 7: 589-602.

Kodavanti, U.P.; Jaskot,  R.; Costa, D.; Dreher, K.L. (1995)  Fibronectin and collagen:
       biomarkers of chronic lung injury inducted b residual fly ash.  Am J Respir Crit Care
       Med 1995; 151:A265.

Koenig, J. Q.; Dumler, K.; Rebolledo, V.; Williams, P. V.; Pierson, W. E. (1993)
       Respiratory effects of inhaled sulfuric acid on senior asthmatics and nonasthmatics.
       Arch. Environ. Health 48:  171-175.

Koenig, J.Q.; Covert, D.S.;  Pierson,  W.E. (1989)  Effects of inhalation of acidic compounds
       on pulmonary function in allergic adolescent subjects.   In:  Symposium on the health
       effects of acid aerosols; October 1987; Research  Triangle Park,  NC.  Environ.  Health
       Perspect.  79:173-178.

Koenig, J.Q.; Pierson, W.E.; Horike,M.  (1983) The effects of inhaled sulfuric acid on
       pulmonary function in adolescent asthmatics.  Am. Rev. Respir. Dis. 128: 221-225.

Koenig, J.Q.; Pierson, W.E.; Horike, M.; Frank, R.  (1981).  Effects of SO,  plus NaCl
       aerosol combined  with moderate exercise on pulmonary function in asthmatic
       adolescents.  Environ. Res. 25: 340-348.

Koutrakis, P.; Briggs, S.L.K.; Leaderer,  B.P. (1992) Source  apportionment of indoor
       aerosols in Suffolk and Onondaga Counties, NY.  Environ. Sci. Technol. 26:521-527.

Lanting, R.W. (1986) Black smoke and soiling.  In: Lee, S.D.; Schneider, T.;  Grant, L.D.;
       Verkerk, P.J., eds. Aerosols: research, risk assessment and control strategies:
       proceedings of the second U.S.-Dutch international symposium;  May 1985;
       Williamsburg, VA. Chelsea, MI: Lewis Publishers, Inc.; pp. 923-932.

Lawther, P.J. (1973)  Compliance with the Clean Air Act: Medical Aspects.  J. of the Instit.
       of Fuel: Volume XXXVI, No.  271:341-344.

Lawther, P. J.; Waller, R. E.; Henderson, M. (1970) Air pollution and exacerbations  of
       bronchitis. Thorax 25: 525-539.

Leaderer, B.P.; Tanner, R.L.; Holford, T.R. (1982)  Diurnal  variations,  chemical
       composition and relation to meteorological variables of summer aerosol in  the New
       York subregion. Atmos. Environ.  16:2075-2087.

-------
                                         R-16

Leikauf, G.D.; Spektor, D.M.; Albert, R.E.; Lippmann, M. (1984) Dose-dependent effects
       of submicrometer sulfuric acid aerosol on particle clearance from ciliated human lung
       airways. Am. Ind. Hyg. Assoc.  J. 45: 285-292.

Li, Y.; Roth, H.D. (1995).  Daily Mortality analysis by using different regression  models in
       Philadelphia County, 1973-1990. Inhalation Toxicol 7:45-58.

Ligocki, M. P.; Salmon. L. G.; Fall, T.; Jones, M. C.  ; Nazaroff, W. W.; Cass G. R.
       (1993) Characteristics of airborne particles  inside southern California museums.
       Atmos. Environ. Part A 27: 697-711.

Linn, W.S.; Shamoo, D.A.; Anderson, K.R.; Peng, R.C.; Avol, E.L.; Hackney, J.D.
       (1994)  Effects of prolonged, repeated exposure  to ozone, sulfuric acid,  and their
       combination on healthy and asthmatic volunteers. Am. J.  Respir.  Crit Care Med.
       150:431-440.

Lioy, PJ. (1990)  Assessing total human exposure to contaminants.  Environ. Sci.  Technol.
       24: 938-945.

Lioy, P.J.; Waldman, J.M.; Buckley, T.; Butler, J.; Pietarinen, C. (1990)  The personal,
       indoor and outdoor concentrations of PM-10 measured in an industrial community
       during the winter.  Atmos. Environ. Part B 24: 57-66.

Lipfert, F.W.; Wyzga, R.E (1995b) Air pollution  and mortality: issues and uncertainties. J.
       Air Waste Mangage. Assoc. 45:949-966.

Lipfert, F. W. (1994a) Air Pollution and Community Health - a Critical Review and Data
       Source book. Chapter 5 -  The Air  Pollution Disasters. New York, NY:  Van
       Nostrand Reinhold.  Pgs. 111-142.

Lipfert, F. W. (1994b) Application of spatial filtering techniques  to cross-sectional  analysis
       of air pollution-mortality relationships. Am. J. Respir. Crit. Care Med.  149: A661.

Lipfert, F. W. (1994c) A simulation study of the effect  of measurement error on the
       determination of empirical  dose-response functions [draft]. Palo Alto, CA: Electric
       Power Research Institute

Lipfert, F. W. (1994d) Filter artifacts associated with particulate  measurements: recent
       evidence and effects on statistical relationships. Atmos. Environ. 28: 3233-3249.

Lipfert, F. W. (1993) Community air pollution and mortality: analysis of 1980  data from
       US metropolitan areas.  I. Particulate air pollution.  Upton, NY:  U.S. Department of
       Energy, Brookhaven National  Laboratory; report no. BNL 48446-R.

-------
                                         R-17

Lipfert, F. W. (1984) Air pollution and mortality: specification searches using SMSA-based
       data. J. Environ. Econ. Manage.  11: 208-243.

Lipfert, F. W.; Malone, R. G.; Daum, M. L.; Mendell, N. R.; Yang, C.-C. (1988)
       A statistical study of the macroepidemiology of air pollution and total mortality.
       Upton, NY: Brookhaven National Laboratory; report no. BNL-52122.

Lippmann, M.; Shy, C.;  Stolwijk, J.;  Speizer, F. (1996) Letter to Administrator Carol M.
       Browner.  Supplement to the Closure Letter from the Clean Air Scientific Advisory
       Committee.  March 20, 1996.

Lippmann, M. (1989a) Progress, prospects, and research needs on the health effects of acid
       aerosols.  In: Symposium on the health effects of acid aerosols; October 1987;
       Research Triangle Park, NC. Environ. Health Perspect. 79: 203-205.

Lippmann, M. (1989b)  Background on health effects of acid aerosols.  Environ. Health
       Perspect. 79:3-6.

Lippman,  M.; Schlesinger, R.B.; Leikanf,  G.; Spektor, D.; Albert, R.E. (1981).  Effects of
       sulfuric acid aerosols on respiratory tract airways. Inhaled Particles V.

Lippmann, M.; Thurston,  G.D. (1996).  Sulfate concentrations as an indicator of ambient
       paniculate matter air pollution for health risk calculations. J. Exposure Anal.
       Environ.  Epidemiol.: accepted.

Lowe, D.O. (1993)  Laboratory investigation  of pre-thrombotic states.  In: Potter, L.;
       Thompson, J.M.; eds.  Thrombosis and its management. Edinburgh: Churchill
       Livingstone, 31-46.

Lundgren, D.A.; Rich, T.A.; Hlaing,  D.N. (1996) PM,, PM25, and PM10 Aerosol
       Chemistry vs. Meteorology for Phoenix, Arizona.  Paper presented at Utah
       Conference.

Lunn, J. E.; Knowelden, J.; Handyside, A. J. (1967) Patterns of respiratory illness in
       Sheffield infant schoolchildren. Br. J. Prev. Soc. Med. 21: 7-16.

MacNee, W.; Selby, C. (1993) Neutrophil traffic in the lungs: role of haemodynamics, cell
       adhesion, and deformability. Thorax; 48:79-88.

Malm, W.C.; Gebhart, K.A.; Molenar, J.; Cahill, T.A.; Eldred, R.,; Huffman, D.  (1994a)
       Examining the relationship between atmospheric aerosols and light extinction at
       Mount Rainier and North Cascades  National Parks.  Atmos.  Environm. 28:347-360.

Malm, W.C; Sisler, J.F.;  Huffman,  D.; Eldred, R.,; Cahill, T.A. (1994b) Spatial and
       seasonal trends in particle concentration and optical extinction in the United States.  J.
       Geophys. Res. 29:1347-1370.

-------
                                         R-18

Martin, A. E. (1964) Mortality and morbidity statistics and air pollution. Proc. R. Soc.  Med.
       57: 969-975.

Martin, A. E.; Bradley, W. H. (1960) Mortality, fog and atmospheric pollution:
       an investigation during  the winter of 1958-59. Mon. Bull.  Minist. Health Public
       Health Lab. Serv. (GB) 19: 56-73.

