United States                         May 2008

Agency menta' Pr°teCli°n                 EPA/600/R-08/070
  Integrated Science
  Assessment for Sulfur
  Oxides - Health Criteria
  Annexes
  (Second External Review Draft)

-------
                             Disclaimer

     This document is a second external review draft being released for review purposes
only and does not constitute U.S. Environmental Protection Agency policy. Mention of
trade names or commercial products does not constitute endorsement or recommendation
for use.

-------
                             Table of Contents
List of Tables	iii

List of Figures	v

Abbreviations and Acronyms	vi

Annex A. Literature Selection	A-1
   A.1. Literature Search and Retrieval	 A-1
   A.2. General Criteria for Study Selection	 A-1
      A.2.1. Criteria for Selecting Epidemiological Studies	A-2
      A.2.2. Criteria for Selecting Animal and Human Toxicological Studies	 A-4
   A.3. Evaluation Guidelines	 A-5
      A.3.1. Background  on Causality Decision Framework	 A-5
      A.3.2. Approaches  to the Determination of Causality	 A-7
      A.3.3. Surgeon General's Report: The Health Consequences of Smoking (CDC, 2004)	 A-7
      A.3.4. The EPA Guidelines for Carcinogen Risk Assessment	 A-10
      A.3.5. Improving the Presumptive Disability Decision-Making Process for Veterans	 A-14
      A.3.6. Guidelines for Formulation of Scientific Findings to be Used for Policy Purposes	 A-19
         A.3.6.1. International Agency for Research on Cancer Guidelines for Scientific Review
                 and Evaluation	 A-20
         A.3.6.2. National Toxicology Program Criteria	 A-27

Annex B. Additional Information on the Atmospheric Chemistry of Sulfur Oxides	 B-1
   B.1. Introduction	 B-1
      B.1.1. Multiphase Chemical Processes Involving Sulfur Oxides and Halogens	 B-3
      B.1.2. Mechanisms for the Aqueous  Phase Formation of Sulfate	 B-4
      B.1.3. Multiphase Chemical Processes Involving Sulfur Oxides and Ammonia	 B-5
   B.2. Transport of Sulfur Oxides in the Atmosphere	 B-7
   B.3. Emissions of SO2	 B-7
   B.4. Methods Used to  Calculate SOX and Chemical Interactions in the Atmosphere	 B-14
   B.5. Chemical-transport Models	 B-15
      B.5.1. Regional Scale Chemical-transport Models	 B-16
      B.5.2. Global-scale CTMs	 B-22
      B.5.3. Modeling the Effects of Convection	 B-23
      B.5.4. CTM Evaluation	 B-24
   B.6. Sampling and Analysis of Sulfur Oxides	 B-25
      B.6.1. Sampling and Analysis for SO2	 B-25
         B.6.1.1. Other  Techniques for Measuring SO2	 B-26
      B.6.2. Sampling and Analysis for Sulfate, Nitrate, and Ammonium	 B-27

Annex C. Modeling Human  Exposure	 C-32
   C.1. Introduction	 C-32
   C.2. Population Exposure Models: Their Evolution and Current  Status	C-37
   C.3. Characterization of Ambient Concentrations of SO2 and Related Air Pollutants 	 C-41
   C.4. Characterization of Microenvironmental Concentrations	C-42
      C.4.1. Characterization of Activity Events	 C-44
      C.4.2. Characterization of Inhalation Intake and Uptake	 C-44

Annex D. Controlled Human Exposure	 D-1

Annex E. Toxicological Studies	 E-3

Annex F. Epidemiological Studies	F-1
May 2008                                  i                 DRAFT—DO NOT QUOTE OR CITE

-------
                             List of Tables
Table B-1
Table B-2
Table B-3
Atmospheric lifetimes of SO2 and reduced sulfur species with respect to
reaction with OH, NO3, and Cl radicals.	
                                                                                     B-1
Relative contributions of various reactions to the total S(IV) oxidation rate
within a sunlit cloud, 10 min after cloud formation.	
Emissions of NOX, ammonia, and SO2 in the U.S. by source and
category, 2002.
                                                                                     B-4
                                                                                     B-9
Table C-1.
Table D-1
Table D-2
Table E-1.
Table E-2.
Table E-3.
Table E-4.
Table E-5.
Table E-6.
Table E-7.
Table E-8.
Table E-9.
Table E-1 0.
Table E-1 1.
Table E-1 2.
Table E-1 3.
Table E-1 4.
Table E-1 5.
Table E-1 6.
Table E-1 7.
Table E-1 8.
Table E-1 9.
Table E-20.
Table E-21 .
Table E-22.
Table E-23.
Table E-24.
Table F-1 .
The Essential Attributes of the pNEM, HAPEM, APEX, SHEDS, and
MENTOR-1A
Effects of medications on SO2-induced changes in lung function among
human subjects.
Summary of new studies of controlled human exposure to SO?.
Physiological effects of SO? exposure.
Inflammatory responses following SO? exposure.
Effects of SO? exposure on host lung defenses.
Effects of SO? exposure on hypersensitivity/allergic reactions.
Effects of SO? exposure on cardiovascular endpoints.
Neurophysiology and biochemistry effects of SO? and derivatives.
Reproductive and developmental effects of SO?.
Hematological effects of SO?.
Endocrine system effects of SO?.
Effects of SO? exposure on respiratory system morphology.
Carcinogenic effects of SO?.
Respiratory system biochemistry effects of SO?.
Respiratory system effects of SO? in disease models.
Effects of SO? layered on metallic or carbonaceous particles.
Effects of sulfite and mixtures of sulfite and sulfate.
Effects of mixtures containing SO? and ozone.
Effects of SO? and sulfate mixtures.
Effects of actual or simulated air pollution mixtures.
Effects of meteorological conditions on SO? effects.
In vitro or ex vivo respiratory system effects of SO? and metabolites.
Genotoxic effects of SO? and metabolites.
Liver and gastrointestinal effects of SO?.
Renal effects of SO?.
Lymphatic system effects of SO? and SO? mixtures.
Associations of short-term exposure to SO2 with respiratory morbidity in
field/panel studies.
C-39
D-1
D-1
E-3
E-4
E-4
E-5
E-6
E-7
E-10
E-11
E-1 2
E-1 2
E-1 3
E-1 4
E-1 5
E-1 6
E-1 8
E-1 9
E-20
E-21
E-23
E-24
E-25
E-27
E-28
E-28
F-1
May 2008
                                            DRAFT—DO NOT QUOTE OR CITE

-------
Table F-2.

Table F-3.

Table F-4.

Table F-5.
Table F-6.
Table F-7.

Table F-8.

Table F-9.
Table F-10.
Associations of short-term exposure to SO2 with emergency department
visits and hospital admissions for respiratory diseases	
Associations of short-term exposure to SO2 with cardiovascular morbidity
in field/panel studies.	
Associations of short-term exposure to SO2 with emergency department
visits and hospital admissions for cardiovascular diseases.	
Associations of short-term exposure to SO2 with mortality.	
.F-18

.F-58

.F-64
.F-77
 F-88
Associations of long-term exposure to SO2 with respiratory morbidity.	
Associations of long-term exposure to SO2 with lung cancer incidence
and mortality.	F-100
Associations of long-term exposure to SO2 with prenatal and neonatal
outcomes.	F-102
Associations of long-term exposure to SO2 with mortality.	F-108
Associations of long-term exposure to SO2 with lung cancer.	F-112
May 2008
                             IV
                                               DRAFT—DO NOT QUOTE OR CITE

-------
                             List  of  Figures
Figure A-1.
Figure A-2.

Figure A-3.
Figure A-4.

Figure A-5.
Figure B-1
Figure B-2
Figure B-3

Figure C-1.
Selection process for studies included in the ISA._
  A-3
Focusing on unmeasured confounders/covariates, or other sources of
spurious association from bias.	
Example posterior distribution for the determination of Sufficient._
Example posterior distribution for the determination of Equipoise and
Above.	
Example posterior distribution for the determination of Against.	
Transformations of sulfur compounds in the atmosphere.	
Comparison of aqueous-phase oxidation paths.	
.A-16
.A-17

.A-17
.A-18
_B-2
  B-5
Sulfate wet deposition (mg(S)m"2yr1) of the mean model versus
measurements for the North American Deposition Program (NADP)
network.	B-24
Schematic description of a general framework identifying the processes
(steps or components) involved in assessing inhalation exposures and
doses for individuals and populations.	C-36
May 2008
                                             DRAFT—DO NOT QUOTE OR CITE

-------
              Abbreviations  and Acronyms

                    8-OHdG       8-hydroxy-2D-deoxyguanosine
                    AHH          aryl hydrocarbon hydroxylase
                    ALT          alanine-amino-transferase
                    AM           alveolar macrophages
                    AMMN        N-nitroso-acetoxymethylmethylamine
                    AP           alkaline phosphatase
                    AST          aspartate-amino-transferase
                    B[a]P         benzo[a]pyrene
                    BAL          bronchoalveolar lavage
                    BC           black carbon
                    BHPN        N-bis(2-hydroxypropyl) nitrosamine
                    bw           body weight
                    C            carbon or carbon black particles
                    CA           chromosome aberrations
                    CAT          catalase
                    Choi          cholesterol
                    CMD         count median diameter
                    CYP          Cytochrome P450
                    Dae          aerodynamic diameterAM      alveolar macrophage
                    DEcCBP      DEP extract coated carbon black particles
                    DEN          diethylnitrosamine
                    DEP          diesel exhaust  particles
                    DEP+C       diesel exhaust  particle extract adsorbed to C
                    dG           2D-deoxyguanosine
                    DMBA        7, 12-dimethylbenzanthracene
                    DMSO        dimethyl sulfoxide
                    EC           elemental carbon
                    FHLC        fetal hamster lung cells
                    GCS          y-Qlutarnylcysteine synthetase
                    GPx          glutathione peroxidase
                    GPx          Se-dependent glutathione peroxidase
                    GPx          glutathione peroxidase
                    GRed        glutathione reductase
                    GSD          geometric standard deviation
                    GSH          glutathione
                    GSSG        glutathione disulfide
                    GSSO3H      glutathione S-sulfonate
                    GST          glutathione-S-transferase
                    GT           v-Qlutarnyl transpeptidase
                    HP           hydrolyzed protein
                    HVA-ICa      high-voltage activated calcium currents
                    IgG           immunoglobulin
May 2008
VI
                 DRAFT—DO NOT QUOTE OR CITE

-------
                     LDH           lactate dehydrogenase
                     MAD          median aerodynamic diameter
                     MMAD         mass median aerodynamic density
                     MMD          mass median diameter
                     MN            micronuclei
                     MNPCE        micronucleated PCE
                     Mo            molybdenum
                     NDMA         N-nitroso-dimethylamine
                     NMBzA        N-nitrosomethylbenzylamine
                     NR            Not Reported
                     OC            organic carbon
                     PAH           polycyclic aromatic hydrocarbons
                     PCE           polychromatic erythrocytes
                     PEC           pulmonary endocrine cells
                     PKA           cyclic AMP-dependent protein kinaseA
                     PKI            synthetic peptide inhibitor of PKA
                     PL            phospholipids
                     PNC           particle number concentration
                     RBC           red blood cell or erythrocyte
                     RH            relative humidity
                     SCE           sister chromatid exchanges
                     SEPs          somatosensory-evoked potentials
                     SOD           superoxide dismutase
                     SPF           specific pathogen free
                     SPM           suspended particulate matter extract
                     SQCA         squamous cell carcinoma
                     SSO           seabuckthorn seed oil
                     SV40          simian virus 40
                     TEARS        thiobarbituric acid-reactive substance
                     TOC           potassium channel transient outward currents
                     TTX           tetrodotoxin
                     TTX-R         tetrodotoxin-resistant
                     TTX-S         tetrodotoxin-sensitive
                     VE            ventilation rate
                     VEPs          visual-evoked potentials
                     W            tungsten
May 2008
VII
                  DRAFT—DO NOT QUOTE OR CITE

-------
                      Annex A.  Literature Selection

 1        Annex A of this second draft ISA includes detailed information on the methods used to
 2   identify and select studies, and on frameworks for evaluating scientific evidence relative to
 3   causality determination. While the overarching framework is outlined in the introduction to
 4   Chapter 1, this Annex provides supporting information for that framework, including excerpts
 5   from decision frameworks or criteria developed by other organizations.

     A.1. Literature Search and Retrieval
 6        Literature searches are conducted continuously, to identify studies published since the last
 7   review. The current review includes studies published subsequent to the 1982 AQCD for Sulfur
 8   Oxides (U.S. 1986). Search strategies are iteratively modified in an effort to optimize the
 9   identification of pertinent publications. Additional publications are identified for inclusion in
10   several ways: review of pre-publication tables of contents for journals in which relevant papers
11   may be published; independent identification of relevant literature by expert authors; and
12   identification by the public and CAS AC during the external review process. Generally, only
13   information that has undergone scientific peer review and has been published, or accepted for
14   publication, in the open literature is considered. Studies identified are further evaluated by EPA
15   staff and outside experts to determine if they merit inclusion. Criteria used for study selection are
16   summarized below.

     A.2. General Criteria for Study Selection
17        In assessing the scientific quality and relevance of epidemiological and animal or  human
18   toxicological studies, the following considerations have been taken into account.

19        •  Were the study populations adequately selected and are they sufficiently  well defined to
20           allow for meaningful comparisons between study groups?

21        •  Are the statistical analyses appropriate, properly performed, and properly interpreted?
     May 2008                             A-1               DRAFT—DO NOT QUOTE OR CITE

-------
 1         •  Are likely covariates (i.e., potential confounders or effect modifiers) adequately
 2            controlled or taken into account in the study design and statistical analysis?

 3         •  Are the reported findings internally consistent, biologically plausible, and coherent in
 4            terms of consistency with other known facts?

 5         "To what extent are the aerometric data, exposure, or dose metrics of adequate quality
 6            and sufficiently representative to serve as indicators of exposure to ambient SC>2?

 7    Consideration of these issues informs our judgments on the relative quality of individual studies
 8    and allows us to focus the assessment on the most pertinent studies.

      A.2.1.  Criteria for Selecting Epidemiological Studies
 9         In selecting epidemiological studies for this assessment, EPA considered whether a given
10    study contains information on (1) associations with measured sulfur oxides concentrations using
11    short- or long-term exposures at or near ambient levels of sulfur oxides, (2) health effects of
12    specific  sulfur oxides species or indicators related to sulfur oxides sources (e.g., combustion-
13    related particles), (3) health endpoints and populations not previously extensively researched, (4)
14    multiple pollutant analyses and other approaches to address issues related to potential
15    confounding and modification of effects, and/or (5) important methodological issues (e.g., lag of
16    effects, model specifications, thresholds, mortality displacement) related to interpretation of the
17    health evidence. Among the epidemiological studies, particular emphasis has been placed on
18    those most relevant to reviews of the NAAQS. Specifically, studies conducted in the United
19    States or Canada may be discussed in more detail than those from other geographic regions.
20    Particular emphasis has been placed oN: (A) recent multeity studies that employ standardized
21    methodological analyses for evaluating effects of sulfur oxides and that provide overall estimates
22    for effects based  on combined analyses of information pooled across multiple cities, (B) recent
23    studies that provide quantitative effect estimates for populations of interest, and (C)  studies that
24    consider sulfur oxides as a component of a complex mixture of air pollutants.
      May 2008                               A-2               DRAFT—DO NOT QUOTE OR CITE

-------
                               Identification of Studies for Inclusion in the ISA
                   Continuous,
                   comprehensive
                   literature review
                   of peer-reviewed
                   journal articles
                  Informative studies
                    are identified.
                                                Studies added
                                                to the docket
                                                during public
                                                comment period.
        Studies identified
        during EPA
        sponsored kickoff
        meeting (including
        studies in
        preparation).
       Studies that do
       not address
       exposure and/or
       effects of air
       pollutant(s) under
       review are
       excluded.
Selection of
studies
discussed and
additional studies
identified during
CASAC peer
review of draft
document.
        INFORMATIVE studies are well-designed/^
          properly implemented, thoroughly
          described.

          HIGHLY INFORMATIVE studies reduce
          uncertainty on critical issues, may include
          analyses of confounding or effect
          modification by copollutants or other
          variables, analyses of concentration-
          response or dose-response relationships,
          analyses related to time between
          exposure and response, and offer
          innovation in method or design.

          POLICY-RELEVANT studies may include
          those conducted at or near ambient
          concentrations and studies conducted in
        i U.S. and Canadian airsheds.
                         tudies are
                       evaluated for
                    inclusion in the ISA
                       and included
                          in the
                         Annexes
Studies
summarized
in figures are
included
because they are
sufficiently
comparable
to be displayed
together. A study
highlighted in the
text may not
appear in a
summary figure.
A figure from a
highlighted study
maybe
reproduced
in its entirely.
Studies
REFERENCED
in the text include
those that provide
a basis for or
describe
the association
between the
criteria pollutant
and effects. In
addition, policy
relevant
and highly
informative
studies
are discussed
in the text.
Figure A-1.  Selection process for studies included in the ISA.
May 2008
          A-3
              DRAFT—DO NOT QUOTE OR CITE

-------
 1         Not all studies were accorded equal weight in the overall interpretive assessment of
 2    evidence regarding SCVassociated health effects. Among studies with adequate control for
 3    confounding, increasing scientific weight is accorded in proportion to the precision of their effect
 4    estimates. Small-scale studies without a wide range of exposures generally produce less precise
 5    estimates compared to larger studies with a broad exposure gradient. For time-series studies, the
 6    size of the study, as indicated by the duration of the study period and total  number of events, and
 7    the variability of SC>2 exposures are important components that help to determine the precision of
 8    the health effect estimates. In evaluating the epidemiologic evidence in this chapter, more weight
 9    is accorded to estimates from studies with narrow confidence bands.
10         The goal was to perform a balanced and objective evaluation that summarizes, interprets,
11    and synthesizes the most important studies and issues in the epidemiologic database pertaining to
12    sulfur oxides exposure, illustrated by using newly created or previously  published summary
13    tables and figures. For each study presented, the quality of the exposure and outcome data, as
14    well as the quality of the statistical analysis methodology, are discussed. The discussion
15    incorporates the magnitude and statistical strengths of observed associations between SC>2
16    exposure and health outcomes.

      A.2.2. Criteria for Selecting Animal and Human Toxicological Studies
17         Criteria for the selection of research evaluating animal toxicological or controlled human
18    exposure studies included a focus on those studies  conducted at levels within about an order of a
19    magnitude of ambient SO2 concentrations and those studies that approximated expected human
20    exposure conditions in terms of concentration and duration. Studies that elucidate mechanisms of
21    action and/or susceptibility, particularly if the studies were conducted under atmospherically
22    relevant conditions, were emphasized whenever possible.
23         The selection of research evaluating controlled human exposures to sulfur oxides was
24    mainly limited to studies in which subjects were exposed to < 5 ppm 862.  For these controlled
25    human exposures, emphasis was placed on studies  that  (1) investigated potentially susceptible
26    populations such as asthmatics, particularly studies that compared  responses in susceptible
27    individuals with those in age-matched healthy controls; (2) addressed issues such as
28    concentration-response or time-course of responses; (3) investigated  exposure to SC>2 separately
      May 2008                              A-4                DRAFT—DO NOT QUOTE OR CITE

-------
 1    and in combination with other pollutants such as 63 and NO2; (4) included control exposures to
 2    filtered air; and (5) had sufficient statistical power to assess findings.

      A.3. Evaluation  Guidelines

      A.3.1. Background on Causality Decision Framework
 3         The critical assessment of health evidence presented in that ISA was conceptually based
 4    upon consideration of salient aspects of the evidence so as to reach fundamental judgments about
 5    the likely causal significance of the observed associations. It is appropriate to draw from those
 6    aspects initially presented in Hill's classic monograph (Hill, 1965) and widely used by the
 7    scientific community in conducting such evidence-based reviews. A number of these aspects
 8    were judged to be particularly salient in evaluating the body of evidence available in this review,
 9    including the aspects described by Hill as strength, experiment, consistency, plausibility, and
10    coherence. Other aspects identified by Hill, including temporality and biological gradient, were
11    also relevant and considered here (e.g., in characterizing lag structures and concentration-
12    response relationships), but were more directly addressed in the design and analyses of the
13    individual epidemiologic studies included in this assessment. (As noted below, Hill's remaining
14    aspects of specificity and analogy were not considered to be particularly salient in this
15    assessment.) As discussed below, these salient aspects were interrelated and considered
16    throughout the evaluation of the evidence presented in this chapter, and were more generally
17    reflected in the ISA.
18         In the following sections, the general evaluation of the strength of the epidemiological
19    evidence reflects consideration not only of the magnitude of reported sulfur oxides effects
20    estimates and their statistical significance, but also of the precision of the effects estimates and
21    the robustness of the effects associations. Consideration of the robustness of the associations
22    took into account a number of factors, including in particular the impact of alternative models
23    and model specifications and potential  confounding by copollutants, as well issues related to the
24    consequences of measurement error. Another aspect that is related to the strength of the  evidence
25    in this assessment was the availability of evidence from "found experiments," or so-called
26    intervention studies, which  have the potential to provide particularly strong support for making
27    causal inferences.

      May 2008                              A-5               DRAFT—DO NOT QUOTE OR CITE

-------
 1         Consideration of the consistency of the associations, as discussed in the following sections,
 2    involved looking across the results of multi- and single-city studies conducted by different
 3    investigators in different places and times. In this assessment, it is important to consider the
 4    aspect of consistency. Other relevant factors are also known to exhibit much variation across
 5    studies. These include, for example, the presence and levels of copollutants, the relationships
 6    between central measures of sulfur oxides and exposure-related factors, relevant demographic
 7    factors related to sensitive subpopulations, as well as climatic and meteorological conditions.
 8    Thus, in this case, consideration of consistency,  and the issue of related heterogeneity of effects,
 9    was appropriately understood as an evaluation of the similarity or general concordance of results,
10    rather than an expectation of finding quantitative results within a very narrow range. Particular
11    weight was given in this assessment,  consistent with Hill's views, to the presence of "similar
12    results reached in quite different ways, e.g., prospectively and retrospectively" (Hill, 1965). On
13    the other hand, in light of complexities of exposure and surrogate issues and its spatial and
14    temporal variations, Hill's specificity of effects and analogy aspects were not viewed as being
15    particularly salient here.
16         Looking beyond the epidemiological evidence, evaluation of the biological plausibility of
17    the associations observed in epidemiologic studies reflected consideration of both exposure-
18    related factors and dosimetric/toxicologic evidence relevant to identification of potential
19    biological mechanisms. Similarly, consideration of the coherence of health effects associations
20    reported in the epidemiologic literature reflected broad consideration of information pertaining to
21    the nature of the various respiratory-  and cardiac-related mortality and morbidity effects and
22    biological markers evaluated in toxicologic and epidemiologic studies.
23         In identifying these aspects as being particularly salient in this assessment, it is also
24    important to recognize that no one aspect was either necessary or sufficient for drawing
25    inferences of causality. As Hill emphasized:

26              "None of my nine viewpoints can bring indisputable evidence for or
27              against the cause-and-effect hypothesis and none can be required as a sine
28              qua non. What they can do, with greater or less strength, is to help us to
29              make up our minds on the fundamental question - is there any other way
30              of explaining the set of facts before us, is there any other answer equally,
31              or more, likely than cause and effect?"
      May 2008                                A-6                DRAFT—DO NOT QUOTE OR CITE

-------
 1         Thus, while these aspects frame considerations were weighed in assessing the
 2   epidemiologic evidence, they do not lend themselves to being considered in terms of simple
 3   formulas or hard-and-fast rules of evidence leading to answers about causality (Hill, 1965). One,
 4   for example, cannot simply count up the numbers of studies reporting statistically significant
 5   results for sulfur oxides and health endpoints evaluated in this assessment and reach credible
 6   conclusions about the relative strength of the evidence and the likelihood of causality. Rather,
 7   these important considerations were taken into account throughout this assessment with a goal of
 8   producing an objective appraisal of the evidence (informed by peer and public comment and
 9   advice), which included the weighing of alternative views on controversial issues.

     A.3.2. Approaches to  the Determination of Causality
10         The following sections  include excerpts from several reports that have documented
11   approaches for the determination of causality, or related decision-making processes. These
12   sections provide supplementary documentation of approaches that are similar in nature to EPA's
13   framework for evaluation of health evidence.

     A.3.3. Surgeon General's Report: The Health  Consequences of
     Smoking (CDC,  2004)
14         The Surgeon General's  Report (U.S. Surgeon General, 2004) evaluated the health effects of
15   smoking; it built upon the first Surgeon General's report published in 1964 (U.S. Surgeon
16   General, 1964). It also updated the methodology for evaluating evidence that was first presented
17   in the 1964 report. The 2004 report acknowledged the effectiveness of the previous methodology,
18   but attempted to standardize the language surrounding causality of associations.
19         The Surgeon General's  Reports on Smoking played a central role in the translation of
20   scientific evidence into policy. As such, it is important that scientific evidence was presented in a
21   manner that conveys most succinctly the link between  smoking and a health  effect. Specifically,
22   the report stated:

23             The statement that an exposure "causes" a disease in humans represents a
24             serious claim, but  one that carries with it the possibility of prevention.
25             Causal determinations may also carry substantial economic implications
26             for society and for those who might be held responsible for the exposure
27             or for achieving its prevention.

     May 2008                              A-7               DRAFT—DO NOT QUOTE OR CITE

-------
 1         To address the issue of identifying causality, the 2004 report provided the following
 2    summary of the earlier 1964 report:

 3              When a relationship or an association between smoking... and some
 4              condition in the host was noted, the significance of the association was
 5              assessed.

 6              The characterization of the assessment called for a specific term. ... The
 7              word cause is the one in general usage in connection with matters
 8              considered in this study, and it is capable of conveying the notion of a
 9              significant, effectual relationship between an agent and an associated
10              disorder or disease in the host.

11              No member was so naive as to insist upon mono-etiology in pathological
12              processes or in vital phenomena. All were thoroughly aware...  that the end
13              results are the net effect of many actions and counteractions.

14              Granted that these complexities were recognized, it is to be noted clearly
15              that the Committee's considered decision to use the words "a cause,"  or "a
16              major cause," or "a significant cause," or "a causal association" in certain
17              conclusions about smoking and health affirms their conviction  (U.S.
18              Surgeon General, 2004, p. 21).

19    This 2004 report created uniformly labeled conclusions that were used throughout the document.

20    The following excerpts from the report also include a description of the methodology and the
21    judgments used to reach a conclusion:

                Terminology of Conclusions and Causal Claims

22              The first step in introducing this revised approach is to outline the
23              language that will be used for summary conclusions regarding causality,
24              which follows hierarchical language used by Institute of Medicine
25              committees (IOM, 2007) to couch causal conclusions, and by IARC to
26              classify carcinogenic substances (IARC, 2006). These entities use a four-
27              level hierarchy for classifying the strength of causal inferences based  on
28              available evidence as follows:

29              Evidence is sufficient to infer a causal relationship.

30              Evidence is suggestive but not sufficient to infer a causal relationship.

31              Evidence is inadequate to infer the presence or absence of a causal
32              relationship (which encompasses evidence that is sparse, of poor quality,
33              or conflicting).
      May 2008                               A-8                DRAFT—DO NOT QUOTE OR CITE

-------
 1              Evidence is suggestive of no causal relationship.

 2              For this report, the summary conclusions regarding causality are expressed
 3              in this four-level classification. Use of these classifications should not
 4              constrain the process of causal inference, but rather bring consistency
 5              across chapters and reports, and greater clarity as to what the final
 6              conclusions are actually saying. As shown in Table 1.1 [see original
 7              document], without a uniform classification the precise nature of the final
 8              judgment may not always be obvious, particularly when the judgment is
 9              that the evidence falls below the "sufficient" category. Experience has
10              shown that the "suggestive" category is often an uncomfortable one for
11              scientists,  since scientific culture is such that any evidence that falls short
12              of causal proof is typically deemed inadequate to make a causal
13              determination. However, it is very useful to distinguish between evidence
14              that is truly inadequate versus that which just falls short of sufficiency.

15              There is no category beyond "suggestive of no causal relationship" as it is
16              extraordinarily difficult to prove the complete absence of a causal
17              association. At best, "negative"  evidence is suggestive, either strongly or
18              weakly. In instances where this  category is used, the strength of evidence
19              for no relationship will be indicated in the body of the text. In this new
20              framework, conclusions regarding causality will be followed by a section
21              on implications. This section will separate the issue of causal inference
22              from recommendations for research, policies, or other actions that might
23              arise from the causal conclusions. This section will assume a public health
24              perspective, focusing on the population consequences of using or not
25              using tobacco and also a scientific perspective, proposing further research
26              directions. The proportion of cases in the population as a result of
27              exposure (the population attributable risk), along with the total prevalence
28              and seriousness of a disease, are more relevant for deciding on actions
29              than the relative risk estimates typically used for etiologic determinations.
30              In past reports, the failure to sharply separate issues of inference from
31              policy issues resulted in inferential statements that were sometimes
32              qualified with terms for action. For example, based on the evidence
33              available in 1964, the first Surgeon General's report on smoking and
34              health contained the following statement about the relationship between
35              cardiovascular diseases and smoking:

36              It is established that male cigarette smokers have a higher death rate from
37              coronary artery disease than non-smoking males. Although the causative
38              role of cigarette smoking in deaths from coronary  disease is not proven,
39              the Committee considers it more prudent from the public health viewpoint
40              to assume that the established association has causative meaning, than to
41              suspend judgment until no uncertainty remains (U.S. Surgeon General,
42              2004, p. 32).
      May 2008                               A-9                DRAFT—DO NOT QUOTE OR CITE

-------
 1             Using this framework, this conclusion would now be expressed
 2             differently, probably placing it in the "suggestive" category and making it
 3             clear that although it falls short of proving causation, this evidence still
 4             makes causation more likely than not. The original statement makes it
 5             clear that the 1964 committee judged that the evidence fell short of
 6             proving causality but was sufficient to justify public health action. In this
 7             report, the rationale and recommendations for action will be placed in the
 8             implications section, separate from the causal conclusions. This separation
 9             of inferential from action-related statements clarifies the degree to which
10             policy recommendations are driven by the strength of the evidence and by
11             the public health consequences acting to reduce exposure. In addition, this
12             separation appropriately reflects the differences between the processes and
13             goals of causal inference and decision making.

      A.3.4. The  EPA Guidelines for Carcinogen Risk Assessment

14         The EPA Guidelines for Carcinogen Risk Assessment, published in 2005 (U.S. EPA, 2005),

15    was an update to the previous risk assessment document published in 1986. This document

16    served to guide EPA staff and public about the Agency's risk assessment development and

17    methodology. In the 1986 Guidelines, a step-wise approach was used to evaluate the scientific

18    findings. However, this newer document was similar to the Surgeon General's Report on

19    Smoking in that it used single integrative  step after assessing all of the individual lines of

20    evidence. Five standard descriptors were used to evaluate the weight of evidence:

           1.  Carcinogenic to Humans

           2.  Likely to Be Carcinogenic to Humans

           3.  Suggestive Evidence  of Carcinogenic Potential

           4.  Inadequate Information to Assess Carcinogenic Potential

           5.  Not Likely to Be Carcinogenic to Humans.

21         The 2005 Guidelines recommend that a separate narrative be prepared on the weight of

22    evidence and the descriptor. The Guidelines further recommend that the descriptors should only

23    be used in the context of a weight-of-evidence discussion.

24         The following excerpt describes how a weight of evidence narrative should be developed

25    and a how a descriptor should be selected (U.S. EPA, 2005):
      May 2008                              A-10               DRAFT—DO NOT QUOTE OR CITE

-------
 1              The weight of the evidence should be presented as a narrative laying out
 2              the complexity of information that is essential to understanding the hazard
 3              and its dependence on the quality, quantity, and type(s) of data available,
 4              as well as the circumstances of exposure or the traits of an exposed
 5              population that may be required for expression of cancer. For example, the
 6              narrative can clearly state to what extent the determination was based on
 7              data from human exposure, from animal experiments, from some
 8              combination of the two, or from other data. Similarly, information on
 9              mode of action can specify to what extent the data are from in vivo or in
10              vitro exposures or based on similarities to other chemicals. The extent to
11              which an agent's mode of action occurs only on reaching a minimum dose
12              or a minimum duration should also be presented.  A hazard might also be
13              expressed disproportionately in individuals possessing a specific gene;
14              such characterizations may follow from a better understanding of the
15              human genome. Furthermore, route of exposure should be used to qualify
16              a hazard if, for example, an agent is not absorbed by some routes.
17              Similarly, a hazard can be attributable to exposures during a susceptible
18              lifestage on the basis of our understanding of human development.

19              The weight of evidence-of-evidence narrative should highlight:

20                 •  the quality and quantity of the data;

21                 "all key decisions and the basis for these major decisions; and

22                 •  any data, analyses, or assumptions that are unusual for or new to EPA.

23              To capture this complexity, a weight of evidence  narrative generally
24              includes

25                 •  conclusions about human carcinogenic potential (choice of descriptor(s),
26                    described below)

27                 •  a summary of the key evidence  supporting these conclusions (for each
28                    descriptor used), including information on the type(s) of data (human and/or
29                    animal, in vivo and/or in vitro) used to support the conclusion(s)

30                 •  available information on the epidemiologic or experimental conditions that
31                    characterize expression of carcinogenicity (e.g., if carcinogenicity is possible
32                    only by one exposure route or only above a certain human exposure level),

33                 "a summary of potential modes of action and how they reinforce the
34                    conclusions,

35                 •  indications of any susceptible populations or lifestages, when available, and

36                 "a summary of the key default options invoked when the available information
37                    is inconclusive.
      May 2008                              A-11               DRAFT—DO NOT QUOTE OR CITE

-------
 1              To provide some measure of clarity and consistency in an otherwise free-
 2              form narrative, the weight of evidence descriptors are included in the first
 3              sentence of the narrative. Choosing a descriptor is a matter of judgment
 4              and cannot be reduced to a formula. Each descriptor may be applicable to
 5              a wide variety of potential data sets and weights of evidence. These
 6              descriptors and narratives are intended to permit sufficient flexibility to
 7              accommodate new scientific understanding and new testing methods as
 8              they are developed and accepted by the scientific community and the
 9              public. Descriptors represent points along a continuum of evidence;
10              consequently, there are gradations and borderline cases that are clarified
11              by the full narrative. Descriptors, as well as an introductory paragraph, are
12              a short summary of the complete narrative that preserves the complexity
13              that is an essential part of the hazard characterization. Users of these
14              cancer guidelines and of the risk assessments that result from the use
15              of these cancer guidelines should consider the entire range of
16              information included in the narrative rather than focusing  simply on
17              the descriptor

18              In borderline cases, the narrative explains the case for choosing one
19              descriptor and discusses the arguments for considering but not  choosing
20              another. For example, between "suggestive" and "likely" or between
21              "suggestive" and "inadequate," the explanation clearly communicates the
22              information needed to consider appropriately the  agent's carcinogenic
23              potential in subsequent decisions.

24              Multiple descriptors can be used for a single agent,  for example, when
25              carcinogenesis is dose- or route-dependent. For example, if an  agent
26              causes point-of-contact tumors by one exposure route but adequate testing
27              is negative by another route, then the agent could be described  as likely to
28              be carcinogenic by the first route but not likely to be carcinogenic by the
29              second. Another example is when the mode of action is sufficiently
30              understood to conclude that a key event in tumor  development  would not
31              occur below a certain dose range. In this case, the agent could be
32              described as likely to be carcinogenic above a certain dose range but not
33              likely to be carcinogenic below that range.

34              Descriptors can be selected for an agent that has not been tested in a
35              cancer bioassay if sufficient other information, e.g., toxicokinetic and
36              mode of action information, is available to make a strong, convincing, and
37              logical case through scientific inference. For example, if an agent is  one of
38              a well-defined class of agents that are understood to operate through a
39              common mode of action and if that agent has the  same mode of action,
40              then in the narrative the untested agent would have  the same descriptor as
41              the class. Another example is when an untested agent's effects are
42              understood to be caused by a human metabolite, in which case  in the
43              narrative the untested agent could have the same descriptor as the
44              metabolite. As new testing methods are developed and used, assessments
      May 2008                              A-12               DRAFT—DO NOT QUOTE OR CITE

-------
 1              may increasingly be based on inferences from toxicokinetic and mode of
 2              action information in the absence of tumor studies in animals or humans.

 3              When a well-studied agent produces tumors only at a point of initial
 4              contact, the descriptor generally applies only to the exposure route
 5              producing tumors unless the mode of action is relevant to other routes.
 6              The rationale for this conclusion would be explained in the narrative.