MathTech, Inc.(1983)  Benefit and net benefit analysis of alternative national ambient air
       quality standards for paniculate matter (Volume III). Prepared for the U.S. EPA,
       Office of Air Quality Planning and Standards, Research Triangle Park, NC. January
       1983.

Mauderly, J. L.; Snipes, M. B.;  Barr, E. B.; Belinsky, S. A.; Bond, J. A.; Brooks,  A.  L.;
       Chang, I.-Y.; Cheng, Y. S.; Gillett, N. A.; Griffith, W.  C.; Henderson, R. F.;
       Mitchell,  C. E.;  Nikula, K. J.; Thomassen,  D. G. (1994)  Pulmonary toxicity  of
       inhaled diesel exhaust and carbon black in chronically exposed rats. Part I: neoplastic
       and nonneoplastic lung  lesions. Cambridge,  MA:  Health Effects Institute; research
       report no. 68.

Mauderly, J.L.;  Bice, D.E.; Cheng, Y.S.; Gillet, N.A.;  Griffith, W.C.;  Henderson,  R.F.;
       Pickerell, J.A.; Wolff,  R.K. (1990)  Influence of pre-existing pulmonary emphysema
       on susceptibilty of rats to diesel exhaust. Am. Rev. Respir. Dis.  141:1333-1341.

Mazumdar,  S.; Schimmel, H.; Higgins, I.T.T. (1982) Relation of daily mortality to air
       pollution: an analysis of 14 London winters,  1958/59 -  1971/72. Arch. Environ.
       Health 37: 213-220.

Mazumdar,  S.; Schimmel, H.; Higgins, I.  (1981) Daily mortality, smoke and SO2 in
       London, England 1959  to 1972.  In: Frederick, E. R., ed.  A  specialty conference on:
       the proposed SOX and paniculate standard; September 1980;  Atlanta, GA. Pittsburgh,
       PA:  Air Pollution Control Association; pp. 219-239.

Mazumdar,  S. and N. Sussman (1981). Relationships of air pollution to health:  results  from
       the Pittsburgh study.  Proceedings of the 74th Annual Meeting,  Air Pollution  Control
       Association, Philadelphia, PA.  June 21-26, 1981.

McClelland, G.; Schulze, W.;  Waldman,  D; Irwin, J.; Schenk, D.; Stewart,  T.; Deck,  L.:
       Thayer, M. (1991) Valuing eastern visibility: a field test of the contingent valuat;on
       method. Washington, DC: draft report to the U.S. Environmental Protection Age  cy;
       cooperative agreement no. CR-815183-01-3.

McFarland, A.R.; Ortiz, C.A.; Rodes,  C.E. (1981).  Wind tunnel evaluation of British
       smoke shade sampler.   Atmos.  Environ.

-------
                                         R-19

McFarland, A.R. (1979).  Wind tunnel evaluation of British smoke shade sampler.  EPA
       Contract No. 68-02-2720.  Air Quality Laboratory Report 3565/05/79 ARM.  Civil
       Engineering Department, Texas A&M University.

McKinney, N.; Hermance, H. W. (1969) Stress corrosion cracking rates of a nickel-brass
       alloy under applied potential.  In:  Stress corrosion testing:  a symposium presented at
       the sixty-ninth annual meting of the American Society for Testing and Materials;
       June-July 1966; Atlantic City, NJ.  Philadelphia, PA: American Society for Testing
       and Materials; pp. 274-291.  (ASTM special technical publication no. 425).

McJilton, C.E.; Frank, R.; Charlson,  R.J. (1976) Influence of relative humidity on
       functional effects of an inhaled SO2-aerosol mixture.  Am. Rev. Respir. Dis. 113:
       163-169.

Middleton, P.  (1993) Brown Cloud II:  The Denver Air Quality Modeling Study,  Final
       Summary Report.   Metro Denver Brown Cloud Study, Inc.  Denver,  CO.

Miller, F.J.; Anjiluel, S.;  Menache, M.G.; Asgharin, B.; Gerrity,  T.R. (1995)  Dosimetric
       considerations relating to paniculate toxicity. Inhalation Toxicology 7:615-632.

Miller, F.J.; Gardner, D.E.; Graham, J.A.; Lee, R.E.; Wilson, W.E.; Bachmann, J.D.
       (1979)  Size considerations for establishing a standard for inhalable particles.  J. of the
       Air Pollution Control Association, June  1979; Vol. 29, 6:610-615.

Ministry of Health - London - Her Majesty's Stationary Office (1954).  Reports on Public
       Health  and Medical Subjects No. 95 Mortality and Morbidity During  the London Fog.
       December 1992. Pages 1-60.

Moolgavkar, S. H.; Luebeck, E. G. (1996) Paniculate Air Pollution and Mortality:
       A Critical Review of the Evidence.  Epidemiology: Volume 7, pgs. 420-428.

Moolgavkar, S. H.; Luebeck, E. G.;  Hall, T.  A.; Anderson, E. L. (1995a) Paniculate air
       pollution, sulfur dioxide, and daily mortality: a  reanalysis of the Steubenville data.
       Inhalation Toxicol.  7:  35-44.

Moolgavkar, S. H.; Luebeck, E. G.;  Hall, T.  A.; Anderson, E. L. (1995b) Air pollution and
       daily mortality in Philadelphia. Epidemiology 6: 476-484.

Morgan, W. K. C. (1978) Magnetite  pneumoconiosis. J. Occup. Med. 20: 762-763.

Morgan, A;  Moores, S.R.; Homes, A; Evans J.C.; Evans, N.H.; Black, A.  (1980) The
       effects  of quartz administered by intratracheal instillation on the rat lung. I the
       cellular response. Environ. Res. 22:1-12.

-------
                                        R-20

Morrow, P.E.; Utell, M.J.; Bauer, M.A.; Speers, D.M.; Gibb, F.R. (1994) Effects of near
       ambient levels of sulphuric acid aerosol on lung function in exercising subjects with
       asthma and chronic obstructive pulmonary disease.  In: Dodgson, J.; McCallum, R.I.,
       eds. Inhaled particles VII: proceedings of an international symposium; September
       1991; Edinburgh, United Kingdom.  Ann. Occup. Hyg. 38 (suppl. 1): 933-938.

Morrow ,P.E. (1988)  Possible mechanisms to explain dust overloading of the lungs.
       Fundam. Appl. Toxicol.  10:369-384.

MTP International Review of Science Physiology Series One: Vol. 2, Respiratory
       Physiology. Ed. J.G. Widdicombe University Park Press, Baltimore, MD, (1974).

Mueller, P.  K.; Hidy, G. M. (1983) The sulfate regional experiment (SURE): report of
       findings. Palo Alto,  CA: Electric Power Research Institute; EPR1 report no.
       EA-1901. 3v.

National Academy of Sciences.  (1985)  Epidemiology and Air Pollution. Available from:
       National Academy Press, 2101 Constitution Ave., NW, Washington, DC  20418.

National Academy of Sciences.  (1980)  Controlling Airborne Particles.  University Press,
       Baltimore, MD.

National Acid Precipitation Assessment  Program (NAPAP), (1991).  Office of the Director,
       Acid Deposition: State of Science and Technology. Report 24, Visibility: Existing
       and Historical Conditions - Causes and Effects.  Washington, D.C.

National Center for  Health  Statistics; Vital  and  Health Statistics:  Current Estimates from the
       National Health Interview Survey, 1993 (DHHS Publication No. (PHS) 95-1518;
       December 1994.

National Center for  Health  Statistics (1994) Current estimates from the  National Health
       Inteview Survey, 1992.  Hyattsville, MD: U.S.  Department of Health and Human
       Services, Public Health Service,  Centers for Disease Control and Prevention;  DHHS
       publication no. (PHS) 94-1517.  (Data from the National Health Survey:  series 10,
       no. 189).

National Research Council  (1979)  Airborne particles. University Park Press, Baltimore,
       MD.

National Research Council, Committee  on  Haze in National Parks and Wilderness Areas
       (1993) Protecting Visibility in National Parks and Wilderness Areas.  National
       Academy Press, Washington, D.C.

-------
                                         R-21

National Toxicology Program (1992). Toxicology and Carcinogenisis Studies of Talc in
       D44/N rats and B6C3f, mice (inhalation studies) NTP TR 421 NIH Publications No.
       92-3152.U.S. Department of Health and Human Services.

Neas, L.M.; Dockery, D.W.; Koutrakis, P.; Tollerud, D.J.; Speizer, F.E. (1995) The
       association of ambient air pollution with twice daily peak expiratory flow rate
       measurements in children. Am. T. Epidemiol. 141:111-122.

Neas, L.M.; Dockery, D.W.; Ware, J.H.; Spengler, J.D.; Ferris, B.C., Jr.; Speizer, F.E.
       (1994) Concentration of indoor paniculate matter as a determinant of respiratory
       health in children. Am. J. Epidemiol. 139: 1088-1099.