 7              When tumors occur at a site other than the point of initial contact, the
 8              descriptor generally applies to all exposure routes that have not been
 9              adequately tested at sufficient doses. An exception occurs when there is
10              convincing information, e.g., toxicokinetic data that absorption does not
11              occur by another route.

12              When the response differs qualitatively as well as quantitatively with dose,
13              this information should be part of the characterization of the hazard. In
14              some cases reaching a certain dose range can be a precondition for effects
15              to occur, as when cancer is secondary to another toxic effect that appears
16              only above a certain dose. In other cases exposure duration can be a
17              precondition for hazard if effects occur only after exposure is sustained for
18              a certain duration.  These considerations differ from the issues of relative
19              absorption or potency at different dose levels because they may represent
20              a discontinuity in a dose-response function.

21              When multiple bioassays are inconclusive, mode of action data are likely
22              to hold the key to resolution of the more appropriate descriptor. When
23              bioassays are few,  further bioassays to replicate a study's results or to
24              investigate the potential for effects in another sex, strain, or species may
25              be useful.

26              When there are few pertinent data, the descriptor makes a statement about
27              the database, for example, "Inadequate Information to Assess
28              Carcinogenic Potential," or a database that provides "Suggestive Evidence
29              of Carcinogenic Potential." With more information, the descriptor
30              expresses a conclusion about the agent's carcinogenic potential to humans.
31              If the conclusion is positive, the agent could be described as "Likely to Be
32              Carcinogenic to Humans" or, with strong evidence, "Carcinogenic to
33              Humans." If the conclusion is negative, the agent could be described as
34              "Not Likely to Be Carcinogenic to Humans."

35              Although the term  "likely" can have a probabilistic connotation in other
36              contexts, its use as a weight of evidence descriptor does not correspond to
37              a quantifiable probability of whether the chemical is carcinogenic. This is
38              because the data that support cancer assessments generally are not suitable
39              for numerical calculations of the probability that an agent is a carcinogen.
40              Other health agencies have expressed a comparable weight of evidence
      May 2008                               A-13               DRAFT—DO NOT QUOTE OR CITE

-------
 1              using terms such as "Reasonably Anticipated to Be a Human Carcinogen"
 2              (NTP) or "Probably Carcinogenic to Humans" (IARC, 1989).

      A.3.5. Improving the Presumptive  Disability Decision-Making Process
      for Veterans

 3         A recent publication by the Institute of Medicine also provided foundation for the causality

 4    framework adapted in this ISA (IOM, 2007). The Committee on Evaluation of the Presumptive

 5    Disability Decision-Making Process for Veterans was charged by the Veterans Association to

 6    describe how presumptive decisions are made for veterans with health conditions arising from

 7    military service currently, as well as recommendations for how such decisions could made in the

 8    future. The committee proposed a multiple-element approach that includes a quantification of the

 9    extent of disease attributable to an exposure. This process involved a review of all relevant data

10    to decide the strength  of evidence for causation, using one of four categories:

11          •   Sufficient:  the evidence is sufficient to conclude that a causal relationship exists.

12          •   Equipoise and Above: the evidence is sufficient to conclude that a causal relationship
13              is at least as likely as not, but not sufficient to conclude that a causal relationship
14              exists.

15          •   Below Equipoise: the evidence is not sufficient to conclude that a causal relationship
16              is at least as likely as not, or is not sufficient to make a scientifically informed
17             judgment.

18          •   Against: the evidence suggests the lack of a causal relationship.

19    The following is an excerpt from this report and describes these four categories in detail:

20              In light of the categorizations used by other health organizations and
21              agencies as well as considering the particular challenges of the
22              presumptive disability decision-making process, we propose a four-level
23              categorization of the strength of the overall evidence for or against a
24              causal relationship from exposure to disease.

25              We use the term "equipoise" to refer to the point at which the evidence is
26              in balance between favoring and not favoring causation. The term
27              "equipoise" is widely used in the biomedical literature, is a concept
28              familiar to  those concerned with evidence-based decision-making and is
29              used in VA processes for rating purposes as well as being a familiar term
30              in the veterans'community.
      May 2008                              A-14               DRAFT—DO NOT QUOTE OR CITE

-------
 1              Below we elaborate on the four-level categorization which the Committee
 2              recommends.

                Sufficient

 3              If the overall evidence for a causal relationship is categorized as
 4              Sufficient, then it should be scientifically compelling. It might include:

 5                  •      replicated and consistent evidence of a causal association: that is, evidence
 6                     of an association from several high-quality epidemiologic studies that cannot
 7                     be explained by plausible noncausal alternatives (e.g., chance, bias, or
 8                     confounding)

 9                  •      evidence of causation from  animal  studies and mechanistic knowledge

10                  •      compelling evidence from animal studies and strong mechanistic evidence
11                     from studies in exposed humans, consistent with (i.e., not contradicted by) the
12                     epidemiologic evidence.

13              Using the Bayesian framework to illustrate the evidential support and the
14              resulting state of communal scientific opinion  needed for reaching the
15              Sufficient category (and the lower categories that follow), consider again
16              the causal diagram in Figure A-2. In this model, used to help clarify
17              matters conceptually, the  observed association between exposure and
18              health is the result of: (1)  measured confounding, parameterized by a; (2)
19              the causal relation, parameterized by P; and (3) other, unmeasured sources
20              such as bias or unmeasured confounding, parameterized by y. The belief
21              of interest, after all the evidence has been weighed, is in the size of the
22              causal parameter p. Thus, for decision  making, what matters is how
23              strongly the evidence supports the proposition that P is above 0. As it is
24              extremely unlikely that the types of exposures considered for
25              presumptions reduce the risk of developing disease, we exclude values of
26              P below 0. If we consider the evidence as supporting degrees of belief
27              about the size of P, and we have a posterior distribution over the possible
28              size of P, then a posterior like Figure A-2 illustrates a belief state that
29              might result when the evidence for causation is considered Sufficient.

30              As the "mass" over a positive effect (the area under the curve to the right
31              of the zero) vastly "outweighs" the small mass over no effect (zero), the
32              evidence is considered sufficient to conclude that the association is causal.
33              Put another way, even though the scientific community might be uncertain
34              as to the size of P, after weighing all the evidence, it is highly confident
35              that the probability that P  is greater than zero is substantial; that is, that
36              exposure causes disease.
      May 2008                               A-15               DRAFT—DO NOT QUOTE OR CITE

-------
                Equipoise and Above

 1              To be categorized as Equipoise and Above, the scientific community
 2              should categorize the overall evidence as making it more confident in the
 3              existence of a causal relationship than in the non-existence of a causal
 4              relationship, but not sufficient to conclude causation.

 5              For example, if there are several high-quality epidemiologic studies, the
 6              preponderance of which show evidence of an association that cannot
 7              readily be explained by plausible noncausal alternatives (e.g., chance, bias,
 8              or confounding), and the causal relationship is consistent with the animal
 9              evidence and biological knowledge, then the overall evidence might be
10              categorized as Equipoise and Above.  Alternatively, if there is strong
11              evidence from animal studies or mechanistic evidence, not contradicted by
12              human or other evidence, then the overall evidence might be categorized
13              as Equipoise and Above. Equipoise is a common term employed by VA
14              and the courts in deciding disability claims (see Appendix D).
                                               Measured
                                          Confounders/Covariates
                                          I      '       I
                                     Exposure   	^.    Health
                                    to Substance     p     Outcome
                                                   Y
                                    Unmeasured Confounders/Covariates, or
                                 Other Sources of Spurious Association from Bias
      Figure A-2.  Focusing on unmeasured confounders/covariates, or other sources of spurious
      association from bias.
      May 2008                               A-16               DRAFT—DO NOT QUOTE OR CITE

-------
                                                                 Posterior Mass
                                                                 Over an Effect
                                      Size of the Causal Effect (3
      Figure A-3.  Example posterior distribution for the determination of Sufficient.

      Source: IOM (2007).
1
2
                Again, using the Bayesian model to illustrate the idea of Equipoise and
                Above, Figure A-4 shows a posterior probability distribution that is an
                example of belief compatible with the category Equipoise and Above.
                                                                 Posterior Mass
                                                                 Over an Effect
                                      Size of the Causal Effect 6
      Figure A-4.  Example posterior distribution for the determination of Equipoise and Above.

      Source: IOM (2007).
 4
 5
 6
 1
 8
 9
10
               In this figure, unlike the one for evidence classified as Sufficient, there is
               considerable mass over zero, which means that the scientific community
               has considerable uncertainty as to whether exposure causes disease at all;
               that is, whether P is greater than zero. At least half of the mass is to the
               right of the zero, however, so the community judges causation to be at
               least as likely as not, after they have seen and combined all the evidence
               available.
      May 2008
                                            A-17
DRAFT—DO NOT QUOTE OR CITE

-------
                Below Equipoise
 1
 2
 3
 5
 6
 7
 9

10

11
12

13
14
15
16
17
18
19
20
21
To be categorized as Below Equipoise, the overall evidence for a causal
relationship should either be judged not to make causation at least as likely
as not, or not sufficient to make a scientifically informed judgment.

This might occur:

    •      when the human evidence is consistent in showing an association, but the
       evidence is limited by the inability to rule out chance, bias, or confounding
       with confidence, and animal or mechanistic evidence is weak

    •      when animal evidence suggests a causal relationship, but human and
       mechanistic

    •      evidence is weak or inconsistent

    •      when mechanistic evidence is suggestive but animal and human evidence
       is weak or inconsistent
          when the evidence base is very thin.
                Against
To be categorized as Against, the overall evidence should favor belief that
there is no causal relationship from exposure to disease. For example, if
there is human evidence from multiple studies covering the full range of
exposures encountered by humans that are consistent in showing no causal
association, or there are is animal or mechanistic evidence supporting the
lack of a causal relationship, and combining all of the evidence results in a
posterior resembling Figure A-5 then the scientific community should
categorize the evidence as Against causation.
                            0
                       Size of the Causal Effect (3 •
                                                                Posterior Over (3

                                                                Posterior Mass
                                                                Over an Effect
      Figure A-5.  Example posterior distribution for the determination of Against.

      Source: IOM (2007).
      May 2008
                            A-18
DRAFT—DO NOT QUOTE OR CITE

-------
      A.3.6. Guidelines for Formulation of Scientific Findings to be Used for
      Policy Purposes

 1         The following guidelines in the form of questions were developed and published in 1991

 2    by the NAPAP Oversight Review Board for the National Acid Precipitation Assessment Program

 3    (Washington, 1991) to assist scientists in formulating presentations of research results to be used

 4    in policy decision processes.

 5              Is the statement sound? Have the central issues been clearly identified?
 6              Does each statement contain the distilled essence of present scientific and
 7              technical understanding of the phenomenon or process to which it applies?
 8              Is the statement consistent with all relevant evidence - evidence developed
 9              either through NAPAP research or through analysis of research conducted
10              outside of NAPAP? Is the statement contradicted by any important
11              evidence developed through research inside or outside of NAPAP? Have
12              apparent contradictions or interpretations of available  evidence been
13              considered in formulating the statement of principal findings?

14              Is the statement directional and, where appropriate, quantitative? Does the
15              statement correctly quantify both the direction and magnitude of trends
16              and relationships in the phenomenon or process to which the statement is
17              relevant? When possible, is a range of uncertainty given for each
18              quantitative result? Have various sources of uncertainty been identified
19              and quantified, for example, does the statement include or acknowledge
20              errors in actual measurements, standard errors of estimate, possible biases
21              in the availability of data, extrapolation of results beyond the
22              mathematical, geographical, or temporal relevancy of available
23              information, etc. In  short, are there numbers in the statement? Are the
24              numbers correct? Are the numbers relevant to the general meaning of the
25              statement?

26              Is the degree of certainty or uncertainty of the statement indicated clearly?
27              Have appropriate statistical tests been applied to the data used in drawing
28              the conclusion set forth in the statement? If the statement is based on a
29              mathematical or novel conceptual model, has the model or concept been
30              validated? Does the statement describe the model or concept on which it is
31              based and the degree of validity of that model or concept?

32              Is the statement correct without qualification? Are there limitations of
33              time, space, or other special circumstances in which the statement is true?
34              If the statement is true only in some circumstances, are these limitations
35              described adequately and briefly?

36              Is the statement clear and unambiguous? Are the words and phrases used
37              in the statement understandable by the decision makers of our society? Is
      May 2008                             A-19               DRAFT—DO NOT QUOTE OR CITE

-------
 1              the statement free of specialized jargon? Will too many people
 2              misunderstand its meaning?

 3              Is the statement as concise as it can be made without risk of
 4              misunderstanding? Are there any excess words, phrases, or ideas in the
 5              statement which are not necessary to communicate the meaning of the
 6              statement? Are there so many caveats in the statement that the statement
 7              itself is trivial, confusing, or ambiguous?

 8              Is the statement free of scientific or other biases or implications of societal
 9              value judgments? Is the  statement free of influence by specific schools of
10              scientific thought? Is the statement also free of words, phrases, or concepts
11              that have political, economic, ideological, religious, moral, or other
12              personal-, agency-, or organization-specific values, overtones, or
13              implications? Does the choice of how the statement is expressed rather
14              than its specific words suggest underlying biases or value judgments? Is
15              the tone impartial and free of special pleading? If societal value judgments
16              have been discussed, have these judgments been identified as such and
17              described both clearly and objectively?

18              Have societal implications been described objectively? Consideration of
19              alternative courses of action  and their consequences inherently involves
20              judgments of their feasibility and the importance of effects. For this
21              reason, it is important to ask if a reasonable range of alternative policies or
22              courses of action have been evaluated? Have societal implications of
23              alternative courses of action  been stated in the following general form?

24              "If this [particular option] were adopted then that [particular outcome]
25              would be expected."

26              Have the professional biases of authors and reviewers been described
27              openly? Acknowledgment of potential sources of bias is important so that
28              readers can judge for themselves the credibility of reports and
29              assessments.

      A.3.6.1. International Agency for Research on Cancer Guidelines for Scientific
      Review and Evaluation

30         The following is excerpted from the International Agency for Research on Cancer

31    Monographs on the evaluation of carcinogenic risks to humans (IARC, 2006)

32              The available  studies are summarized by the Working Group, with
33              particular regard to the qualitative aspects discussed below. In general,
34              numerical findings are indicated as they appear in the original report; units
35              are converted when necessary for easier comparison. The Working Group
36              may conduct additional analyses of the published data and use them in
      May 2008                              A-20              DRAFT—DO NOT QUOTE OR CITE

-------
 1              their assessment of the evidence; the results of such supplementary
 2              analyses are given in square brackets. When an important aspect of a study
 3              that directly impinges on its interpretation should be brought to the
 4              attention of the reader, a Working Group comment is given in square
 5              brackets.

 6              The scope of the IARC Monographs programme has expanded beyond
 7              chemicals to include complex mixtures, occupational exposures, physical
 8              and biological agents, lifestyle factors and other potentially carcinogenic
 9              exposures. Over time, the structure of a Monograph has evolved to include
10              the following sections:

11              1. Exposure data

12              2. Studies of cancer in humans

13              3. Studies of cancer in experimental animals

14              4. Mechanistic and other relevant data

15              5. Summary

16              6. Evaluation and rationale

17              In addition, a section of General Remarks at the front of the volume
18              discusses the reasons the agents were scheduled for evaluation and some
19              key issues the Working Group encountered  during the meeting.

20              This part  of the Preamble discusses the types of evidence considered and
21              summarized in each section of a Monograph, followed by the scientific
22              criteria that guide the evaluations.

                Evaluation and  rationale

23              Evaluations of the strength of the evidence for carcinogenicity arising
24              from human and experimental animal data are made, using  standard terms.
25              The strength of the mechanistic evidence is  also characterized.

26              It is recognized that the criteria for these evaluations, described below,
27              cannot encompass all of the factors that may be relevant to  an evaluation
28              of carcinogenicity. In considering all of the  relevant scientific data, the
29              Working Group may assign the agent to a higher or lower category than a
30              strict interpretation of these criteria would indicate.

31              These categories refer only to the strength of the evidence that an exposure
32              is carcinogenic and not to the extent of its carcinogenic activity (potency).
33              A classification may change as new information becomes available.
      May 2008                              A-21               DRAFT—DO NOT QUOTE OR CITE

-------
 1              An evaluation of the degree of evidence is limited to the materials tested,
 2              as defined physically, chemically or biologically. When the agents
 3              evaluated are considered by the Working Group to be sufficiently closely
 4              related, they may be grouped together for the purpose of a single
 5              evaluation of the degree of evidence.

 6              (a) Carcinogenicity in humans

 7              The evidence relevant to carcinogenicity from studies in humans is
 8              classified into one of the following categories:

 9              Sufficient evidence of carcinogenicity. The Working Group considers that
10              a causal relationship has been established between exposure to the agent
11              and human cancer. That is, a positive relationship has been observed
12              between the exposure and cancer in studies in which chance, bias and
13              confounding could be ruled out with reasonable confidence. A statement
14              that there is sufficient evidence is followed by a separate sentence that
15              identifies the target organ(s) or tissue(s) where an increased risk of cancer
16              was observed in humans. Identification of a specific target organ or tissue
17              does not preclude the possibility that the agent may cause cancer at  other
18              sites.

19              Limited evidence of carcinogenicity. A positive association has been
20              observed between exposure to the agent and cancer for which a causal
21              interpretation is considered by the Working Group to be credible, but
22              chance, bias or confounding could not be ruled out with reasonable
23              confidence.

24              Inadequate evidence of carcinogenicity. The available studies are of
25              insufficient quality, consistency or statistical power to permit a conclusion
26              regarding the presence or absence of a causal association between
27              exposure and cancer, or no data on cancer in humans are available.

28              Evidence suggesting lack of carcinogenicity. There are several  adequate
29              studies covering the full range of levels of exposure that humans are
30              known to encounter, which are mutually consistent in not showing a
31              positive association between exposure to the agent and any studied cancer
32              at any observed level of exposure. The results from these studies alone or
33              combined should have narrow confidence intervals with an upper limit
34              close to the null value (e.g. a relative risk of 1.0). Bias and confounding
35              should be ruled out with reasonable confidence, and the studies should
36              have an adequate length of follow-up. A conclusion of evidence
37              suggesting lack of carcinogenicity is inevitably limited to the cancer sites,
38              conditions and levels of exposure, and length of observation covered by
39              the available studies. In addition, the possibility of a very small risk at the
40              levels of exposure studied can never be excluded.
      May 2008                               A-22               DRAFT—DO NOT QUOTE OR CITE

-------
 1              In some instances, the above categories may be used to classify the degree
 2              of evidence related to carcinogenicity in specific organs or tissues.

 3              When the available epidemiological studies pertain to a mixture, process,
 4              occupation or industry, the Working Group seeks to identify the specific
 5              agent considered most likely to be responsible for any excess risk. The
 6              evaluation is focused as narrowly as the available data on exposure and
 7              other aspects permit.

 8              (b) Carcinogenicity in experimental animals

 9              Carcinogenicity in experimental animals can be evaluated using
10              conventional bioassays, bioassays that employ genetically modified
11              animals, and other in-vivo bioassays that focus on one or  more of the
12              critical stages of carcinogenesis. In the absence of data from conventional
13              long-term bioassays or from assays with neoplasia as the end-point,
14              consistently positive results in several models that address several stages
15              in the multistage process of carcinogenesis should be considered in
16              evaluating the degree of evidence of carcinogenicity in  experimental
17              animals.

18              The evidence relevant to carcinogenicity in experimental  animals is
19              classified into one of the following categories:

20              Sufficient evidence of carcinogenicity. The Working Group considers that
21              a causal relationship has been established between the agent and an
22              increased incidence of malignant neoplasms or of an appropriate
23              combination of benign and malignant neoplasms in (a) two or more
24              species of animals or (b) two or more independent studies in one species
25              carried out at different times or in different laboratories or under different
26              protocols. An increased incidence of tumours in both sexes of a single
27              species in a well-conducted study, ideally conducted under Good
28              Laboratory Practices, can also provide sufficient evidence.

29              A single study in one species and sex might be considered to provide
30              sufficient evidence of carcinogenicity when malignant neoplasms occur to
31              an unusual degree with regard to incidence, site, type of tumour or age at
32              onset, or when there are strong findings of tumours at multiple sites.

33              Limited evidence of carcinogenicity. The  data suggest a carcinogenic
34              effect but are limited for making a definitive evaluation because, e.g. (a)
35              the evidence of carcinogenicity is restricted to a single experiment; (b)
36              there are unresolved questions regarding the adequacy of the design,
37              conduct or interpretation of the studies; (c) the agent increases the
38              incidence only of benign neoplasms or lesions of uncertain neoplastic
39              potential; or (d) the evidence of carcinogenicity is restricted to studies that
      May 2008                               A-23               DRAFT—DO NOT QUOTE OR CITE

-------
 1              demonstrate only promoting activity in a narrow range of tissues or
 2              organs.

 3              Inadequate evidence of carcinogenicity. The studies cannot be interpreted
 4              as showing either the presence or absence of a carcinogenic effect because
 5              of major qualitative or quantitative limitations, or no data on cancer in
 6              experimental animals are available.

 7              Evidence suggesting lack of carcinogenicity: Adequate studies involving
 8              at least two species are available which show that, within the limits of the
 9              tests used, the agent is not carcinogenic. A conclusion of evidence
10              suggesting lack of carcinogenicity is inevitably limited to the species,
11              tumour sites, age at exposure, and conditions and levels of exposure
12              studied.

13              (c) Mechanistic and other relevant data

14              Mechanistic and other evidence judged to be relevant to an evaluation of
15              carcinogenicity and of sufficient importance to affect the overall
16              evaluation is highlighted. This may include data on preneoplastic lesions,
17              tumour pathology, genetic and related effects, structure-activity
18              relationships, metabolism and toxicokinetics, physicochemical parameters
19              and analogous biological agents.

20              The strength of the evidence that any carcinogenic effect observed is due
21              to a particular mechanism is evaluated, using terms such as 'weak,'
22              'moderate' or 'strong.' The Working Group then assesses whether that
23              particular mechanism is likely  to be operative in humans. The strongest
24              indications that a particular mechanism operates in humans derive from
25              data on humans or biological specimens obtained from exposed humans.
26              The data may be considered to be especially relevant if they show that the
27              agent in question has caused changes in exposed humans that are on the
28              causal pathway to carcinogenesis. Such data may, however, never become
29              available, because it is at least  conceivable that certain compounds may be
30              kept from human use solely on the basis of evidence of their toxicity
31              and/or carcinogenicity in experimental systems.

32              The conclusion that a mechanism operates in experimental animals is
33              strengthened by findings of consistent results in different experimental
34              systems, by the demonstration  of biological plausibility and by coherence
35              of the overall database. Strong support can be obtained from studies that
36              challenge the hypothesized mechanism experimentally, by demonstrating
37              that the suppression of key  mechanistic processes leads to the suppression
38              of tumour development. The Working Group considers whether multiple
39              mechanisms might contribute to tumour development, whether different
40              mechanisms might operate  in different dose ranges, whether separate
41              mechanisms might operate  in humans  and experimental animals and
      May 2008                              A-24               DRAFT—DO NOT QUOTE OR CITE

-------
 1              whether a unique mechanism might operate in a susceptible group. The
 2              possible contribution of alternative mechanisms must be considered before
 3              concluding that tumours observed in experimental animals are not relevant
 4              to humans. An uneven level of experimental support for different
 5              mechanisms may reflect that disproportionate resources have been focused
 6              on investigating a favoured mechanism.

 7              For complex exposures, including occupational and industrial exposures,
 8              the chemical composition and the potential contribution of carcinogens
 9              known to be present are considered by the Working Group in its overall
10              evaluation of human carcinogenicity. The Working Group also determines
11              the extent to which the materials tested in experimental systems are related
12              to those to which humans are exposed.

13              (d) Overall evaluation

14              Finally, the body of evidence is considered as a whole, in order to reach an
15              overall evaluation of the carcinogenicity of the agent to humans.

16              An evaluation may be made for a group of agents that have been evaluated
17              by the Working Group. In addition, when supporting data indicate that
18              other related agents, for which there is no direct evidence of their capacity
19              to induce cancer in humans or in animals, may also be carcinogenic, a
20              statement describing the rationale for this conclusion is added to the
21              evaluation narrative; an additional evaluation may be made for this
22              broader group of agents if the strength of the evidence warrants it.

23              The agent is described according to the wording of one of the following
24              categories, and the designated group is given. The categorization of an
25              agent is a matter of scientific judgement that reflects the strength of the
26              evidence derived from studies in humans and in experimental animals and
27              from mechanistic and other relevant data.

28              Group 1: The agent is carcinogenic to humans.

29                    This category is used when there is sufficient evidence of
30                    carcinogenicity in humans. Exceptionally, an agent may be placed
31                    in this category when evidence of carcinogenicity in humans is less
32                    than sufficient but there  is sufficient evidence of carcinogenicity in
33                    experimental animals and strong evidence in exposed humans that
34                    the agent acts through a  relevant mechanism of carcinogenicity.

35              Group 2.

36                    This category includes agents for which, at one extreme, the degree
37                    of evidence of carcinogenicity in humans is almost sufficient, as
38                    well as those for which,  at the other extreme, there are no human
39                    data but for which there  is evidence of carcinogenicity in
      May 2008                              A-25              DRAFT—DO NOT QUOTE OR CITE

-------
 1                    experimental animals. Agents are assigned to either Group 2A
 2                    (probably carcinogenic to humans) or Group 2B (possibly
 3                    carcinogenic to humans) on the basis of epidemiological and
 4                    experimental evidence of carcinogenicity and mechanistic and
 5                    other relevant data. The terms probably carcinogenic and possibly
 6                    carcinogenic have no quantitative significance and are used simply
 7                    as descriptors of different levels of evidence of human
 8                    carcinogenicity, with probably carcinogenic signifying a higher
 9                    level of evidence than possibly carcinogenic.

10              Group 2A: The agent is probably carcinogenic to humans.

11                    This category is used when there is limited evidence of
12                    carcinogenicity in humans and sufficient evidence of
13                    carcinogenicity in experimental animals. In some cases, an agent
14                    may be classified in this category when there is inadequate
15                    evidence of carcinogenicity in humans and sufficient evidence of
16                    carcinogenicity in experimental animals and strong evidence that
17                    the carcinogenesis is mediated by a mechanism that also operates
18                    in humans. Exceptionally, an agent may be classified in this
19                    category solely on the basis of limited evidence of carcinogenicity
20                    in humans. An agent may be assigned to this category if it clearly
21                    belongs, based on mechanistic considerations, to a class of agents
22                    for which one or more members have been classified in Group 1 or
23                    Group 2A.

24              Group 2B: The agent is possibly carcinogenic to humans.

25                    This category is used for agents for which there is limited evidence
26                    of carcinogenicity in humans and less than sufficient evidence of
27                    carcinogenicity in experimental animals. It may also be used when
28                    there is inadequate evidence of carcinogenicity in humans but
29                    there is sufficient evidence of carcinogenicity in experimental
30                    animals. In some instances, an agent for which there is inadequate
31                    evidence of carcinogenicity in humans and less than sufficient
32                    evidence of carcinogenicity in experimental animals together with
33                    supporting evidence from mechanistic and other relevant data may
34                    be placed in this group. An agent may be classified in this category
35                    solely on the basis of strong evidence from mechanistic and other
36                    relevant data.

37              Group 3:  The agent is not classifiable as to its carcinogenicity to humans.

38                    This category is used most commonly for agents for which the
39                    evidence of carcinogenicity is inadequate in humans and
40                    inadequate or limited in  experimental animals.
      May 2008                              A-26               DRAFT—DO NOT QUOTE OR CITE

-------
 1                     Exceptionally, agents for which the evidence of carcinogenicity is
 2                     inadequate in humans but sufficient in experimental animals may
 3                     be placed in this category when there is strong evidence that the
 4                     mechanism of carcinogenicity in experimental animals does not
 5                     operate in humans.

 6                     Agents that do not fall into any other group are also placed in this
 7                     category.

 8                     An evaluation in Group 3 is not a determination of non-
 9                     carcinogenicity or overall safety. It often means that further
10                     research is needed, especially when exposures are widespread or
11                     the cancer data are consistent with differing interpretations.

12              Group 4: The agent is probably not carcinogenic to humans.

13                     This category is used for agents for which there is evidence
14                     suggesting lack of carcinogenicity in humans and in experimental
15                     animals. In some instances, agents for which there is inadequate
16                     evidence of carcinogenicity in humans but evidence suggesting
17                     lack of carcinogenicity in experimental animals, consistently and
18                     strongly supported by a broad range of mechanistic and other
19                     relevant data, may be classified in this group.

20              (e) Rationale

21              The reasoning that the Working Group used to reach its evaluation is
22              presented and discussed. This section integrates the major findings from
23              studies of cancer in humans,  studies of cancer in experimental animals,
24              and mechanistic and other relevant data. It includes concise statements of
25              the principal line(s) of argument that emerged, the conclusions of the
26              Working Group on the strength of the evidence for each group of studies,
27              citations to indicate which studies were pivotal to these conclusions, and
28              an explanation of the reasoning of the Working Group in weighing data
29              and making evaluations. When there are significant differences of
30              scientific interpretation among Working Group Members, a brief summary
31              of the alternative interpretations is provided, together with their scientific
32              rationale and an indication of the relative degree of support for each
33              alternative.

      A.3.6.2. National Toxicology Program Criteria

34         The criteria for listing an agent, substance, mixture,  or exposure circumstance in the

35    National Toxicology Program's Report on Carcinogens (NTP, 2005) were as follows:
      May 2008                              A-27               DRAFT—DO NOT QUOTE OR CITE

-------
 1              Known to Be Human Carcinogen:

 2              There is sufficient evidence of carcinogenicity from studies in humans*,
 3              which indicates a causal relationship between exposure to the agent,
 4              substance, or mixture, and human cancer.

 5              Reasonably Anticipated to Be Human Carcinogen:

 6              There is limited evidence of carcinogenicity from studies in humans*,
 7              which indicates that causal interpretation is credible, but that alternative
 8              explanations, such as chance, bias, or confounding factors, could not
 9              adequately be excluded,

10              or

11              there is sufficient evidence of carcinogenicity from studies in experimental
12              animals, which indicates there is an increased incidence of malignant
13              and/or a combination of malignant and benign tumors (1) in multiple
14              species or at multiple tissue sites, or (2) by multiple routes of exposure, or
15              (3) to an unusual degree with regard to incidence, site, or type of tumor, or
16              age at onset,

17              or

18              there is less than sufficient evidence of carcinogenicity in humans or
19              laboratory animals; however,  the agent, substance, or mixture belongs to a
20              well-defined, structurally related class of substances whose members are
21              listed in a previous Report on Carcinogens as either known to be a human
22              carcinogen or reasonably anticipated to be a human carcinogen, or there is
23              convincing relevant information that the agent acts through mechanisms
24              indicating it would likely cause  cancer in humans.

25              Conclusions regarding carcinogenicity in humans or experimental animals
26              are based on scientific judgment, with consideration given to all relevant
27              information. Relevant information includes, but is not limited to, dose
28              response, route of exposure, chemical structure, metabolism,
29              pharmacokinetics, sensitive sub-populations, genetic effects, or other data
30              relating to mechanism of action or factors that may be unique to a given
31              substance. For example, there may be substances for which there is
32              evidence of carcinogenicity in laboratory animals, but there are
33              compelling data indicating that the agent acts through mechanisms which
34              do not operate in humans and would therefore not reasonably be
35              anticipated to cause cancer in humans.

36              *This evidence can include traditional cancer epidemiology studies, data
37              from clinical studies, and/or data derived from the  study of tissues or cells
38              from humans exposed to the substance in question that can be useful for
39              evaluating whether a relevant cancer mechanism is operating in people.
      May 2008                              A-28               DRAFT—DO NOT QUOTE OR CITE

-------
 4
 5
 6
 7
         Annex  B. Additional Information on  the
       Atmospheric Chemistry of Sulfur Oxides

B.1. Introduction
     SC>2 is chiefly but not exclusively primary in origin; it is also produced by the
photochemical oxidation of reduced sulfur compounds such as dimethyl sulfide (CH3-S-CH3),
hydrogen sulfide (H2S), carbon disulfide (€82), carbonyl sulfide (OCS), methyl mercaptan
(CH3-S-H), and dimethyl disulfide (CH3-S-S-CH3) which are all mainly biogenic in origin. Their
sources are discussed in Section B.3. Table B-l lists the atmospheric lifetimes of reduced sulfur
species with respect to reaction with various oxidants. Except for OCS, which is lost mainly by
photolysis (x~6 months), these species are lost mainly by reaction with OH and NO3 radicals.
Because OCS is relatively long-lived in the troposphere, it can be transported upwards into the
stratosphere.
     Table B-1   Atmospheric lifetimes of SO2 and reduced sulfur species with respect to reaction with
                OH, NO3, and Cl radicals.
COMPOUND
SO2
CH3-S-CH3
H2S
CS2
OCS
CH3-S-H
CH3-S-S-CH3
OH
KX1012
1.6
5.0
4.7
1.2
0.0019
33
230
T
7.2d
2.3d
2.2 d
9.6 d
17y
8.4 h
1.2h
N03
KX1012
NA
1.0
NA
< 0.0004
< 0.0001
0.89
0.53
T

1.1-h

> 116d
> 1.3y
1.2h
2.1-h
CL
KX1012
NA
400
74
< 0.004
< 0.0001
200
NA
T

29 d
157 d
NR
NR
58 d

     NA = Reaction rate coefficient not available.
     NR = Rate coefficient too low to be relevant as an atmospheric loss mechanism. Rate coefficients were calculated at 298 K and 1 atmosphere.
     y=year h = hour OH = 1 * 106/cm3 NO3 = 2.5 * 108/cm3    Cl = 1 * 103/cm3.    1Rate coefficients were taken from JPL Chemical Kinetics Evaluation No. 14
     (JPL, 2003)
     Source: Seinfeld and Pandis (
10        Crutzen (1976) proposed that its oxidation serves as the major source of sulfate in the
11   stratospheric aerosol layer sometimes referred to the "Junge layer," (Junge et al., 1961) during
12   periods when volcanic plumes do not reach the stratosphere. However, the flux of OCS into the
13   stratosphere is probably not sufficient to maintain this stratospheric aerosol layer. Myhre et al.
14   (2004) proposed instead that SO2 transported upwards from the troposphere is the most likely
     May 2008
                                     B-1
DRAFT—DO NOT QUOTE OR CITE

-------
 1   source, as the upward flux of OCS is too small to sustain observed sulfate loadings in the Junge
 2   layer. In addition, in situ measurements of the isotopic composition of sulfur do not match those
 3   of OCS (Leung, 2002). Reaction with NO3 radicals at night most likely represents the major loss
 4   process for dimethyl  sulfide and methyl mercaptan. The mechanisms for the oxidation of DMS
 5   are still not completely understood. Initial attack by NOs and OH radicals involves H atom
 6   abstraction, with a smaller branch leading to OH addition to the S atom. The OH addition branch
 7   increases in importance as temperatures decrease, becoming the major pathway below
 8   temperatures of 285 K (Ravishankara, 1997). The adduct may either decompose to form methane
 9   sulfonic acid (MSA), or undergo further reactions in the main pathway, to yield dimethyl
10   sulfoxide (Barnes et al., 1991). Following H atom abstraction from DMS, the main reaction
11   products include MSA and SO2. The ratio of MSA to SO2 is strongly temperature dependent,
12   varying from about 0.1 in tropical waters to about 0.4 in Antarctic waters (Seinfeld and Pandis,
13   1998b). Excess sulfate (over that expected from the sulfate in seawater) in marine aerosol is
14   related mainly to the  production of SO2 from the oxidation of DMS. Transformations among
15   atmospheric sulfur compounds are summarized in Figure B-l.
     Figure B-1   Transformations of sulfur compounds in the atmosphere.
     Source: Adapted from Berresheim et al. (1995).
     May 2008
B-2
DRAFT—DO NOT QUOTE OR CITE

-------
      B.1.1. Multiphase Chemical Processes Involving Sulfur Oxides and
      Halogens
 1         Chemical transformations involving inorganic halogenated compounds effect changes in
 2    the multiphase cycling of sulfur oxides in ways analogous to their effects on NOX. Oxidation of
 3    dimethylsulfide (CH3)2S by BrO produces dimethylsulfoxide (CH3)2SO (Barnes et al., 1991;
 4    Toumi, 1994), and oxidation by atomic chloride leads to formation of SO2 (Keene et al., 1996).
 5    (CH3)2SO and SO2 are precursors for methanesulfonic acid (CH3SO3H) and H2SO4. In the MBL,
 6    virtually all H2SO4 and CH3SO3H vapor condenses onto existing aerosols or cloud droplet, which
 7    subsequently evaporate, thereby contributing to aerosol growth and acidification. Unlike
 8    CH3SO3H, H2SO4 also has the potential to produce new particles (Korhonen et al., 1999;
 9    Kulmala et al., 2000), which in marine regions is thought to occur primarily in the free
10    troposphere. Saiz-Lopez et al. (2004) estimated that observed levels of BrO at Mace Head
11    Atmospheric Research Station in Ireland, would oxidize (CH3)2S about six times faster than OH
12    and thereby substantially increase production rates of H2SO4 and other condensible S species in
13    the MBL.  Sulfur dioxide is also scavenged by deliquesced aerosols and oxidized to H2SO4 in the
14    aqueous phase by several  strongly pH-dependent pathways (Chameides and Stelson, 1992;
15    Keene et al.,  1998; Vogt et al., 1996). Model calculations indicate that oxidation of S(IV) by O3
16    dominates in fresh, alkaline sea salt aerosols, whereas oxidation by hypohalous acids (primarily
17    HOC1) dominates in moderately acidic solutions. Additional  particulate non-sea salt (nss) SO42"
18    is generated by  SO2 oxidation in cloud droplets (Clegg and Toumi, 1998). Ion-balance
19    calculations indicate that most nss SO42" in short-lived (two to 48 h) sea salt size fractions
20    accumulates in acidic aerosol solutions and/or in acidic aerosols processed through clouds (e.g.,
21    Keene et al., 2004). The production, cycling, and associated radiative effects of S-containing
22    aerosols in marine and coastal air are regulated in part by chemical transformations involving
23    inorganic halogens (Von Glasow et al., 2002). These transformations include: dry-deposition
24    fluxes of nss  SO42" in marine air dominated, naturally, by the sea salt size fractions (Huebert et
25    al., 1996; Turekian et al., 2001); HC1 phase partitioning that regulates sea salt pH and associated
26    pH-dependent pathways for S(IV) oxidation (Keene et al., 2002; Pszenny et al., 2004); and
27    potentially important oxidative reactions with reactive halogens for (CH3)2S and S(IV).
28    However, both the absolute magnitudes and relative importance of these processes in MBL S
29    cycling are poorly understood.