Nichols, M.D.  (1996)  Memorandum to EPA Air Division Directors regarding Areas
       Affected by PM-10 Natural Events, dated May 30, 1996.

Niinimaa, V.; Cole,  P.; Mintz, S.; Shephard, R.J.  (1980)  The switching point from nasal to
       oronasal breathing.  Respir. Physiol. 42: 61-71.

Nriagu, J.O. (1978)  Deteriorative effects of sulfur pollution on  materials.  In: Nriagu, J.O.,
       ed. Sulfur in the environment, part II: ecological impacts.  New  York, NY: John
       Wiley & Sons; pp. 1-59.

Oberdorster, G.; Gelein, R.; Ferin,  J.;  Weiss,  B. (1995a) Association of paniculate air
       pollution and  acute mortality: involvement of ultrafine particles?  In: Colloquium on
       paniculate air pollution and human mortality and morbidity; January; Irvine, CA.
       Inhalation  Toxicol. 7:  111-124.

Oberdorster, G.; Ferin, J.;  Gelein, R.;  Mercer, P.; Corson, N.; Godleski, J. (1995b)  Low-
       level ambient air paniculate levels and acute mortality/morbidity: studies with
       ultrafine Teflon™ particles.  Am. J. Respir.  Crit. Care Med. 151 (suppl.):  A66.

Oberdorster et al. (1995c)  In: Advances  in combustion toxicology: proceedings of an
       international colloquium;  Oklahoma City, OK.

Oberdorster, G.; Ferin, J.;  Lehnert, B.  E. (1994) Correlation between particle size, in vivo
       particle persistence,  and lung  injury. Environ. Health Perspect. 102(suppl. 5):
       173-179.

Oberdorster, G.; Ferin, J.;  Gelein, R.;  Soderholm, S. C.; Finkelstein, J. (1992) Role of the
       alveolar  macrophage in lung  injury: studies with ultrafine particles. Environ. Health
       Perspect. 97:  193-199.

Osebold, J.W.;  Gershwin, L.J.;  Zee, Y.C. (1980)  Studies on the enhancement of allergic
       lung  sensitization by inhalation of ozone and sulfuric aerosol. J. Environ. Pathol.
       Toxicol. 221-234.

-------
                                         R-22

Ostro, B. (1993) The association of air pollution and mortality: examining the case for
      inference. Arch. Environ. Health 48: 336-342.

Ostro, B. D.  (1987) Air pollution and morbidity revisited: a specification test. J. Environ.
      Econ.  Manage. 14: 87-98.

Ostro, B. (1984) A search for a threshold in the relationship of air pollution to mortality: a
      reanalysis of data on London winters. Environ.  Health  Perspect. 58: 397-399.

Ostro, B. D.  (1983) The effects of air pollution on work loss and morbidity. J.  Environ.
      Econ.  Manage. 10: 371-382.

Ostro, B. D.; Sanchez, J. M.;  Aranda, C.; Eskeland, G. S. (1996) Air pollution and
      mortality: results from a study of Santiago, Chile.  In:  Lippmann, M., ed. Papers
      from the ISEA-ISEE annual meeting; September 1994;  Research Triangle Park, NC.
      J. Exposure Anal. Environ.  Epidemiol.: Vol. 6, no.  1, January-March 1996, pp. 97-
       114.

Ostro, B. D.; Lipsett, M. J.; Mann, J. K.; Braxton-Owens, H.; White, M. C. (1995) Air
      pollution and asthma exacerbations among African-American children in Los Angeles.
      In: Phalen,  R.F.; Bates, D.V., eds. Proceedings of the colloquium on paniculate air
      pollution and human mortality and morbidity, part II; January 1994; Irvine, CA.
      Inhalation Toxicol. 7:711-722.

Ostro, B.D.;  Lipsett, M.J.; Mann,  J.K.; Krupnick, A.; Harrington, W. (1993)  Air pollution
      and respiratory morbidity among adults in Southern California.  Am. J. of Epidem.
       137: p. 691.

Ostro, B. D.; Lipsett, M. J.; Wiener, M. B.;  Seiner, J. C. (1991) Asthmatic responses to
      airborne acid aerosols. Am. J. Public Health 81: 694-702.

Ostro, B. D.; Rothschild, S. (1989) Air pollution and acute respiratory morbidity:
      an observational study of multiple pollutants.  Environ.  Res. 50: 238-247.

Ottaway, C.A.; Warren, R.E.; Saibil, F.G.; et al. (1984)  Monocyte procoagulant activity in
      Whipple's disease.  J.  Clin. Immunol.  4:348-58.

Ozkaynak, H.; Xue, J.;  Spengler, J.D.; Wallace, L.A.; Pellizzari, E.D.; Jenkins,  P. (1996)
       Personal exposures to airborne particles and metals; results from the particle TEAM
       study  in Riverside, CA.   In Lippmann, M., ed. Papers from the ISEA-ISEE annual
       meeting; September 1994; Research Triangle Park, NC, J. Exp. Analy.  Environ.
       Epidemiol.: Vol. 6, No.  1,  January-March 1996, pp. 57-78.

-------
                                         R-23

Ozkaynak, H.; Xue, J.; Severance, P.; Burnett, R.;  Raizenne, M. (1994) Associations
       between daily mortality, ozone, and paniculate air pollution in Toronto, Canada.
       Presented at: Colloquim on paniculate air pollution and human mortality and
       morbidity; January; Irvine, CA. Irvine, CA:  University of California at Irvine, Air
       Pollution Health Effects Laboratory; report no. 94-02.

Ozkaynak, H.; Thurston, G. D. (1987) Associations between 1980 U.S. mortality rates and
       alternative measures of airborne particle concentration. Risk Anal. 7: 449-461.

Penna, M. L. F.; Duchiade, M. P. (1991) Contamination del aire y  mortalidad infantil por
       neumom'a [Air pollution and infant mortality  due to pneumonia]. Bol. Of. Sanit.
       Panam. 110: 199-207.

Perry,  G.B.; Chai, H.; Dickey, D.W.; Jones, R.H.; Kinsman, R.A.; Morrill,  C.G.; Specter,
       S.L.; Weiser, R.C. (1983)  Effects of paniculate air pollution on asthmatics. Am. J.
       Public Health 73: 50-56.

Pinto, M.; Birnbaum, S.C.;  Kadar, T.; Goldberg, G.M. (1979)   Lung injury in mice induced
       by factors acting synergistically with inhaled  paniculate antigen. Clin.  Immunol.
       Immunopathol.  13: 361-368.

Pitchford, M.; Malm, W. (1994) Development and Applications of a Standard Visual  Index.
       Atmospheric Environment, vol. 28, no. 5, pp. 1049-1054.

Pope, C. A., III. (1991) Respiratory hospital admissions associated with PM10  pollution in
       Utah, Salt Lake, and  Cache Valleys. Arch. Environ. Health 46: 90-97.

Pope, C. A., Ill; Dockery,  D. W.; Schwartz, J. (1995) Review of epidemiological evidence
       of health effects of paniculate air pollution. Inhalation Toxicol. 7: 1-18.

Pope, C. A., Ill; Kalkstein,  L. S. (1996) Synoptic weather modeling and estimates of the
       exposure-response relationship  between daily  mortality and paniculate air pollution.
       Environ. Health Perspect.  104: in press.

Pope, C. A., Ill; Thun, M. J.; Namboodiri, M. M.; Dockery, D. W.;  Evans, J. S.;
       Speizer, F.  E.;  Heath,  C. W.,  Jr. (1995) Paniculate air pollution as a predictor of
       mortality in a prospective study of U.S. adults. Am. J. Respir.  Crit.  Care Med.
       151: 669-674.

Pope, C. A., Ill; Kanner, R. E. (1993) Acute effects of PM10 pollution on pulmonary
       function of smokers with mild to moderate chronic obstructive pulmonary disease.
       Am. Rev. Respir. Dis. 147: 1336-1340.

-------
                                         R-24

Pope, C. A., Ill; Schwartz, J.; Ransom, M. R. (1992a) Daily mortality and PM10 pollution
       in Utah valley. Arch. Environ. Health 47: 211-217.

Pope, C. A., III.; Dockery, D. W. (19925) Acute health effects of PM10 pollution on
       symptomatic and asymptomatic children. Am. Rev. Respir. Dis. 145: 1123-1128.

Pope, C. A., Ill; Dockery,  D. W.; Spengler, J. D.; Raizenne, M. E. (1991) Respiratory
       health and PM10 pollution: a daily time series analysis. Am. Rev. Respir. Dis.
       144:  668-674.

Pope, C. A., Ill (1989).  Respiratory Disease Associated with Community Air Pollution and
       a Steel Mill, Utah Valley.  Am.  J. Public Health 79:623-628.

Pope, C. A., III. (1994) Paniculate pollution and mortality in Utah valley. Prepared for:
       Critical evaluation workshop on  paniculate matter— mortality epidemiology studies;
       November;  Raleigh, NC. Provo, UT: Brigham Young University.

Pope, C.A., III. (1995)  Combustion Source Paniculate Air Pollution and Health and
       Regulatory Issues.  Proceedings  of an International Specialty Conference.   Pittsburgh,
       PA April 4/6 1995.

Quackenboss, J. J.; Krzyzanowski, M.; Lebowitz, M. D. (1991) Exposure assessment
       approaches  to evaluate respiratory health effects of paniculate matter and nitrogen
       dioxide.  J. Exposure Anal. Environ. Epidemiol.  1: 83-107.

Raabe, O.G.; Wilson, D.W.; Al-Bayati, M.A.; Hornof, W.J.; Rosenblatt, L.S. (1994)
       Biological effects of inhaled pollutant aerosols. In: Dodgson, J.; McCallum, R.I.,
       eds. Inhaled particles VII: proceedings of an international symposium;  September
       1991; Edinburgh,  United Kingdom. Ann. Occup. Hyg. 38 (suppl.  1): 323-330.

Raizenne, M.; Neas, L. M.; Damokosh, A. I.; Dockery, D. W.; Spengler, J. D.; Koutrakis.
       P.; Ware, J. H.; Speizer, F. E.  (1996) Health effects of acid aerosols on North
       American children: pulmonary function. Environ. Health Perspect.: accepted.

Ransom, M. R.; Pope, C. A., III. (1992) Elementary school absences and PMUI pollution in
       Utah Valley. Environ. Res. 58:  204-219.

Raub, J.A.; Hatch, G.E.; Mercer, R.R.; Grady, M.; Hu, P.-C. (1985) Inhalation studies of
       Mt. St. Helens volcanic ash in animals: II. lung function, biochemistry, and
       histology. Environ. Res.  37: 72-83.
Rodes, C.E.; Evans, E.G. (1985) Preliminary assessment of 10 pm paniculate sampling at
       eight locations in the United States.  Atmos. Environ. Vol.  19, No. 2, pp. 293-303.

-------
                                         R-25

Roemer, W.; Hoek, G.; Brunekreef, B.  (1993) Effect of ambient winter air pollution on
       respiratory health of children with chronic respiratory symptoms. Am. Rev. Respir.
       Dis.  147: 118-124.

Rosenberg, L.; Kaufman, D.W.; Helmrich, S.P.; Shapiro, S. (1985)  The risk of myocardial
       infarction after quitting smoking in men under 55 years of aage. N.  Engl. J.  Med.
       313:1511-1514.

Rosenberg, L.; Palmer, J.R.; Shapiro, S. (1990) Decline in the risk of myocardial infarction
       among women who stop smoking.  N. Engl. J. Med. 322:213-217.

Russell, C.A.  (1976) How environmental pollutants diminish contact reliability.  Insul.
       Circuits 22: 43-46.

SAI (1996) Statistical Support for the Particulate Matter NAAQS. Prepared  by Systems
       Applications International, 101 Lucas Valley Road, San Rafael,  CA 94903. April
       1996.

Saldiva, P.H.N.; Pope, C.A.,  III; Schwartz, J.; Dockery, D.W.; Licntenfels, A.J.; Salge,
       J.M.; Barone, I.; Bohm, G.M. (1995). Air Pollution and Mortality  in Elderly
       People: A Time-Series Study in Sao Paulo, Brazil. Arch, of Environ. Health 50:159-
       163.

Samet, J.M.; Zeger, S.L.; Kelsall, J.E.;  Xu, J. (1996a) Air pollution and mortality in
       Philadelphia, 1974-1988,  report to the Health  Effects Institute on phase IB: Particle
       Epidemiology Evaluation  Project.  Cambridge, MA: Health  Effects  Institute;
       accepted.

Samet, J. M.; Zeger, S. L.; Kelsall, J. E.; Xu, J.; Kalkstein, L. S. (1996b) Weather,  air
       pollution and mortality  in Philadelphia, 1973-1980, report to the Health  Effects
       Institute on phase IB, Particle Epidemiology Evaluation Project. Cambridge, MA:
       Health Effects Institute; review draft.

Samet, J.M.; Zeger, S.L.; Berhane, K. (1995). The Association of Mortality and Particulate
       Air Pollution, in:  Particulate Air Pollution and Daily Mortality:  Replication  and
       Validation of Selected Studies. Prepared by:  Health Effects Institute, 141 Portland
       Street, Suite 7300,  Cambridge, MA 02139.  August 1995, pp.  1-104.

Samet, J. M.; Tager, I. B.; Speizer, F.  E.  (1983) The relationship  between respiratory
       illness  in childhood and chronic air-flow obstruction in adulthood. Am. Rev. Respir.
       Dis.  127: 508-523.

-------
                                         R-26

Samet, J.M.; Speizer, F.E.; Bishop, J.D.; Spengler, J.D.; Ferris, B.J. (1981). The
       relationship between air pollution and emergency room visits in an industrial
       community.  J. Air Pollut. Control Assoc. 31:236-240.

Sant1 Ambrogio, G. (1982) Information arizing from the tracheobronchial tree of
       mammmals.  Physiol. Rev. 62:531-569.

Sanyal, B.; Singhania, O.K. (1956) Atmospheric corrosion of metals; part I.J. Sci. Ind. Res.
       Sect. B 15: 448-455.

Schimmel, H. (1978) Evidence for possible acute health effects of ambient air pollution from
       time series analysis: methodological questions and some new results based on New
       York City daily mortality, 1963-1976. Bull.  N. Y. Acad.  Med. 54: 1052-1108.

Schimmel, H., and T.J.  Murawski (1976).  The relation of air pollution to mortality. J.
       Occup. Med.  18:316-333.

Schlesinger, R.B.; Gunnison,  A.F.; Zelikoff, J.T. (1990)  Modulation of pulmonary
       eicosanoid metabolism following exposure to sulfuric  acid. Fundam. Appl. Toxicol.
       15: 151-162.

Schrenk, H.H.; Heimann, H.; Clayton, G.D.; Gafafer, W.M.; Wexler, H. (1949)  Air
       pollution  in Donora, PA.  Epidemiology of the unusual smog episode of October
       1948: preliminary report.   Washington,  DC: Public Health Service; Public Health
       Service bulletin no. 306.

Schulze,  W.D.; Brookshire, D.S.; Walther, E.G.; MacFarland,  K.K.; Thayer, M.A.;
       Whitworth, R.L.; Ben-David,  S.; Malm, W.; Molenar. Jr. (1983)  The economic
       benefits of preserving visibility in the national parklands of the southwest.  Nat.
       Resour. J. 23: 149-173.

Schwartz, H. (1972) On the effect of magnetite on atmospheric  rust and  on  rust  under a coat
       of paint.  Werkst. Korros. 23: 648-663.

Schwartz, J.  (1995). Short term fluctuations in  air pollution and hospital admissions  of the
       elderly for respiratory  disease. Thorax 50:531-538.

Schwartz, J.  (1996). Air pollution and hospital admissions for respiratory disease.
       Epidemiology 7:20-28.

-------
                                         R-27

Schwartz, J. (1995b) Health effects of air pollution from traffic: ozone and paniculate
       matter. In: Fletcher, T., ed. Health at the crossroads: transportation policy and urban
       health, proceedings of the fifth annual public health forum of the London School of
       Hygiene and Tropical Medicine; April; London, United Kingdom.  New  York, NY:
       John Wiley & Sons, Inc.; in preparation.

Schwartz, J. (1994a) Total suspended paniculate matter and daily mortality in Cincinnati,
       Ohio.  Environ. Health Perspect. 102: 186-189.

Schwartz, J. (1994b) Air pollution and daily mortality: a review and meta analysis. Environ.
       Res. 64: 36-52.

Schwartz, J. (1994c) What are people dying of on high air pollution days? Environ. Res.
       64:  26-35.

Schwartz, J. (1994d) Air pollution and hospital admissions for the elderly  in Detroit,
       Michigan. Am. J. Respir. Crit. Care Med. 150: 648-655.

Schwartz, J. (1994e) Air pollution and hospital admissions for the elderly  in Birmingham,
       Alabama. Am. J.  Epidemiol. 139: 589-598.

Schwartz, J. (1994f) PM10,  ozone, and hospital admissions for the elderly  in Minneapolis,
       MN. Arch. Environ. Health 49: 366-374.

Schwartz, J. (1994g) Nonparametric smoothing in the analysis of air pollution and respiratory
       illness. Can. J. Stat. 22: 1-17.