      May 2008                               B-3               DRAFT—DO NOT QUOTE OR CITE

-------
 1         Iodine chemistry has been linked to ultrafine particle bursts at Mace Head (O'Dowd et al.,
 2    1999; 2002). Observed bursts coincide with the elevated concentrations of IO and are
 3    characterized by particle concentrations increasing from background levels to up to 300,000 cm"3
 4    on a time scale of seconds to minutes. This newly identified source of marine aerosol would
 5    provide additional aerosol surface area for condensation of sulfur oxides and thereby presumably
 6    diminish the potential for nucleation pathways involving H^SO/t. However, a subsequent
 7    investigation in polluted air along the New England, USA coast found no correlation between
 8    periods of nanoparticle growth and corresponding concentrations of I oxides (Fehsenfeld et al.,
 9    2006). The potential importance of I chemistry in aerosol nucleation and its associated influence
10    on sulfur cycling remain highly uncertain.

      B.1.2. Mechanisms for the Aqueous Phase Formation of Sulfate
11         Warneck (1999) constructed a box model describing the chemistry of the oxidation of SC>2
12    and NC>2 including the interactions of N and S species and minor processes in sunlit cumulus
13    clouds. The relative contributions of different reactions to the oxidation of S(IV) species to  S(VI)
14    and NC>2 to NCV 10 min after cloud formation are given in Table B-2. The two columns show the
     Table B-2    Relative contributions of various reactions to the total S(IV) oxidation rate within a
                 sunlit cloud, 10 min after cloud formation.
REACTION
% OF TOTAL3
% OF TOTAL"
Gas Phase
OH + SO2
3.5
3.1
Aqueous Phase
O3 + HSO3"
03 + S032"
H2O2 + SO3"
CH3OOH + HSO3"
HNO4 + HSO3"
HOONO + HSO3"
HSO5" + HSO3"
S05" + S032"
HSO5" + Fe2+
0.6
7.0
78.4
0.1
9.0
<0.1
1.2
<0.1

0.7
8.2
82.1
0.1
4.4
<0.1
<0.1
<0.1
0.6
     sln the absence of transition metals.   In the presence of iron and copper ions.
     Source: Adapted from Warneck (1999).
     May 2008
B-4
DRAFT—DO NOT QUOTE OR CITE

-------
1   relative contributions with and without transition metal ions. As can be seen from Table B-2, SCh
2   within a cloud (gas + cloud drops) is oxidized mainly by H2O2 in the aqueous phase, while and
3   the gas-phase oxidation by OH radicals is small by comparison. A much smaller contribution in
4   the aqueous phase is made by methyl hydroperoxide (CHsOOH) because it is formed mainly in
5   the gas phase and its Henry's Law constant is several orders of magnitude smaller that of H2O2.
6   After H2O2, HNC>4 is the major contributor to S(IV) oxidation.

    B.1.3. Multiphase Chemical Processes Involving Sulfur Oxides and
    Ammonia
7        The phase partitioning of NH3 with deliquesced aerosol solutions is controlled primarily by
8   the thermodynamic properties of the system expressed as follows:
                A',.
           NH,
                       aqi
                                1Q-6
                               10-
                               to-1
                               10'1
                                    x
                                                 pH
    Figure B-2  Comparison of aqueous-phase oxidation paths. The rate of conversion of S(IV) to
    S(VI) is shown as a function of pH. Conditions assumed are: [SO2(g)] = 5 ppb; [NO2(g)] = 1 ppb;
    [H202(g)] = 1 ppb; [03(g)] = 50 ppb; [Fe(lll)(aq)] = 0.3 uM; [Mn(ll)(aq)] = 0.3 uM.
    Source: Seinfeld and Pandis (1998a).
    May 2008
B-5
DRAFT—DO NOT QUOTE OR CITE

-------
 1    where KH and Kb are the temperature-dependent Henry's Law and dissociation constants
 2    (62 M atm"1) (1.8 x 10"5 M), respectively, for NH3, and Kw is the ion product of water (1.0 x 10"14
 3    M) (Chameides, 1984). It is evident that for a given amount of NHX (NH3 + particulate NH4+) in
 4    the system, increasing aqueous concentrations of particulate H+ will shift the partitioning of NH3
 5    towards the condensed phase. Consequently, under the more polluted conditions characterized by
 6    higher concentrations of acidic sulfate aerosol, ratios of gaseous NH3 to particulate NH4+
 7    decrease (Smith et al., 2007). It also follows that in marine air, where aerosol acidity varies
 8    substantially as a function of particle size, NH3 partitions preferentially to the more acidic sub-
 9    micron size fractions (e.g., Keene et al., 2004; Smith et al., 2007).
10         Because the dry-deposition velocity of gaseous NH3 to the surface is substantially greater
11    than that for the sub-micron, sulfate aerosol size factions with which most particulate NH4+ is
12    associated, dry-deposition fluxes of total NH3 are dominated by the gas phase fraction (Russell et
13    al., 2003; Smith et al., 2007). Consequently, partitioning with highly acidic sulfate aerosols
14    effectively increases the  atmospheric lifetime of total NH3 against dry deposition. This shift has
15    important consequences  for NH3 cycling and potential ecological effects. In coastal New
16    England during summer, air transported from rural eastern Canada contains relatively low
17    concentrations of particulate non-sea salt (nss) SC>42" and total NH3 (Smith et al., 2007). Under
18    these conditions, the roughly equal partitioning of total NH3 between the gas and particulate
19    phases sustains substantial dry-deposition fluxes of total NH3 to the coastal ocean (median of
20    10.7 jiM m"2 day"1). In contrast, heavily polluted air transported from the industrialized
21    midwestern United States contains concentrations of nss SC>42" and total NH3 that are about a
22    factor of 3 greater, based on median values. Under these conditions, most total NH3 (> 85%)
23    partitions to the highly acidic sulfate aerosol size fractions and,  consequently, the median dry-
24    deposition flux of total NH3 is 30% lower than that under the cleaner northerly flow regime. The
25    relatively longer atmospheric lifetime of total NH3 against dry deposition under more polluted
26    conditions implies that, on average, total NH3 would accumulate to higher atmospheric
27    concentrations under these conditions and also be subject to atmospheric transport over longer
28    distances. Consequently, the importance of NHX removal via wet deposition would also increase.
29    Because of the inherently sporadic character of precipitation, we might expect greater
30    heterogeneity in NH3 deposition fields and any potential responses in sensitive ecosystems
31    downwind of major S-emission regions.

      May 2008                                B-6                DRAFT—DO NOT QUOTE OR CITE

-------
      B.2. Transport of Sulfur Oxides in the Atmosphere
 1         Crutzen and Gidel (1983), Gidel (1983), and Chatfield and Crutzen (1984) hypothesized
 2    that convective clouds played an important role in rapid atmospheric vertical transport of trace
 3    species and first tested simple parameterizations of convective transport in atmospheric chemical
 4    models. At nearly the same time, evidence was shown of venting the boundary layer by shallow,
 5    fair weather cumulus clouds (Greenhut et al., 1984; 1986). Field experiments were conducted in
 6    1985 which resulted in verification of the hypothesis that deep convective clouds are
 7    instrumental in atmospheric transport of trace constituents (Dickerson et al., 1987). Once
 8    pollutants are lofted to the middle and upper troposphere, they typically have a much longer
 9    chemical lifetime and with the generally stronger winds at these altitudes, they can be
10    transported large distances from their source regions.

      B.3. Emissions of  SO2
11         As can be seen from Table B-3, emissions  of 862 are due mainly to the combustion of
12    fossil fuels by electrical utilities and industry. Transportation related sources make only a minor
13    contribution. As a result, most SO2 emissions originate from point sources. Since sulfur is a
14    volatile component of fuels, it is almost quantitatively released during combustion and emissions
15    can be calculated on the basis of the sulfur content of fuels to greater accuracy than for other
16    pollutants such as NOx or primary PM.
17         The major natural sources of SC>2 are volcanoes  and biomass burning and DMS oxidation
18    over the oceans. SC>2 constitutes a relatively minor fraction (0.005% by volume) of volcanic
19    emissions (Holland, 1978). The ratio of H2S to 862 is  highly variable in volcanic gases. It is
20    typically much less than one, as in the Mt. Saint  Helen's eruption (Turco et al., 1983). However,
21    in addition to being degassed from magma, H2S  can be produced if ground waters, especially
22    those containing organic matter, come into contact with volcanic gases. In this case, the ratio of
23    H2S to SC>2 can be greater than one. H2S produced this way would more likely be emitted
24    through side vents than through  eruption columns (Pinto et al., 1989). Primary particulate sulfate
25    is a component of marine aerosol and is also produced by wind erosion of surface soils.
26         Volcanic sources of SC>2 are limited to the  Pacific Northwest, Alaska, and Hawaii. Since
27    1980, the Mount St. Helens volcano in the Washington Cascade Range (46.20 N, 122.18 W,

      May 2008                              B-7              DRAFT—DO NOT QUOTE OR CITE

-------
 1    summit 2549 m asl) has been a variable source of 862. Its major effects came in the explosive
 2    eruptions of 1980, which primarily affected the northern part of the mountainous western half of
 3    the United States. The Augustine volcano near the mouth of the Cook Inlet in southwestern
 4    Alaska (59.363 N, 153.43 W, summit 1252 m asl) has had variable SC>2 emission since its last
 5    major eruptions in 1986. Volcanoes in the Kamchatka peninsula of eastern region of Siberian
 6    Russia do not significantly effect surface SO2 concentrations  in northwestern North America.
 7    The most serious effects in the United States from volcanic SC>2 occurs on the island of Hawaii.
 8    Nearly continuous venting of 862 from Mauna Loa and Kilauea produces 862 in such large
 9    amounts that > 100 km downwind of the island SC>2 concentrations can exceed 30 ppbv
10    (Thornton and Bandy, 1993). Depending on wind direction, the west coast of Hawaii (Kona
11    region) has had significant deleterious effects from SO2 and acidic sulfate aerosols for the past
12    decade.
13         Emissions of SC>2 from burning vegetation are generally in the range of 1 to 2% of the
14    biomass burned (e.g., Levine et al., 1999). Sulfur is bound in amino acids in vegetation. This
15    organically bound sulfur is released during combustion. However, unlike nitrogen, about half of
16    the sulfur initially present in vegetation is found in the ash (Delmas,  1982). Gaseous emissions
17    are mainly in the form of 862 with much smaller amounts of H2S and OCS. The ratio of gaseous
18    nitrogen to sulfur emissions is about 14, very close to their ratio in plant tissue (Andreae, 1991).
19    The ratio of reduced nitrogen and sulfur species such as NH3  and H2S to their more oxidized
20    forms, such as NO and SO2,  increases from flaming to smoldering phases of combustion, as
21    emissions of reduced species are favored by lower temperatures and reduced C>2 availability.
22         Emissions of reduced sulfur species are associated typically with marine organisms living
23    either in pelagic or coastal zones and with anaerobic bacteria in marshes and estuaries.
24    Mechanisms for their oxidation were discussed in Section B.I. Emissions of dimethyl sulfide
25    (DMS) from marine plankton represent the largest single source of reduced sulfur species to the
26    atmosphere (e.g., Berresheim et al., 1995). Other sources such as wetlands and terrestrial plants
27    and soils probably account for less than 5% of the DMS  global flux, with most of this coming
28    from wetlands.
29         The coastal and wetland sources of DMS have a dormant period in the fall/winter from
30    senescence of plant growth. Marshes die back in fall and winter, so dimethyl sulfide emissions
31    from them are lower, reduced light levels in winter at mid to high latitudes reduce phytoplankton

      May 2008                              B-8               DRAFT—DO NOT QUOTE OR CITE

-------
 1   growth which also tends to reduce DMS emissions. Western coasts at mid to high latitudes have
 2   reduced levels of the light that drive photochemical production and oxidation of DMS. Freezing
 3   at mid and high latitudes affects the release of biogenic sulfur gases,  particularly  in the nutrient-
 4   rich regions around Alaska. Transport of SC>2 from regions of biomass burning seems to be
 5   limited by heterogeneous losses that accompany convective processes that ventilate the surface
 6   layer and the lower boundary layer (Thornton et al., 1996, TRACE-P data archive).
 7         However, it should be noted that reduced sulfur species are also produced by industry. For
 8   example, DMS is used in petroleum refining and in petrochemical production processes to
 9   control the formation of coke and carbon monoxide. In addition, it is used to control dusting in
10   steel mills. It is also used in a range of organic syntheses. It also has a use as a food flavoring
11   component.  It can also be oxidized by natural or artificial means to dimethyl sulfoxide (DMSO),
12   which has several important solvent properties.
     Table B-3   Emissions of NOX, ammonia, and SO2 in the U.S. by source and category, 2002.
2002 EMISSIONS (TG/YR)
Total All Sources
Fuel Combustion Total
Fuel Combustion Electrical Utilities
Coal
Bituminous
Subbituminous
Anthracite & Lignite
Other
Oil
Residual
Distillate
Gas
Natural
Process
Other
Internal Combustion
Fuel Combustion Industrial
Coal
Bituminous
NOX1
23.19
9.11
5.16
4.50
2.90
1.42
0.18
<0.01
0.14
0.13
0.01
0.30
0.29
0.01
0.05
0.17
3.15
0.49
0.25
NH2
4.08
0.02
<0.01
<0.01




<0.01


<0.01


<0.01
<0.01
<0.01
<0.01

S02
16.87
14.47
11.31
10.70
8.04
2.14
0.51

0.38
0.36
0.01
0.01


0.21
0.01
2.53
1.26
0.70
     May 2008
B-9
DRAFT—DO NOT QUOTE OR CITE

-------
2002 EMISSIONS (TG/YR)
Subbituminous
Anthracite & Lignite
Other
Oil
Residual
Distillate
Other
Gas
Natural
Process
Other
Other
Wood/Bark Waste
Liquid Waste
Other
Internal Combustion
Fuel Combustion Other
Commercial/Institutional Coal
Commercial/Institutional Oil
Commercial/Institutional Gas
Miscellaneous Fuel Combustion
(Except Residential)
Residential Wood
Residential Other
Distillate Oil
Bituminous/Subbituminous
Other
INDUSTRIAL PROCESS TOTAL
Chemical & Allied Product Mfg
Organic Chemical Mfg
Inorganic Chemical Mfg
Sulfur Compounds
Other
Polymer & Resin Mfg
Agricultural Chemical Mfg
Ammonium Nitrate/Urea Mfg.
Other
Paint, Varnish, Lacquer, Enamel Mfg
NOX1
0.07
0.04
0.13
0.19
0.09
0.09
0.01
1.16
0.92
0.24
<0.01
0.16
0.11
0.01
0.04
1.15
0.80
0.04
0.08
0.25
0.03
0.03
0.36
0.06
0.26
0.04
1.10
0.12
0.02
0.01


<0.01
0.05


0.00
NH2



<0.01



<0.01



<0.01



<0.01
<0.01
<0.01
<0.01
<0.01
<0.01





0.21
0.02
<0.01
<0.01


<0.01
0.02
<0.01
0.02

S02
0.10
0.13
0.33
0.59
0.40
0.16
0.02
0.52



0.15



0.01
0.63
0.16
0.28
0.02
0.01
<0.01
0.16
0.15
<0.01
<0.01
1.54
0.36
0.01
0.18
0.17
0.02
<0.01
0.05


0.00
May 2008
B-10
DRAFT—DO NOT QUOTE OR CITE

-------
2002 EMISSIONS (TG/YR)
Pharmaceutical Mfg
Other Chemical Mfg
Metals Processing
Non-Ferrous Metals Processing
Copper
Lead
Zinc
Other
Ferrous Metals Processing
Metals Processing
Petroleum & Related Industries
Oil & Gas Production
Natural Gas
Other
Petroleum Refineries & Related Industries
Fluid Catalytic Cracking Units
Other
Asphalt Manufacturing
Other Industrial Processes
Agriculture, Food, & Kindred Products
Textiles, Leather, & Apparel Products
Wood, Pulp & Paper, & Publishing Products
Rubber & Miscellaneous Plastic Products
Mineral Products
Cement Mfg
Glass Mfg
Other
Machinery Products
Electronic Equipment
Transportation Equipment
Miscellaneous Industrial Processes
Solvent Utilization
Degreasing
Graphic Arts
Dry Cleaning
Surface Coating
Other Industrial
NOX1
0.00
0.03
0.09
0.01




0.07
0.01
0.16
0.07


0.05


0.04
0.54
0.01
<0.01
0.09
<0.01
0.42
0.24
0.01
0.10
<0.01
<0.01
<0.01
0.01
0.01
<0.01
<0.01
<0.01
<0.01
<0.01
NH2

<0.01
<0.01
<0.01




<0.01
<0.01
<0.01
<0.01


<0.01
<0.01
<0.01

0.05
<0.01
<0.01
<0.01
<0.01
<0.01



<0.01
<0.01

0.05
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
S02
0.00
0.12
0.30
0.17
0.04
0.07
0.01
<0.01
0.11
0.02
.38
0.11
0.11
0.01
0.26
0.16
0.07
0.01
0.46
0.01
<0.01
0.10
<0.01
0.33
0.19

0.09
<0.01
<0.01
<0.01
0.02
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
May 2008
B-11
DRAFT—DO NOT QUOTE OR CITE

-------
2002 EMISSIONS (TG/YR)
Nonindustrial
Solvent Utilization Nee
Storage & Transport
Bulk Terminals & Plants
Petroleum & Petroleum Product Storage
Petroleum & Petroleum Product Transport
Service Stations: Stage II
Organic Chemical Storage
Organic Chemical Transport
Inorganic Chemical Storage
Inorganic Chemical Transport
Bulk Materials Storage
Waste Disposal & Recycling
Incineration
Industrial
Other
Open Burning
Industrial
Land Clearing Debris
Other
Public Operating Treatment Works
Industrial Waste Water
Treatment, Storage, And Disposal Facility
Landfills
Industrial
Other
Other
TRANSPORTATION TOTAL
Highway Vehicles
Light-Duty Gas Vehicles & Motorcycles
Light-Duty Gas Vehicles
Motorcycles
Light-Duty Gas Trucks
Light-Duty Gas Trucks 1
Light-Duty Gas Trucks 2
Heavy-Duty Gas Vehicles
Diesels
NOX1
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
0.01
<0.01
<0.01
0.01
0.17
0.06


0.10



<0.01
<0.01
<0.01
<0.01


<0.01
12.58
8.09
2.38
2.36
0.02
1.54
1.07
0.47
0.44
3.73
NH2


<0.01
<0.01
<0.01
<0.01

<0.01

<0.01

<0.01
0.14
<0.01


<0.01



0.14
<0.01
<0.01
<0.01


<0.01
0.32
0.32
0.20


0.10


<0.01
<0.01
S02


0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
0.03
0.02

<0.01
<0.01
<0.01

<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
<0.01
0.76
0.30
0.10
0.10
0.00
0.07
0.05
0.02
0.01
0.12
May 2008
B-12
DRAFT—DO NOT QUOTE OR CITE

-------
2002 EMISSIONS (TG/YR)
Heavy-Duty Diesel Vehicles
Light-Duty Diesel Trucks
Light-Duty Diesel Vehicles
Off-Highway
Non-Road Gasoline
Recreational
Construction
Industrial
Lawn & Garden
Farm
Light Commercial
Logging
Airport Service
Railway Maintenance
Recreational Marine Vessels
Non-Road Diesel
Recreational
Construction
Industrial
Lawn & Garden
Farm
Light Commercial
Logging
Airport Service
Railway Maintenance
Recreational Marine Vessels
Aircraft
Marine Vessels
Diesel
Residual Oil
Other
Railroads
Other
Liquefied Petroleum Gas
Compressed Natural Gas
Miscellaneous
Agriculture & Forestry
NOX1
3.71
0.01
0.01
4.49
0.23
0.01
0.01
0.01
0.10
0.01
0.04
<0.01
<0.01
<0.01
0.05
1.76
0.00
0.84
0.15
0.05
0.57
0.08
0.02
0.01
<0.01
0.03
0.09
1.11
1.11


0.98
0.32
0.29
0.04
0.39
<0.01
NH2



<0.01
<0.01










<0.01
















<0.01


3.53
3.45
S02



0.46
0.01










0.22










0.01
0.18



0.05
0.00


0.10
<0.01
May 2008
B-13
DRAFT—DO NOT QUOTE OR CITE

-------
2002 EMISSIONS (TG/YR)
Agricultural Crops
Agricultural Livestock
Other Combustion
Health Services
Cooling Towers
Fugitive Dust
Other
Natural Sources
NOX1







3.10
NH2
<0.01
2.66
0.08




0.03
S02


0.10





      Emissions are expressed in terms of NO2.
     2 Emissions based on Guenther et al. (2000)
     Source: U.S. Environmental Protection Agency (2006a)
     B.4. Methods  Used to Calculate  SOX and Chemical
     Interactions  in the Atmosphere
 1        Atmospheric chemistry and transport models are the major tools used to calculate the
 2   relations among 63, other oxidants, and their precursors, the transport and transformation of air
 3   toxics, the production of secondary organic aerosol, the evolution of the particle size distribution,
 4   and the production and deposition of pollutants affecting ecosystems. Chemical transport models
 5   are driven by emissions inventories for primary species such as the precursors for Os and PM and
 6   by meterological fields produced by other numerical models. Emissions of precursor compounds
 7   can be divided into anthropogenic and natural source categories. Natural sources can be further
 8   divided into biotic (vegetation, microbes, animals) and abiotic (biomass burning, lightning)
 9   categories. However, the distinction between natural sources and anthropogenic sources is often
10   difficult to make as human activities affect directly,  or indirectly, emissions from what would
11   have been considered natural sources during the preindustrial era. Emissions from plants and
12   animals used in agriculture have been referred to as anthropogenic or natural in different
13   applications. Wildfire emissions may be considered to be natural,  except that forest management
14   practices may have led to the buildup of fuels on the forest floor, thereby altering the frequency
15   and severity of forest fires. Needed meteorological quantities such as winds and temperatures are
16   taken from operational analyses, reanalyses, or circulation models. In most cases, these are off-
17   line analyses, i.e., they are not modified by radiatively active species such as Os and particles
18   generated by the model.
     May 2008
B-14
DRAFT—DO NOT QUOTE OR CITE

-------
 1         A brief overview of atmospheric chemistry-transport models is given in Section B.5.
 2    A discussion of emissions inventories of precursors used by these models is given in Section B.5.
 3    Uncertainties in emissions estimates have also been discussed in Air Quality Criteria for
 4    Particulate Matter (U.S. EPA, 1996). Chemistry-transport model evaluation and an evaluation of
 5    the reliability of emissions inventories are also presented in Section B.5.

      B.5.  Chemical-transport Models
 6         Atmospheric CTMs have been developed for application over a wide range of spatial
 7    scales ranging from neighborhood to global. Regional scale CTMs are used: (1) to obtain better
 8    understanding of the processes controlling the formation, transport, and destruction of gas-and
 9    particle-phase criteria and hazardous air pollutants; (2) to understand the relations between Os
10    concentrations and concentrations of its precursors such as NOx and VOCs, the factors leading to
11    acid deposition, and hence to possible damage to ecosystems; and (3) to understand relations
12    among the concentration patterns of various pollutants that may exert adverse health effects.
13    Chemistry Transport Models are also used for determining control strategies for 63 precursors.
14    However, this application has met with varying degrees of success because of the highly
15    nonlinear relations between Os and emissions of its precursors, and uncertainties in emissions,
16    parameterizations of transport, and chemical production and loss terms. Uncertainties in
17    meteorological variables and emissions can be large enough to lead to significant errors in
18    developing control strategies (e.g., Russell and Dennis, 2000; Sillman, 1995).
19         Global scale CTMs are used to address issues associated with climate change, stratospheric
20    63 depletion, and to provide boundary conditions for regional scale models. CTMs include
21    mathematical (and often simplified) descriptions of atmospheric transport, the transfer of solar
22    radiation through the atmosphere, chemical reactions, and removal to the surface by turbulent
23    motions and precipitation for pollutants emitted into the model domain. Their upper boundaries
24    extend anywhere from the top of the mixing layer to the mesopause (about 80 km in height), to
25    obtain more realistic boundary conditions for problems involving stratospheric dynamics. There
26    is a trade-off between the size of the modeling domain and the grid resolution used in the CTM
27    that is imposed by computational resources.
28         There are two major formulations of CTMs in current use. In the first approach, grid-
29    based, or Eulerian, air quality models, the region to be modeled (the modeling domain) is

      May 2008                              B-15               DRAFT—DO NOT QUOTE OR CITE

-------
 1    subdivided into a three-dimensional array of grid cells. Spatial derivatives in the species
 2    continuity equations are cast in finite-difference there are also some finite-element models, but
 3    not many applications form over this grid, and a system of equations for the concentrations of all
 4    the chemical species in the model are solved numerically at each grid point. Time dependent
 5    continuity (mass conservation) equations are solved for each species including terms for
 6    transport, chemical production and destruction, and emissions and deposition (if relevant), in
 7    each cell. Chemical processes are simulated with ordinary differential equations, and transport
 8    processes are simulated with partial differential equations. Because of a number of factors such
 9    as the different time scales inherent in different processes, the coupled, nonlinear nature of the
10    chemical process terms, and computer storage limitations, all of the terms in the equations are
11    not solved simultaneously in three dimensions. Instead, operator splitting, in which terms in the
12    continuity equation involving individual processes are solved sequentially, is used. In the second
13    CTM formulation, trajectory or Lagrangian models, a large number of hypothetical air parcels
14    are specified as following wind trajectories. In these models, the original system of partial
15    differential equations is transformed into a system of ordinary differential equations.
16         A less common approach is to use a hybrid Lagrangian/Eulerian model, in which certain
17    aspects of atmospheric chemistry and transport are treated with a Lagrangian approach and
18    others are treaded in an Eulerian manner (Stein et al., 2000). Each approach has its advantages
19    and disadvantages. The Eulerian approach is more general in that it includes processes that mix
20    air parcels and allows integrations to be carried out for long periods during which individual air
21    parcels lose their identity. There are, however, techniques for including the effects of mixing in
22    Lagrangian models such as FLEXPART (e.g., Zanis et al., 2003), ATTILA (Reithmeier and
23    Sausen, 2002), and CLaMS (McKenna et al., 2002).

      B.5.1.  Regional Scale Chemical-transport Models
24         Major modeling efforts within the U.S. Environmental Protection Agency center on the
25    Community Multiscale Air Quality modeling system (CMAQ) (Byun and Ching, 1999; Byun
26    and Schere,  2006). A number of other modeling platforms using Lagrangian and Eulerian
27    frameworks have been reviewed in the 96 AQCD for Oj (U.S. EPA, 2007), and in Russell and
28    Dennis (2000). The capabilities of a number of CTMs designed to study local- and regional-scale
29    air pollution problems are summarized by Russell and Dennis (2000). Evaluations of the

      May 2008                              B-16              DRAFT—DO NOT QUOTE OR CITE

-------
 1    performance of CMAQ are given in Arnold et al. (2003), Eder and Yu (2006), Appel et al.
 2    (2005), and Fuentes and Raftery (2005). The domain of CMAQ can extend from several hundred
 3    km to the hemispherical scale. In addition, both of these classes of models allow the resolution of
 4    the calculations over specified areas to vary. CMAQ is most often driven by the MM5 mesoscale
 5    meteorological model (Seaman, 2000), though it may be driven by other meteorological models
 6    (e.g., RAMS). Simulations of Os episodes over regional domains have been performed with a
 7    horizontal resolution as low as 1 km, and smaller calculations over limited domains have been
 8    accomplished at even finer scales. However, simulations at such high resolutions require better
 9    parameterizations of meteorological processes such as boundary layer fluxes, deep convection
10    and clouds (Seaman, 2000), and finer-scale emissions. Finer spatial resolution is necessary to
11    resolve features such as urban heat island circulations;  sea, bay, and land breezes; mountain and
12    valley breezes, and the nocturnal low-level jet.
13         The most common approach to setting up the horizontal domain is to nest a finer grid
14    within a larger domain of coarser resolution. However, there are other strategies such as the
15    stretched grid (e.g., Fox-Rabinovitz et al., 2002) and the adaptive grid. In a stretched grid, the
16    grid's resolution continuously varies throughout the domain, thereby eliminating any potential
17    problems with the sudden change from one resolution to another at the boundary. Caution should
18    be exercised in using such a formulation, because certain parameterizations that are valid on a
19    relatively coarse grid scale (such as convection)  may not be valid on finer scales. Adaptive grids
20    are not fixed at the start of the simulation, but instead adapt to the needs of the simulation as it
21    evolves (e.g., Hansen et al., 1994). They have the advantage that they can resolve processes at
22    relevant spatial scales. However, they can be very slow if the situation to be modeled is complex.
23    Additionally, if adaptive grids are used for separate meteorological, emissions, and
24    photochemical models, there is no reason a priori why the resolution of each grid should match,
25    and the gains realized from increased resolution  in one model will be wasted in the transition to
26    another model. The use of finer horizontal resolution in CTMs will necessitate finer-scale
27    inventories  of land use and better knowledge of the exact paths of roads, locations of factories,
28    and, in general, better methods for locating  sources and estimating their emissions.
29         The vertical resolution of these CTMs is variable, and usually configured to have higher
30    resolution near the surface and decreasing aloft.  Because the height of the boundary layer is of
31    critical importance in simulations of air quality, improved resolution of the boundary layer height

      May 2008                              B-17               DRAFT—DO  NOT QUOTE OR CITE

-------
 1   would likely improve air quality simulations. Additionally, current CTMs do not adequately
 2   resolve fine scale features such as the nocturnal low-level jet in part because little is known about
 3   the nighttime boundary layer.
 4         CTMs require time-dependent, three-dimensional wind fields for the period of simulation.
 5   The winds may be either generated by a model using initial fields alone or with four-dimensional
 6   data assimilation to improve the model's performance, fields (i.e., model equations can be
 7   updated periodically or "nudged," to bring results into agreement with observations. Modeling
 8   efforts typically focus on simulations of several days' duration, the typical time scale for
 9   individual 63 episodes, but there have been several attempts at modeling longer periods. For
10   example, Kasibhatla and Chameides (2000) simulated a four-month period from May to
11   September of 1995 using MAQSIP.  The current trend in modeling applications is towards annual
12   simulations. This trend is driven in part by the need to better understand observations of periods
13   of high wintertime PM (e.g., Blanchard et al., 2002) and the need to simulate Os episodes
14   occurring outside of summer.
15         Chemical kinetics mechanisms (a set of chemical reactions) representing the important
16   reactions occurring in the atmosphere are used in CTMs to estimate the rates of chemical
17   formation and destruction of each pollutant simulated as a function of time. Unfortunately,
18   chemical mechanisms that explicitly treat the reactions of each individual reactive species are too
19   computationally demanding to be incorporated into CTMs. For example, a master chemical
20   mechanism includes approximately  10,500 reactions involving 3603  chemical species (Dentener
21   et al., 2005). Instead, "lumped" mechanisms, that group compounds of similar chemistry
22   together, are used. The  chemical mechanisms used in existing photochemical Os models contain
23   significant uncertainties that may limit the accuracy of their predictions; the accuracy of each of
24   these mechanisms is also limited by missing chemistry. Because of different approaches to the
25   lumping of organic compounds into surrogate groups, chemical mechanisms can produce
26   somewhat different results under similar conditions. The CB-IV chemical mechanism (Gery et
27   al., 1989), the RADM II mechanism (Stockwell et  al., 1990), the SAPRC (e.g., Carter, 1990;
28   Wang et al., 2000b; a) and the RACM mechanisms can be used in CMAQ. Jimenez et al. (2003)
29   provide brief descriptions of the features of the main mechanisms in use and they compared
30   concentrations of several key species predicted by  seven chemical mechanisms in a box model
31   simulation over 24 h. The average deviation from the average of all mechanism predictions for

     May 2008                               B-18              DRAFT—DO NOT QUOTE OR CITE

-------
 1    Os and NO over the daylight period was less than 20%, and was 10% for NO2 for all
 2    mechanisms. However, much larger deviations were found for HNOs, PAN, HO2, H2O2, C2H4,
 3    and C5H8 (isoprene). An analysis for OH radicals was not presented. The large deviations shown
 4    for most species imply differences between the calculated lifetimes of atmospheric species and
 5    the assignment of model simulations to either NOx-limited or radical quantity limited regimes
 6    between mechanisms. Gross and Stockwell (2003) found small  differences between mechanisms
 7    for clean conditions, with differences becoming more significant for polluted conditions,
 8    especially for NO2 and organic peroxy radicals. They caution modelers to consider carefully the
 9    mechanisms they are using. Faraji et al. (2005) found differences of 40% in peak 1-h Os in the
10    Houston-Galveston-Brazoria area between simulations using SAPRAC and CB4. They attributed
11    differences in predicted Oj concentrations to differences in the mechanisms of oxidation of
12    aromatic hydrocarbons.
13         CMAQ and other CTMs  (e.g., PM-CAMx) incorporate processes and interactions of
14    aerosol-phase chemistry (Mebust et al., 2003). There have also been several attempts to study the
15    feedbacks of chemistry on atmospheric dynamics using meteorological models, like MM5 (e.g.,
16    Grell et al., 2000; Liu et al., 2001b; Lu et al., 1997; Park et al., 2001b). This coupling is
17    necessary to simulate accurately feedbacks such as may be caused by the heavy aerosol loading
18    found in forest fire plumes (Lu et al., 1997; Park et al., 2001b),  or in heavily polluted areas.
19    Photolysis rates in CMAQ can  now be calculated interactively with model produced Os, NO2,
20    and aerosol fields (Binkowski et al., 2007).
21         Spatial and temporal characterizations of anthropogenic and biogenic precursor emissions
22    must be  specified as inputs to a CTM. Emissions inventories have been compiled on grids of
23    varying resolution for many hydrocarbons, aldehydes, ketones,  CO, NET?, and NOx. Emissions
24    inventories for many species require the application of some algorithm for calculating the
25    dependence of emissions on physical variables such as temperature and to convert the
26    inventories into formatted emission files required by a CTM. For example, preprocessing of
27    emissions data for CMAQ is done by the Spare-Matrix Operator Kernel Emissions (SMOKE)
28    system. For many species, information concerning the temporal variability of emissions is
29    lacking,  so long-term (e.g., annual or Os-season) averages are used in short-term, episodic
30    simulations. Annual emissions  estimates are often modified by the emissions model to produce
31    emissions more characteristic of the time of day and season. Significant errors in emissions can