Schwartz, J. (1994h) The use of generalized additive models in epidemiology. Proc.  Int.
       Conf.  Biometric Soc. 17: 55-80.

Schwartz, J. (1993a) Air pollution and daily mortality  in Birmingham,  Alabama.
       Am. J. Epidemiol. 137:  1136-1147.

Schwartz, J. (1993b) Particulate air pollution and chronic respiratory disease. Environ.  Res.
       62:  7-13.

Schwartz, J.; Dockery, D. W.; Neas, L. M. (1996a).  Is Daily Mortality  Associated
       Specifically with Fine Particles?  J. Air Waste Manage. Assoc.: accepted.

Schwartz, J.; Spix,  C.; Touloumi, G.; Bacharova, L.; Barumamdzadeh, T.; Le Tertre, A.;
       Piekarksi, T.; Ponce de  Leon, A.; Ponka, A.; Rossi,  G.; Saez, M.; Shouten, J. P.
       (1996b) Methodological  issues in studies of air pollution and daily  counts of deaths or
       hospital admissions.  J. Epidemiol. Commun. Health: in press.

-------
                                         R-28

Schwartz, J.; Morris, R. (1995) Air pollution and hospital admissions for cardiovascular
       disease in Detroit, Michigan. Am. J. Epidemiol. 142:23-35.

Schwartz, J.; Dockery, D.W.; Neas, L.M.; Wypij, D.; Ware, J.H.; Spengler, J.D.;
       Koutrakis, P.; Speizer, F.E.; Ferris, E.G., Jr. (1994) Acute effects of summer air
       pollution on respiratory symptom reporting in children.  Am. J. Respir. Crit. Care
       Med.  150: 1234-1242.

Schwartz, J.; Slater, D.; Larson, T. V.; Pierson, W. E.; Koenig, J. Q. (1993) Paniculate air
       pollution and hospital emergency room visits for asthma in Seattle. Am. Rev. Respir.
       Dis. 147: 826-831.

Schwartz, J.; Dockery, D. W. (1992a)  Increased mortality in Philadelphia associated with
       daily air pollution concentrations. Am. Rev. Respir. Dis. 145:  600-604.

Schwartz, J.; Dockery, D. W. (1992b)  Paniculate air pollution  and daily mortality in
       Steubenville, Ohio. Am. J. Epidemiol. 135:  12-19.

Schwartz, J.; Spix, C.; Wichmann, H.  E.; Malin,  E. (1991) Air pollution and acute
       respiratory illness in five German communities. Environ. Res.  56: 1-14.

Schwartz, J.; Marcus, A.H.  (1986)  Statistical reanalyses of data relating mortality to air
       pollution during London winters 1958-1972.  Washington, DC: U.S. Environmental
       Protection Agency, Office of Policy, Planning and Evaluation.

Schwartz, L.W.; Moore, P.P.; Chang,  D.P.; Tarkmyton, B.K.; Dungworth, D.L.; Tyler,
       W.S.  (1977) Short-term effects  of sulfuric acid aerosols on the respiratory tract.  A
       morphological study in guinea pigs, mice, rats,  and monkeys.  In: Lee, S.D., ed.
       Biochemical effects of environmental pollutants.  Ann Arbor, MI:  Ann Arbor Science
       Publishers, Inc.; pp. 257-271.

Seaton, A.; MacNee, W.; Donaldson, K.; Godden, D. (1995) Paniculate air pollution  and
       acute  health effects. Lancet (8943): 176-178.

Shaw,  R.W., Jr.; Paur, RJ. (1983) Composition of aerosol  particles collected at rural sites
       in the Ohio River Valley. Atmos. Environ. 17:2031-2044.

Sherwin, R.P., M.L. Barman, and J.L. Abraham (1979).  Silicate pneumoconiosis of farm
       workers. Lab. Invest. 40:576-582.

Shumway, R. H.  (1988) Applied statistical time series analysis. Englewood Cliffs, NJ:
       Prentice Hall, Inc.

-------
                                        R-29

Shumway, R. H.; Azari, A. S.; Pawitan, Y. (1988) Modeling mortality fluctuations in Los
       Angeles as functions of pollution and weather effects. Environ. Res. 45: 224-241.

Sinclair, J.D. (1992) The relevance of particle contamination to corrosion of electronics  in
       processing and field environments. Proc.  Electrochem. Soc. 93-1:  325-335.

Sioutas, C.; Koutrakis, P.; Burton, R.M. (1995)  A technique to expose animals to
       concentrated fine ambient aerosols.  Environmental Health  Perspectives 103:172-177.

Sisler, J.; Malm, W.; Molenar, J.; Gebhardt, K.  (1996)  Spatial and Seasonal Patterns and
       Long Term Variability of the Chemical Composition of the Haze in the U.S.: An
       Analysis of Data from the IMPROVE Network. Fort Collins, CO: Cooperative
       Institute for Research in the Atmosphere, Colorado State University.

Sisler, J.; Malm, W. (1994) The Relative Importance of Soluble Aerosols to Spatial and
       Seasonal Trends of Impaired Visibility in the United States. Atmospheric
       Environment,  vol. 28, no. 5, pp. 851-862.

Sisler, J.; Huffman, D.; Latimer,  D. (1993) Spatial and Temporal Patterns and the
       Chemical Composition of the Haze in the United States:  An Analysis of Data from
       the IMPROVE Network, 1988-1991, Fort Collins, CO, 1993.

Skerry, B.S.; Johnson, J.B.; Wood, G.C. (1988a) Corrosion in smoke, hydrocarbon,  and
       S02 polluted atmospheres—I. General behaviour of iron. Corros.  Sci.  28:  657-695.

Skerry, B.S.; Wood, J.C.; Johnson, J.B.; Wood, G.C. (1988b) Corrosion in  smoke,
       hydrocarbon, and SO2 polluted atmospheres—II. Mechanistic implications for iron
       from surface analytical and allied techniques.  Corros.  Sci. 28: 697-719.

Skornick, W.A.; Brain, J.D.  (1983)  Relative toxicity of inhaled  metal sulfate salts for
       pulmonary macrophages.  Am Rev Resp Dis 1983; 128: 297-303.

Spektor, D.M.; Hofmeister, V.A.; Artaxo, P.; Brague, J.A.P.; Echelar, F.; Noguieira,
       D.P.; Hayes, C.; Thurston, G.D.; Lippmann, M.  (1991)  Effects of heavy  industrial
       pollution on respiratory function  in the children of Cubatao, Brazil: a preliminary
       report.  Environ. Health Perspect. 94: 51-54.

Spence, J. W.; Haynie, F. H. (1975) Design of a laboratory experiment to identify the
       effects of environmental pollutants on materials. In:  Corrosion in natural
       environments.  Philadelphia PA:   American Society for Testing and Materials;
       ASTM  special  technical publication no.  558; p. 279-291.

Spencer, H. (1977) Pathology of the lung (excluding pulmonary tuberculosis). 3rd  ed.
       Oxford, United Kingdom: Pergamon Press. 2v.

-------
                                         R-30

Spengler, J.D.; Koutrakis, P.; Dockery, D.W.; Raizenne, M.; Speizer,  F.E. (1996)  Health
       effects of acid aerosols on North American children:  air pollution exposures.
       Environ. Health Perspect.: in press.

Spengler, J.D.; Dockery, D.W.; Turner, W.A.; Wolfson, J.M.; Ferris,  E.G., Jr. (1981)
       Long-term measurements of respirable sulfates and particles inside and outside homes.
       Atmos. Environ. 15:23-30.

Spix, C.; Heinrich, J.;  Dockery, D.;  Schwartz, J.; Volksch, G.; Schwinkowski, K.;  Collen,
       C.; Wichmann,  H. E. (1993) Air pollution and daily  mortality in Erfurt, East
       Germany,  1980-1989. Environ. Health Perspect. 101: 518-526.

Stern, B. R.; Raizenne, M. E.; Burnett, R.  T.; Jones, L.; Kearney, J.; Franklin, C.  A.
       (1994) Air pollution and childhood respiratory health: exposure to sulfate and  ozone
       in 10 Canadian rural  communities. Environ. Res.  66: 125-142.

Stober, W.; Mauderly,  J.L.: (1994) Model-inferred hypothesis of a critical dose for
       overload tumor reduction by diesel soot and carbon black. Inhalation Toxicology,
       6:427-457.

Stringer, B.K.; Imrich, A.; Kobzik, L.  (1995)  Lung epithelial cell (A549) uptake of
       opsonized environmental particles. Am. J. Respiratory Cr. Care Med.  151:A264.