      May 2008                              B-19               DRAFT—DO NOT QUOTE OR CITE

-------
 1    occur if an inappropriate time dependence or a default profile is used. Additional complexity
 2    arises in model calculations because different chemical mechanisms are based on different
 3    species, and inventories constructed for use with another mechanism must be adjusted to reflect
 4    these differences. This problem also complicates comparisons of the outputs of these models
 5    because one chemical mechanism may produce some species not present in another mechanism
 6    yet neither may agree with the measurements.
 7         In addition to wet deposition, dry deposition (the removal of chemical species from the
 8    atmosphere by interaction with ground-level  surfaces) is an important removal process for
 9    pollutants on both urban and regional scales and must be included in CTMs. The general
10    approach used in most models is the resistance in series method, in which where dry deposition
11    is parameterized with a Vd, which is represented as Vd = (ra + rb + r^"1 where ra, rb, and rc
12    represent the resistance due to atmospheric turbulence, transport in the fluid sublayer very near
13    the elements of surface such as leaves or soil, and the resistance to uptake of the surface itself.
14    This approach works for a range of substances, although it is inappropriate for species with
15    substantial emissions from the surface or for  species whose deposition to the surface depends on
16    its concentration at the surface itself. The approach is also modified  somewhat for aerosols: the
17    terms rb and rc are replaced with a surface Vd to account for gravitational settling. In their review,
18    Wesley and Hicks (2000) point out several shortcomings of current knowledge of dry deposition.
19    Among those shortcomings are difficulties in representing dry deposition over varying terrain
20    where horizontal advection plays a significant role in determining the magnitude of ra and
21    difficulties in adequately determining a Vd for extremely stable conditions such as those
22    occurring at night (e.g., Mahrt, 1998). Under the best of conditions, when a model is exercised
23    over a relatively small area where dry deposition measurements have been made, models still
24    commonly show uncertainties at least as large as ± 30% (e.g., Brook et al.,  1996; Massman et al.,
25    1994; Padro, 1996). Wesely and Hicks (2000) state that an important result of these comparisons
26    is that the current level of sophistication of most dry deposition models is relatively low, and that
27    deposition estimates therefore must rely heavily on empirical data. Still larger uncertainties exist
28    when the surface features in the built environment are not well known or when the surface
29    comprises a patchwork of different surface types, as is common in the eastern United States.
30         The initial conditions, i.e., the concentration fields of all species computed by a model, and
31    the boundary conditions, i.e., the concentrations of species along the horizontal and upper

      May 2008                              B-20               DRAFT—DO NOT QUOTE OR CITE

-------
 1    boundaries of the model domain throughout the simulation must be specified at the beginning of
 2    the simulation. It would be best to specify initial and boundary conditions according to
 3    observations. However, data for vertical profiles of most species of interest are sparse. The
 4    results of model simulations over larger, preferably global, domains can also be used. As may be
 5    expected, the influence of boundary conditions depends on the lifetime of the species under
 6    consideration and the time scales for transport from the boundaries to the interior of the model
 7    domain (Liu etal., 2001).
 8         Each of the model components described above has an associated uncertainty, and the
 9    relative importance of these uncertainties varies with the modeling application. The largest errors
10    in photochemical modeling are still thought to arise from the meteorological and emissions
11    inputs to the model (Russell  and Dennis, 2000). Within the model itself, horizontal advection
12    algorithms are still thought to be significant source of uncertainty (e.g., Chock and Winkler,
13    1994), though more recently, those errors are thought to have been reduced (e.g., Odman and
14    Ingram, 1996). There are also indications that problems with mass conservation continue to be
15    present in photochemical and meteorological models (e.g., Odman and Russell, 2000); these  can
16    result in significant simulation errors. The effects of errors in initial conditions can be minimized
17    by including several days "spin-up" time in a simulation to allow the model to be driven by
18    emitted species before  the simulation of the period of interest begins.
19         While the effects of poorly specified boundary conditions propagate through the model's
20    domain, the effects of these errors remain undetermined. Because many meteorological processes
21    occur on spatial scales  which are smaller than the model grid spacing (either horizontally or
22    vertically) and thus are not calculated explicitly,  parameterizations of these processes must be
23    used and these introduce additional uncertainty.
24         Uncertainty also  arises in modeling the chemistry of Os formation because it is highly
25    nonlinear with respect  to NOx concentrations. Thus, the volume of the grid cell into which
26    emissions are injected is important because the nature of 63 chemistry (i.e.,  63 production or
27    titration) depends in a complicated way on the concentrations of the precursors and the OH
28    radical as noted earlier. The use of ever-finer grid spacing allows regions of O3 titration to be
29    more clearly separated from  regions of Os production. The use of grid spacing fine enough to
30    resolve the chemistry in individual power-plant plumes is too demanding of computer resources
31    for this to be attempted in most simulations. Instead,  parameterizations of the effects of sub-grid-

      May 2008                               B-21               DRAFT—DO NOT QUOTE OR  CITE

-------
 1    scale processes such as these must be developed; otherwise serious errors can result if emissions
 2    are allowed to mix through an excessively large grid volume before the chemistry step in a
 3    model calculation is performed. In light of the significant differences between atmospheric
 4    chemistry taking place inside and outside of a power plant plume (e.g., Ryerson et al., 1998;
 5    Sillman, 2000), inclusion of a separate, meteorological module for treating large, tight plumes is
 6    necessary. Because the photochemistry of Os and many other atmospheric species is nonlinear,
 7    emissions correctly modeled in a tight plume may be incorrectly modeled in a more dilute plume.
 8    Fortunately, it appears that the chemical mechanism used to follow a plume's development need
 9    not be as detailed as that used to simulate the rest of the domain, as the inorganic reactions are
10    the most important in the plume see (e.g., Kumar and Russell, 1996). The need to include
11    explicitly plume-in-grid chemistry only down to the level of the smallest grid disappears if one
12    uses the adaptive grid approach mentioned previously, though such grids are more
13    computationally intensive. The differences in simulations are significant because they can lead to
14    significant differences in the calculated sensitivity of Os to its precursors (e.g., Sillman, 1995).
15         Because the chemical production and loss terms in the continuity equations for individual
16    species are coupled, the chemical calculations must be performed iteratively until calculated
17    concentrations converge to within some preset criterion. The number of iterations and the
18    convergence criteria chosen also can introduce error.

      B.5.2.  Global-scale CTMs
19         The importance of global transport of O3 and O3 precursors and their contribution to
20    regional O3 levels in the United States is slowly becoming apparent. There are presently on the
21    order of 20 three-dimensional global models that have been developed by various groups to
22    address problems in tropospheric chemistry. These models resolve synoptic meteorology,
23    Os-NOx-CO-hydrocarbon photochemistry, have parameterizations for wet and dry deposition,
24    and parameterize sub-grid scale vertical mixing processes such as convection. Global models
25    have proven useful for testing and advancing scientific understanding beyond what is possible
26    with observations alone. For example, they can calculate quantities of interest that cannot be
27    measured directly, such  as the export of pollution from one continent to the global atmosphere or
28    the response of the atmosphere to future perturbations to anthropogenic emissions.
      May 2008                              B-22               DRAFT—DO NOT QUOTE OR CITE

-------
 1         Global simulations are typically conducted at a horizontal resolution of about 200 km2.
 2    Simulations of the effects of transport from long-range transport link multiple horizontal
 3    resolutions from the global to the local scale. Finer resolution will only improve scientific
 4    understanding to the extent that the governing processes are more accurately described at that
 5    scale. Consequently, there is a critical need for observations at the appropriate scales to evaluate
 6    the scientific understanding represented by the models.
 7         During the recent IPCC-AR4 tropospheric chemistry study coordinated by the European
 8    Union project Atmospheric Composition Change: the European Network of excellence
 9    (ACCENT), 26 atmospheric CTMs were used to estimate the impacts of three emissions
10    scenarios on global atmospheric composition, climate, and air quality in 2030 (Dentener et al.,
11    2006b). All models were required to use anthropogenic emissions developed at IIASA (Dentener
12    et al., 2005) and GFED version 1 biomass burning emissions (Van der Werf et al., 2003) as
13    described in Stevenson et al. (2006). The base simulations from these models were evaluated
14    against a suite of present-day observations. Most relevant to this assessment report are the
15    evaluations with ozone and NC>2, and for nitrogen and sulfur deposition (Dentener et al., 2006b;
16    Stevenson et al., 2006; van Noije et al., 2006); see Figure B-3.

      B.5.3. Modeling the Effects of Convection
17         The effects of deep convection can be simulated using cloud-resolving models,  or in
18    regional or global models in which the convection is parameterized. The Goddard Cumulus
19    Ensemble (GCE) model (Tao and Simpson, 1993) has been used by Pickering et al. (1991;
20    1992a; 1992b; 1993;  1996), Scala et al. (1990), and Stenchikov et al. (1996) in the analysis of
21    convective transport of trace gases. The cloud model is nonhydrostatic and contains a detailed
22    representation of cloud microphysical processes. Two- and three-dimensional versions of the
23    model have been applied in transport analyses. The initial  conditions for the model are usually
24    from a sounding of temperature, water vapor and winds representative of the region of storm
25    development. Model-generated wind fields can be used to perform air parcel trajectory analyses
26    and tracer advection calculations.
      May 2008                              B-23               DRAFT—DO NOT QUOTE OR CITE

-------
1000





800





600

"3
o



/
/
/
/
/
_ ^
i
i
i
i • |
/
• ' . i-
m t • ~ •
' I
/• "B •
i / m*i *
400- /•"••''i*
L ' i ^ •!••
i / • •* • •
L.£/u . ,,
I JP^EfB _ _ ™ ^ *^
200 rSE. * *"

tp%V---''
B*y * •
n R x i i
0 200 400


I



• •
• •
| ..."
• i if ' •
• " • , -
• • " jP • *
* ri i _ •
• ^B
" »
1 • •
^ ; "B •
11 **
• * 1
" •
^
«*•
***
•»*
1 ^ "
*f
V *

Ave, model: 383 Ave. Meas: 322 r; 0.87 n = 226
2 param fit: y = 114.0 + 0.77x
1 param fit: y = 1,00x
percentage within ± 50%: 66.0
•
I I


























600 800 1000
Measurement
Figure B-3 Sulfate wet deposition (mg(S)m"2yr"1) of the mean model versus measurements for the
North American Deposition Program (NADP) network. Dashed lines indicate factor of 2. The gray
line is the result of a linear regression fitting through 0.
Source: Dentener et al. (2006).

     B.5.4. CTM Evaluation
1         The comparison of model predictions with ambient measurements represents a critical task
2    for establishing the accuracy of photochemical models and evaluating their ability to serve as the
3    basis for making effective control strategy decisions. The evaluation of a model's performance,
4    or its adequacy to perform the tasks for which it was designed can only be conducted within the
5    context of measurement errors and artifacts. Not only are there analytical problems, but there are
6    also problems in assessing the representativeness of monitors at ground level for comparison
7    with model values which represent typically an average over the volume of a grid box.
     May 2008
B-24
DRAFT—DO NOT QUOTE OR CITE

-------
 1        Evaluations of CMAQ are given in Arnold et al. (2003) and Fuentes and Raftery (2005).
 2   Discrepancies between model predictions and observations can be used to point out gaps in
 3   current understanding of atmospheric chemistry and to spur improvements in parameterizations
 4   of atmospheric chemical and physical processes. Model evaluation does not merely involve a
 5   straightforward comparison between model predictions and the concentration field of the
 6   pollutant of interest. Such comparisons may not be meaningful because it is difficult to determine
 7   if agreement between model predictions and observations truly represents an accurate treatment
 8   of physical and chemical processes in the CTM or the effects of compensating errors in complex
 9   model routines. Ideally, each of the model components (emissions inventories, chemical
10   mechanism, meteorological driver) should be evaluated individually. However, this is rarely done
11   in practice.

     B.6. Sampling and Analysis of Sulfur Oxides

     B.6.1. Sampling  and Analysis  for SO2
12        SC>2 molecules absorb ultraviolet (UV) light at one wavelength and emit UV light at longer
13   wavelengths. This process is known as fluorescence, and involves the excitation of the SC>2
14   molecule to a higher energy (singlet) electronic state. Once excited, the molecule decays non-
15   radiatively to a lower energy electronic state from which it then decays to the original, or ground,
16   electronic state by emitting a photon of light at a longer wavelength (i.e., lower energy) than the
17   original, incident photon. The process can be summarized by the following equations

            SOj+hVj -»SO2*

            SO2*-»SO2+hv2
18   where SC>2* represents the excited state of SC>2, h2 molecules in the  sample gas.
21        In commercial analyzers,  light from a high intensity UV lamp passes through a bandwidth
22   filter, allowing only photons with wavelengths around the SC>2 absorption peak (near 214 nm) to
23   enter the optical chamber. The light passing through the source bandwidth filter is collimated
     May 2008                             B-25              DRAFT—DO NOT QUOTE OR CITE

-------
 1    using a UV lens and passes through the optical chamber, where it is detected on the opposite side
 2    of the chamber by the reference detector. A photomultiplier tube (PMT) is offset from and placed
 3    perpendicular to the light path to detect the SO2 fluorescence. Since the SO2 fluorescence (330
 4    nm) is at a wavelength that is different from the excitation wavelength, an optical bandwidth
 5    filter is placed in front of the PMT to filter out any stray light from the UV lamp. A lens is
 6    located between the filter and the PMT to focus the fluorescence onto the active area of the
 7    detector and optimize the fluorescence signal. The LOD for a non-trace level SO2 analyzer is 10
 8    parts per billion (ppb) (Code of Federal Regulations, Title 40, Part  53.23c). The SO2
 9    measurement method is subject to both positive and negative interference.

      B.6.1.1. Other Techniques  for Measuring SO2

10         A more sensitive SO2 measurement method than the UV-fluorescence method was reported
11    by Thornton et al. (2002). Thornton et  al. reported use of an atmospheric pressure ionization
12    mass spectrometer.  The high measurement precision and instrument sensitivity were achieved by
13    adding isotopically  labeled SO2 (34S16O2) continuously to the manifold as an internal standard.
14    Field studies showed that the method precision was better than  10% and the limit of detection
15    was less than 1 ppt for a sampling interval of Is.
16         Sulfur dioxide can be measured by LIF at around 220 nm (Matsumi et al., 2005). Because
17    the laser wavelength is alternately tuned to an SO2 absorption peak at 220.6 and trough at
18    220.2 nm, and the difference signal at the two wavelengths is used  to extract the SO2
19    concentration, the technique eliminates interference from either absorption or fluorescence by
20    other species and has high sensitivity (5 ppt in 60 sec). Sulfur dioxide can also be measured by
21    the same DO AS instrument that can measure NO2.
22         Photoacoutsic techniques have been employed for SO2 detection, but they generally have
23    detection limits suitable only for source monitoring (Gondal, 1997; Gondal and Mastromarino,
24    2001).
25         Chemical Ionization Mass Spectroscopy (CIMS) utilizes ionization via chemical reactions
26    in the gas phase to determine an unknown sample's mass spectrum and identity. High sensitivity
27    (10 ppt or better) has been achieved with uncertainty of-15% when a charcoal scrubber is used
28    for zeroing and the  sensitivity is measured with isotopically labeled 34SO2 (Hanke et al., 2003;
29    Hennigan et al., 2006; Huey et  al., 2004).

      May 2008                               B-26                DRAFT—DO NOT QUOTE OR CITE

-------
      B.6.2. Sampling and Analysis for Sulfate, Nitrate, and Ammonium
 1         Sulfate is commonly present in PM2.5. Most PM2.5 samplers have a size-separation device
 2    to separate particles so that only those particles approximately 2.5 jim or less are collected on the
 3    sample filter. Air is drawn through the sample filter at a controlled flow rate by a pump located
 4    downstream of the sample filter. The systems have two critical flow rate components for the
 5    capture of fine particulate: (1) the flow of air through the sampler must be at a flow rate that
 6    ensures that the size cut at 2.5 jim occurs; and (2) the flow rate must be optimized to capture the
 7    desired amount of particulate loading with respect to the analytical method detection limits.
 8         When using the system described above to collect sulfate sampling artifacts can occur
 9    because of: (1) positive sampling artifact for sulfate, nitrate, and particulate ammonium due to
10    chemical reaction; and (2) negative sampling artifact for nitrate and ammonium due to the
11    decomposition and evaporation.
12         There are two major PM speciation ambient air-monitoring networks in the United States:
13    the Speciation Trend Network (STN), and the Interagency Monitoring of Protected Visual
14    Environments (IMPROVE) network. The current STN samplers include three filters: (1) Teflon
15    for equilibrated mass and elemental  analysis including elemental sulfur; (2) a HNO3 denuded
16    nylon filter for ion analysis including NOs and SO/t, (3) a quartz-fiber filter for elemental and
17    organic carbon. The IMPROVE sampler, which collects two 24-h samples per week,
18    simultaneously collects one sample of PMio on a Teflon filter, and three samples of PM2.5 on
19    Teflon, nylon, and quartz filters. PM2 5 mass concentrations are determined gravimetrically from
20    the PM2.5  Teflon filter sample. The PM2.5 Teflon filter sample is also used to determine
21    concentrations of selected elements. The PM2.s nylon filter sample, which is preceded by a
22    denuder to remove acidic gases, is analyzed to determine nitrate and sulfate aerosol
23    concentrations. Finally, the PM2.s  quartz filter sample is analyzed for OC and EC using the
24    thermal-optical reflectance (TOR) method.  The STN and the IMPROVE networks represent a
25    major advance in the measurement of nitrate, because the combination of a denuder (coated with
26    either Na2COs or MgO) to remove HNOs vapor and a Nylon filter to  adsorb HNOs vapor
27    volatilizing from the collected ammonium nitrate particles overcomes the loss of nitrate from
28    Teflon filters.
29         The extent to which sampling  artifacts for particulate NH3+ have been adequately
30    addressed in the current networks is not clear. Recently, new denuder-filter sampling systems

      May 2008                              B-27              DRAFT—DO NOT QUOTE OR CITE

-------
 1    have been developed to measure sulfate, nitrate, and ammonium with an adequate correction of
 2    ammonium sampling artifacts. The denuder-filter system, Chembcomb Model 3500 speciation
 3    sampling cartridge developed by Rupprecht & Patashnick Co, Inc. could be used to collect
 4    nitrate, sulfate, and ammonium simultaneously. The sampling system contains a single-nozzle
 5    size-selective inlet, two honeycomb denuders, the aerosol filter and two backup filters (Keck and
 6    Wittmaack, 2005). The first denuder in the system is coated with 0.5% sodium carbonate and 1%
 7    glycerol and  collects acid gases such as HCL,  SC>2, HONO, and HNOs. The second denuder is
 8    coated with 0.5% phosphoric acid in methanol for collecting NHs. Backup filters collect the
 9    gases behind denuded filters. The backup filters are coated with the same solutions as the
10    denuders. A similar system based on the same principle was applied by Possanzini et al. (1999).
11    The system contains two NaCl-coated annular denuders followed by other two denuders coated
12    with NaCOs/glycerol and citric acid, respectively. This configuration was adopted to remove
13    HNOs quantitatively on the first NaCl denuder. The third and forth denuder remove SC>2 and
14    NHa, respectively. A polyethylene cyclone and a two-stage filter holder containing three filters is
15    placed downstream of the denuders. Aerosol fine particles are collected on a Teflon membrane. A
16    backup nylon filter and a subsequent citric acid impregnated filter paper collect dissociation
17    products (HNOs and NHs) of ammonium nitrate evaporated from the filtered particulate matter.
18         Several traditional and new methods could be used to quantify elemental S collected on
19    filters: energy dispersive X-ray fluorescence, synchrotron induced X-ray fluorescence, proton
20    induced X-ray emission (PIXE), total reflection X-ray fluorescence, and scanning electron
21    microscopy. Energy dispersive X-ray fluorescence (EDXRF) (Method IO-3.3, U.S.
22    Environmental Protection Agency,  1997; see 2004 PM CD for details) and PIXE are the most
23    commonly used methods. Since sample filters often contain very small amounts of particle
24    deposits, preference is given to methods that can accommodate small sample sizes and require
25    little or no sample preparation or operator time after the samples are placed into the analyzer.
26    X-ray fluorescence (XRF) meets these needs and leaves the sample intact after analysis so it can
27    be submitted for additional examinations by other methods as needed. To obtain the greatest
28    efficiency and sensitivity, XRF typically places the filters in a vacuum which may cause volatile
29    compounds (nitrates and organics) to evaporate. As a result, species that can volatilize such as
30    ammonium nitrate and certain organic compounds can be lost during the analysis. The effects of
      May 2008                              B-28               DRAFT—DO NOT QUOTE OR CITE

-------
 1    this volatilization are important if the PTFE filter is to be subjected to subsequent analyses of
 2    volatile species.
 3         Polyatomic ions such as sulfate, nitrate, and ammonium are quantified by methods such as
 4    ion chromatography (1C) (an alternative method commonly used for ammonium analysis is
 5    automated colorimetry). All ion analysis methods require a fraction of the filter to be extracted in
 6    deionized distilled water for sulfate and NaCOs/NaHCOs solution for nitrate and then filtered to
 7    remove insoluble residues prior to analysis. The extraction volume should be as small as possible
 8    to avoid over-diluting the solution and inhibiting the detection of the desired constituents at
 9    levels typical of those found in ambient PM2.5 samples. During analysis, the sample extract
10    passes through an ion-exchange column which separates the ions in time for individual
11    quantification,  usually by an electroconductivity detector.  The ions are identified by their
12    elution/retention times and are quantified by the conductivity peak area or peak height.
13         In a side-by-side comparison of two of the major aerosol monitoring techniques (Hains et
14    al., 2007), PM2.5 mass and major contributing species were well correlated among the different
15    methods with r-values in excess of 0.8. Agreement for mass, sulfate, OC, TC, and ammonium
16    was good while that for nitrate and BC was weaker. Based on reported uncertainties, however,
17    even  daily concentrations of PM2.5 mass and major contributing species were often significantly
18    different at the 95% confidence level. Greater values of PM2.5  mass and individual species were
19    generally reported from Speciation Trends Network methods than from the Desert Research
20    Institute Sequential Filter Samplers. These differences can only be partially accounted for by
21    known random errors. The authors concluded that the current uncertainty estimates used in the
22    STN  network may underestimate the actual uncertainty.
23         The reaction of SO2 (and other acid gases) with basic sites on glass fiber filters or with
24    basic coarse particles on the filter leads to the formation of sulfate (or other nonvolatile salts,
25    e.g., nitrate, chloride). These positive artifacts lead to the overestimation of total mass, and
26    sulfate, and probably also nitrate concentrations. These problems were largely overcome by
27    changing to quartz fiber or Teflon filters and by the separate collection of PM2.5. However, the
28    possible reaction of acidic gases with basic coarse particles remains a possibility, especially with
29    PMio and PMio-2.s measurements. These positive artifacts could be effectively eliminated by
30    removing acidic gases in the sampling line with denuders coated with NaCl or Na2CC>3.
      May 2008                               B-29               DRAFT—DO NOT QUOTE OR CITE

-------
 1         Positive sampling artifacts also occur during measurement of particulate NH4. The reaction
 2    of NH3 with acidic particles (e.g. 2NH3 + H2SO4 D (NH4)2SO4), either during sampling or
 3    during transportation, storage, and equilibration could lead to an overestimation of parti culate
 4    NH4 concentrations. Techniques have been developed to overcome this problem: using a denuder
 5    to remove NH3 during sampling and to protect the collected PM from NH3 (Brauer et al., 1991;
 6    Keck and Wittmaack, 2006; Koutrakis et al., 1988a; 1988b; Possanzini et al., 1999; Suh et al.,
 7    1992; 1994; Winberry et al., 1999). Hydrogen fluoride, citric acid, and phosphorous acids have
 8    been used as coating materials for the NH3 denuder. Positive artifacts for particulate NH4 can
 9    also be observed during sample handling due to contamination. No chemical analysis method, no
10    matter how accurate or precise,  can adequately represent atmospheric concentrations if the filters
11    to which these methods are applied are improperly handled. Ammonia is emitted directly from
12    human  sweat, breath and smoking. It can then react with acidic aerosols on the filter to form
13    ammonium sulfate, ammonium  bisulfate and ammonium nitrate if the filter was not properly
14    handled (Sutton et al., 2000). Therefore, it is important to keep filters away from ammonia
15    sources, such as human breath, to minimize neutralization of the acidic compounds. Also, when
16    filters are handled, preferably in a glove box, the analyst should wear gloves that are antistatic
17    and powder-free to act as an effective contamination barrier.
18         Continuous methods for the quantification of aerosol sulfur compounds first remove
19    gaseous sulfur (e.g., 862, H2S) from the sample stream by a diffusion tube denuder followed by
20    the analysis of particulate sulfur (Cobourn et al., 1978; Durham et al., 1978; Huntzicker et al.,
21    1978; Mueller and Collins,  1980; Tanner et al., 1980). Another approach is to measure total
22    sulfur and gaseous sulfur separately by alternately removing particles from the sample stream.
23    Particulate sulfur is obtained as  the difference between the total and gaseous sulfur (Kittelson et
24    al., 1978). The total sulfur content is measured by a flame photometric detector (FPD) by
25    introducing the sampling stream into a fuel-rich, hydrogen-air flame (e.g., Farwell and
26    Rasmussen, 1976; Stevens et al., 1969) that reduces sulfur compounds and measures the intensity
27    of the chemiluminescence from electronically excited sulfur molecules (S2*). Because the
28    formation of S2*  requires two sulfur atoms, the intensity of the chemiluminescence is
29    theoretically proportional to the square of the concentration of molecules that contain a single
30    sulfur atom. In practice, the exponent is between 1 and 2 and depends on the sulfur compound
31    being analyzed (Dagnall et  al., 1967; Stevens et al., 1971). Calibrations are performed using both

      May 2008                               B-30               DRAFT—DO NOT QUOTE OR CITE

-------
 1   particles and gases as standards. The FPD can also be replaced by a chemiluminescent reaction
 2   with ozone that minimizes the potential for interference and provides a faster response time
 3   (Benner and Stedman, 1989; 1990). Capabilities added to the basic system include in situ thermal
 4   analysis and sulfuric acid speciation (Cobourn et al., 1978; Cobourn and Husar, 1982;
 5   Huntzicker et al., 1978; Tanner et al.,  1980). Sensitivities for particulate sulfur as low as 0.1
 6   Hg/rn3, with time resolution ranging from  1 to 30 min,  have been reported. Continuous
 7   measurements of particulate sulfur content have also been obtained by on-line XRF analysis with
 8   resolution of 30 min or less (Jaklevic  et al., 1981). During a field-intercomparison study of five
 9   different sulfur instruments, Camp et  al. (1982)  reported four out of five FPD systems agreed to
10   within ± 5% during a 1-week sampling period.
     May 2008                               B-31               DRAFT—DO NOT QUOTE OR CITE

-------
                Annex  C. Modeling  Human Exposure

      C.1.  Introduction
 1         Predictive (or prognostic) exposure modeling studies1, specifically focusing on 862, could
 2    not be identified in the literature, though, often, statistical (diagnostic) analyses have been
 3    reported using data obtained in various field exposure studies. However, existing prognostic
 4    modeling systems for the assessment of inhalation exposures can in principle be directly applied
 5    to, or adapted for, SC>2 studies; specifically, such systems include APEX, SHEDS, and
 6    MENTOR-1 A, to be discussed in the following sections. Nevertheless, it should be mentioned
 7    that such applications will be constrained by data limitations, such as the degree of ambient
 8    concentration characterization (e.g., concentrations at the local level) and quantitative
 9    information on indoor sources and sinks.
10         Predictive models of human exposure to ambient air pollutants such as 862 can be
11    classified and differentiated based upon a variety of attributes. For example, exposure models can
12    be classified as:

13         • models of potential (typically maximum) outdoor exposure versus models of actual
14           exposures (the latter including locally modified microenvironmental exposures, both
15           outdoor and indoor);
16         • Population Based Exposure Models (PBEM) versus Individual Based Exposure Models
17           (IBEM);
18         • deterministic versus probabilistic (or statistical) exposure models; and
19         • observation-driven versus mechanistic air quality models (see Section C.4 for discussions
20           about the construction, uses and limitations of this class of mathematical models.
21         Some points should be made regarding terminology and essential concepts in exposure
22    modeling, before proceeding to the overview of specific developments reported in the current
23    research literature:
24         First, it must be understood that there is significant variation in the definitions of many of
25    the terms used in the exposure modeling literature; indeed, the science of exposure modeling is a
      i.e. assessments that start from emissions and demographic information and explicitly consider the physical and chemical processes of environmental
       and microenvironmental transport and fate, in conjunction with human activities, to estimate inhalation intake and uptake.
      May 2008                              C-32              DRAFT—DO NOT QUOTE OR CITE

-------
 1    rapidly evolving field and the development of a standard and commonly accepted terminology is
 2    an ongoing process (see, e.g., WHO, 2004).
 3         Second, it should also be mentioned that, very often, procedures that are called exposure
 4    modeling, exposure estimation, etc. in the scientific literature, may in fact refer to only a sub-set
 5    of the complete set of steps or components required for a comprehensive exposure assessment.
 6    For example, certain self-identified exposure modeling studies focus solely on refining the sub-
 7    regional or local spatio-temporal dynamics of pollutant concentrations (starting from raw data
 8    representing monitor observations or regional grid-based model estimates). Though not exposure
 9    studies per se, such efforts have value  and are included in the discussion of the next sub-section,
10    as they provide potentially useful tools that can be used in a complete exposure assessment. On
11    the other hand, formulations which are self-identified as exposure models but actually focus only
12    on ambient air quality predictions,  such as chemistry-transport models, are not included in the
13    discussion that follows.
14         Third, the process of modeling human exposures to ambient pollutants (traditionally
15    focused on ozone) is very often identified explicitly with population-based modeling, while
16    models describing the specific mechanisms affecting the exposure of an actual individual (at
17    specific locations) to an air contaminant (or to a group of co-occurring gas and/or aerosol phase
18    pollutants) are usually associated with studies focusing specifically on indoor air chemistry
19    modeling.
20         Finally, fourth, the concept of microenvironments, introduced in earlier sections of this
21    document, should be clarified further,  as it is critical in developing procedures for exposure
22    modeling. In the past, microenvironments have typically been defined as individual or aggregate
23    locations (and sometimes  even as activities taking place within a location) where a homogeneous
24    concentration of the pollutant is encountered. Thus a microenvironment has often been identified
25    with an ideal (i.e. perfectly mixed) compartment of classical compartmental modeling. More
26    recent and general definitions view the microenvironment as a control  volume,  either indoors or
27    outdoors, that can be fully characterized by a set of either mechanistic  or phenomenological
28    governing equations, when appropriate parameters are available, given necessary initial and
29    boundary conditions. The  boundary conditions typically would reflect interactions with ambient
30    air and with other microenvironments. The parameterizations of the governing equations
31    generally include the information on attributes of sources and sinks within each

      May 2008                              C-33               DRAFT—DO NOT QUOTE OR CITE

-------
 1    microenvironment. This type of general definition allows for the concentration within a
 2    microenvironment to be non-homogeneous (non-uniform), provided its spatial profile and
 3    mixing properties can be fully predicted or characterized. By adopting this definition, the number
 4    of microenvironments used in a study is kept manageable, but variability in concentrations in
 5    each of the microenvironments can still be taken into account. Microenvironments typically used
 6    to determine exposure include indoor residential microenvironments, other indoor locations
 7    (typically occupational microenvironments), outdoors near roadways, other outdoor locations,
 8    and in-vehicles. Outdoor locations near roadways are segregated from other outdoor locations
 9    (and can be further classified into street canyons, vicinities of intersections, etc.) because
10    emissions from automobiles alter local concentrations significantly compared to background
11    outdoor levels. Indoor residential microenvironments (kitchen, bedroom, living room, etc. or
12    aggregate home microenvironment) are typically separated from other indoor locations because
13    of the time spent there and potential differences between the residential environment and the
14    work/public environment.
15         Once the actual individual and relevant activities and locations (for Individual Based
16    Modeling),  or the sample population and associated spatial (geographical) domain (for
17    Population Based Modeling) have been defined along with the temporal framework of the
18    analysis (time period and resolution), the comprehensive modeling of individual/population
19    exposure to SO2 (and related pollutants) will in general require seven steps (or components, as
20    some of them  do not have to be performed in sequence) that are listed below. This list represents
21    a composite based on approaches and frameworks described in the literature over the last twenty -
22    five years (WHO 2005; U.S. EPA, 1992; 1997; Georgopoulos and Lioy, 1994; Georgopoulos et
23    al., 2005; 2006; Ott, 1982; Price et al., 2003) as well on the  structure of various inhalation
24    exposure models (see Annex Section C.2 that have been used in the past or in current studies to
25    specifically assess inhalation exposures. Figure C-l, adapted from (Georgopoulos et al., 2005),
26    schematically  depicts the sequence of steps summarized here.

         1)  Estimation of the background or ambient levels of both SO2 and related pollutants. This is
             done through either (or a combination of):
                a)  multivariate spatio-temporal analysis of fixed monitor data, or
      May 2008                              C-34               DRAFT—DO NOT QUOTE OR CITE

-------
          b)  emissions-based, photochemical, air quality modeling (typically with a regional,
              grid-based model such as Models-3/CMAQ or CAMx) applied in a coarse
              resolution mode.

   2)  Estimation of local outdoor pollutant levels of both SO2 and related pollutants. These
       levels could typically characterize the ambient air of either an administrative unit (such as
       a census tract, a municipality, a county, etc.) or a conveniently defined grid cell of an
       urban scale air quality model. Again, this may involve either (or a combination of):

          a)  spatio-temporal  statistical analysis of monitor data, or

          b)  application of an urban multi-scale, grid based model (such as CMAQ or CAMx)
              at its highest resolution (typically around 2-4 km), or

          c)  correction  of the estimates of the regional model using some scheme that adjusts
              for observations and/or for subgrid chemistry and mixing processes.

   3)  Characterization of relevant attributes of the individuals or populations under study
       (residence and work locations, occupation, housing data, income, education, age, gender,
       race, weight, and other physiological characteristics). For Population Based Exposure
       Modeling (PBEM) one  can either:

          d)  select a fixed-size sample population of virtual individuals in a way that
              statistically reproduces essential demographics (age, gender, race, occupation,
              income,  education) of the administrative population unit used in the assessment
              (e.g., a sample of 500 people is typically used to represent the demographics of a
              given census tract, whereas a sample  of about 10,000 may be needed to represent
              the demographics of a county), or

          e)  divide the population-of interest into a set of cohorts representing selected
              subpopulations where the cohort is defined by characteristics known to influence
              exposure.

   4)  Development of activity event (or exposure event) sequences for each member of the
       sample population (actual or virtual) or for each cohort for the exposure period. This
       could utilize:

          f)  study-specific information, if available

          g)  existing  databases based on composites of questionnaire information from past
              studies

          h)  time-activity databases, typically in a format compatible with EPAs Consolidated
              Human Activity Database (McCurdy  et al., 2000)

   5)  Estimation of levels and temporal profiles of both 862 and related pollutants in various
       outdoor and indoor microenvironments such as street canyons, roadway intersections,
       parks, residences,  offices, restaurants, vehicles, etc. This is done through either:
May 2008                               C-35              DRAFT—DO NOT QUOTE OR CITE

-------
               i)  linear regression of available observational data sets,

               j)  simple mass balance models (with linear transformation and sinks) over the
                  volume (or a portion of the volume) of the microenvironment,

               k)  lumped (nonlinear) gas or gas/aerosol chemistry models, or

               1)  detailed combined chemistry and Computational Fluid Dynamics modeling.

        6) Calculation of appropriate inhalation rates for the members of the sample population,
           combining the physiological attributes of the (actual or virtual) study subjects and the
           activities pursued during the individual exposure events.