Studnicka, M.J.; Frischer, T.; Meinert, R.;  Studnicka-Benke, A.; Hajek, K.; Spengler, J.D.;
       Neumann, M.G. (1995) Acidic  particles and lung function in children:  a summer
       camp study in the  Austrian Alps. Am. J.  Respir. Crit. Care Med. 151: 423-430.

Styer, P.; McMillan, N.; Gao, F.;  Davis, J.; Sacks, J. (1995)  Effect of outdoor airborne
       paniculate matter on  daily death counts. Environ.  Health  Perspect.

Suh, H. H.; Allen,  G.  A.; Koutrakis, P.; Burton,  R. M.  (1995)  Spatial variation in acidic
       sulfate and ammonia  concentrations within  metropolitan Philadelphia. J.  Air Waste
       Manage. Assoc. 45: 442-452.

Suh, H.H.; Koutrakis,  P.; Spengler, J.D. (1994) The relationship between airborne acidity
       and ammonia in indoor environments.  J. Expos.  Assess.  Environ. Epidemiology
       1994; 4:1-23.

Suh, H.H.; Koutrakis,  P.; Spengler, J.D. (1993) Validation  of personal  exposure models for
       sulfate and aerosol strong acidity.  J.  Air Waste Manage. Assoc.  43:845-850.

-------
                                         R-31

Sunyer, J.; Saez, M.; Murillo, C.; Castellsague, J.; Martinez, F.; Anto, J. M.  (1993) Air
       pollution and emergency room admissions for chronic obstructive pulmonary disease:
       a 5-year study. Am. J. Epidemiol. 137: 701-705.

Svartengren, K.; Lindstad, P.A.; Svartengren, M.; Bylin, G; Philipson, K. Camner, P.
       (1994)  Deposition of inhaled particles in the mouth and throat of asthmatic subjects.
       Am. Respir. J. 1467-1473.

Sydberger, T.;  Ericsson, R.  (1977) Laboratory testing of the atmospheric corrosion of steel.
       Werkst.  Korros.  28:  154-158.

Takenaka, S.; Dornhofer-Takenaka, H.; Muhle, H. (1986) Alveolar distribution of fly ash
       and of titanium dioxide after long-term inhalation by Wistar rats. J.  Aerosol Sci.  17:
       361-364.

Tamura, K.; Ando, M.; Sagai, M.;  Matsumoto, Y. (1996) Estimation of levels of personal
       exposure to suspended paniculate matter and nitrogen dioxide in Tokyo. Environ. Sci.
       (Tokyo) 6: accepted.

Tamura, K.; Ando, M.  (1994) Suspended paniculate matter in indoor and outdoor air along
       a main road  and  personal exposure assessment. Presented at: Sixth conference of the
       International Society  for Environmental Epidemiology/fourth conference of the
       International Society  for Exposure Analysis joint conference; September; Research
       Triangle Park, NC. Chapel Hill, NC:  University of North Carolina, School of Public
       Health; abstract no.  116.

Tepper, J.  S.; Lehmann, J. R.; Winsett, D. W.; Costa, D. L.; Ghio, A. J. (1994) The role
       of surface-complexed iron in the development of acute lung inflammation and airway
       hyperresponsiveness.  Am. Rev.  Respir. Dis. 149(4, pt. 2): A839.

Terrat,  M.-N.;  Joumard, R.  (1990)  The measurement of soiling. Sci. Total Environ. 93:
       131-138.

Thatcher,  T. L.;  Layton, D. W. (1995) Deposition, resuspension, and penetration of
       particles within a residence. Atmos.  Environ. 29: 1487-1497.

Thurston,  G. D.; Gorczynski, J. E., Jr.;  Currie, J. H.; He, D.; Ito, K.; Hipfner, J.;
       Waldman, J.; Lioy, P. J.; Lippmann,  M. (1994a) The nature and origins of acid
       summer haze air pollution in metropolitan Toronto, Ontario. Environ. Res.
       65:  254-270.

Thurston,  G. D.; Ito, K.; Hayes, C. G.;  Bates, D. V.; Lippmann,  M. (1994b) Respiratory
       hospital  admissions and summertime haze air pollution in Toronto, Ontario:
       consideration of  the role of acid aerosols. Environ.  Res. 65: 271-290.

-------
                                         R-32

Thurston, G. D.; Ito, K.; Kinney, P. L.; Lippmann, M. (1992) A multi-year study of air
       pollution and respiratory hospital admissions in three New York State metropolitan
       areas: results for 1988 and 1989 summers.  J. Exposure  Anal. Environ. Epidemiol.
       2: 429-450.

Thurston, G. D.; Ito, K.; Lippmann, M.; Hayes, C. (1989) Mortality in  relation to exposure
       to acidic aerosols during 1963-1972 winters. Environ. Health Perspect. 79:  73-82.

Thurston, G. D.; Kinney, P. L. (1995) Air pollution epidemiology: considerations in
       time-series modeling. In: Phalen, R. F.; Bates, D. V., eds. Proceedings of the
       colloquium on paniculate air pollution and  human mortality and morbidity;  January
       1994; Irvine, CA. Inhalation Toxicol.  7: 71-83.

Touloumi, G.;  Pocock, S. J.; Katsouyanni, K.; Trichopoulos, D. (1994)  Short-term effects
       of air pollution on daily mortality in Athens: a  time-series analysis.  Int. J. Epidemiol.
       23:  1-11.

Toyama, T. (1964)   Air pollution  and its health effects in Japan.  Arch. Environ. Health
       8:153-173.

Trijonis, J. (1982) Existing and natural background levels of visibility and fine particles in
       the rural East. Atmos. Environ.  16:2431-2445.

U.S. Bureau  of Census (1992) Statistical abstact of the United States 1992. 112th  ed.
       Washington  DC: U.S. Department of Commerce.

U.S. Department of Health and Human Services (1992). Vital and Health  Statistics, Detailed
       Diagnoses an Procedures, National Hospital Discharge Survey,  1990.  HyattsviHe,
       MD: Center for Disease Control. June 1992.

U.S. Department of Health, Education, and Welfare (1964).  Smoking and health: report of
       the Advisory Committee to the Surgeon General of  the Public Health Service.
       Washington,  DC: Public Health Service; p. 60.

U.S. Environmental Protection Agency (1996). Air Quality Criteria for  Paniculate Matter.
       Research Triangle Park, NC: National Center for Environmental  Assessment.  Office
       of Research  and Development.  Final draft. April 12, 1996.

U.S. Environmental Protection Agency (1995a). National  Air  Quality and Emissions Trends
       Report  1994.  EPA  report  no. EPA-44/R-95-014. October 1995.

-------
                                        R-33

U.S. Environmental Protection Agency (19955).  National Air Pollutant Emission Trends,
       1900-1994.  EPA report no. 454/R-95-011.  October 1995.

U.S. Environmental Protection Agency (1995c).  Ambient levels and noncancer health effects
       of inhaled crystalline al a morphons silica.  Research Triangle Park, NC: National
       Center for Environmental Assessment. Office of Research and Development. EPA
       report no. EPA 600/R-95/115.  Final draft.

U.S. Environmental Protection Agency (1994a).  National Air Quality and Emissions Trends
       Report, 1993. Office of Air Quality Planning and Standards. EPA454/R-94-026.
       October 1994.

U.S. Environmental Protection Agency (1994b).  Review of the Ambient Air Quality
       Standards for Sulfur Oxides:  Updated Assessment of Scientific and Technical
       Information, Supplement to the  1986 OAQPS Staff Paper Addendum, Office of Air
       Quality Planning and Standards, Research  Triangle Park, NC, EPA/452/R-94-013.

U.S. Environmental Protection Agency (1994c).  Review of the Ambient Air Quality
       Standards for Sulfur Oxides:  Assessment of Scientific and Technical Information,
       Supplement  to the 1986 OAQPS Staff Paper Addendum, Office of Air Quality
       Planning and Standards,  Research Triangle Park, NC, EPA/452/R-94-013.

U.S. Environmental Protection Agency (1993). Office of Air Quality Planning and
       Standards Effects of the  1990 Clean Air Act Amendments on Visibility in Class I
       Areas: An EPA Report  to Congress.   Research Triangle Park, N.C. 1993.

U.S. Environmental Protection Agency (1992). Respiratory Health Effects of Passive
       Smoking:  Lung Cancer and Other Disorders.  Office of Health and Environmental
       Assessment. EPA/600/6-90/006F.

U.S. Envrionmental Protection Agency (1991) Air Quality Criteria for Carbon Monoxide.
       Office of Research and Development. EPA/600/8-90/045f.

U.S. Environmental Protection Agency. (1989a) Exposure factors handbook. Washington,
       DC: Office of Health and Environmental Assessment; EPA report no.
       EPA/600/8-89/043.  Available from: NTIS, Springfield, VA; PB90-106774.

U.S. Environmental Protection Agency (1989b).  An acid aerosols issue paper:  health
       effects and aerometrics.  Research Triangle Park, NC: Office of Health and
       Environmental Assessment,  Environmental Criteria and Assessment Office; EPA
       report no. EPA-600/8-88-005F.  Available from: NTIS, Springfield, VA; PB91-
       25864.

-------
                                        R-34

U.S. Environmental Protection Agency (1988). Report of the Clean Air Scientific Advisory
       Committee (CASAC) - Recommendations for Future Research on Acid Aerosols.
       Office of the Administrator, Science Advisory Board.  EPA report  no: EPA-
       SAB/CASAC-89-002.

U.S. Environmental Protection Agency. (1986a) Second addendum to air quality criteria for
       paniculate  matter and sulfur oxides (1982): assessment of newly  available health
       effects information. Research Triangle Park, NC: Office of Health  and Environmental
       Assessment, Environmental Criteria and Assessment Office; EPA report no.
       EPA-600/8-86-020F. Available from: NTIS, Springfield, VA; PB87-176574.

U.S. Environmental Protection Agency. (1986b) Review of the national  ambient air quality
       standards for paniculate matter: updated assessment of scientific  and technical
       information, addendum to the 1982 OAQPS staff paper. Research Triangle Park, NC:
       Office of Air Quality Planning and Standards, Strategies and Air Standards Division;
       report no. EPA/450/05-86/012. Available from: NTIS, Springfield, VA;
       PB87-176871/XAB.

U.S. Environmental Protection Agency. (1986c) Air quality criteria for  ozone and other
       photochemical oxidants. Research Triangle  Park, NC: Office of  Health and
       Environmental Assessment, Environmental Criteria and  Assessment Office; EPA
       report nos. EPA-600/8-84-020aF-eF. 5v. Available from: NTIS,  Springfield, VA;
       PB87-142949.

U.S. Environmental Protection Agency. (1982a) Air quality criteria for  paniculate matter and
       sulfur oxides. Research Triangle Park,  NC: Office  of Health and Environmental
       Assessment, Environmental Criteria and Assessment Office; EPA report no.
       EPA-600/8-82-029aF-cF. 3v. Available from: NTIS, Springfield, VA; PB84-156777.

U.S. Environmental Protection Agency. (1982b) Review of the national  ambient air quality
       standards for paniculate matter: assessment of scientific and technical information.
       Research Triangle Park,  NC: Office of Air Quality Planning and Standards, Strategies
       and Air Standards Division; report no.  EPA-450/5-82-001.  Available from: NTIS,
       Springfield, VA;  PB82-177874.

U.S. Environmental Protection Agency, Office of Air Quality Planning  and Standards (1979)
       Protecting  Visibility:  An EPA Report to Congress.  Research Triangle Park,  N.C.

U.S. Environmental Protection Agency.  (1969) Air Quality Criteria for Particulate Matter.
       U.S.  Government Printing Office,  Washington, DC.  AP-49.

-------
                                         R-35

Utell, MJ. and Frampton, M.W. (1995) Particles and mortality: a clinical perspective.
       Inhal. Tox. 7:  645-655.

Utell, M.J.; Frampton, M.W.; Morrow, P.E. (1993)  Quantitative clinical studies with
       defined exposures atmospheres.  In: Gardner, D.E.; Crapo, J.D.; McClellan, R.O.,
       eds. Toxicology of the lung. 2nd ed. New York, NY: Raven Press pp. 283-309.
       (Hayes,  A.W.; Thomas, J.A.; Gardner, D.E.; eds. Target organ toxicology series).

Utell, M.J.; Mariglio, J.A.;  Morrow, P.E.; Gibb, F.R.; Speers, D.M. (1989)  Effects of
       inhaled acid aerosols  on  respiratory function: the role of endogenous ammonia.  J.
       Aerosol  Med. 2: 141-147.

Utell, M.J.; Morrow,  P.E.;  Speers, D.M.; Darling, J.;  Hyde, R.W. (1983) Airway
       responses to sulfate and sulfuric acid aerosols in  asthmatics:  an exposure-response
       relationship. Am. Rev. Respir.  Dis. 128:444-450.

Utell, M.J.; Morrow,  P.E.;  Hyde, R.W. (1982)  Comparison of normal and asthmatic
       subjects' responses to snephate pollutant aerosols. In: Walton, W.H.; ed. Inhaled
       particles V. proceedings of an international symposium organized by the British
       Occupational Hygiene Society; September 1980;  Cardiff, Wales. Ann. Occup.  Hyg.
       26:691-697.

Utell, M.J.; Morrow,  P.E.;  Hyde, R.W. (1981). Comparison of normal and asthmatic
       subjects  responses to  sulfate pollutant aerosols, in: Walton, W.H.; ed.  Inhaled
       particles V. proceedings of an international symposium organized by the British
       Occupational Hygiene Society; September 1980;  Cardiffk Wales. Ann. Occup. Hyg.
       26: 691-697.

Utell, M. J.;  Aquilina, A. T.; Hall, W. J.; Speers,  D. M.; Douglas, R. G., Jr.; Gibb, F.
       R.; Morrow, P. E.; Hyde, R. W. (1980) Development of airway reactivity to  nitrates
       in subjects with influenza.  Am. Rev. Respir. Dis. 121: 233-241.

Wallace, L. (1996) Indoor particles: a review. J.  Air Waste Manage. Assoc. 46: 98-126.

Waller, R.E.  (1980).  The assessment of suspended particulates in relation to health.  Atmos.
       Environ. 14:1115-1118.

Walton, J.R.; Johnson, J.B.; Wood, G.C. (1982) Atmospheric corrosion initiation by  sulphur
       dioxide and paniculate matter—I. Test-Cell  apparatus for simulated atmospheric
       corrosion studies.  Br. Corros. J. 17: 59-64.

-------
                                         R-36

Ware, J. H.; Ferris, B. G., Jr.; Dockery, D. W.; Spengler, J. D.; Stram, D. O.;
      Speizer, F. E. (1986) Effects of ambient sulfur oxides and suspended particles on
      respiratory health of preadolescent children. Am. Rev. Respir.  Dis.  133: 834-842.

Ware, J. H.; Dockery,  D. W.; Spiro, A., Ill; Speizer,  F. E.; Ferris, B. G., Jr. (1984)
      Passive smoking, gas cooking, and respiratory health of children living in six cities.
      Am.  Rev.  Respir.  Dis.  129: 366-374.

Ware, J. H.; Thibodeau,  L. A.; Speizer, F. E.; Colome, S.; Ferris, B. G., Jr. (1981)
      Assessment of the health effects of atmospheric  sulfur oxides and paniculate matter:
      evidence from observational studies.  Environ. Health Perspect.  41: 255-276.

Watson, W. D.; Jaksch, J. A.  (1982) Air pollution:  household soiling and consumer welfare
      losses. J. Environ. Econ. Manage. 9: 248-262.

Watson, W. D., Jr.; Jaksch, J. A. (1978) Household cleaning costs and air pollution.
      Presented at the 71st Annual  Meeting, Air Pollution Control Association,  Houston,
      Texas, June 25-30. Paper No.  78-52.3.

Weiss, S.T.; Sparrow,  D.; O'Connor,  G.T. (1993)  The interrelationship among  allergy,
      airways  responsiveness, and asthma.  J. Asthma 30: 329-349.

Whitby, K.T.; Sverdrup,  G.M. (1980) California aerosols:  their physical  and chemical
      characteristics.  In: Hidy, G.M.; Mueller, P.K.; Grosjean, D.;  Appel, B.R.;
      Wesolowski, J.J.,  eds.  The character and origins of smog aerosols: a digest of
      results from the California Aerosol Characterization Experiment (ACHEX).  New
      York, NY: John Wiley  & Sons, Inc.; pp. 477-517.  (Advances in  environmental
      science and technology: v. 9).

Whitby, K.T.; Charlson,  R.E.; Wilson, W.E.; Stevens, R.K. (1975) The size of suspended
      particle matter in air. Science (Washington, DC) 183:  1098-1099.

Whitby, K.T.; Husar, R.B.; Liu, B.Y.H. (1972) The aerosol size distribution of Los Angeles
      smog.  J. Colloid. Interface Sci. 39:  177-204.

White and Roth.  Progressive Lung Injury in Pulmonary Hibrination from Monocrotaline.
      In: J.O. Cantor, ed. (Handbook of Am. Model of Pulmonary Disease, Vol. 2, Boca
      Raton, FL; CRC Press, 1989).

Whittemore, A. S.; Korn, E. L. (1980) Asthma and air pollution in the Los Angeles area.
      Am.  J. Public Health 70: 687-696.