        7)  Calculation of target tissue dose through biologically based modeling estimation
           (specifically, respiratory dosimetry modeling in the case of SC>2 and related reactive
           pollutants) if sufficient information is available.
r
Calculate
potential
outdoor
exposures
L
Figure C-1.
component
population!
SHEDS, AP

i.a. Emissions: NEI (NET, NTI), State; Processing with SMOKE,
EMS-HAP, MOBILE/MOVES, NONROAD, FAAED, BEIS, etc.
i.b. Meteorology; NWS, NCDC; Modeling MMS, RAMS, CALMET
i.e. Land Use/Land Cover, Topography: NLDC (USGS), etc.
* * .
Estimate background j
levels of air pollutants ;
throng!) j
a. muitivariate spatio- i
temporal analysis of i
monitor data j
b. emissions-based air W^
quality modeling i
(with regional, j
grid-based models: i
Models-3/ CMAQ, CAMx i
and RE MS AD) i
•" Estimate local outdoor
*C pollutant levels that
characterize the ambient air of
an administrative unit (such as
census tract) or a conveniently
defined grid through
a. spatiotemporal statistical
analysis of monitor data
b. application of urban scale
model at high resolution
c. subgrid (e.g. plume-in -grid)
modeling
d. data/model assimilation
_
'"v Develop database of j
;-,? individual subjects i
attributes (residence & j
work location, housing i
characteristics, age, i
gender, race, income, etc.) j.
a. collect study-specific "*
information
b. supplement with
available relevant local,
regional, and national
demographic
information
, f Develop activity event
" (or exposure event)
sequences for each individual
of Hie study for the exposure
period
a. collect study-specific
information
b, supplement wrth other
available data
c. organize time-activity
database in format
compatible with CHAD
* *
/ Study-specific survey / Study-specific survey
(also US Census/ I (or default from
US Housing Survey) | CHAD, NHAPS)

ii.a. Emissions: EMS-HAP
ii.b. Local Meteorology — Local
Effects: RAMS, FLUENT
<- ..^^^ZI^l
-*
s
-+
^ Estimate levels and
- ' temporal profiles of
pollirtattts in various
microefivironments (streets,
residences, offices, restaurants,
vehicles, etc.} through
a. regression of observational
data
b. simple linear mass balance
c. lumped (nonlinear} uC
gas/aerosol chemistry models
d. combined chemistry & CFD
(DNS, LES, RANS) models
*
Calculate appropriate
1 inhalation rates for the
members of the sample
population, combining the
physiological attributes of the
study subjects and the
activities pursued during the
individual exposure events
t
riCRP and Other
Physiological & METS
Databases

N
Calcu ate "7 Biologically
exposures/ if based
intakes target tissue
dose modeling
J

Schematic description of a general framework identifying the processes (steps or
s) involved in assessing inhalation exposures and doses for individuals and
5. In general terms, existing comprehensive exposure modeling systems such as
EX, and MENTOR-1 A follow this framework.
1         Implementation of the above framework for comprehensive exposure modeling has

2   benefited significantly from recent advances and expanded availability of computational
     May 2008
C-36
DRAFT—DO NOT QUOTE OR CITE

-------
 1   technologies such as Relational Database Management Systems (RDBMS) and Geographic
 2   Information Systems (GIS) (Georgopoulos et al., 2005; Purushothaman and Georgopoulos, 1997;
 3   1999b; a).
 4         In fact, only relatively recently comprehensive, predictive, inhalation exposure modeling
 5   studies for ozone, PM, and various air toxics, have attempted to address/incorporate all the
 6   components of the general framework described here. In practice, the majority of past exposure
 7   modeling studies have either incorporated only subsets of these components or treated some of
 8   them in a simplified manner, often focusing on the importance of specific factors affecting
 9   exposure. Of course, depending on the objective of a particular modeling study, implementation
10   of only a limited number of steps may be necessary. For example, in a regulatory setting, when
11   comparing the relative effectiveness of emission control strategies, the focus can be on expected
12   changes in ambient levels (corresponding to those observed at NAAQS monitors) in relation to
13   the density  of nearby populations. The outdoor levels of pollutants, in conjunction with basic
14   demographic information,  can thus be used to calculate upper bounds of population exposures
15   associated with ambient air (as opposed to total exposures that would include contributions from
16   indoor sources) useful in comparing alternative control strategies. Though the metrics derived
17   would not be quantitative indicators of actual human exposures, they can serve as surrogates of
18   population exposures associated with outdoor air, and thus  aid in regulatory decision making
19   concerning pollutant standards and in studying the efficacy of emission control strategies.  This
20   approach has been used in studies performing comparative evaluations of regional and local
21   emissions reduction strategies in the eastern United States (Foley et al., 2003; Georgopoulos et
22   al., 1997; Purushothaman and Georgopoulos, 1997).

     C.2. Population Exposure  Models:  Their  Evolution and
     Current Status
23         Existing comprehensive inhalation exposure models  consider the trajectories  of individual
24   human subjects (actual or virtual), or of appropriately defined cohorts, in space and time as
25   sequences of exposure events. In these sequences, each event is defined by time, a geographic
26   location,  a microenvironment, and the activity of the subject.  EPA offices (OAQPS  and NERL)
27   have supported the  most comprehensive efforts in developing models implementing this general
28   concept (see, e.g., Johnson, 2002). These  families of models are the result: National Exposure

     May 2008                             C-37              DRAFT—DO NOT QUOTE OR CITE

-------
 1   Model and Probabilistic National Exposure Model (NEM/pNEM, Whitfield et al., 1997);
 2   Hazardous Air Pollutant Exposure Model (HAPEM, Rosenbaum, 2005); Simulation of Human
 3   Exposure and Dose System (SHEDS, Burke et al., 2001); Air Pollutants Exposure Model
 4   (APEX, U.S. EPA, 2006a; 2006b); and Modeling Environment for Total Risk Studies
 5   (MENTOR, Georgopoulos et al., 2005; Georgopoulos and Lioy, 2006). European efforts have
 6   produced some formulations with similar general attributes as the above U.S. models but,
 7   generally, involving simplifications in some of their components. Examples of European models
 8   addressing exposures to photochemical oxidants (specifically, ozone) include the Air Pollution
 9   Exposure Model (AirPEx, Freijer et al., 1998), which basically replicates the pNEM approach
10   and has been applied to the Netherlands, and the Air Quality Information System Model
11   (AirQUIS, Clench-Aas et al., 1999).
12         The NEM/pNEM, SHEDS, APEX, and MENTOR for One-Atmosphere studies
13   (MENTOR-1A) families of models provide exposure estimates defined by concentration and
14   breathing rate for each individual exposure event, and then average these estimates over periods
15   typically ranging from one hour to one year. These models allow simulation of certain aspects of
16   the variability and uncertainty in the principal factors affecting exposure. An alternative approach
17   is taken by the HAPEM family of models that typically provide annual average exposure
18   estimates based on the quantity of time spent per year in each combination  of geographic
19   locations and microenvironments. The NEM, SHEDS, APEX, and MENTOR-type models are
20   therefore expected to be more appropriate for pollutants with complex chemistry such as SO2,
21   and could provide useful information for enhancing related health assessments.
22         More specifically, regarding the consideration of population demographics and activity
23   patterns:

24         • pNEM divides the population of interest into representative cohorts based on the
25            combinations of demographic characteristics (age, gender, and employment),
26            home/work district, residential cooking fuel and replicate number, and then assigns an
27            activity diary record from the CHAD to each cohort according to demographic
28            characteristic, season, day-type (weekday/weekend) and temperature.
29         • HAPEM6 divides the population of interest into demographic groups based on  age,
30            gender and race, and then for each demographic group/day-type (weekday/weekend)
31            combination, selects multiple activity patterns  randomly (with replacement) from
32            CHAD and combines them to find the averaged annual time allocations for group
33            members in each census tract for different day types.

     May 2008                             C-38              DRAFT—DO NOT QUOTE OR CITE

-------
 1         • SHEDS, APEX, and MENTOR-1A generate population demographic files, which contain
 2           a user-defined number of person records for each census tract of the population based
 3           on proportions of characteristic variables (age, gender, employment, and housing)
 4           obtained from the population of interest, and then assign a matching activity diary
 5           record from CHAD to each individual record of the population based on the
 6           characteristic variables. It should be mentioned that, in the formulations of these
 7           models, workers may commute from one census tract to another census tract for work.
 8           So, with the specification of commuting patterns, the variation of exposure
 9           concentrations due to commuting between different census tracts can be captured.

10         The conceptual approach originated by the SHEDS models was modified and expanded for

11   use in the development of MENTOR-1 A. Flexibility was incorporated into this modeling system,

12   such as the option of including detailed indoor chemistry and other relevant microenvironmental

13   processes, and providing interactive linking with CHAD for consistent definition of population

14   characteristics and activity events (Georgopoulos et al., 2005).
     Table C-1.  The Essential Attributes of the pNEM, HAPEM, APEX, SHEDS, and MENTOR-1A

Exposure Estimate
Characterization of the
High-End Exposures
Typical Spatial
Scale/Resolution
Temporal
Scale/Resolution
Population Activity
Patterns Assembly
Microenvironment
Concentration
Estimation
Microenvironmental
(ME) Factors
Specification of Indoor
Source Emissions
Commuting Patterns
Exposure Routes
Potential Dose
Calculation
PNEM
Hourly averaged
Yes
Urban areas/Census
tract level
A yr/one h
Top-down approach
Non-steady-state and
steady-state mass
balance equations
(hard-coded)
Random samples
from probability
distributions
Yes (gas-stove,
tobacco smoking)
Yes
Inhalation
Yes
HAPEM
Annual averaged
No
Ranging from
urban to national/
Census tract
level
A yr/one h
Top-down
approach
Linear
relationship
method (hard-
coded)
Random
samples from
probability
distributions
Available; set to
zero in HAPEM6
Yes
Inhalation
No
APEX
Hourly averaged
Yes
Urban area/Census
tract level
A yr/one h
Bottom-up "person-
oriented" approach
Non-steady-state mass
balance and linear
regression (flexibility of
selecting algorithms)
Random samples from
probability distributions
Yes (multiple sources
defined by the user)
Yes
Inhalation
Yes
SHEDS
Activity event based
Yes
Urban areas/Census tract
level
A yr/event based
Bottom-up "person-
oriented" approach
Steady-state mass
balance equation
(residential) and linear
regression (non-
residential) (hard-coded)
Random samples from
probability distributions
Yes (gas-stove, tobacco
smoking, other sources)
Yes
Inhalation
Yes
MENTOR-1A
Activity event based
Yes
Multiscale/ Census tract
level
A yr/activity event based
time step
Bottom-up "person-oriented"
approach
Non-steady-state mass
balance equation with indoor
air chemistry module or
regression methods
(flexibility of selecting
algorithms)
Random samples from
probability distributions
Yes (multiple sources
defined by the user)
Yes
Multiple (optional)
Yes
     May 2008
C-39
DRAFT—DO NOT QUOTE OR CITE

-------

Physiologically Based
Dose
Variability/Uncertainty
PNEM
No
Yes
HAPEM
No
No
APEX
No
Yes
SHEDS
Yes
Yes
MENTOR-1A
Yes
Yes (Various "Tools")
 1         The essential attributes of the pNEM, HAPEM, APEX, SHEDS, and MENTOR-1A models
 2    are elaborated in Table C-l.
 3         NEM/pNEM implementations have been extensively applied to ozone studies in the 1980s
 4    and 1990s. The historical evolution of the pNEM family of models of OAQPS started with the
 5    introduction of the first NEM model in the 1980s (Biller et al., 1981). The first such
 6    implementations of pNEM/Os in the 1980s used a regression-based relationship to estimate
 7    indoor ozone concentrations from outdoor concentrations. The second generation of pNEM/Os
 8    was developed in 1992 and included a simple mass balance model to estimate indoor ozone
 9    concentrations. A report by Johnson et al. (2000) describes this version of pNEM/Os and
10    summarizes the results of an initial application of the model to 10 cities. Subsequent
11    enhancements to pNEM/Os and its input databases included revisions to the methods used to
12    estimate equivalent ventilation rates, to determine commuting patterns, and to adjust ambient
13    ozone levels to simulate attainment of proposed NAAQS. During the mid-1990s, the
14    Environmental Protection Agency applied updated versions of pNEM/Os to three different
15    population groups in selected cities: (1) the general population of urban residents, (2) outdoor
16    workers, and (3) children who tend to spend more time outdoors than the average child. This
17    version of pNEM/Os used a revised probabilistic mass balance model to determine ozone
18    concentrations over one-h periods in indoor and in-vehicle microenvironments (Johnson, 2001).
19         In recent years, pNEM has been replaced by (or "evolved to") the Air Pollution Exposure
20    Model (APEX). APEX differs from earlier pNEM models in that the probabilistic features of the
21    model are incorporated into a Monte Carlo framework (U.S. EPA, 2006a; 2006c; Langstaff,
22    2007). Like SHEDS and MENTOR-1 A, instead of dividing the population-of-interest into a  set
23    of cohorts, APEX generates individuals as if they were being randomly sampled from the
24    population. APEX provides each generated individual with a demographic profile that  specifies
25    values for all parameters required by the model. The values are selected from distributions and
26    databases  that are specific to the age, gender, and other specifications stated in the demographic
      May 2008
C-40
DRAFT—DO NOT QUOTE OR CITE

-------
 1   profile. The EPA has applied APEX to the study of exposures to ozone and other criteria
 2   pollutants; APEX can be modified and used for the estimation of 862 exposures, if required.
 3         Reconfiguration of APEX for use with SO2 or other pollutants would require significant
 4   literature review, data analysis, and modeling efforts. Necessary steps include determining spatial
 5   scope and resolution of the model; generating input files for activity data, air quality and
 6   temperature data; and developing definitions for microenvironments and pollutant-
 7   microenvironment modeling parameters (penetration and proximity factors, indoor source
 8   emissions rates, decay rates, etc.) (ICF Consulting, 2005). To take full advantage of the
 9   probabilistic capabilities of APEX, distributions of model input parameters should be used
10   wherever possible.

     C.3. Characterization of Ambient Concentrations  of
     SO2 and Related Air Pollutants
11         As mentioned earlier, background and regional outdoor concentrations of pollutants over a
12   study domain may be estimated through emissions-based mechanistic modeling, through  ambient
13   data based modeling, or through a combination of both. Emissions-based models calculate the
14   spatio-temporal fields of the pollutant concentrations using precursor emissions and
15   meteorological conditions as inputs and using numerical representations of transformation
16   reactions to drive outputs.  The ambient data based models typically calculate spatial or spatio-
17   temporal distributions of the pollutant through the use of interpolation schemes, based on either
18   deterministic or stochastic models for allocating monitor station observations to the nodes of a
19   virtual regular grid covering the region of interest. The geostatistical technique of kriging
20   provides various standard procedures for generating an interpolated spatial distribution for a
21   given time, from data at a  set of discrete points. Kriging approaches were evaluated by
22   Georgopoulos et al. (1997) in relation to the calculation of local ambient ozone concentrations
23   for exposure assessment purposes, using either monitor observations or regional/urban
24   photochemical model outputs. It was found that kriging is severely limited by the nonstationary
25   character of the concentration patterns of reactive pollutants; so the advantages of this method in
26   other fields of geophysics  do not apply here. The above study showed that the appropriate
27   semivariograms had to be hour-specific, complicating the automated reapplication of any purely
28   spatial interpolation over an extended time period.

     May 2008                             C-41               DRAFT—DO NOT QUOTE OR CITE

-------
 1         Spatio-temporal distributions of pollutant concentrations such as ozone, PM, and various
 2   air toxics have alternatively been obtained using methods of the Spatio-Temporal Random Field
 3   (STRF) theory (Christakos and Vyas, 1998a; b). The STRF approach interpolates monitor data in
 4   both space and time simultaneously. This method can thus analyze information on temporal
 5   trends which cannot be incorporated directly in purely spatial interpolation methods such as
 6   standard kriging. Furthermore, the STRF method can optimize the use of data which are not
 7   uniformly sampled in either space or time. STRF was further extended within the Bayesian
 8   Maximum Entropy (BME) framework and applied to ozone interpolation studies (Christakos and
 9   Hristopulos, 1998; Christakos and Kolovos, 1999; Christakos, 2000). It should be noted that
10   these studies formulate an over-arching scheme for linking air quality with population dose and
11   health effects; however, they are limited by the fact that they do not include any
12   microenvironmental effects. MENTOR has incorporated STRF/BME methods as one  of the steps
13   for performing a comprehensive analysis of exposure to ozone and PM (Georgopoulos et al.,
14   2005).
15         The issue of subgrid variability (SGV) from the perspective of interpreting and  evaluating
16   the outcomes of grid-based, multiscale, photochemical air quality simulation models is discussed
17   in (Ching et al., 2006), who suggest a framework that can provide for qualitative judgments on
18   model performance based on comparing observations to the grid predictions and its SGV
19   distribution. From the perspective of Population Exposure Modeling, the most feasible/practical
20   approach for treating subgrid variability of local concentrations is probably through 1) the
21   identification and proper  characterization of an adequate number of outdoor microenvironments
22   (potentially related to different types of land use within the urban area as well as to proximity to
23   different types of roadways) and 2) then, concentrations in these microenvironments will have to
24   be sadjusted from the corresponding local background ambient concentrations through either
25   regression of empirical data or various types of local atmospheric dispersion/transformation
26   models. This is discussed further in the next section.

     C.4. Characterization of Microenvironmental
     Concentrations
27         Once the background and local ambient spatio-temporal concentration patterns  have been
28   derived, microenvironments that can represent either outdoor or indoor settings when individuals

     May 2008                             C-42              DRAFT—DO NOT QUOTE OR CITE

-------
 1    come in contact with the contaminant of concern (e.g., 862) must be characterized. This process

 2    can involve modeling of various local sources and sinks, and interrelationships between ambient

 3    and microenvironmental concentration levels. Three general approaches have been used in the

 4    past to model microenvironmental concentrations:

 5         • Empirical (typically linear regression) fitting of data from studies relating ambient/local
 6            and microenvironmental concentration levels to develop analytical relationships.

 7         • Parameterized mass balance modeling over, or within, the volume of the
 8            microenvironment. This type of modeling has ranged from very simple formulations,
 9            i.e. from models assuming ideal (homogeneous) mixing within the microenvironment
10            (or specified portions of it) and only linear physicochemical transformations (including
11            sources and sinks), to models incorporating analytical solutions of idealized dispersion
12            formulations (such as Gaussian plumes), to models that take into account aspects of
13            complex multiphase chemical and physical interactions and nonidealities in mixing.

14         • Detailed Computational Fluid Dynamics (CFD) modeling of the outdoor or indoor
15            microenvironment, employing either a Direct Numerical Simulation (DNS) approach, a
16            Reynolds Averaged Numerical Simulation (RANS) approach, or a Large Eddy
17            Simulation (LES) approach, the latter typically for outdoor situations (see, e.g., Chang
18            and Meroney, 2003; Chang, 2006; Milner et al., 2005).

19         Parameterized mass balance modeling is the approach currently preferred for exposure

20    modeling for populations. As discussed earlier, the simplest microenvironmental setting

21    corresponds to a homogeneously mixed compartment, in contact with possibly both

22    outdoor/local environments as well as other microenvironments. The air quality of this idealized

23    microenvironment is affected mainly by the following processes:

24         •  Transport processes: These can include advection/convection and dispersion that are
25            affected by local processes and obstacles such as vehicle induced turbulence, street
26            canyons, building structures, etc.

27         •  Sources and sinks: These can include local outdoor emissions, indoor emissions, surface
28            deposition, etc.

29         •  Transformation processes: These can include local outdoor as well as indoor gas and
30            aerosol phase chemistry, such as formation of secondary organic and inorganic aerosols.

31         Exposure modeling also requires information on activity patterns to determine time spent

32    in various microenvironments and estimates of inhalation rates to characterize dose. The next

33    two subsections describe recent work done in these areas.
      May 2008                              C-43              DRAFT—DO NOT QUOTE OR CITE

-------
      C.4.1.       Characterization of Activity Events
 1         An important development in inhalation exposure modeling has been the consolidation of
 2    existing information on activity event sequences in the Consolidated Human Activity Database
 3    (CHAD) (McCurdy, 2000; McCurdy et al., 2000). Indeed, most recent exposure models are
 4    designed (or have been re-designed) to obtain such information from  CHAD which incorporates
 5    24-h time/activity data developed from numerous  surveys. The surveys include probability-based
 6    recall studies conducted by Environmental Protection Agency and the California Air Resources
 7    Board, as well as real-time diary studies conducted in individual U.S. metropolitan areas using
 8    both probability-based and volunteer subject panels. All ages of both genders are represented in
 9    CHAD. The data for each subject consist of one or more days of sequential activities, in which
10    each activity is defined by start time, duration, activity type (140 categories), and
11    microenvironment classification (110 categories). Activities vary from one min to one h in
12    duration, with longer activities being subdivided into clock-hour durations to facilitate exposure
13    modeling. A distribution of values for the ratio of oxygen uptake rate  to body mass (referred to as
14    metabolic equivalents or METs) is provided for each activity type listed in CHAD. The forms
15    and parameters of these distributions were determined through an extensive review of the
16    exercise and nutrition literature. The primary source of distributional data was Ainsworth et al.
17    (1996), a compendium developed specifically to facilitate the coding  of physical activities and to
18    promote comparability across studies.

      C.4.2.  Characterization of Inhalation  Intake and  Uptake
19         Use of the information in CHAD provides a rational way for incorporating realistic intakes
20    into exposure models by linking inhalation rates to activity information. As mentioned earlier,
21    each cohort of the pNEM-type models, or each (virtual or actual) individual of the SHEDS,
22    MENTOR, APEX, and HAPEM models, is assigned an exposure event sequence derived from
23    activity diary data. Each exposure event is typically defined by a start time, a duration,
24    assignments to a geographic location and microenvironment, and an indication of activity level.
25    The most recent versions of the above models have defined activity levels using the activity
26    classification coding scheme incorporated into CHAD. A probabilistic module within these
27    models converts the activity classification code  of each exposure event to an energy expenditure
28    rate, which in turn is converted into an estimate of oxygen uptake rate. The oxygen uptake rate is

      May 2008                              C-44               DRAFT—DO NOT QUOTE OR CITE

-------
 1    then converted into an estimate of total ventilation rate (VE), expressed in liters min'l. Johnson
 2    (2001) reviewed briefly the physiological principles incorporated into the algorithms used in
 3    pNEM to convert each activity classification code to an oxygen uptake rate and describes the
 4    additional steps required to convert oxygen uptake to (VE)
 5         McCurdy (1997b; a; 2000) has recommended that the ventilation rate should be estimated
 6    as a function of energy expenditure rate. The energy expended by an individual during a
 7    particular activity can be expressed as EE = (MET)(RMR) in which EE is the average energy
 8    expenditure rate (kcal min"1) during the activity and RMR is the resting metabolic rate of the
 9    individual expressed in terms of number of energy units expended per unit of time (kcal min"1).
10    MET (the metabolic equivalent of tasks) is a ratio specific to the activity and is dimensionless. If
11    RMR is specified for an individual, then the above equation requires only an activity-specific
12    estimate of MET to produce an estimate of the energy expenditure rate for a given activity.
13    McCurdy et al. (2000) developed distributions of MET for the activity classifications appearing
14    in the CHAD database.
15         An issue that should be mentioned in closing is that of evaluating comprehensive
16    prognostic exposure modeling studies, for either individuals or populations, with field data.
17    Although databases that would be adequate for performing a comprehensive evaluation are not
18    expected to be available any time soon, there have been a number of studies, reviewed in earlier
19    sections  of this chapter, which can be used to start building the necessary information base. Some
20    of these studies report field observations of personal, indoor, and outdoor levels and have also
21    developed simple semi-empirical personal exposure models that were parameterized using the
22    observational data and regression techniques.
23         In  conclusion, though existing inhalation exposure  modeling  systems have evolved
24    considerably in recent years, limitations of available modeling methods and data in relation to
25    potential SC>2 studies should be taken into account. Existing prognostic modeling systems for
26    inhalation exposure can in principle be directly applied to, or adapted for, 862 studies; APEX,
27    SHEDS, and MENTOR-1A are candidates. However, such applications would be constrained by
28    data limitations such as ambient characterization at the local scale and by lack of quantitative
29    information for indoor sources and sinks.
      May 2008                              C-45               DRAFT—DO NOT QUOTE OR CITE

-------
       Annex D. Controlled Human Exposure
Table D-1  Effects of medications on SO2-induced changes in lung function among human
        subjects.
STUDY
Bigby and
Boushey
(1993)
Lazarus
etal.
(1997)
Field et al.
(1996)
Gong
etal.
(1996)
Gong
etal.
(2001)
CONC.
0.25-
8.0 ppm
0.25-
8.0 ppm
0.25-
8.0 ppm
0.75
ppm
0.75
ppm
DURATION
4 min
4 min
3 min
10 min
10 min
SUBJECTS
10
asshmatics
12
asthmatics
31
asthmatics
10
asthmatics
11 asthmatics
EXPOSURE STATUS
Increasing concentrations of SO2 during
voluntary eucapnic hyperpnea (20 L/min)
preceded by administration of nedocromil
sodium (baseline, placebo, 2 mg, 4 mg, 8
mg).
Subjects exposed using a mouthpiece to
filtered air and increasing concentrations
of SO2 during eucapnic hyperventilation
(20 L/min). Exposures occurred following
pretreatment with zafirlukast (20 mg) or
placebo.
Increasing concentrations of SO2
(including clean air exposure) in an
exposure chamber during voluntary
eucapnic hyperpnea (35 L/min) preceded
by administration of placebo, ipratropium
bromide (15 subjects), morphine (15
subjects), or indomethacin (16 subjects).
Subjects exposed to SO2 or clean air in a
chamber while performing light exercise
(29 L/min) at 1, 12, 18, and 24 h after
pretreatment with salmeterol xinafoate or
placebo (each subject exposed 4 times).
Exposure to SO2 or clean air following
three days of treatment with montelukast
or placebo (each subject exposed 4
times). Exposures conducted in an
exposure chamber during moderate levels
of exercise (35 L/min).
EFFECTS
Treatment with the inhaled anti-inflammatory agent,
nedocromil sodium, significantly increased the
concentration of SO2 required to produce an 8 unit
increase in sRaw. Increasing the dose of nedocromil
sodium from 2 mg to 8 mg did not significantly affect
the response.
Compared with placebo, zafirlukast significantly
increased the SO2 concentration required to
produce an 8 unit increase in sRaw. This effect was
observed with challenges occurring both at 2 and 10
h following treatment.
Both ipratropium bromide and morphine reduced the
responsiveness to SO2, significantly increasing the
SO2 concentration required to reduce specific
airway conductance by 35%. Similarly, indomethacin
was observed to attenuate airway responsiveness
to SO2, however, this effect was smaller than what
was observed with either ipratropium bromide or
morphine.
Observed a significant protective effect of salmeterol
xinafoate at 1 and 12-h post-dosing. Following
exercise/SO2 exposure at 1, 12, 18, and 24 h, FEV,
decreased (versus preexposure) by 7, 12, 25, and
26%, respectively. Exercise with SO2 resulted in an
approximate 26% decrease in FEVi at all time
points with placebo.
Reported a statistically significant SO2-induced
increase in eosinophil count in induced sputum.
Measures of lung function (FEVi and sRaw), as well
as respiratory symptoms and eosinophil count all
showed significant improvement after pretreatment
with montelukast.
Table D-2  Summary of new studies of controlled human exposure to SO2.
STUDY
Trenga
etal.
(2001)
Devalia
etal.
(1994)
CONC.
0.1,0.25
ppm
0.2 ppm
DURATION
10 min
6h
SUBJECTS
17
asthmatics
10
asthmatics
EXPOSURE STATUS
SO2-sensitive asthmatics exposed to SO2
via mouthpiece while performing mild to
moderate levels of exercise. Exposures
preceded by 45 min exposures to filtered
air or ozone (0.12 ppm), with or without
pretreatment with dietary antioxidants.
Exposures to filtered air, as well as 0.2
ppm SO2 and 0.4 ppm NO2, conducted
separately and in combination in an
exposure chamber (subjects at rest). All
subjects sensitive to inhaled house dust
mite antigen.
EFFECTS
Exposure to ozone slightly increased bronchial
responsiveness to SO2 as measured by FEV1 and
peak expiratory flow. Pretreatment with dietary
antioxidants was shown to have a protective effect
on respiratory response, particularly among
individuals with greater sensitivity to SO2.
Neither SO2 nor NO2, alone or in combination,
significantly affected FEV,. The combination of SO2
and NO2 significantly reduced the amount of inhaled
allergen (60.5% change, p = 0.015) required to
produce a 20% decrease in FE\/i (PD20FEVO. Both
SO2 and NO2 alone reduced PD20FEV!, but this
reduction was not statistically significant (32.2% (p =
0.506), and 41 .2% (p = 0.125), respectively).
May 2008
D-1
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Rusznak
etal.
(1996)
Tunnicliffe
etal.
(2001)
Tunnicliffe
etal.
(2003)
Routledge
etal.
(2006)
Nowak
etal.
(1997)
Trenga
etal.
(1999)
Winterton
etal.
(2001)
Gong et
al. (1995)
CONC.
0.2 ppm
0.2 ppm
0.2 ppm
0.2 ppm
0.25-
2.0 ppm
0.5 ppm
0.5 ppm
0.5, 1.0
ppm
DURATION
6h
1 h
1 h
1 h
3 min
10 min
10 min
10 min
SUBJECTS
10
asthmatics
12 healthy
adults, 12
asthmatics
12 healthy
adults, 12
asthmatics
20 older
adults with
coronary
artery
disease (age
52-74), 10
healthy older
adults (age
56-75)
786 adults
47 asthmatic
62
asthmatics
14
asthmatics
EXPOSURE STATUS
Exposures to filtered air and a
combination of 0.2 ppm SO2 and 0.4 ppm
NO2 in an exposure chamber (subjects at
rest). All subjects sensitive to inhaled
house dust mite antigen.
Exposures (head dome) at rest to filtered
air and 0.2 ppm SO2.
Exposures (head dome) at rest to filtered
air and 0.2 ppm SO2.
Exposures (head dome) at rest to filtered
air, as well as 0.2 ppm SO2 and ultra-fine
carbon particles (50 pg/m3), separately
and in combination.
Mouthpiece exposures to filtered air and
increasing concentrations of SO2 during
eucapnic hyperventilation (40 L/min).
Subjects exposed to SO2 via mouthpeice
while performing light to moderate levels
of exercise.
Subjects exposed to SO2 via mouthpiece
while performing light to moderate levels
of exercise.
Exposure to SO2 and filtered air were
conducted in an exposure chamber during
low, moderate, and heavy levels of
exercise (target ventilation ranges of 20-
29, 30-39, and 40-49 L/min).
EFFECTS
Confirmed findings of Devalia et al. and further
observed that the combination of SO2 and NO2
enhanced airway responsiveness to an inhaled
allergen up to 48 h post-exposure (maximal
response at 24 h).
Among healthy subjects, an SO2-induced increase
in heart rate variability (total power) was observed,
while a reduction in heart rate variability with SO2
versus air was observed in asthmatics.
Exposures to SO2 at 0.2 ppm did not have a
significant effect on lung function, respiratory
symptoms, markers of inflammation, or antioxidant
levels in healthy adults or mild asthmatics.
In healthy subjects, exposure to SO2 alone
significantly decreased heart rate variability 4 h
post-exposure compared to clean air. No effect was
observed in subjects with coronary artery disease.
The combination of SO2 and carbon particles did not
affect heart rate variability in either group. SO2 was
not observed to affect markers of inflammation or
coagulation.
Among individuals who were not hyperresponsive to
methacholine, less than 1% were found to be
hyperresponsive to SO2. However, more than 22%
of the individuals who were hyperresponsive to
methacholine were also hyperresponsive to SO2.
Individuals were considered hyperresponsive to SO2
when exposure resulted in a 20% or greater
decrease in FEVi versus baseline.
An SO2-induced decrease in FEVi of at least 8%
was observed in 53% of the subjects (range 8-44%).
Increases in respiratory symptoms were significantly
associated with decreases in FEV,. Among SO2-
sensitive subjects, severity of asthma (as defined by
medication use) was not a significant predictor of
the level of response. It is not clear whether the
response was adjusted for the effects of exercise in
clean air.
Subjects who experienced at least at 12% decrease
in FEVi following exposure were considered to be
sensitive to SO2. Out of 58 subjects who were
genotyped for the polymorphism at position -308 in
the promoter region of TNF-a, 21% (N: 12) were
sensitive to SO2. Sensitivity to SO2 was found to be
associated with the homozygous wild type allele
(GG) (12 of 12 responders versus 28 of 46 subjects
who were not responsive to SO2).
For the average individual, increasing SO2
concentration resulted in a significant decrement in
lung function (decrease in FEV1 and increase in
sRaw) as well as a significant increase in
respiratory symptoms. Increasing SO2 concentration
had a greater effect on lung function and respiratory
symptoms than did increasing level of exercise.
May 2008
D-2
DRAFT—DO NOT QUOTE OR CITE

-------
           Annex E. Toxicological Studies
Table E-1.  Physiological effects of SO2 exposure.
STUDY
CONC.
DURATION
SPECIES
EFFECTS
Acute and Subacute Exposures
Lewis and
Kirch ner
(1984)
Barthelemy
etal.
(1988)
Amdur
etal.
(1983)
Conner et
al. (1985)
Douglas
etal.
(1994)
10 or 30 ppm
(26.2 or 78.6
mg/m3);
intratracheal
0.5 or 5 ppm
(1.3 or 13.1
mg/m3);
intratracheal
~1 ppm
(2.62 mg/m3);
head only
1 ppm
(2.62 mg/m3);
nose only
5 ppm
(13.1 mg/m3);
whole body
5 min
45 min
1 h
3 h/day for 6 days;
animals evaluated
up to 48 h post-
exposure
2 h/day for 13wks
from birth
Mongrel dogs; male
and female; age and
weight NR; N: 5-15
/group
Rabbit; sex NR;
adult; mean 2.0 kg;
N: 5-9/ group; rabbits
were mechanically
ventilated
Hartley guinea pig,
male, age NR,
200-300 g,N:
8-23/group
Hartley guinea pig,
male, age NR,
250-320 g,N:< 18
group/time point
New Zealand White
rabbit, male and fe-
male, 1 day old,
weight NR, N: 3-4/
group, immunized
against Alternaria
tenuis
Initial transient bronchoconstriction approximately 10 min in duration
followed by a gradual change in pulmonary mechanics (43% increase in
airway resistance and 30% decrease in dynamic compliance) 4 hrs
following 30 ppm but not 10 ppm SO2.
Lung resistance increased by 16% and 50% in response to 0.5 and 5 ppm
SO2, respectively. Bivagotomy had no effect on 5 ppm SO2-induced
increases in lung resistance. Reflex bronchoconstrictive response to
phenyldiguanide (intravenously administered) was eliminated by exposure
to SO2 but SO2 had no effect on lung resistance induced by intravenously-
administered histamine. Authors concluded that (1) vagal reflex is not
responsible for SO2-induced increase in lung resistance at 45 min;
(2) transient alteration in tracheobronchial wall following SO2 exposure
may have reduced accessibility of airway nervous recaptors to
phenyldiguanide.
An 11% increase in pulmonary resistance and 12% decrease in dynamic
compliance were observed. Neither effect persisted into the 1 h period
following exposure. No effects were observed for breathing frequency,
tidal volume, or min volume.
No effect was observed on residual volume, functional reserve capacity,
vital capacity, total lung capacity, respiratory frequency, tidal volume,
pulmonary resistance, pulmonary compliance, diffusing capacity for CO or
alveolar volume at 1 or 48 h after last exposure.
No effects on lung resistance, dynamic compliance, transpulmonary
pressure, tidal volume, respiration rate, or min volume.
Subchronic and Chronic Exposure
Scanlon
etal.
(1987)
Smith et al.
(1989)
15 or 50 ppm
(39.3 or
131 mg/m3);
intratracheal
exposure
1 ppm
(2.62 mg/m3);
whole body
2 h/day, 4 or 5
days/wk, for 5 mos
(low dose group)
or 10-11 mos (high
dose group);
authors stated that
physiological
changes were
observed within 5
mos; 7-9 mo
recovery period
5 h/day, 5 days/wk
for 4 mos
Mongrel dogs, adult,
sexNR, 10-20 kg; N:
3-4/group (3 hyper-
responsive, 3
hyporesponsive, and
1 avg responsive)
Sprague-Dawley rat,
male, young adult,
initial weight NR, N:
1 2-1 5/ data point
At 15 ppm, there was no clinical evidence of bronchitis; pulmonary
resistance increased by 35-38% in 2 of 3 dogs, and dynamic lung
compliance decreased in 1 of 3 dogs, but the physiological changes were
not significant for the group as a whole. At 50 ppm, cough and mucous
hypersecretion were observed; the symptoms ceased during the recovery
period. Pulmonary resistance increased by 56% during the treatment
period and an additional 28% during the recovery period for a total
increase of 99%; dynamic lung compliance decreased in 2 of 4 dogs and
increased in 1 of 4 dogs during treatment but there were no significant
changes in the group as a whole. Authors considered 15 ppm to be the
lower limit of exposure that failed to produce physiological changes.
Physiological tests were conducted in anesthetized animals, many while
rat breathed spontaneously and during paralysis. SO2 exposure resulted
in 11% decrease in residual volume during paralysis and reduced
quasistatic compliance in paralyzed animals. Authors noted that because
residual volume was only decreased in paralyzed rats and magnitude of
effect was very small, it may have been due to chance. Quasistatic com-
pliance values observed to be very high in controls; may have accounted
for effect in treatment group.
May 2008
E-3
DRAFT—DO NOT QUOTE OR CITE