Widdicombe, J.G. (1988) Vagal Reflexes in the Airways in: The airways neural control in
      Health and disease. Ed. A. Kaliner, P.J. Barnes.  Marcel Dekker  Inc.

-------
                                         R-37

Widdicombe, J.G.  (1974) Reflex control of breathing. Respiratory Physiology. Vol II
       Widdicombe, J.G. ed. University Park Press. Baltimore.

Willeke, K.; Whitby, K.T. (1975) Atmospheric aerosols: size distribution interpretation.  J.
       Air Pollut. Control Assoc. 25: 529-534.

Wilson, W.E.; Suh, H.H. (1996).  Fine and Coarse Particles: Concentration Relationships
       Relevant to Epidemiological Studies. J. Air Waste Manage.  Assoc.: accepted.

Wilson, W. E.; Suh, H. H. (1995) Differentiating fine and coarse particles: definitions and
       exposure relationships relevant to epidemiological studies. In: Schmidt-Ott, A., ed.
       Trends in aerosol research IV: new approaches in aerosol science and technology,
       proceedings of the seminar; January; Gerhard Mercator University, Duisburg,
       Germany. Duisburg, Germany: Gerhard Mercator University of Duisburg;  pp. 57-71.

Wolff, G.T. (1996a) Letter from George T. Wolff, Chair, Clean Air Scientific Advisory
       Committee (CAS AC) to Administrator Carol M. Browner.  Closure letter by CAS AC
       on draft Air Quality Criteria for Particulate Matter.  March 15,  1996.

Wolff, G.T. (1996b) Letter from George T. Wolff, Chair, Clean Air Scientific Advisory
       Committee (CASAC) to Administrator Carol M. Browner. Re: Clean Air Scientific
       Advisory Committee (Clean Air Scientific Advisory Committee (CASAC) Comments
       on the November  1995 Drafts of the Air Quality Criteria for Particulate Matter and
       the Review of the National Ambient Air Quality Standards for Particulate Matter:
       Policy Assessment of Scientific and Technical Information (OAQPS Staff Paper).
       Closure letter by CASAC on draft Air Quality Criteria for Particulate Matter.
       January 5, 1996.

Wolff, G.T. (1996c) Letter from George T. Wolff, Chair, Clean Air Scientific Advisory
       Committee (1996). Closure by the Clean Air Scientific Advisory Committee
       (CASAC) on the Staff Paper for Particulate Matter.  June 13, 1996.

Wolff, G.T.; Collins, D.C.; Rodgers,  W.R.; Verma,  M.H.;  Wong, C.A. (1990) Spotting of
       automotive finishes from the interactions between dry deposition of crustal  material
       and wet deposition of sulfate.  J. Air Waste Manage. Assoc.  40:  1638-1648.

Wolff, G.T.; Kelly, N.A.; Ferman, M.A.; Morrissey, M.L. (1983) Rural measurements of
       the chemical composition of airborne particles  in the eastern United States.  J.
       Geophys. Res. C: Ocean Atmos. 88:10,769-10,775.

Wolff, G.T.; Korsog, P.E.;  Stroup, D.P.; Ruthkosky, M.S.; Morrissey, M.L. (1985) The
       influence of local and regional sources on the concentration of inhalable paniculate
       matter in southeastern Michigan. Atmos. Environ.  19:305-313.

-------
                                         R-38

Wyzga, R. E.; Lipfert, F. W. (1995a) Ozone and daily mortality: the ramifications of
       uncertainties and interactions and some initial regression results. Presented at:
       AWMA specialty conference on tropospheric ozone; May  1994; Orlando, FL.
       Pittsburgh, PA: Air & Waste Management Association; in press.

Wyzga, R. E.; Lipfert, F. W. (1995b) Temperature-pollution  interactions with daily
       mortality in Philadelphia.  In: Paniculate matter: health and regulatory issues:
       proceedings of an international specialty conference; April; Pittsburgh, PA.
       Pittsburgh, PA: Air & Waste Management Association; pp.  3-42. (A&WMA
       publication VIP-49).

Yocom, I.E.; Grappone,  N. (1976) Effects of power plant emissions on materials.  Palo
       Alto, CA: Electric Power Research Institute: report no. EPRI/EC-139. Available
       from NTIS, Springfield, VA; PB-257 539.

Yocom, I.E.; Upham, J.B. (1977) Effects of economic materials and structures.  In:   Air
       pollution: v. 11, the effects of air pollution. 3rd ed. New York, NY: Academic Press,
       Inc.; pp. 93-94.

Yoshimura, I. (1990)  The effect of measurement  error on the dose-response curve.
       Environ. Health Perspect.  87: 173-178.

Zelikoff, J.T.; Sisco, M.P.; Yang, Z.;  Cohen, M.D.;  Schlesinger,  R.B. (1994)
       Immunotoxicity of sulfuric acid aerosol: effects on pulmonary macrophage effector
       and functional activities critical for maintaining host resistance against infectious
       diseases. Toxicology 92: 269-286.

Zelikoff, J.T.; Schlesinger, R.B. (1992)  Modulation of pulmonary immune defense
       mechanisms by  sulfuric acid: effects on macrophage-derived tumor necrosis  factor and
       superoxide.  Toxicology 76: 271-281.

-------
                                       TECHNICAL REPORT DATA
                                 (Please read Instructions on reverse before completing!
 \  REPORT NO
   EPA-452/R-96-013
                                                                          3  RECIPIENT'S ACCESSION NO
 4  TITLE AND SUBTITLE
 Review of the National Ambient Air Quality Standards
 for Participate Matter: Policy Assessment of Scientific and
 Technical Information -- OAQPS Staff Paper
                                               5 REPORT DATE
                                                 July  1996
                                               6 PERFORMING ORGANIZATION CODE
 7. AUTHOR(S)
 Bachmann, J.D.;  Caldwell, J.C.; Damberg, R.J.; Edwards, C.;
 Koman, T.; Martin, K.; Polkowsky, B.;  Richmond, H.M.;
 Smith, E.; Woodruff, T.
                                                                          8  PERFORMING ORGANIZATION REPORT NO
 9. PERFORMING ORGANIZATION NAME AND ADDRESS
 Office of Air and Radiation
 Office of Air Quality Planning and Standards
 U.S. Environmental Protection Agency
 Research Triangle Park, North Carolina 27711
                                                                          10  PROGRAM ELEMENT NO
                                               II. CONTRACT/GRANT NO.
 12. SPONSORING AGENCY NAME AND ADDRESS
                                               13 TYPE OF REPORT AND PERIOD COVERED
                                               Final
                                                                          14  SPONSORING AGENCY CODE
 15. SUPPLEMENTARY NOTES
 16. ABSTRACT
 This staff paper evaluates and interprets the updated scientific and technical information that EPA staff believes is most
 relevant to the review of primary and secondary national ambient air quality standards for paniculate matter (PM). This
 assessment is intended to bridge the gap between the scientific review in the 1996 cntena document and the judgements
 required of the Administrator in setting ambient air quality standards for PM.  The major staff recommendations presented in
 the staff paper for consideration by the Administrator include: (1) the current PM standards should be revised in light of
 evidence showing effects in areas that attain current NAAQS; (2) PM10 remains an appropriate indicator, but the fine
 (PM^and coarse fractions of PMW should be regulated separately; (3) two PM:.5 standards should be established: a 24-hour
 standard with a  more  robust form and a level selected from a range of 20-65 /ig/m3, and an annual expected mean standard
 selected from a  range of 12.5-20 /ig/m3; (4) consideration  should be given to the use of spatial averaging across multiple
 monitors for PM2J standards; (5) an  annual PM,0 standard should be retained for control of coarse fraction particles, alone or
 in combination with a 24-hour PMIO  standard; (6) the level of the annual standard should be selected from a range of 40-50
 /tg/m3; if a 24-hour standard is retained, the level should remain at 150 /xg/m3, but with a more robust form; and, (7)
 secondary standards for PM should be set equal to the primary standards to address  soiling and nuisance; consideration should
 be given to addressing remaining visibility impairment issues through regional haze regulations.             	
 17.
                                          KEY WORDS AND DOCUMENT ANALYSIS
                     DESCRIPTORS
                                                       b. IDENTIFIERS/OPEN ENDED TERMS
                                                                                               c. COSATI Field/Group
 Particulate Matter
 PM
 Air Pollution
 Health Effects
 Welfare Effects
Mortality
Morbidity
Exposure Assessment
 Risk Assessment
Air Quality Standards
  18 DISTRIBUTION STATEMENT
                                                       19 SECURITY CLASS (Report!
                                                                                               21  NO OF PAGES
                                                       20 SECURITY CLASS (Page)
                                                                                               22 PRICE
WA froi-m 2220-1 (Rev. 4-77)
                       PREVIOUS EDITION IS OBSOLETE

-------