-------
Table E-2.   Inflammatory responses following SO2 exposure.
STUDY
CONC.
DURATION
SPECIES
EFFECTS
Acute/Subacute/Subchronic
Clarke
etal.
(2000)
Meng
etal.
(2005a)
Langley-E
vans et al.
(1996)
Conner et
al. (1989)
Park etal.
(2001 a)
Li et al.
(2007)
10 ppm
(26.2 mg/m3);
nose only
14, 28, or 56
mg/m3; (5.35,
10.7, or
21.4 ppm); whole
body
5, 50, or 100 ppm
(13.1, 131, or
262 mg/m3); whole
body
1 ppm
(2.62 mg/m3);
nose only
0.1 ppm
(0.26 mg/m3);
whole body; with
and without
exposure to
ovalbumin
2 ppm (5.24
mg/m3) with and
without exposure
to ovalbumin
4h
4 h/day for 7
days
5 h/day for 7-28
days
3 h/day for 5
days; bronchial-
veolar lavage
performed daily
5 h/day for 5
days
1 h/day for 7
days
Outbred Swiss
mouse, female, age,
weight NR, N: 10/
experimental value
Kunming albino
mouse, male, age
NR, 18-22 g, N:
10/group
Wistar rat, male,
7 wks old, weight
NR, N: 4-5/
treatment group, 8
controls
Hartley guinea pig,
male, age NR,
250-320 g,N: 4
Dunkin-Hartley
guinea pig, male,
age NR, 250-350 g,
N: 7-12/group
Wistar rats, male,
age NR
No evidence was seen of inflammatory response in terms of total cell
number, lymphocyte/polymorphonuclear leukocytes differentials, or total
protein level taken from BAL fluid.
In lung tissue, in vivo SO2 exposure (low, mid concentrations)
significantly elevated levels of the pro-inflammatory cytokines
interleukin-6 and tumor necrosis factor-a, but did not affect levels of the
anti-inflammatory cytokine transforming growth factor-p1 . In serum, the
only effect observed was a low-dose elevation of tumor necrosis factor-a.
No lung injury was observed and evidence of inflammatory response was
only observed in the 100 ppm group. A 4-fold increase in BAL fluid
leukocyte numbers was observed in the 100 ppm group at day 14; the
increase lessened at days 21 and 28 but remained higher than controls.
The number of macrophages in BAL fluid was increased at day 28 in the
100 ppm group. Neutrophil numbers were 120 times higher than controls
at day 14 in the 100 ppm group but returned to normal by day 21. Blood
neutrophils were depleted in rats exposed to 50 ppm on days 7-21 but
were increased in rats exposed to 5 ppm (significant) and 100 ppm
(non-significant) at day 14. Lung epithelial permeability was not affected.
No change in numbers of total cells and neutrophils, protein levels or
enzyme activity in lavage fluid following SO2 exposure.
After bronchial challenge, the ovalbumin/SO2.exposed group had
significantly increased eosinophil counts in BAL fluids compared with all
other groups, including the SO2 group. The bronchial and lung tissue of
this group showed infiltration of inflammatory cells, bronchiolar epithelial
damage, and mucus and cell plug in the lumen.
Increased number of inflammatory cells in BALfluid, increased levels of
MUC5AC and ICAM-1 and an enhanced histopathological response
compared with those treated with ovalbumin or SO2 alone
Table E-3.   Effects of SO2 exposure on host lung defenses.
STUDY
CONC.
DURATION
SPECIES
EFFECTS
Clearance - Subchronic
Wolff etal.
(1989)
5 ppm (13.1
mg/m3); nose
only
2 h/day, 5
days/wk for 4
wks
F344/CM rat, male
and female, 10-11
wks old, weight NR,
N: 6/sex/group
There was no effect on pulmonary clearance of radiolabeled alumi-
nosilicate particles (MMAD 1.0 pM).
Immune Responses - Acute/Subacute
Jakab et al.
(1996)
10 ppm (26.2
mg/m3); nose
only
4h
Specific patho-
gen-free white Swiss
mice, female, 5 wks
old, 20-23 g,N:5/
group
No effect was observed on in situ Fc-receptor-mediated phagocytosis of
sheep red blood cells by AM, which was assessed 3 days after exposure
to SO2
May 2008
E-4
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Clarke et al.
(2000)
Azoulay-Du
puis et al.
(1982)
CONC.
10 ppm (26.2
mg/m3) SO2;
nose only
10 ppm (26.2
mg/m3); whole
body
DURATION
4h
24 h, 1 wk, 2
wks, or 3 wks
SPECIES
Outbred Swiss
mouse, female, age
and weight NR, N:
10/experimental
value
OFi mice, female,
age NR, mean 20.6
g, N: 768 (32/group)
EFFECTS
No effect on in situ AM phagocytosis (data not shown) or on intrapul-
monary bactericidal activity toward Staphylococcus aureus.
Respiratory challenge with Klebsiella pneumoniae resulted in increased
mortality and decreased survival time in the 1,2, and 3 wk SO2 exposure
groups compared to controls. Differences did not correlate with exposure
length.
Table E-4.   Effects of SO2 exposure on hypersensitivity/allergic reactions.
STUDY
CONC.
DURATION
SPECIES
EFFECTS
Antigen Sensitization/Allergic Reactions - Acute/Subacute
Amdur et al.
(1988)
Abraham
etal. (1981)
Parketal.
(2001 a)
Riedel et al.
(1988)
1 ppm
5 ppm (13.1
mg/m3); head
only
0.1 ppm
(0.26 mg/m3);
whole body; with
and without
exposure to
ovalbumin
0.1, 4.3, or
16.6 ppm (0,
0.26, 11.3, or
43.5 mg/m3);
whole body;
animals were
sensitized to
ovalbumin on the
last 3 days of
exposure.
1-h
4h
5 h/day for 5 days.
8 h/day for 5 days
Guinea pig, n=8
Sheep, sex and age
NR, mean weight 38
± 7 kg, N: 7/group
Dunkin-Hartley
guinea pig, male,
age NR, 250-350 g,
N: 7-12/group
Perlbright-White
Guinea pig, female,
age NR, 300-350 g,
N: 5 ore/group (14
controls)
Airway responsiveness to acetylcholine was measured 2 h following SO2
exposure. No changes were observed.
Acute exposure to 5 ppm SO2 did not produce significant airway changes
(pulmonary resistance, static compliance, dynamic compliance, tidal
volume, breathing frequency) in either normal or allergic (sensitized to
Ascaris suum antigen) sheep, nor increase airway reactivity (measured as
pulmonary resistance increase after aerosolized carbachol provocation) in
normal sheep. However, 5 ppm SO2 did significantly increase airway
reactivity in allergic sheep, which have antigen-induced airway responses
similar to humans with allergic airway disease; may model airway
responses to SO2 in a sensitive human subpopulation.
After bronchial challenge, the ovalbumin/SO2-exposed group had signifi-
cantly increased enhanced pause (indicator of airway obstruction) com-
pared with all other groups, including the SO2 group. Authors concluded
low level SO2 may enhance the development of ovalbumin-induced
asthmatic reactions in guinea pigs.
Bronchial provocation with ovalbumin was conducted every other day for 2
wks, starting at 1 wk after last exposure. Numbers of animals displaying
symptoms of bronchial obstruction after ovalbumin provocation increased
in all SO2 groups compared to air-exposed groups. Anti-ovalbumin
antibodies (IgG total and lgG1) were increased in BAL fluid and serum of
SO2exposed compared to air-exposed controls; statistical significance
obtained for IgG total in BAL fluid at > 4.3 ppm SO2 and in serum at all
SO2 concentrations. Results indicate subacute exposure to even low
concentrations of SO2 can potentiate allergic sensitization of the airway.
Antigen Sensitization/Allergic Reactions - Subchronic
Kitabatake
etal. (1992;
1995)
5 ppm
(13.1 mg/m3);
whole body;
sensitized with
Candida albicans
on day 1 and wk
4
4 h/day, 5
days/wk, 6 wks
Hartley guinea pig,
male, age NR, -200
g, N: 12/group
Respiratory challenge to Candida albicans 2 wks after last exposure. At
15 h after challenge increased number of SO2-exposed animals displayed
prolonged expiration, inspiration, or both. Authors concluded SO2
exposure increased dyspneic symptoms.
General Bronchial Reactivity Studies - Acute
Douglas
etal. (1994)
5 ppm
(13.1 mg/m3);
whole body
2 h
New Zealand White
rabbit, sex NR,
apparently 3 mos
old, 2.2-3.1 kg,
n=6/group
No effect on airway responsiveness to inhaled histamine, as measured by
provocation concentrations of histamine required to increase pulmonary
resistance by 50% and decrease dynamic compliance by 35%.
May 2008
E-5
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Lewis and
Kirch ner
(1984)
CONC.
10 or 30 ppm
(26.2 or
78.6 mg/m3);
intratracheal
DURATION
5 min; 2nd expo-
sure 20 days later,
after exposure to
the antiallergic
drug
SPECIES
Mongrel dogs, male
and female, age and
weight NR; N:
5-15/group
EFFECTS
No effect observed at 10 ppm. At 30 ppm hyperresponsiveness and
hypersensitivity to aerosolized methacholine and 5-hydroxytryptamine
observed for up to 24 h following exposure. 20 days later, pretreatment
with aerosolized 4% Wy-41 ,195 or disodium cromoglycate (antiallergic
drugs) at high doses lessened the methacholine-induced hypersensitivity
observed after exposure to 30 ppm SO2. Calculations used to determine
hyperresponsive and hyperreactivity were not clear.
General Bronchial Reactivity Studies - Chronic
Scanlon
etal. (1987)
15 or 50 ppm
(39.3 or
131 mg/m3);
intratracheal
2 h/day, 4 or 5
days/wk for 5 mos
(low dose group)
or 10-11 mos
(high dose group);
physiological
changes observed
within 5 mos; 7-9
mo recovery
period.
Mongrel dogs, adult,
sexNR, 10-20 kg; N:
3-4/ group (3 hyper-
responsive, 3 hypo-
responsive, and
1 avg responsive)
Bronchial reactivity in response to inhaled histamine or methacholine was
not affected in either treatment group, as determined by the concentration
of histamine or methacholine required to double pulmonary resistance or
the concentrations required to decrease dynamic compliance by 65%
(ED65).
Table E-5.   Effects of SO2 exposure on cardiovascular endpoints.
STUDY
CONC.
DURATION
SPECIES
EFFECTS
In Vitro Exposure
Nie and
Meng
(2005)



Nie and
Meng
(2006)



Bisulfite/sulfite,
1 :3 molar/molar,
10 pM



Bisulfite/sulfite,
1 :3 molar/molar,
10 pM



NR





NR





Ventricular myocytes
isolated from Wistar
rats, adult,
200-300 g, N: 8


Ventricular myocytes
isolated from
200-300 g, N: 8



Effects of the 10 [M bisulfite/sulfite mixture on sodium current included a
shift of steady state inactivation curve to a more positive potential, a shift
of the time- dependent recovery from inactivation curve to the left,
accelerated recovery, and shortened inactivation and activation time
constants. It was concluded the bisulfite/sulfite mixture stimulated cardiac
sodium channels.
Effects of the 10 pM bisulfite/sulfite mixture on voltage- dependent L-type
calcium currents included a shift of steady-state activation and inactivation
to more positive potentials, accelerated recovery from inactivation, and
shortened fast and slow time inactivation constants. Authors stated that
their results suggested the possibility cardiac injury following SO2
inhalation.
Acute/Subacute Exposure
Halinen
et al. (2000)














Halinen
etal.
(2000a)






1.0, 2.5, or 5 ppm
(2.62, 6.55, or
13.1 mg/m3) in
cold dry air;
apparently intra-
tracheal










1 ppm (2.62
mg/m3) in cold dry
air; apparently
intratracheal





In pre-exposure
period 15-min ex-
posure to warm
humid air, 10-min
to cold dry air,
and 15-min to
warm humid air.
In exposure
period, 10-min to
each SO2 con-
centration or cold
dry air, preceded
and followed by
15-min exposure
to warm, humid
air.
60 min








Duncan-Hartley
guinea pigs, male,
age and weight NR,
N: 7-12/ group, me-
chanically ventilated;
animals were hyper-
ventilated during cold
air and SO2 exposure
to simulate exercise







Duncan-Hartley
guinea pigs, male,
age and weight NR,
N: 8-9/group, me-
chanically ventilated;
animals were hyper-
ventilated during cold
air and SO2 exposure
to simulate exercise
Arterial blood pressure increased transiently during exposure to 5 ppm
SO2 in cold dry air. No analyses were done to determine the effects on
blood pressure were caused by exposure to cold air or SO2.













Blood pressure and heart rate increased similarly with exposure to cold
dry air or SO2 in cold dry air. Blood pressure generally increased during
the first 10-20 min of exposure and remained steady from that point
forward. The increase in heart rate was gradual. No analyses were done
to determine if the effects on blood pressure were caused by exposure to
cold air or SO2.



May 2008
E-6
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Nadziejko
et al. (2004)
Meng et al.
(2003a)
Meng et al.
(2003a)
Langley-
Evans et al.
(1996)
Meng et al.
(2003a)
Wu and
Meng
(2003)
CONC.
1 ppm
(2.62 mg/m3);
nose only
10, 20, or 40 ppm
(26.2, 52.4 or
105 mg/m3);
whole body
10, 20, or 40 ppm
(26.2, 52.4, or
104.8 mg/m3);
whole body
5, 50, or 100 ppm
(13.1, 131,or262
mg/m3); whole
body
22, 56, or 112
mg/m3 (8.4, 21, or
43 ppm); whole
body
22, 64, or 148
mg/m3 (8.4, 24.4,
or 56.5 ppm);
whole body
DURATION
4h
6h
6 h/day for 7 days
5 h/day for
7-28 days
6 h/day for 7 days
6 h/day for 7 days
SPECIES
F344rat, male, 18
mos old, weight NR,
N: 20 (crossover
design)
Wistar rat, male, 7-8
wksold, 180-200 g;
N: 10/group
Wistar rat, male, 7-8
wksold, 180-200 g;
N: 10/group
Wistar rat, male, 7
wks old, weight NR,
N: 4-5/treatment
group, 8 controls
Kunming albino mice,
male and female, 5
wksold, 19 ±2 g, N:
10/sex/group
Kunming-strain mice,
male, age NR, 18-20
g, N: 10/group
EFFECTS
SO2 exposure had no effect on spontaneous arrythmia frequency in aged
rats. Authors urged caution in the interpretation of effects because
occurrence of arrhythmias in aged rats was sporadic and variable from
day to day.
A dose-related decrease in blood pressure was observed at > 20 ppm.
Dose-related decreases in blood pressure were observed on exposure
day 3 in the 10 ppm group, exposure days 2-6 in the 20 ppm group, and
all exposure days in the 40 ppm group. The authors noted possible
adaptive mechanism in the low but not the high dose group.
GSH was depleted in the heart at 5 and 100 ppm. At 50 ppm, GSH level
decreased in heart at 7 days and returned to normal by 14 days. No
effects observed for other GSHrelated enzymes. Injury not assessed in
heart, but assessment in lung revealed no effect.
Changes observed in heart (concentrations of effect) included: lower SOD
activity in males and females (> 8.4 ppm), higher TEARS level in males
and females (> 8.4 ppm), lower GPx activity in males (8.4 and 21 ppm;
also 43 ppm according to text) and lower GSH level in males (43 ppm).
Authors concluded that SO2 induced oxidative damage in hearts of mice.
GSH, GST, and glucose-6-phosphate dehydrogenase activities were
decreased in the heart at 148 mg/m3.
Table E-6.   Neurophysiology and biochemistry effects of SO2 and derivatives.
STUDY
CONC.
DURATION
SPECIES
EFFECTS
In Vitro/Ex Vivo
Du and
Meng
(2004)






Du and
Meng
(2004b)










1, 10, 50, or 100
pMSO2
derivatives (1 :3,
NaHSO3to
Na2S03)




1 or10pMSO2
derivatives
(1:3, NaHSO3to
Na2SO3)









Not specified








2-4 min












Wistar rat, sex NR,
6-12 days old, weight
and number NR;
typical observations
made on 60 isolated
hippocampal neurons
per concentration


Wistar rat, both
sexes, 10-15 days
old, weight and
number NR; N:6-13
isolated dorsal root
ganglion neurons
avgd per endpoint






Exposure to SO2 derivatives (sulfite, bisulfite) reversibly increased the
amplitude of potassium channel TOCs in a dose-dependent and volt-
age-dependent manner. Compared to controls, 10 pM SO2 shifted
inactivation of depolarization toward more positive potentials without
significantly affecting the activation process. By increasing maximal TOC
conductance and delaying TOC inactivation, micromolar concentrations of
SO2 derivatives may increase the excitability of hippocampal neurons and
thus contribute to the enhanced neuronal activity associated with SO2
intoxication.
Maximum sodium current amplitudes for both TTX-S and TTX-R channels
were increased by exposure to SO2 derivatives (10 or 1 [M, respectively),
with amplitudes diminished at more negative evoking potentials and
enhanced at less negative or positive potentials. SO2 derivatives (a)
slowed both current activation and inactivation for both types of sodium
channels; (b) shifted activation currents to more positive potentials,
increasing threshold voltages for action potential generation and
contributing to reduced neuron excitability; and (c) caused even larger
counteracting positive shifts in inactivation voltages tending to increase
dorsal root ganglion neuron excitability. On balance, the data suggest
micromolar concentrations of sulfite/bisulfite can increase the excitability
of dorsal root ganglion neurons, providing a basis for SO2-associated
neurotoxicity.
May 2008
E-7
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Du and
Meng
(2006)













CONC.
0.01,0.1,0.5, or 1
[iMSO2
derivatives (1 :3,
NaHSO3to
Na2SO3)











DURATION
Not specified, but
brief ("added to
the external
solution just
before each
experiment")










SPECIES
Wistar rat, both
sexes, 10-15 days
old, weight and
number NR; N:6-15
isolated dorsal root
ganglion neurons
avgd per endpoint









EFFECTS
In isolated dorsal root ganglion neurons, SO2 derivatives increased
HVA-/Ca amplitudes in a concentration- and depolarizing volt-
age-dependent manner (EC50 was ~0.4 pM) by altering Ca channel
properties. This effect was partially reversible by SO2 derivative washout,
and was PKI-inhibitable, indicating involvement of PKA and secondary
messengers. Additionally, exposure caused a positive shift in reversal
potential. SO2 derivatives also delayed activation and inactivation of Ca
channels, but the latter was more pronounced, thus overall prolonging
action potential duration and increasing HVA-/Ca. Exposure also slowed
the fast component and accelerated the slow component of recovery from
Ca channel inactivation. Thus, < 1 [M sulfite/bisulfite caused prolonged
opening and altered properties of Ca channels, elevated HVA-/Ca, and
abnormal Ca signaling with neuronal cell injury. Authors speculate these
effects may correlate to SO2 inhalation toxicity, perhaps leading to
abnormal regulation via peripheral neuron Ca channels of nociceptive
impulse transmission.
Acute/Subacute/Subchronic Exposure
Wu and
Meng
(2003)

Haider
etal.
(1981)








Haider
etal.
(1982)











Haider and
Hasan
(1984)











22,64, or 148
mg/m3 (8.4, 24.4,
or 56.5 ppm);
whole body
10 ppm (26.2
mg/m3); whole
body








10 ppm (26.2
mg/m3); whole
body











10 ppm (26.2
mg/m3) SO2
alternated with
20 ppm
(14.7 mg/m3) H2S;
whole body








6 h/day for 7
days


1 h/day for 21 or
24 days









1 h/day for
30 days












1 h/day for
30 days (alter-
nating SO2 or
H2S)










Kunming-strain mice,
male, age NR, 18-20
g, N: 10/group

Guinea pig, sex NR,
adult, 250-500 g, N:
12/group
(6/subgroup)







Charles Foster rat,
male, adult, 150-200
g, N: 12/group
(6/subgroup)










Guinea pig, sex and
age NR, 250-400 g,
N: 18/group in 2
groups (6/group in
some subgroups)









Decreased glutathione, glucose-6-phosphate dehydrogenase, and GST
activities were observed in the brain at 64 and 148 mg/m3.


The effects of SO2 exposure on lipid profiles, lipid peroxidation and lipase
activity in three regions of the brain (cerebral hemisphere, CH;
cerebellum, CB; brain stem, BS) were examined. Significant
(p < 0.001-0.05) findings include reductions in total lipids (CH, BS; also
CB, but nonsignificant) and free fatty acids (CH, CB, BS). PL were
elevated in CH, but reduced in CB; Choi was elevated in CH, but reduced
in CB and BS; and esterified fatty acids were elevated in CB, but reduced
in CH and BS. Levels of malonaldehyde and lipase activity were elevated
in all regions. Results indicate that subacute brief exposures to SO2 can
lead to degradation of brain lipids, with the exact nature of the lipid
alterations dependent upon brain region.
The effects of SO2 exposure on lipid profiles, lipid peroxidation and lipase
activity in three regions of the brain (cerebral hemisphere, CH;
cerebellum, CB; brain stem, BS) were examined. Significant (p <
0.001-0.05) findings include reductions in total lipids (CH, BS, CB), while
PL were elevated only in CB. Choi was elevated in CH and CB, but not
BS; and gangliosides were elevated in CB and BS, but reduced in CH.
Lipid peroxidation (malonaldehyde formation) was elevated in whole brain
and all regions (although nonsignificantly in BS), as was lipase activity in
CH, the only tissue examined. Despite regional differences in PL and Choi
changes, Choi/PL ratios were elevated in all three brain regions (again
nonsignificantly in BS). Results are somewhat different than those seen in
guinea pig (Haider et al., 1981), but again suggest that subacute brief
exposures to SO2 can lead to degradation of brain lipids, with the exact
nature of the lipid alterations dependent upon brain region.
The effects of alternating SO2 + H2S exposure on lipid profiles, lipid
peroxidation and lipase activity in four regions of the brain (cerebral
hemisphere, CH; basal ganglia, BG; cerebellum, CB; brain stem, BS) and
in the spinal cord (SC) were examined. Significant (p < 0.001-0.05)
findings include reductions in total lipids and Choi, and elevated lipid
peroxidation (malonaldehyde formation) and lipase activity, in all brain
regions and SC. Choi/PL ratios were also reduced in all tissues (but
nonsignificantly in BG and CB). For other parameters (PL, free fatty acids,
esterified fatty acids, and gangliosides), changes were observed in most
tissues but were region-specific. Results indicate that subacute brief,
alternating exposures to SO2 or H2S lead to degradation of brain lipids,
again with the exact nature of the lipid alterations dependent upon
brain/spinal cord region. Additionally, some of the effects observed for this
mixture vary from those seen with SO2 alone (Haider et al., 1981; 1982).
May 2008
E-8
DRAFT—DO NOT QUOTE OR CITE

-------
 STUDY
                CONC.
                               DURATION
                                                   SPECIES
                                                                                              EFFECTS
Agar et al.
(2000)
10 ppm (26.2
mg/m3) (± iv
alloxan to induce
experimental type
1 diabetes); whole
body
1 h/day,
7 days/wk for
6wks
Swiss albino rat,
male, 3 mos old,
weight NR, N:
10/group in 4 groups
In retina tissue, exposure elevated SOD activity and reduced GPx and
catalase activities. TEARS were elevated only in non-diabetic rats
exposed to SO2. In brain tissue, exposure elevated SOD and reduced
GPx activities in both non-diabetics and diabetics, while catalase activities
were not affected; TEARS were elevated in both non-diabetics and
diabetics. With respect to VEPs, exposure prolonged latencies in 4 of 5
VEP components in non-diabetics and 5 of 5 in diabetics, while reducing
virtually all peak-to-peak amplitudes in non-diabetics and diabetics. For
many endpoints, SO2 effects were additive to those resulting from the
induced diabetic condition. In summary, brain and retinal anti-oxidant and
lipid peroxidation status, as well as neuro-visual performance were
affected by subchronic exposure to brief periods of 10 ppm SO2, and
these effects were exacerbated by a diabetic condition.
Subchronic/Chronic Exposure
Kiiciikatay
etal.
(2003)
10 ppm (26.2
mg/m3) (± iv
alloxan to induce
experimental type
1 diabetes); whole
body
1 h/day,
7 days/wk for
6wks
Rat, male, 3 mos old,
weight not reported,
N: 10/group in 4
groups
In brain tissue, SO2 exposure elevated SOD and reduced GPx activities in
both non-diabetics and diabetics, while catalase activities were not
affected; TEARS were elevated in both non-diabetics and diabetics. With
respect to afferent peripheral nerve pathways (SEPs), exposure prolonged
latencies in 4 of 4 SEP components in both non-diabetics and diabetics;
also altered were some inter-peak latencies (non-diabetics and diabetics)
and some peak-to-peak amplitudes (non-diabetics only). In some cases,
SO2 effects were additive to those resulting from the induced diabetic
condition. In summary,  brain anti-oxidant and lipid peroxidation status, as
well as afferent peripheral nerve pathways, were affected by subchronic
exposure to 10 ppm SO2, and these effects were exacerbated by a
diabetic condition. Authors suggest that SO2 exposure could potentiate
the  incidence and/or severity of diabetes.
Yargifoglu
etal.
(1999)
10 ppm (26.2
mg/m3); whole
body
1 h/day,
7 days/wk for
6wks
Swiss albino rat,
male, 3, 12, or24
mos old, weight not
reported, N: 10/group
in 6 groups
Effects of aging ± SO2 exposure on levels of lipid peroxidation (TEARS),
antioxidant enzyme status (catalase, GPx, SOD), and afferent peripheral
nerve pathways (SEPs) were monitored in the brain of young (Y, 3 mo),
middle-aged (M, 12 mo) and old (O, 24 mo) rats. In addition to age-related
changes, SO2 exposure significantly (p < 0.0001-0.02) elevated TEARS
and SOD, while reducing  GPx (Y, M, O); catalase levels were not affected.
Of 4 monitored SEP component peaks, SO2 significantly (p < 0.01-0.05)
prolonged latencies in groups Y (4/4) and  M (1/4), but not in O (0/4).
Peak-to-peak amplitudes  were  decreased in Y, (2/3) and increased in M
(1/3),  but not affected in O (0/3). Taken together, these data indicate that
subchronic exposure to brief periods of 10 ppm SO2 can impact afferent
peripheral nerve pathways and the lipid peroxidation and antioxidant
enzyme status of the brain.
Kilic (2003)
10 ppm
(26.2 mg/m3);
whole body
1 h/day,
7 days/wk for
6wks
Swiss albino rat,
male, 3, 12, or
24 mos old, weight
not reported, N:
10/group in 6 groups
Effects of aging ± SO2 exposure on levels of lipid peroxidation (TEARS),
antioxidant enzyme status (catalase, GPx, SOD), and visual system
function (VEPs) were monitored in the brain and eye (retina and lens) of
young (Y, 3 mo), middle-aged (M, 12 mo) and old (O, 24 mo) rats. In
addition to age-related changes, SO2 exposure significantly
(p < 0.0001-0.04) elevated TEARS in brain and lens (Y, M, O), and in
retina (Y); reduced GPx in brain (Y) and lens (Y, M, O); reduced catalase
in retina (Y, M, O); and elevated SOD in brain (Y, M), retina (Y, M, O) and
lens (M, O). Of 5 monitored VEP component peaks, SO2 prolonged
latencies in groups Y (4/5), M (3/5) and O (1/5). Taken together, these
data indicate that subchronic exposure to brief periods of 10 ppm SO2 can
impact the visual system and the lipid peroxidation and antioxidant
enzyme status of the brain and eye.
Neurodevelopment/Neurobehavior
Singh
(1989)
32 or 65 ppm
(83.8 or
170 mg/m3); whole
body
Gestation day
7-18
CD-1 mouse dams
were exposed; num-
bers of dams
exposed and offspring
evaluated not indi-
cated
Righting and negative geotaxis reflexes were delayed at both concen-
trations.
May 2008
                                               E-9
                                                         DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Petruzzi
etal.
(1996)












Fiore et al.
(1998)









CONC.
5, 12, or 30 ppm
(13.1,31.4, or
78.6 mg/m3);
whole body











5, 12, or 30 ppm
(13.1,31.4, or
78.6 mg/m3);
whole body







DURATION
Near continuous
(80% of time)
exposure from
9 days before
mating through
the 12-1 4th day
of pregnancy








Near continuous
(90% of time) ex-
posure from
9 days before
mating through
the 14th day of
pregnancy




SPECIES
CD-1 mouse, adult
male and female
parental animals were
exposed (N:
10/group/sex) and
male and female
offspring (N:
8 litters/group,
fostered by
unexposed dams at
birth) were evaluated
at 2-18 days of age;
adult male offspring
also evaluated (N:
8/group)
CD-1 mouse, adult
male and female
parental animals were
exposed and adult
male offspring
(fostered by
unexposed dams at
birth) were evaluated
at ~120 days of age,
N: 11 -12 offspring/
group
EFFECTS
Offspring: No effects observed for birth weight, postnatal body weight
gain, somatic and neurobehavioral development (e.g., eyelid and ear
opening, incisor eruption, and reflex development); no postnatal de-
velopmental data were shown by authors. No effects observed in passive
avoidance testing of adult males. Adults: Observation of behavior outside
the exposure chamber on exposure days 3, 6, and 9 revealed
dose-related increases in digging and decreases in grooming by females
in the 30 ppm group on exposure day 9; non-dose related increases were
observed for crossing and wall rearing by females in the 30 ppm group on
exposure day 9. Observance of behaviors in 2 breeding pairs/group in the
12 and 30 ppm groups revealed increased rearing and social interaction in
the 30 ppm group shortly after the start of exposure, followed by return to
baseline levels; effects were generally of greater magnitude in males.


In 20-min encounters with unexposed males, prenatally-exposed males
compared to controls displayed (dose(s) of effect, time of testing effect
observed) increased duration of self grooming (5 ppm, 15-20 min),
decreased frequency and duration of tail rattling (> 5 ppm at 5-10 min and
12 ppm at 10-15 min), and decreased duration of defensive postures
(> 12 ppm, 0-5 min). Authors also noted a non-significant decrease in
freezing (apparently at all dose levels) and non-significant increases in
social exploration (apparently at all doses) and rearing (apparently at
> 12 ppm).


Table E-7.   Reproductive and developmental effects of SO2.
STUDY
CONC.
DURATION
SPECIES
EFFECTS
Reproductive Organ Effects - Subacute/Subchronic
Meng and
Bai
(2004)
Gunnison
etal.
(1987)
Singh
(1989)
Petruzzi
etal.
(1996)
22,56, or
112 mg/m3
(8.4, 21, or
43 ppm); whole
body
10 or 30 ppm (26.2
or 78.6 mg/m3);
whole body
32 or 65 ppm (83.8
or 170 mg/m3);
whole body
5, 12, or 30 ppm
(13.1,31.4, or
78.6 mg/m3); whole
body
6 h/day for 7 days
6 h/day,
~5 days/wkfor 21
wks (total of 99
days)
Gestation day
7-18
Near continuous
(80% of time)
exposure from
9 days before
mating through
the 12-1 4th day of
pregnancy
Kunming albino mice,
male, 5 wks old, 19 ±
2 g, N: 10/group
Sprague-Dawley CD
rat, male, 8 wks old,
weight NR, N:
70/group in 3 groups
(inhalation series)
CD-1 mouse dams
were exposed;
numbers of dams
exposed and offspring
evaluated not
indicated
CD-1 mouse, adult
male and female
parental animals were
exposed (N:
10/group/sex) and
male and female
offspring (N:
8 litters/group,
fostered by
unexposed dams at
birth) were evaluated
at 2-18 days of age;
adult male offspring
also evaluated (N:
8/group)
Changes observed in mouse testes (concentrations of effects) included
decreased activities of SOD (43 ppm, possibly at 21 ppm according to
text) and GPx (> 21 ppm), increased catalase activity (8.4 and 21 ppm),
decreased GSH level (> 21 ppm), and increased TEARS levels
(> 8.4 ppm). The authors concluded that SO2 can induce oxidative
damage in testes of mice.
No significant (p < 0.05) effect on testes histopathology was found,
although there was a very slight and probably biologically insignificant
increase in relative testes weight. (0.61 ± 0.02 vs. 0.56 ± 0.02, % body
weight.).
No significant effects were observed for number of live pups born/litter.
Pup birth weight was lower at 65 ppm. Righting and negative geotaxis
reflexes were delayed at both concentrations.
Decreased food and water intake were observed in parental males and
females of the 12 and 30 ppm groups at the start of mating (exposure
days 9-13). No effects observed for mating or successful pregnancies.
There were no effects on litter sizes, sex ratio, or neonatal mortality (data
not shown by authors). No effects observed for birth weight, postnatal
body weight gain, somatic and neurobehavioral development (e.g., eyelid
and ear opening, incisor eruption, and reflex development); no postnatal
developmental data were shown by authors. No effects observed in
passive avoidance testing of adult males.
May 2008
E-10
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Fiore
etal.
(1998)
Douglas
etal.
(1994)
CONC.
5, 12, or 30 ppm
(13.1,31.4, or
78.6 mg/m3); whole
body
5 ppm
(13.1 mg/m3);
whole body
DURATION
Near continuous
(90% of time)
exposure from
9 days before
mating through
the 14th day of
pregnancy
2 h/day for
13wks
SPECIES
CD-1 mouse, adult
male and female
parental animals were
exposed and adult
male offspring
(fostered by
unexposed dams at
birth) were evaluated
at ~120 days of age,
N:
11-12 offspring/group
New Zealand White
rabbit, male and
female, N: 3-4/group,
1-day-old, immunized
against Alternaria
tenuis
EFFECTS
In 20-min encounters with unexposed males, prenatally-exposed males
compared to controls displayed (dose(s) of effect, time of testing effect
observed) increased duration of self grooming (5 ppm, 15-20 min),
decreased frequency and duration of tail rattling (> 5 ppm at 5-10 min
and 12 ppm at 10-15 min), and decreased duration of defensive postures
(> 12 ppm, 0-5 min). Authors also noted a non-significant decrease in
freezing (apparently at all dose levels) and non-significant increases in
social exploration (apparently at all doses) and rearing (apparently at
> 12 ppm).
Following subchronic exposure beginning in the neonatal period, there
were no effects on lung resistance, dynamic compliance, transpulmonary
pressure, tidal volume, respiration rate, or min volume.
Table E-8.   Hematological effects of SO2.
STUDY
CONC.
DURATION
SPECIES
EFFECTS
Acute/Subacute Exposure
Baskurt
(1988)
Giimiislii
etal.
(1998)
Yargifogl
u et al.
(2001)
0.87 ppm
(2 .36 mg/m3);
whole body
10 ppm
(26.2 mg/m3);
whole body
10 ppm (26.2
mg/m3); whole
body
24 h
1 h/day, 7
days/wk for 8 wks
1 h/day, 7
days/wk for 6 wks
Swiss Albino rat, male,
age NR, 250-300 g, N:
51,50
Swiss-Albino rat, male,
2.5-3.0 mos old, weight
NR, N: 30 (14 controls,
16 treated)
Albino rat, male, 3, 12,
and 24 mos old, mean
weight213-448g, N:
10/ group
Effects of SO2 exposure included increased hematocrit, sulfhemoglobin
and osmotic fragility and decreased whole blood and packed cell
viscosities. RBC number, hemoglobin, mean corpuscular volume, mean
corpuscular hemoglobin concentration, and plasma viscosity were not
significantly altered.
Decreased Cu, Zn-SOD activity, increased GPx and GST activity, and
increased TEARS formation were observed in RBC of treated rats. No
significant effect on glucose-6-phosphate dehydrogenase or catalase
levels was observed.
Enzyme and GSH activity (GPx, catalase, GSH, and GST) were
increased and copper-zinc SOD activity was decreased in RBCs of all
experimental groups compared to controls. RBCs in older rats had lower
levels of all antioxidants enzymes and increased TBARS activity
compared to younger rats.
Subchronic Exposure
Langley-E
vans et al.
(1997;
2007)
Etlik et al.
(1995)
Agar et al.
(2000)
286 mg/m3 (100
ppm); whole body.
Units were initially
reported as pg/m
but were corrected
per
correspondence
w/author.
10 ppm
(26.2 mg/m3);
whole body
10 ppm
(26.2 mg/m3);
whole body
5 h/day for
28 days
1 h/day for
30 days
1 h/day, 7 days/
wk for 6 wks
Wistar rat, male, 7 wks
old, weight NR, N:4-16
Guinea pig, sex and
age NR, 250-450 g, N:
12/group
Swiss Albino rat, male,
3 mos old, weight NR,
N: 10 per group in 4
groups
Dams were fed diets containing casein at 180 [control], 120, 90, or 60
g/kg during pregnancy and their offspring were exposed to air or SO2 as
adults. In blood of offspring, SO2 exposure significantly reduced the
numbers of circulating total leukocytes and lymphocytes in the 180 and
120 g/kg dietary groups; neutrophils numbers were not affected in any
group. GSH levels in the 180 and 60 g/kg (but not the two intermediate)
dietary groups were reduced by SO2 exposure.
SO2 exposure resulted in RBC membrane lipoperoxidation (elevated
levels of malonyldialdehyde) and other oxidative damage (elevated
osmotic fragility ratios and levels of methemoglobin and sulfhemoglobin).
All effects significantly (p < 0.05) mitigated by injections of Vitamin E+C
three times per wk.
RBC parameters were monitored in non-diabetic rats, non-diabetic rats
exposed to SO2, alloxan-induced diabetic rats, and diabetic rats exposed
to SO2. In both non-diabetic and diabetic rats exposed to SO2, levels of
GPx, catalase, GSH, GST, and TBARS were elevated in RBC while
those of SOD were reduced.
May 2008
E-11
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Etlik et al.
(1997)
CONC.
10 ppm
(26.2 mg/m3);
whole body
DURATION
1 h/day for
45 days
SPECIES
Rat, sex and age NR,
214-222 g, N: 6-8 per
group
EFFECTS
SO2 exposure significantly elevated levels of methemoglobin,
sulfhemoglobin and malonyldialdehyde, the latter of which was
substantially reversed by Vitamin E+C treatment. RBC osmotic fragility
was increased by SO2, and again partially mitigated by Vitamin E+C.
SO2 elevated RBC, white blood cell, hemoglobin and hematocrit values,
but not mean corpuscular volume, mean corpuscular hemoglobin or
mean corpuscular hemoglobin concentration. Vitamin E+C exposure did
not affect these parameters.
Table E-9.   Endocrine system effects of SO2.
STUDY
Lovati et
al. (1996)
Agar et al.
(2000)
Kucukatay
etal.
(2003)
CONC.
5 or 10 ppm (13.1
or 26. 2 mg/m3);
whole body
10 ppm (26.2
mg/m3); whole
body
10 ppm (26.2
mg/m3); whole
body
DURATION
24 h/day for
15 days
1 h/day,
7 days/wk for 6
wks
1 h/day,
7 days/wk for 6
wks
SPECIES
Sprague-Dawley CD
rat, male, age NR,
250-275 g, N:
9/subgroup in 9
subgroups
Swiss Albino rat, male,
3 mos old, weight NR,
N: 10/group
Rat, male, 3 mos old,
weight NR, N: 10/group
in 4 groups
EFFECTS
Subjects were rats fed standard diet (normal) or high cholesterol diet, and
rats with streptozotocin-induced diabetes fed standard diet. In diabetic
rats, there was no effect on glucose levels. Exposure to > 5 ppm lowered
plasma insulin level in normal and hypercholesterolemic diet groups, but
elevated it (non-significantly) in diabetic rats. In each rat model, inhalation
of SO2 at levels without overt effects affected insulin levels. Specific
effects varied according to diet or diabetes.
Effects were compared in non-diabetic rats and rats with alloxan induced
diabetes. SO2 increased blood glucose in diabetic and non-diabetic rats.
Effects were compared in normal rats and rats with alloxan induced
diabetes. SO2 elevated blood glucose levels in both non-diabetics and
diabetics.
Table E-10.  Effects of SO2 exposure on respiratory system morphology.
STUDY
CONC.
DURATION
SPECIES
EFFECTS
Acute/Subacute Exposure
Conner
etal.
(1985)
1 ppm (2.6
mg/m3); nose
only
3 h/day for 6 days;
animals evaluated
for up to 72 h
following exposure
Hartley guinea pig, male,
age NR, 250-320 g, N:
14/group/time point
In combined group of SO2 exposed animals and furnace gas controls, no
alveolar lesions were observed.
Subchronic/Chronic Exposure
Wolff
etal.
(1989)
Smith
etal.
(1989)
Gunnison
et al.,
(1987)
5 ppm (13
mg/m3); nose
only
1 ppm (2.62
mg/m3); whole
body
10 and 30 ppm
2 h/day, 5 days/wk
for 4 wks
5 h/day, 5 days/wk
for 4 or 8 mos;
half the animals in
the 8-mo group
were allowed to
recover for 3 mos.
6 h/day and 5
day /wk for 21 wks
F344/CM rat, male and
female, 10-11 wks old,
weight NR, N: 3/sex/ group
Sprague-Dawley rat, male,
young adult, initial weight
NR, N: 1 2-1 5/data point
Sqrague-Dawley CD rat, 8
week of age
No nasal or pulmonary lesions.
At 4 mos of SO2 exposure, increases were observed for incidence of
bronchial epithelial hyperplasia (80 vs. 40% in controls) and numbers of
nonciliated epithelial cells (31.1 vs. 27.7% in controls); neither effect
persisted past 4 mos of exposure.
Mild epithelial hyperplasia in the trachea and larger bronchi, mucoid
degeneration and desquamation of epithelium of the larger bronchi
May 2008
E-12
DRAFT—DO NOT QUOTE OR CITE

-------
Table E-11.  Carcinogenic effects of SO2.
STUDY
CONC.
DURATION
SPECIES
EFFECTS
Pulmonary Effects
Gunnison
etal.
(1988)


















Ohyama
etal.
(1999)
















Ito et al.
(1997)







Heinrich
etal.
(1989)












0, 10, or
30 ppm (0,
26.2, or 78.6
mg/m3) SO2
(whole body)
± 1 mg B[a]P
0 100 or

"" '
[400 ppm W +
40 ppm Mo] in
a low-Mo diet
± B[a]P (See

e°\ S D^ "in
umn) ± B[aJP






0, 0.2 mLC, or
(0.2 mL
DEP+C
± [4 ppm
(10. 48 mg/m3)
SO2 or 6 ppm
(1 1 .28 mg/m3)
NO2 or 4 ppm
SO2 + 6 ppm
NO2]; whole
body
[Note' 0 2 ml_
CBP = 1mg'
0 2 ml_
DEcCBP =1
mg CBP + 25
mn nFPM
rng ucr^jj


0, C, or (25 mg
SPM+C ± [4
ppm (10.48
mg/m ) SO2 or
6 ppm (11.28
mg/m3) NO2 or
4 ppm SO2
+ 6 ppm NOJ);
whole body
0 or [10 ppm
(26.2 mg/m3)
SO2 + 5 ppm
(9.4 mg/m3)
NO2] ± [3 or
6 mg/kg bw of
DEN];
exposure to
gases whole
body





SO2: 21 wk, 5
day/wk (minus
holidays), 6 h/day
High W low Mo
diet' 21 wk 7
dav/wk

B[a]P: 15 wk,
once perwk
starting wk 4











SO2 and/or NO2:
10 mo, 16 h/day
CBP or DEcCBP:
4 wk, once/wk by
inrarac ea

infusion













SO2 and/or NO2:
11 mo, 16 h/day
Q + SPM' 4 wk

It'
'



SO2 + NO2: 6,
10.5, 15, or 18
mo, 5 day/wk,
19 h/day
DEN* ones bv
s c inisction
"2 wk aftsr ths
s a o in a a ion

sxposurs





Rat, Sprague- Dawley,
male, 9 wk old, -315- 340
g, N: 20-74/group


















Rat, SPF F344/Jcl, male, 6
wkold.wtNR, N: 23-30
per group in 6 groups
















Rat, SPF Fisher 344, male,
Swkold, wt NR, N: 5 per
group in 6 groups






Hamster, Syrian golden,
both sexes, 10 wk old, bw
NR, N: 40/sex per each of
12 exposure groups











Purpose was to investigate arcinogenic/cocarcinogenic effects of SO2
inhalation or dietary-induced high levels of systemic sulfite/bisulfite in
conjunction with tracheal installation of B[a]P. High drinking water levels
of Win conjunction with low-Mo feed induce sulfite oxidase deficiency in
rats, and thus high systemic levels of sulfite and bisulfite (at 0, 100 or
400 ppm W, mean plasma sulfite was 0, 0 or 44 pM, while mean tracheal
sulfite + bisulfite was 33, 69 or 550 nmol/g wet wt). Mortality in B[a]P
groups (-50% after -380-430 d) was due almost exclusively to SQCA of
the respiratory tract; survival rate was excellent for other groups (-50%
mortality after -620-700 d). Results indicate no SQCA was induced in
any of the SO2 inhalation or systemic sulfite + bisulfite groups, nor were
incidences in the B[a]P groups enhanced by such coexposures. This lack
of cocarcinogenicity does not support the hypothesis that SO2 exposure
could elevate systemic sulfite/bisulfite, generating GSSO3H, which would
inhibit GST and reduce intracellular GSH, thus interfering with a major
detoxication pathway for B[a]P and enhancing its carcinogenicity.
Authors note that due to the high incidence of animals with tumors in the
two B[a]P only groups (65/72 and 63/72), cocarcinogenicity of SO2 or
sulfite + bisulfite could only have been demonstrated by shortening of
tumor induction time and/or increased rate of SQCA appearance —
neither were observed.
Purpose was to study effects of DEP on rat lung tumorigenesis and
possible tumor promoting effects of SO2 or NO2 singly or together. Alveo-
lar hyperplasia and adenoma were significantly (p < 0.01-0.05) increased
over controls in the CBP group, but not the DEcCBP group. This was
ascribed to induction of alveolitis and AM infiltration (a tumor response
specific to rat and of questionable relevance to humans) in the former
group, but apparently prevented by DEP in the latter. Alveolar
bronchiolization near small hyaline masses of deposited DEcCBP was
observed in all DEcCBP groups, the masses presumably allowing
long-term exposure to DEP extracts by contacted alveolar epithelium.
DNA adducts were found only in the three gas-exposed groups. Dis-
counting the CBA group, elevated alveolar hyperplasia was seen only in
the DEcCBP + NO2 group, and elevated incidences of alveolar adenoma
in the DEcCBP + SO2 and particularly the DEcCBP + NO2 groups; nei-
ther effect was observed with coexposure to both gases — speculated by
the authors to perhaps result from inhibition of the stronger NO2 promo-
tion by HSO3". Thus, SO2 appears to have weaker capacity than NO2 for
promoting tumor induction (and perhaps genotoxicity) by DEP extract,
and may antagonize such effects by NO2 during coexposure of the
gases.
Purpose was to study effects of Tokyo air SPM, with or without coexpo-
sure to SO2 or NO2 or their combination, on the development of prolifera-
tive lesions of PEC. PEC hyperplasia was significantly (p < 05) increased
by exposure to SPM, but coexposure to either gas or their mixture was
without additional effect. No PEC papillomas were observed in control
groups, while a few were seen in the SPM groups, irrespective of gas
coexposures. Thus, SO2 demonstrated no tumor promotion or co-
carcinogenic properties. [Study did not describe the nature of the carbon
(C) used.]
The principle focus of this large study was to examine whether two in-
haled diesel-exhaust emission preparations (± particulates) could en-
hance the tumorigenesis of injected DEN. Ancillary aim was to see
whether inhalation of the irritant SO2 + NO2 mixture could cause similar
enhancement of DEN tumorigenicity. Gas mixture exposure did not affect
bw gain, but slightly shortened survival times (although significantly only
for females). Apart from effects attributed to DEN, serial sacrifices
showed progressive increases in ciliated tracheal cell aberrations and in
number of tracheal mucosal cells, the lung, gas mixture-related effects
were limited to a progressing alveolar lesion involving lining with
bronchiolar epithelium and the presence of some pigment- containing
AM, and to a mild, diffuse thickening of the alveolar septa. SO2 + NO2
exposure did not by itself elevate tumor rate the upper respiratory tract,
did it enhance increases induced by DEN. Thus the mixture appeared to
have no tumor inducing or promoting effects.
May 2008
E-13
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
CONC.
DURATION
SPECIES
EFFECTS
Nonpulmonary Effects
Klein
etal.
(1989)
0 or 6 ppm (0
or 15.72
mg/m3) SO2, ±
0.2 ppm (600
pg/rrO; whole
body; NDMA
20 mo, 5 day/wk,
4 h/day
Rat, Sprague- Dawley, fe-
male, age and wt NR, N:
36 per group in 4 relevant
groups
This is a preliminary report for observations after 20 mo (800 h inhalation
in 200 exposures, with calculated inhaled cumulative doses of 77 mg SO2
and 2-3 mg NDMA per rat). The effects of NDMA ± SO2 inhalation were
studied. Group mortality was as follows: control (3/36), SO2 (5/36),
NDMA (4/36), NDMA + SO2 (7/36). The only tumors observed were
nasal: control (0), SO2 (0), NDMA (1), NDMA + SO2 (3). No observable
parameters, including body wt gain, were affected by the additional SO2
exposure; assessment of tumor incidence effects could not yet be
performed.
Table E-12.  Respiratory system biochemistry effects of SO2.
STUDY
CONC.
DURATION
SPECIES
EFFECTS
In Vitro Exposure
Li et al.
(2007)
0.1 pM-1mM
NaHSO3 and
Na2SO3 1 :3
4 h, followed by
harvest at 0-24 h
BEP2D cell line of human
bronchial epithelial cells
Increased mRNAand protein levels of MUC5AC and IL-13
Oxidation and Antioxidant Defenses - (Subacute/Subchronic Exposure)
Meng
etal.
(2003b)
Wu and
Meng
(2003)
Langley-E
vans et al.
(1996)
Giimuslii
etal.
(2001)
22, 56, or 112
mg/m3 (8.4,
21, or
43 ppm);
whole body
22, 64, or 148
mg/m3 (8.4,
24.4, or
56.5 ppm);
whole body
5, 50, or
100 ppm
(13.1, 131, or
262 mg/m3);
whole body
10 ppm
(26.2 mg/m3);
whole body
6 h per day for 7
days
6 h/day for 7 days
5 h/day for
7-28 days
1 h/day,
7 days/wk for
6 wks
Kunming albino mice, male
and female, 5 wks old, 19 ±
2 g, N: 10/sex/group
Kunming-strain mice, male,
ageNR, 18-20 g, N: 10/
group
Wistar rat, male, 7 wks old,
weight NR, N:
4-5/treatment group,
8 controls
Swiss albino rat, male, 3,
1 2, or 24 mos old, 2 10-450
g, N: 9-11/group in 6
groups
Changes observed in lung tissue (concentrations of effect) included
higher SOD activity in males (8.4 ppm) and females (8.4 and 21 ppm),
lower SOD activity in males (21 and 43 ppm) and females (43 ppm),
increased GPx activity in males and females (8.4 ppm), decreased GPx
activity in males and females (> 21 ppm), decreased catalase activity in
males (43 ppm), decreased reduced GSH level in males and females
(> 8.4 ppm), increased TEARS level in males (> 8.4 ppm) and females
> 21 ppm). Authors concluded that sulfur dioxide induced oxidative
damage in lungs of mice.
Glucose-6-phosphate dehydrogenase and GST activity were decreased
in lung at 64 and 148 mg/m3. Lung GSH levels were reduced in the 22
and 148 mg/m3 exposure groups. Administration of buckthorn seed oil
increased GST and decreased TEARS activity compared to mice
exposed to 42 mg/m3 SO2 alone.
In the 5 and 100 ppm groups, GSH in BAL fluid decreased at 7 days and
increased at 21 days; at 28 days GSH returned to normal in the 5 ppm
group and further increased in the 100 ppm group. GSH was depleted in
the lung, at 5 and 100 ppm but not at 50 ppm. With respect to
GSH-related enzymes, exposure to 5 ppm lowered GCS, GPx, GST, and
GRed activity in the lung. Effects in the 100 ppm group were similar to the
5 ppm group, except that lung GPx was not reduced. Exposure to 50 ppm
did not affect lung GST, but reduced the number of inflammatory cells in
circulation and decreased GCS, GPx, GRed, and GT in the lung. Authors
concluded that sulfitolysis of glutathione disulphide occurs in vivo during
SO2 exposure and that SO2 is a potent glutathione depleting agent, even
in the absence of pulmonary injury.
Effects of age on SO2-induced oxidative effects in lung tissue were
observed in young (3-mo-old), middle aged (12-mo-old), and old (24-mo
old) rats. SO2 exposure significantly elevated TEARS, SOD, GPx, and
GST in all age groups; reduced catalase in young and middle-aged rats,
but did not affect catalase in old rats. In rats not exposed to SO2, SOD,
GPx and GST increased with age and catalase decreased with age.
General observations in SO2-exposed animals were increases in SOD,
GPx, and TEARS with age. The authors of the AQCD toxicology chapter
noted that while lipid peroxidation increased with age, relative TEARS
increases in response to SO2were inversely correlated with age (i.e.,
largest percent increase seen in young rats).
May 2008
E-14
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Langley-E
vans et al.
(1997;
2007)













CONC.
286 mg/m3
(~101 ppm by
study author
calculations);
whole body
Note' The

*
mistakenly
ud/m and it
was verified

authors that
the units
should have
been listed as
mg/m3.
DURATION
5 h/day for 28
days















SPECIES
Wistar rat, male, 7 wks old,
weight NR, N: 4-16















EFFECTS
This study explored the effects of maternal diet protein restriction during
gestation on offspring lung enzyme responses after SO2 exposure in
adulthood. Adult offspring representing different maternal dietary
concentrations of casein (180 [control], 120, 90 or 60 g/kg) were exposed
either to airorSO2. GSH levels in BAL fluid and the lung were not
affected either by maternal diet or SO2 exposure. In the lung GRed and
GT were not affected by SO2 in any maternal diet group; GPx was
reduced only in the 120 g/kg maternal diet group; GCS was elevated in
the 180 and 60 g/kg groups; and GST was reduced in the 180, 120 and
90 g/kg groups (to the level seen in both the air- and SO2-exposed 60
g/kg maternal diet groups). This study does not provide information
relevant to ambient exposures, but is being mentioned in this table to
record that a low-concentration level study was not overlooked.




Differential Gene Expression - Subacute Exposure
Qin and
Meng
(2005)




Bai and
Meng
(2005a)




14, 28, or 56
mg/m3 (5.35,
10.70, or
21.40 ppm);
whole body


14, 28, or 56
mg/m3 (5.35,
10.70, or
21.40 ppm);
whole body


6 h/day for 7 days






6 h/day for 7 days






Wistar Rat, male, age NR,
180-200 g, N: 6/group in 4
groups




Wistar rat, male, age NR,
180-200 g, N: 6/group in
4 groups




Repeated acute exposure caused significant (p < 0.001-0.05)
concentration-dependent reductions in enzyme activities and gene
expression in the lung for both CYP1A1 and CP1A2. Effects were seen at
the mid and high concentrations, but not the low. Authors speculate that
underlying mechanisms may involve oxidative stress and/or cytokine
release, and may represent an adaptive response to minimize cell
damage.
SO2 exposure caused significant concentration-dependent changes in the
mRNA (mid and high concentrations) and protein expression (all
concentrations in lung, but statistical significance not indicated) of
apoptosis-related genes: increases for bax and p53 apoptosis-promoting
genes, and decreases for the apoptosis-repressing gene bcl-2.
Caspase-3 activity (occurring early in apoptosis process) was also
increased at the mid and high concentration.
Table E-13.  Respiratory system effects of SO2 in disease models.
STUDY
Smith
etal.
(1989)
CONC.
1 ppm
(2.62 mg/m3);
whole body
DURATION
5 h/day, 5 days/wk
for 4 or 8 mos; half
the animals in the
8-mo group were
allowed to recover
for 3 mos.
SPECIES
Sprague-Dawley rat,
male, young adult, initial
weight NR, N: 12-157 data
point
EFFECTS
Respiratory system exposure effects on "normal" and emphysema-like
lungs (elastase induced) were assessed by morphological (e.g.,
histopathology and morphometry) and physiological (e.g., lung function
and volume measured during spontaneous breathing and paralysis) end-
points. At 4 mos of SO2 exposure, bronchial alveolar hyperplasia was
increased in normal animals, but reduced in elastase-treated animals, and
numbers of nonciliated epithelial cells were increased (by 12%) in normal
but not elastase-treated animals; neither morphological observation
persisted past 4 mos of exposure. Physiological tests conducted at 4 mos
of exposure revealed decreased residual volume and quasistatic
compliance in normal SO2-exposed animals during paralyses, and de-
creased residual volume/total lung capacity ratio during spontaneous
breathing and decreased nitrogen washout slope during paralysis in
elastase-treated, SO2-exposed animals. After 8 mos of exposure, lung
volume and incidence of alveolar emphysema were elevated by SO2 only
in the elastase-treated animals; those effects were not observed in the
recovery period. Authors concluded that elastase-induced emphysema
persisted but obscured rather than enhanced SO2 effects. It was indicated
that the model lacked tar residues typically found in the lungs of smokers.
May 2008
E-15
DRAFT—DO NOT QUOTE OR CITE

-------
Table E-14.  Effects of SO2 layered on metallic or carbonaceous particles.
STUDY
Lametal. (1982)
3 h exposure
Hartley guinea pig,
male, age not
reported, 240-300 g,
N: 7-16/group
Amduretal. (1983)
Hartley guinea pig,
male, age not
reported, 200-300 g,
N: 8-23/group
1 h exposure
Chenetal. (1991)
Hartley guinea pig,
male, age not
reported, 275-375 g;
N: 8/group
1 h exposure
Chenetal. (1992)
Hartley guinea pig,
male, age not
reported, 290-410 g,
N: 6-9/group
1 h exposure
S02
-1 ppm
(2.6 mg/m3);
whole body
-1 ppm
(2.6 mg/m3); head
only
1.1 0-1 .25 ppm
(2.9-3.3 mg/m3);
head only
1.02 ppm 2. 7
mg/m3); head only
0
1.10 ppm
(2.9 mg/m3)
1.08 ppm
(2.8 mg/m3)
METAL
Zinc oxide:
0.8, 2. 7, or 6.0 mg/m3
(0.05 [iM projected area
diameter, GSD 2.0)
(sulfate, sulfite, and
sulfur trioxide detected)
7.8 mg/m3
Zinc oxide: ~1-2
(0.05 pM projected area
diameter, GSD 2.0);
mixed at 24 °C and
30% RH
-1-2; mixed at 24 °C
and 30% RH
-1-2; mixed at 480 °C
and 30% RH
-1-2; mixed at 480 °C
and 80% RH with
addition of water vapor
downstream
-1-2; mixed at 480 °C
and 30% RH with
addition of water vapor
during mixing.
Copper oxide
1. 16-2.70 (< 0.1 \iM)
Zinc oxide (0.05 [iM
median diameter, GSD
2.0)
2.76
0.87
2.34
EFFECTS
Vital capacity: No effect with exposure to 7.8 mg/m3 zinc oxide alone and 2.7 mg/m3
zinc oxide in combination with SO2, but decreased with exposure to 0.8 and 6.0
mg/m3 zinc oxide in combination with SO2.
Total lung capacity: No effect with exposure to 7.8 mg/m3 zinc oxide alone, but
decreased with exposure to 6.0 mg/m3 zinc oxide in combination with SO2.
Diffusion capacity for CO and ratio of diffusion capacity for CO to total lung capacity
or alveolar volume: No effect with exposure to 7.8 mg/m3 zinc oxide alone, but
decreased with exposure to 2.7 and 6.0 mg/m3 zinc oxide in combination with SO2.
Alveolar volume: No effect with exposure to 7.8 mg/m3 zinc oxide alone, but
decreased with exposure to 6.0 mg/m3 zinc oxide in combination with SO2.
Pulmonary functioN: SO2 exposure alone resulted in an 11% increase in resistance
and 12% decrease in compliance.
Zinc oxide exposure alone resulted in a 9% decrease in compliance that persisted 1
h after exposure.
Pulmonary functioN: A 12% decrease in compliance and decreased tidal volume that
persisted 1 h after exposure, and decreased min volume. There was no evidence of
new compound formation. Authors concluded that effects on tidal volume and min
volume mostly likely represented an additive effect.
Pulmonary functioN: A 12% decrease in compliance and decreased tidal volume that
persisted 1 h after exposure and a 12% increase in resistance and decreased min
volume. There was no evidence of new compound formation.
Pulmonary functioN: A 13% decrease in compliance that persisted 1 h after
exposure and a 29% increase in resistance. Sulfite formation was observed.
Pulmonary functioN: A 19% increase in resistance that persisted 1 h after exposure,
decreased tidal volume immediately after exposure, and a 26% decrease in
compliance 1 h after exposure. Sulfate, sulfite, and sulfur trioxide formation was
observed.
Pulmonary resistance: Increased 32-47% during exposure and at 1 and 2 h
postexposure when SO2 and copper oxide were mixed at 37 °C, a condition that
resulted in formation of 0.36 pmol/m3 sulfite on the copper oxide particles. No effect
was observed with the compounds were mixed at 1411 °C, a condition that led to
the formation of sulfate on the copper oxide particles.
Dynamic lung compliance: No effect when mixed under conditions that led to the
formation of either sulfate or sulfite on particles.
Baseline pulmonary resistance at 2 h following exposure: No effect in any group.
Airway hyperresponsiveness to acetylcholine: No effect with exposure to SO2 or zinc
oxide alone; compared to furnace controls (3% argon). Hyperresponsiveness
increased in both groups exposed to SO2-layered zinc oxide particles.
May 2008
E-16
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Jakabetal. (1996)
Swiss mics fsmsls
5 wksold, 20-23 g,
N' 5/oroup











Clarke et al. (2000)
Outbred Swiss
mouse, female, age
and weight not
specified, N: 10 or
12 per experimental
value.
4 h exposure once
or for 4, 5, or 6 days















Conner et al. (1985)
Hartley guinea pig,
male, age not
reported, 250-320 g,
N: 5-18/group/
time point
3 h/day for 6 days;
Animals evaluated
for up to 72 h
following exposure















S02
10 ppm (26.2
mg/m3); nose only

0


5 ppm
(13.1 mg/m3)

10 ppm
(26.2 mg/m3)

20 ppm
(52.4 mg/m3)

10 ppm
(26.2 mg/m3);
nose only

0
0





10 ppm
(26.2 mg/m3)


10 ppm
(26.2 mg/m3)


1 ppm
(2.62 mg/m3)


1 ppm
(2.6 mg/m3); nose
only






















METAL
0


Carbon black: 10 mg/m3
(0.3 [iM, GSD 2.7)

10 mg/m3 (formed 6 pg
sulfate at 85% humidity)

10 mg/m3 (formed
13. 7 [ig sulfate at 85%
humidity)
10 mg/m3 (formed 48.7
[ig sulfate at 85%
humidity)
0



Carbon black: 10 mg/m3
(10% humidity)
Carbon black: 10 mg/m3
in 85% humidity to
generate 8 pg/m3 acid
sulfate


Carbon black: 10 mg/m3
in 10% humidity to
generate 41 pg/m3 acid
sulfate
Carbon black: 10 mg/m3
in 85% humidity to
generate 137 pg/m3
acid sulfate
Carbon black: 1 mg/m3
in 85% humidity to
generate 20 pg/m3 acid
sulfate
Zinc oxide: 6 (0.05 [iM
projected area
diameter, GSD 2.0)






















EFFECTS
AM Fc-receptor mediated phagocytosis of sheep red blood cells at 3 days after
exposure: Dose-dependent reductions in AM phagocytosis were observed at each
concentration of SO2 mixed with carbon black aerosol at 85% relative humidity, the
only conditions under which SO2 significantly chemisorbed to carbon black aerosol
and oxidized to sulfate. AM phagocytic activity was reduced somewhat immediately
after exposure (Day 0) was minimal on Days 1 and 3 began increasing on Day 7
and was fully recovered by Day 14. No effects were observed with exposure to SO2
or carbon black alone. The data indicate that environmentally relevant respirable
carbon particles can act as effective vectors for delivering toxic amounts of acid
SO42" to distal parts of the lung.





Inflammatory response after a single 4-h exposure: There was no effect on total cell
number, lymphocyte/PMN differentials, or total protein levels in BAL fluid in any
group.

AM Fc-mediated phagocytosis after a single 4-h exposure: Suppressed by acid
sulfate coated particles (at ~140 Mg/m3) at 1, 3, and 7 days postexposure; values
returned to normal by Day 14.
Intrapulmonary bactericidal activity toward Staphylococcus aureus: Decreased by a
single 4-h exposure to sulfate coated particles (at ~140 pg/m3) at 1 and 3 days
postexposure, with recovery by Day 7. Suppression was also observed after 5 and 6
days of repeated exposure to ~20 pg/m3 sulfate coated particles a condition more
relevant to potential ambient human exposures.












Right lung to body weight ratio: No effect by SO2. Increased for 48 h in group
exposed to SO2-layered zinc oxide.
Right lung wet to dry weight ratio: No effect by SO2. Increased at 1 h after exposure
in SO2-layered zinc oxide group.
Lung morphology: No lesions observed in SO2 group. In group exposed to
SO2-layered zinc oxide, as increased incidence of alveolar duct inflammation
consisting of interstitial cellular infiltrate, increased numbers of macrophages, and
replacement of squamous alveolar epithelium with cuboidal cells. Frequency and
severity of lesions were greatest immediately following exposure and 72 h following
exposure; lesions mild and infrequent.
Tracheal secretory cell concentratioN: No effects with either exposure.
Epithelial permeability: No effects with either exposure scenario.
DNA synthesis (3H-tymidien uptake) terminal bronchial cells: Unaffected by SO2.
Increased at 24 and 72 h after exposure to zinc oxide/SO2.
Lung volumes: Unaffected by SO2 exposure. Functional reserve capacity, vital
capacity, and total lung capacity were decreased from 1 to 72 h following exposure
to zinc oxide/SO2.
Diffusion capacity for carbon monoxide: Unaffected by SO2 exposure. Decreased by
~40-50% from 1 to 24 h following zinc oxide/SO2 exposure.
Alveolar volume: Unaffected by SO2 exposure. Decreased by ~10% from 1 to 24 h
following exposure to zinc oxide/SO2.
Pulmonary mechanics: Respiratory frequency, tidal volume, pulmonary resistance,
pulmonary compliance unaffected by either exposure.
Author conclusioN: Changes were identical to those reported in a previous study in
which guinea pigs were exposed to zinc oxide alone. Sulfur compounds deposited
on surface are less important than zinc oxide particle.
May 2008
E-17
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Amduretal. (1988)
•
Guinea pig, ssx,
age, Weight DOT
reported, N: 8-9
/group

3 h/day for 5 days

1 h exposure




Shamietal. (1985)

and female,
18-19 wks old,
weight not reported,
N: 2/sex/group at
each evaluation time
period
2 h/day for 4 days,
followed by 2 days
without exposure,
followed by 5 more
days of exposure;
evaluated for up to
28 days following
exposure



Wolffetal. (1989)
F344/CN rat male
and fsmals
10-11 wks old
weight not reported,
n=6/sex/group
2 h/d, 5 d/wk, 4 wks
S02
1 ppm
(2.6 mg/m3); head
only










5 ppm
(13 mg/m3); nose
only















5 ppm
(13 mg/m3); nose
only




METAL
Zinc oxide: 1 or 2.5
(0.05 [iM CMD, GSD
2.0)
Sulfate was generated
at 7 and 11 pg/m3at
each respective dose;
sulfuric acid level was
reported at 21 and 33
pg/m3 at each
respective dose.



22 mg/m3 gallium oxide
(0.2 p:M volume median
diameter, GSD not re-
th t riri't'
7 mg/m3
benzo(a)pvrene












Gallium oxide:
27 mg/m3 (-0.20 p:M
MMD, GSD -1.5-2),
with and without
7.5 mg/m3 of
1-nitropyrene and
benzo[a]pyrene

EFFECTS
Pulmonary diffusing capacity: No effect with exposure to 1 ppm SO2 or 2.5 mg/m3
zinc oxide alone (data not shown by authors). Significant and dose related
decreases on exposure days 4 and 5 at 7 pg/m3 sulfate (20% less than control) and
days 2-5 at 11 pg/m3 sulfate (up to 40% less than control).




Bronchial sensitivity to acetylcholine: No effect of 1 ppm SO2 or 2.8 mg/m3 zinc
oxide alone. Increased with SO2 administered in combination with either zinc oxide
dose. The authors noted that responses were similar to those produced by
200 pg/m3 sulfuric acid of similar particle size, thus indicating the importance of
surface layer.
Tracheal and large airways morphology: No effects observed with coexposure to
gallium oxide and SO2.
Pulmonary morphology: Increase numbers of non-ciliated cells in terminal bronchial
epithelium was observed in the SO2/gallium oxide/benzo(a)pyrene group. Mild
peribronchial and perivascular mononuclear inflammatory cell infiltrate and small
hyperplastic epithelial cells in alveoli, and alveolar septal hypertrophy was observed
in the SO2/gallium oxide group, with and without benzo(a)pyrene exposure; effects
were more prominent with benzo(a)pyrene exposure.
Cell proliferation (3H-thymidine intake) in trachea and large airways: In SO2/ gallium
oxide group: increased on days 1 and 14; basal cells primarily labeled, the
SO2/gallium oxide/benzo(a)pyrene group: increased on day 8.
Cell proliferation (3H-thymidine intake) in terminal bronchioles: In SO2/gallium oxide
group: increased on day 14; Clara cells primarily labeled. In the SO2/gallium
oxide/benzo(a)pyrene group: increased on day 11.
Types of 3H-thymidine-labeled cells in the alveolar regioN: In the SO2/gallium oxide
group: type II cells were primarily labeled in the alveolar region through 14 days of
exposure.
In the SO2/gallium oxide/benzo(a)pyrene group: labeling was increased in Type II,
Type I, and endothelial cells on day 8.
Pulmonary particle clearance: No effect was observed with exposure to SO2 alone;
clearance was slowed only by gallium oxide, with or without coexposure to SO2 or
the other compounds; SO2 in combination with the polyaromatic hydrocarbons had
no effect on clearance rate. Authors concluded that toxicity was dominated by
gallium oxide.


Table E-15. Effects of sulfite and mixtures of sulfite and sulfate.
STUDY
SULFITE
SULFATE
DURATION
SPECIES
EFFECTS
Acute
Chen et
al. (1983)




Chen et
al. (1987)






Sodium Sulfite
0.27-1 .95 as
so32-,
submicron in
size, oral
breathing
Sodium Sulfite
0.474-0.972 as
S032",
submicron in
size

















1-h





1-h







Mixed
breed
rabbits, 6
mo, 2.5-
2.7kg, N:
8.
Male
Hartley
guinea
pigs,
200-400 g,
N:7-10


Clearance of tracer aerosol from bronchial tree: Accelerated
clearance at > 1 .2 mg/m3 as SO32~




Pulmonary functioN: A 50% increase in airway resistance and a
19% decrease in compliance were observed at 0.972 mg/m3. All
concentrations resulted in decreased total lung capacity, vital
capacity, functional residual capacity, residual volume, diffusion
capacity for carbon monoxide and increased wet lung weights. The
authors noted that the sulfur of both SO2 and sulfite has a valence
of IV and concluded that aerosols of S(IV) are 6x more potent than
gaseous (IV) in terms of bronchoconstriction.
May 2008
E-18
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
SULFITE
SULFATE
DURATION
SPECIES
EFFECTS
Chronic
Heyder et
al. (1992)
Maier @t
al (1992)

Kreyling
etal.
(1992)
S ch u Iz st
al. (1992)

Takenaka
etal.
(1992)


Heyder et
al. (1999)
Maier et
al (1999)

Kreyling
etal.
(1999)
Schulz et
al. (1999)
Takenaka

et al.
(1999)
Griese
and
Winzinger
(1999)



Neutral sulfite
aerosol, 1.02
mg/m3 with
0.31 mg/m3 as
S(IV)
corresponding
to 0.25 ppm
SO2; some
contamination
with particle-

sulfur (VI) and
gaseous sulfur
(IV) i.e. SO2;
submicron in
size
Neutral sulfite
aerosol, 1.53
mg/m3 with
0.32 mg/m3 as
particle-
associated
S(IV),
corresponding
to 0.25 ppm
SO2; some
minor
contamination

with particle-
associated
sulfur (VI) and
gaseous sulfur
/|\/\ ; _ Q/"i •
(\V) I.e. OU2,
MMAD about a
micron

















Acidic sulfate
aerosol, 5.66
mg/m3, 15.2
pmol/m3
hydrogen ion;
MMAD about one
micron











22.5 h/day for 290
days












Sulfite: 16.5 h/day for
13 mo
Sulfate: 6 h/day for
13 mo

Exposures were
sequential each day
Authors state that
this is equivalent to a
dose received by a
person living for 70 y
in an urban

environment






Beagle
dogs,
male, N: 8












Beagle
dogs,
male, N: 8














Lung mechanics: Decreased specific lung compliance
Alveolar-capillary barrier: Increased permeability
Macrophage-associated defenses: Decreased oxidative defense
and phagocytic capacity. Increased lysosomal activity and
intracellular particle dissolution
Intrapulmonary particle transport: Increased transport to larynx
Cell numbers in BAL fluid: Increased eosinophils and lymphocytes
Oxidant status of extracellular proteins in BAL fluid: Decreased
oxidant status

Structural responses: Proliferative and inflammatory changes in
nasal cavity; loss of cilia in larynx and trachea. Authors conclude
that sulfite aerosols initial a mild histopathological response

Lung mechanics and airway reactivity to carbachol: No significant
effects
Alveolar-capillary barrier: No significant effects
Macrophage-associated defenses: Decreased intracellular particle
dissolution
Intrapulmonary particle transport: Decreased transport to larynx;
increased transport to tracheobronchial lymph nodes
Cell numbers in BAL fluid/cell injury: No significant effects
Antiproteolytic status: Increased elastase inhibitory capacity of
BAL fluid

Oxidant status of extracellular proteins: No significant effects
Structural responses: Increase in volume density of bronchial
glands. Proliferation of Type 2 cells in proximal alveolar region and
thickening of the basal membrane beneath these cells. Increased
volume density of alveolar ducts and alveolar sacs in acinus.
Pulmonary surfactant system: No significant effects. Authors
conclude that these responses were less pronounced than those in
the previous study using sulfite alone.
Table E-16.  Effects of mixtures containing SO2 and ozone.
STUDY
S02
OZONE
DURATION
SPECIES
EFFECTS
Acute/Subacute Exposure
Abraham
etal.
(1986)
3 ppm
(7.9 mg/m3); head
only
0.3 ppm
5 h/day for
3 days
Sheep, sex NR, adult,
23-50 kg, N: 6
Tracheal mucus velocity: Decreased by 40%
immediately after exposure and 25% at 24 h
postexposure to the mixture of the 2 compounds. The
effects of either compound alone were NR.
Ciliary beat frequency: No effect
Chronic/Subchronic Exposure
Aranyi
etal.
(1983)
13.2 mg/m3
(5.0 ppm) in
addition to
1 .04 mg/m3
ammonium sulfate;
whole body
0.2 mg/m3
(0.10 ppm)
5 h/day,
5 days/wk for
up to 103 days
CD1 mice, female,
3-4 wks old, weight NR,
N: 360/group total
(14-154/group in each
assay)
Mortality rate after Streptococcus aerosol challenge:
Increased in groups exposed to ozone alone and
mixture of ozone, SO2, and ammonium sulfate.
Alveolar macrophage bactericidal activity towards
inhaled K. pneumoniae: Increased trend (non-significant)
in ozone group but significantly increased in mixture
group.
Counts, viability, and ATP levels in cells obtained by
pulmonary lavage: No effect of either treatment
May 2008
E-19
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Raub
etal.
(1983)




























S02
1 ppm
(2.62 mg/m3);
whole body




























OZONE
1 ppm in
addition to
3 ppm
trans-2-butene



























DURATION
23 h/day,
7 days/wk, for
4 wks




























SPECIES
Golden hamster, male,
age NR, -105 g, N: 14 or
15/group; mild emphy-
sema was induced in
some animals by
.
elastase
























EFFECTS
Lung volumes: End expiratory volume, residual volume,
total lung capacity and vital capacity were unaffected in
the mixture versus air exposure group in normal or
emphysematous hamsters.
Respiratory system compliance: Unaffected in the
mixture versus air exposure group in normal or
emphysematous hamsters.
Distribution of ventilation (N2 washout slope): The N2
slope decreased in the mixture versus air exposure
group in both normal and emphysematous hamsters.
Diffusion capacity for carbon monoxide: Significantly
increased in the mixture versus air-exposed normal
animals. Although the text reported an increase in the
mixture versus air-exposed emphysematous animals,
Figure 3 of the study indicated that the effect was very
small and did not obtain statistical significance.
Significantly lower in emphysematous versus normal
hamsters exposed to the mixture. The authors noted a
significant interaction between exposure to the mixture
and emphysema.
Histopathology: Inflammatory lesions were found in the
lungs of emphysematous hamsters exposed to air or the
mixture. Hyperplasia incidence was higher in
emphysema hamsters exposed to the mixture versus air.
Inflammatory lesions were similar in emphysematous
hamsters exposed to air or the mixture. Data were not
shown for histopathology data.
Overall author conclusioN: Animals with impaired lung
function may have decreased capacity to compensate
for the pulmonary insult caused by exposure to a
complex pollutant mixture.
Table E-17.  Effects of SO2 and sulfate mixtures.
STUDY
SO2
SULFATE
EFFECTS
Acute
Mannixetal. (1982)
Sprague Dawley rat, male, age
NR, -200 g, N: 8/group
4 h exposure
5 ppm
(13.1 mg/m3);
nose only
Sulfate aerosol 1.5 (0.5
pM MMAD, GSD 1 .6)
Lung clearance of radiolabeled tracer particles: No significant effect was
observed with the mixture of the two compounds at 80-85% humidity.
Chronic/Subchronic
Smith etal. (1989)
Sprague-Dawley rat, male,
young adult, initial weight NR,
N: 12-15/data point
Exposure: 5 h/day, 5 days/wk
for 4 or 8 mos; half the
animals in the 8-mo group
were allowed to recover for
3 mos.
1 ppm
(2.62 mg/m3);
whole body
0
0
(NH4)2SO4:
0.5 mg/m3
(MMAD = 0.42-0.44 ±
0.04pm, GSD 2.2-2.6)
Morphological observations at 4 mos exposure in "normal" rats:
Bronchiolar epithelial hyperplasia and increased numbers of non-ciliated
epithelial cells were observed in rats exposed to either compound alone
but coexposure to both compounds did not magnify the effects. An
increase in alveolar chord length was observed in the (NH4)2SO4 group
and no further changes were observed with coexposure to SO2.
Morphological observations at 4 mos exposure in rats treated with
elastase to induce an emphysema-like condition
Bronchiolar epithelial hyperplasia was decreased in groups exposed to
either compound alone or the mixture of the two compounds. A decrease
in alveolar chord length was observed in the (NH4)2SO4 group and no
further changes were observed with coexposure to SO2.
May 2008
E-20
DRAFT—DO NOT QUOTE OR CITE

-------
          STUDY
                                 SO,
                                                  SULFATE
                                                                                              EFFECTS
                            1 ppm
                            (2.62 mg/m3)
                       0.5 mg/m
Morphological observations at 8 mos exposure in "normal" rats: An
increase in non-ciliated epithelial cells and alveolar birefringence (an
indication of alveolar interstitial fibrosis) was observed only in the group
exposed to (NH4)2SO4.

Morphological observations at 8 mos exposure in rats treated with
elastase: An increase in lung volume per body weight and emphysema
incidence was observed in groups treated with either compound alone or
in combination; alveolar chord length was increased only in the group
exposed to the mixture of compounds.

Morphological observations at 12  mos exposure  in normal rats: Increased
alveolar chord length was observed only in the (NH4)2SO4 group.

Morphological observations at 12  mos exposure  in rats treated with
elastase: In increase in absolute lung volume was observed only in the
group treated with the mixture of both compounds.

Lung function effects at 4 mos exposure in normal rats: A decrease in
residual volume was observed in the SO2 group and decreased
quasistatic compliance was observed in the SO2 group and in the
(NH4)2SO4 group, but the effects were not observed with the mixture.

Lung function effects at 4 mos exposure in elastase-treated rats:  Ratio of
residual volume/total lung capacity and N2 washout was decreased in the
SO2 group and in the (NH4)2SO4 group, but the effects were not observed
with the mixture.

Overall conclusions: In general, pollutant effects were minimal and
transient, and appeared obscured or repressed in elastase-treated
groups; (NH4)2SO4 was more bioactive than SO2, with  little evidence of
mixture additivity (in several instances, effects seen with one or both
pollutants individually were not seen with the mixture).
Table E-18.   Effects of actual or simulated air pollution mixtures.
      STUDY
                              EXPOSED
                                                      CONTROL
                                                                                                  EFFECT
Acute/Subactute
Mautzetal. (1988)

Sprague-Dawley rat,
male, age NR,
240-280 g, N:
6-9/group

Exposure: 4 h
Air pollutant mixture at full con-
centration (tested in 2 studies):
0.35 ppm ozone, 1.3 ppm nitro-
gen dioxide, 2.5 ppm (6.6 mg/m3)
SO2, 10 pg/m3 manganese
sulfate, 500 pg/m3 ferric sulfate,
500 pg/m3 ammonium sulfate,
500 pg/m3 carbon aerosol.
Mixture also tested Vi and 1/J
concentrations. For aerosols:
MMAD = 0.3-0.48 \M with GSD:
2.6-4.6.  Nose-only exposure.
Compounds formed: sulfate, ni-
trate, hydrogen ion, nitric  acid.
                                                   Clean air
 Breathing patterN: Effect of full concentration mixture in 2 studies: increased
 breathing frequency, trend or significant decrease in tidal volume, decreased
 or unaffected oxygen consumption, increased or unaffected ventilation
 equivalent for oxygen. Effect of half concentration mixture: increased min
 ventilation. Quarter concentratioN: no significant effects.

 Histopathology: Full concentratioN: Area of type 1 parenchymal lung lesions
 increased in 1 of 2 experiments; area of type 2 parenchymal lung lesions were
 increased in both experiments. Effects were equivalent to those observed with
 ozone exposure alone. Half and quarter concentrations:  No effects.

 Mucociliary clearance: No effect on early or late clearance of 85Kr-labeled
 polystyrene particles.

 Nasal epithelial injury (measured by tritiated thymidine uptake): No effect at
 any concentration.
Phalen and Kleinman
(1987)

Sprague-Dawley rat,
male, age NR,
200-225 g, N:
5-13/group/
time period

Exposure: 4 h/day for
7 or 21 days
2.55 ppm (6.7 mg/m3) SO2,
0.3 ppm ozone, 1.2 ppm nitrogen
oxide, 150 pg/m3 ferric oxide, 130
pg/m3 nitric acid, 2.0 pM/m3
hydrogen ion, and  500 pg/m3 total
Fe3+, Mn2+, and NH42+ combined;
nose only
                                                   Purified air
 Bronchoalveolar epithelial permeability to   Tc-diethylenetriamine-
 pentaacetate: No effect at either time period.Nasal mucosal permeability to
   mTc-diethylenetriaminepentaacetate: No effect at either time period.

 Macrophage rosette formatioN: Decreased (indicating damage to Fc receptors)
 up to 4 days after 7- or 21-day exposure; magnitude of effect greater following
 21-day exposure. By day 4 after exposure, numbers began increasing and by
 day 7 were equivalent to control values.

 Macrophage phagocytic activity: Rats exposed for 7 days, decreased activity
 observed for 2 days post-exposure. No effects after 21-day exposure.
May 2008
                                     E-21
                   DRAFT—DO NOT QUOTE OR CITE

-------
      STUDY
                              EXPOSED
                                                     CONTROL
                                                                                                 EFFECT
Subchronic/Chronic
Saldivaetal. (1992)

Wistar rat, male, 2
mos old, weight NR,
N: 14-30/group

Exposure: 6 mos
Urban air: Sao Paulo, mean
levels of air pollutants measured
200 m from the police station
where rats were kept: 29.05
pg/m3 (0.011 ppm) SO2;
1.25 ppm carbon monoxide,
11.08 ppb ozone, 35.18 pg/m3
particulates.
Rural air: Atibaia,
an agricultural
town 50 km from
Sao Paulo was
considered the
control; air pollut-
ant levels were
not measured.
Death: 37 of 69 Sao Paulo rats died before study end; autopsy of 10 animals
identified pneumonia; 10/56 Atibaia animals died.

Respiratory mechanics: Nasal resistance higher in Atibaia animals.
No differences for pulmonary resistance or dynamic lung elastance.

Mucus properties: Sao Paulo animals' tracheal mucus output was  lower,
relative speed of tracheal mucus was slower, ratio between viscosity and
elasticity was higher for nasal mucus, and rigidity of tracheal mucus was
increased.

Bronchoalveolar lavage: In lavage fluid from Sao Paulo animals, increased
numbers of cells, lymphocytes, polymorphonuclear.

Histochemical evaluatioN: Hyperplasia was observed in respiratory epithelium
of rats housed in Sao Paulo.

Ultrastructural studies: Animals housed in Sao Paulo had a higher frequency of
cilia abnormalities including composite cilia, microtubular defect, vesiculation,
and decreased microvelocity of luminal  membrane.
Lemosetal. (1994)

Rats from the same
cohort as Saldiva
etal. (1992). N:
15/group

Exposure: 6 mos
Urban air: Sao Paulo, mean lev-
els of pollutants measured 200 m
from police station where rats
were kept: 29.05 pg/m3
(0.011 ppm) SO2; 1.25 ppm car-
bon monoxide, 1.08 pb ozone,
35.18 pg/m3 particulates.
Rural air: Atibaia,
agricultural town
50 km from Sao
Paulo, considered
control; air pollut-
ant levels not
measured.
Nasal passage pathology: Rats housed in Sao Paulo had increased nasal
epithelium volume, larger amounts of mucosubstances stored in epithelium,
and more acidic mucus secretions in lamina propria glands.
Pereiraetal. (1995)

4 groups of rats
housed: 3 mos in Sao
Paulo, 3 mos in Sao
Paulo followed by 3
mos in Atibaia, 3 mos
in Atibaia, or 6 mos at
Atibaia.

Wistar rats, male, 1.0-
1.5 mos old, weight
NR, N: 30/group
Urban air: Sao Paulo, levels of air
pollutants measured were: ~8-50
pg/m3 (0.003-0.019 ppm) SO2,
~0.1-0.45 ppm nitrogen dioxide,
~4.8-7 ppm carbon monoxide,
and ~50-120 pg/m3 particulate
matter.
Rural air: Atibaia,
an agricultural
town 50 km from
Sao Paulo was
considered the
control; air pollut-
ant levels were
not measured.
Lung responsiveness to methacholine: Increased respiratory system elastance
resulting from increased sensitivity to methacholine in rats housed in Sao
Paulo for 3 mos compared to all the other groups. No exposure-related effects
were observed for respiratory system resistance.
May 2008
                                     E-22
                                   DRAFT—DO  NOT QUOTE OR CITE

-------
Table E-19.   Effects of meteorological conditions on SO2 effects.
                STUDY
                                              SO,
                                                           CONDITION
                                                                                                   EFFECT
Barthelemy et al. (1988)

Rabbit, sex NR, adult, mean 2.0 kg, N:
5-10/group; animals were mechanically
ventilated.

Exposure: 45 min
0.5 or 5 ppm (1.31
or 13.1 mg/nr);
intratracheal
Drop in air
temperature from
38°Cto 15 °C
Lung resistance: Exposure to cool air for 20 min resulted in a
~54% mean increase in lung resistance. Exposure to SO2 for 20
min increased lung resistance by 16% at 0.5 ppm and 50% at
5 ppm. The difference in lung resistance from warm to cold air
was halved (27%) by exposure to 0.5 ppm and was not
significant at 5 ppm. The authors concluded that transient
alteration in tracheobronchial wall following SO2 exposure may
have reduced accessibility of airway nervous receptors to cold
Halinen et al. (2000a)

Duncan-Hartley guinea pigs, male, age and
weight NR, N: 7-12/group, mechanically
ventilated; animals were hyperventilated
during cold air and SO2 exposure to
simulate exercise.

In pre-exposure period: 15-min exposure to
warm humid air, 10-min exposure to cold
dry air, and 15-min exposure to warm
humid air. In the SO2 exposure period:
10-min exposures to each SO2
concentration in cold dry air or with cold dry
air alone were preceded and followed by
15-min exposures to warm humid air.
1.0, 2.5, or 5 ppm
(2.62, 6.55  or
13.1  mg/nr);
apparently
intratracheal
Drop in
intratracheal
temperatures from
-35.5 °C to -27
°C
Peak expiratory flow: Percent decreases were significantly
greater with exposures to SO2 in dry air at concentrations of
1.0 ppm (-32.7%) and 2.5 ppm (-35.6%) than with exposure to
cold dry air (-27%); decrease at 5 ppm SO2 in cold dry air
(-25.3%) was similar to that with cold dry air. The effects did not
persist following exposures.

Tidal volume: Percent decreases were significantly greater with
exposure to SO2 in cold dry air at concentrations of 1.0 ppm
(-22.4%) and 2.5 ppm (-28.3%) than with exposure to cold dry
air (-18.1%); decrease at 5 ppm SO2 in cold dry air (-17.8%)
was similar to that of cold dry air. The effects did not persist
following exposures.

Bronchoalveolar lavage: The clean dry air group had significantly
more macrophages, lymphocytes, and increased protein
concentration in lavage than the warm humid air control. The
cold dry air + SO2 group had fewer macrophages than the clean
dry air group and  higher protein concentration than controls.

Histopathology: Increased incidence of eosinophilic infiltration
within and below tracheal epithelium with exposure to cold dry air
or SO2 in cold dry air.
Halinen et al. (2000b)

Duncan-Hartley guinea pigs, male, age and
weight NR, N: 8-9/group, mechanically
ventilated; animals were hyperventilated
during cold air and SO2 exposure to
simulate exercise.

Exposure: 60 min
1 ppm
(2.62 mg/m3);
apparently
intratracheal
Drop in
intratracheal
temperatures from
-37 °C to -26 °C
Peak expiratory flow: Non-significant decreases compared to
baseline (4.5-10.8%) at 10 and 20 min of exposure to cold dry
air. With exposure to SO2 in cold dry air: decreased significantly
(11.4%, i.e., bronchoconstriction) compared to baseline at 10
min of exposure but recovered from 20 to 60 min of exposure.
The effect with SO2 exposure was not statistically significant
compared to that of cold dry air alone.

Tidal volume:  Decreased from baseline throughout most of the
exposure period with cold dry air or SO2 in cold dry air; response
with SO2 was  more shallow than that of cold dry air alone, but
statistical significance compared to cold dry air was obtained
only at 60 min of exposure.

Bronchoalveolar lavage: Decreased neutrophil numbers in the
SO2 group compared to the warm humid air group but no
significant difference compared to the cold dry air group.

Histopathology: No effect in lung or tracheobronchial airway.

General conclusions: Functional effects on the lower respiratory
tract were weaker than in the previous study with 10-min
exposures (Halinen et al., 2000a).
May 2008
                  E-23
                            DRAFT—DO  NOT QUOTE OR CITE

-------
Table E-20.  In vitro or ex vivo respiratory system effects of SO2 and metabolites.
STUDY
CONCENTRATION
DURATION
SPECIES
EFFECTS
In Vitro — Primary/Nonprimary
Blanquart
etal. (1995)





Menzel et al
(1986)













0, 5,10, 20, 30, or 50 ppm (0,
13.1,26.2 52.4, or
131 mg/m )SO2




0,0.1,2,20, or 40 mM (0,4,
80, 800, or 1600 pg/mL)
S032'












1-h






-1 min - 96 h














Fauve de Bourgogne
rabbits, 1 mo old,
tracheal epithelium
explants



Rat, Sprague-Dawley,
200-250g; sex, age,
and n NR; lung cells
and liver cells.

Human lung-derived
cell line, A549








Relative to control cultures, cell viability was not reduced at
5 and 10 ppm, but was at 30 ppm (-70%) and 50 ppm (-60%).
Ciliary beat frequency was significantly reduced (p < 0.05) at
10-30 ppm, and was correlated with swollen mitochondria and
depletion of cellular ATP, as well as with blebbing of ciliated or
microvilli-covered cells and with aggregation and flattening of
cilia.
This study focused on intracellularcovalent reactions of sulfite
with primarily proteinaceous sulfhydryl compounds in cells
isolated from rat lung and rat liver (for some comparative
purposes), as well as in the human lung-derived cell line, A549.
Sulfitolysis of protein disulfide bonds results in formation of
cysteine S-sulfonate, and sulfitolysis of GSSG in formation of
GSSO3H. The latter was formed in dose-dependent fashion
upon the addition of sulfite to A549 cells. In addition to fi-
bronectin and albumin, this study identified a third sulfite-binding
protein in rat lung cytosol. GSSO3H was shown to be a potent
competitive inhibitor of GST in rat lung, liver and A549 cells.
Results suggest that SO2 could affect the detoxication of PAHs
and other xenobiotics via formation of GSSO3H and subsequent
inhibition of GST and enzymatic conjugation of GSH with
reactive electrophiles.
Ex Vivo
Riechelmann
etal. (1995)














Knorst et al.
(1994)




7.5, 15, 22.5, 30, or 37.5
mg/m3 (2.9, 5.7, 8.6, 11. 5, or
14.3 ppm); ex vivo expsoure
of trachea












2.5,5.0,7.5, 10.0, or
12.5 ppm (6.6, 13.1, 19.7,
26.2, or 32.8 mg/m3); ex vivo
exposure of trachea


30 min















30 min





Guinea pig, sex, age,
and weight NR,
N: 4-8/group













Guinea pig, sex, age,
and weight NR, N:
4-7/group



No remarkable morphologic abnormalities in the tracheal
mucociliary system of the 2.9 ppm group, though slight vacuoli-
zation, rare membrane blebs, and slightly widened intercellular
spaces were observed. Abnormalities in the 5.7 and 8.6 ppm
groups were similar and included loosened contact to the basal
membrane, extensive intracellular edema and vacuolization,
swollen mitochondria, polypoid extrusions and huge blebs in the
cell membrane and ciliary membrane, widened intercellular
space, and disrupted tight junctions. Additional abnormalities in
the 11.5 and 14.3 ppm groups included marked epithelial
sloughing, occasionally disrupted cell membranes and
microtubules, and frequently disrupted ciliary membranes.
Tracheal mucociliary activity was significantly decreased in all
exposure groups (from 8.7 ± 1 .0 Hz [controls] to 4.0 ± 1 .1 , 3.4 ±
2.7, 1 .8 ± 2.2, 1 .5 ± 1 .8, and 2.0 ± 1 .2 Hz in the 7.5, 15, 22.5, 30,
and 37.5 mg/m3 groups, respectively).
63% decrease in tracheal mucociliary activity at 2.5 ppm with
dose-dependent decrease to 81% at 7.5 ppm; higher concentra-
tions did not further decrease mucociliary activity. Ciliary beat
frequency decreased by 45% at 5.0 ppm with dose-dependent
decrease to 72% at 12.5 ppm. All reductions are relative to
baseline values; no effect on controls for either parameter.
May 2008
E-24
DRAFT—DO NOT QUOTE OR CITE

-------
Table E-21.  Genotoxic effects of SO2 and metabolites.
STUDY
CONCENTRATION
DURATION
SPECIES/SYSTEM
EFFECTS
In Vitro "Point Mutation"1
Pool-Zobel
etal.
(1990)



0 or 50 ppm (131 mg/m3) SO2
or the equivalent agar
concentration of SO32~,
15 pg/ml)


48 h





Rat, Sprague-Dawley,
female, liver enzyme
preparations



In vitro induction of reverse mutation in cultures of
S. typhimurium strain TA98 was not affected by incubating
the bacterial-B[a]P-liver S9 enzyme activation system in
the presence of SO2/sulfite. An ancillary finding from the 0
pg B[a]P control exposures is that SO2/sulfite itself did not
appear mutagenic.
In Vitro Cytogenetic and DNA Damage2
Pool et al.
(1988)
























Shi and
Mao (1994)


Shi (1994)




0,20, 50 or 200 ppm (0,52.4,
131 or 524 mg/m3) SO2;
0,0.1, 0.2 or 0.4 mM SO32~
0 or 2.5 pmol HSO3" per
microtiter plate well
0 01 02 or 04 mM SO42

0 or 10 pmol MgSO4 per tube




















3 mM SO32-



5 mM SO32" (as Na2SO3)




1-24h

























40 min
(test tube
reactions)

1.5h
(test tube
reaction)


Hamster, Syrian golden,
fetal lung cells (FHLC,
gestational Day 15)
Rat, Sprague-Dawley,
male, age NR, ~200g,
hepatocytes

Chinese hamster ovary
cell line transformed by
SV40, CO60 cells
Precinorm U (human
serum standard)


















dGorDNA



dG




Toxicity and genotoxicity of SO2, sulfite/bisulfite and sulfate
(also NO2/NOX) were variously assessed in several in vitro
test systems. It was noted that medium pH remained
stable at [SO2] £ 200 ppm. Precinorm LDH activity was
substantially inhibited by 50 ppm SO2 after 1-3-h, and by
0.1 mM sulfite ion almost immediately, but not by 0.1 mM
sulfate ion; AST was modestly inhibited after 5 h by
200 ppm SO2; other monitored enzymes were not affected.
While trypan blue exclusion was not affected, SO2
cytotoxicity to FHLC was demonstrated at 20 ppm by
reduced plating efficiency; at 50 ppm, enzyme activity
leaked into culture medium was reduced only for AP and
especially LDH (not other enzymes). 200 ppm SO2 did not
induce DNA damage (single-strand breaks) by itself in
either FHLC or rat hepatocytes, but did somewhat reduce
that induced by AMMN. In hepatocytes, incubation with
MgSO4 also caused a small reduction in AMMN-induced
DNA damage. A 1-h exposure to 200 ppm SO2 did not
induce selective amplification of SV40 DNA in CO60 cells,
nor affect that induced by DMBA or B[a]P. However, while
also not affecting induction by DMBA or B[a]P, HSO3"
added directly to the medium for 24 h did induce SV40
DNA amplification on its own - authors appear to suggest
this might result from arrest of cells in mid-S phase, which
leads to DNA amplification. Thus, principal findings include
inhibition of LDH by SO2 or sulfite that could impair the
cellular energy system; such an impairment could be
responsible (possibly along with SO42" conjugation of
reactive intermediates) for the observed inhibition of
AMMN-induced DNA damage by SO2. Further, SO2 does
not appear by itself to induce DNA damage.
Test tube reaction mixtures that caused sulfite to oxidize to
sulfur trioxide radical (SO3~) resulted in the hydroxylation of
dG (8-OHdG) and the generation of DNA double strand
breaks.
Test tube reaction of sulfite ion with H2O2 shown to
generate OH radicals capable of hydroxylating dG to the
DNA damage marker, 8-OHdG. Furthermore, incubation of
sulfite with nitrite or various transition metal ions was
shown to generate sulfur trioxide anion radical (SO3~).
Acute/Subacute Exposure Cytogenetic and DNA Damage2
Ruan et al.
(2003)







0 mg/m3 (0 ppm) SO2 (+ 0 or
8 mg/kg bw SSO) or 28
mg/m3 (10.7 ppm) SO2 (+ 0, 2,
4, 6 or 8 mg/kg bw SSO);
whole body




± SSO ip on
Days 1-3; then
SO2 for 5 day
(Days 4-8),
6 h/day




Kunming mouse, male
and female, ~6 wk old,
20-25 g, N: 6/sex/conc.






Subacute inhalation of 28 mg/m3 SO2 induced a significant
(p < 0.001) 10-fold increase in mouse bone marrow
MNPCE, which was partially mitigated in dose-dependent
fashion by pretreatment with SSO, a complex natural
anti-oxidant substance. SO2 exposure also resulted in
organ :bw ratios that increased for liver and kidney,
decreased for lung and spleen, and remained unchanged
for heart. Such ratio changes were largely mitigated by
SSO pretreatment.
May 2008
E-25
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Meng et al.
(2002)
Meng et al.
(2005b)
Pool et al.
(1988)
Ohyama
etal.
(1999)
CONCENTRATION
0, 14, 28, 56, or84mg/m3(0,
5.35, 10.7, 21.4, or 32.1 ppm)
SO2; whole body
0, 14, 28, 56, or84mg/m3(0,
5.35, 10.7, 21.4, or 32.1 ppm)
SO2; whole body
0 or 50 ppm (131 mg/m3) SO2
0, 0.2 mLC, or
(0.2 mL DEP+C ± [4 ppm
(10.48 mg/m3) SO2 or 6 ppm
(1 1 .28 mg/m3) NO2 or 4 ppm
SO2 + 6 ppm NOJ); whole
body
[Note: 0.2 mL C = 1 mg; 0.2
mL DEcCBP =1 mg C + 2.5
mg DEP)]
DURATION
7 day, 4 h/day
7 day, 6 h/day
2 wk, 7 day/wk,
24 h/day
SO2 and/or NO2:
10 mo, 16 h/day
C or DEP+C:
4 wk, once/wk by
intratracheal
infusion
SPECIES/SYSTEM
Kunming mouse, male
and female, ~6 wk old,
20-25 g, N: 10/sex/conc.
Kunming mouse, male
and female, ~5 wk old,
18-20 g, N: 6/sex/conc.
Rat, Sprague-Dawley,
female, 4 mo old, wt NR,
N: 5 per group
Rat, SPF F344/Jcl, male,
6 wk old, wt NR, N: 23-30
per group in 6 groups
EFFECTS
In vivo exposure caused significantly (p < 0.01-0.001)
increased frequencies of bone marrow MNPCE similarly in
both sexes at all concentrations in a dose-dependent
manner, and with only minimal cytotoxicity at the 3 highest
concentrations. The level of MNPCE (%) even at the low
SO2 cone, was triple that of the control value. Thus,
subacute inhalation of SO2 at noncytotoxic concentrations
(though still notably higher than most human exposures)
was clastogenic in mice.
Following in vivo exposure to SO2, it was shown by the
single cell gel electrophoresis (comet) assay that such
exposure induced significant (p < .001-. 05) dose-
dependent DNA damage (presumed mostly to be
single-strand breaks and alkali-labile sites) in cells isolated
from brain, lung, liver, intestine, kidney, spleen, and
testicle, as well as in lymphocytes, and beginning at the
lowest concentration (except male intestine — lowest
response at 28 mg/m3). Results demonstrate that SO2, can
cause systemic DNA damage in many organs, not just the
lung. Authors note that potential occupational exposures
and the fact that the obligate nose-breathing mouse
removes ~95% of inhaled SO2 in its nasal passages make
this experimental concentration range relevant to possible
human exposures.
Assessments were conducted on isolated primary lung and
liver cells, or on blood serum. In vivo SO2 exposure did not
affect viability (trypan blue exclusion) of cells either
immediately after isolation or after 1 h incubation with 1%
DMSO (used for enzyme leakage assays). In contrast to
controls, hepatocytes from SO2-exposed rats released no
LDH activity into DMSO-medium after 1-h, and AST
activity was reduced. Other enzyme (AP, ALT, GT) activity
releases were not affected in lung cells, and none were in
hepatocytes. In blood serum, the only effect was a marked
increase in LDH activity. The only significant (p < 0.001-
0.01) exposure effects on lung or liver activities (in x 9000
g supernatants of cell homogenates) of xenobiotic
metabolizing enzymes (AHH, NDMA-D, GST) were
elevated NDMA-D in the liver and reduced GST in the
lung. Single-strand DNA breakage induced by three nitroso
compounds (AMMN, NDMA, NMBzA) was reduced in
hepatocytes from SO2-exposed rats. Authors discuss
possible mechanisms for the observed effects, and note
they are similar to in vitro effects reported elsewhere (Pool
etal., 1988).
Purpose was to study effects of DEP on rat lung
tumorigenesis and possible tumor promoting effects of SO2
or NO2 singly or together. [See Table C-1 1 for
tumor-related effects.] DEP extract-DNA adducts were
found only in the three gas-exposed groups.
Chromatograms revealed two different adducts, one of
which appears somewhat more abundant with SO2
coexposure, the other substantially more so with NO2;
combined coexposure of both gases with DEP+C
produced an adduct chromatogram appearing to be a
composite of those for the individual gases. Thus, SO2 and
NO2 appear capable of promoting the genotoxicity of DEP
extract, though perhaps not in identical fashion.
 Encompasses classical mutant selection assays based upon growth conditions under which mutants (or prototrophic revertants), but not the wild type (or auxotrophic)
population treated with the test agent, can successfully grow (e.g., "Ames test," CHO/HGRPT or mouse lymphoma L5178Y/TK mammalian cell systems, various yeast and
Drosophila systems, etc.); while most viable mutation events detected in these assays are typically "point" mutations (DNA base substitutions, small deletions or frameshifts,
etc.), some may involve larger losses/rearrangements of genetic material.

2Encompasses CA, induction of MN or SCE, aneuploidy/polyploidy, DNA adduct and crosslink formation, DNA strand breakage, etc.
May 2008
E-26
DRAFT—DO  NOT QUOTE OR CITE

-------
Table E-22.  Liver and gastrointestinal effects of SO2.
STUDY
CONCENTRATION
DURATION
SPECIES
EFFECTS
Subacute/Subchronic Exposure
Meng et al.
(2003b)







Wu and Meng
(2003)

Bai and Meng
(2005b)




Qin and Meng
(2005)






Lovati et al.
(1996)



















Langley-Evans
etal. (1996)



Langley-Evans
etal. (1997);
Langley-Evans
(2007)







22,56, or 112 mg/m3
(7.86, 20, or 40 ppm
per author
conversion); whole
body




22,64, or 148 mg/m3
(8.4, 24.4, or
56.5 ppm); whole body
14, 28, or 56 mg/m3
(5.35, 10.70, or
21.40 ppm); whole
body


14, 28, or 56 mg/m3
(5.35, 10.70,
or 21 .40 ppm); whole
body




5 or 10 ppm (13.1 or
26.2 mg/m3); whole
body


















5, 50, or 100 ppm
(13.1, 131, or
262 mg/m3); whole
body

286 mg/m3 (100 ppm);
whole body Units were
incorrectly reported as
pg/m3 in the study but
were corrected
according to
information provided
by study author



6 h/day for 7 days








6 h/day for 7 days


6 h/day for 7 days





6 h/day for 7 days







24 h/day for
15 days



















5 h/day for
7-28 days



5 h/day for
28 days









Kunming albino mouse,
male and female, 5 wks
old, 19±2g, N:
6/sex/subgroup





Kunming-strain mice,
male, age NR, 18-20g,
N: 10/group
Wistar rat, male, age NR,
180-200 g, N: 6/group in
4 groups



Wistar rat, male, age NR,
180-200 g, N: 6/group in
4 groups





Sprague-Dawley CD rat,
male, age NR, 250-275
g, N: 9/subgroup


















Wistar rat, male, 7 wks
old, weight NR, N:
4-5/treatment group,
8 controls

Wistar rat, male, 7 wks
old, weight NR, N:4-16









Effects observed in stomach (concentration of effect)
included: increase in SOD activity (7.86 ppm, males only)
and TEARS level (> 7.86 ppm) and decreases in SOD
(> 20 ppm, males only) and GPx activities (> 20 ppm, males
only) and GSH level (40 ppm). Effects observed in intestine
were increases in catalase activity (> 20 ppm in males,
40 ppm in females) and TEARS level (> 20 ppm) and
decreases in SOD (> 7.86 ppm) and GPx (> 20 ppm)
activities and GSH level (> 7.86 ppm).
No effects were observed in the liver at 22 or 64 mg/m3. GST
and glucose-6-phosphate dehydrogenase activities and GSH
level were decreased at 148 mg/m .
Significant and concentration-dependent changes in mRNA
(mid and high concentrations) and protein expreyssion (all
concentrations) included increases for bax and p53
apoptosis-promoting genes, and decrease for bcl-2
apoptosis-repressing gene. Authors speculated potential
impact on human apoptosis-deficient diseases.
SO2 caused significant concentration-dependent reductions
in liver enzyme activities and gene expression for CYP1A1
and CYP1A2. Effects were seen at the mid and high
concentrations (only high for CYP1A1 enzyme activity), but
not the low. Authors speculate that underlying mechanisms
may involve oxidative stress and/or cytokine release, and
may represent an adaptive response to minimize cell
damage.
Subjects were rats fed standard diet (normal) or high
cholesterol diet, and rats with streptozotocin-induced
diabetes fed standard diet. SO2 (2 5 ppm) elevated plasma
triglycerides in normal and hypercholesterolemic groups,
while 10 ppm lowered plasma high density lipoprotein
cholesterol in hypercholesterolemic rats. In diabetic rats,
10 ppm SO2 lowered triglycerides and free fatty acids without
affecting high density lipoprotein cholesterol or total
cholesterol. In the liver, SO2 elevated triglycerides in normal
and hypercholesterolemic groups (at 10 ppm), but lowered it
in diabetic rats (at > 5 ppm); esterified cholesterol was
elevated in normal rats (at 10 ppm), but lowered in diabetic
rats (at > Sppm), and free cholesterol was unchanged in all
groups. In normal rats, triglycerides secretion rate was
inhibited by 10 ppm SO2. SO2 caused several changes in
plasma apolipoprotein composition in normal and
hypercholesterolemic groups, but not in diabetic rats.
Leukotriene parameters were not affected. Thus, in each rat
model, inhalation of SO2 at levels without overt effects
affected plasma and tissue lipid content. Specific effects
varied according to diet or diabetes.
GSH was depleted in the liver at 5 and 100 ppm but not at
50 ppm. With respect to GSH-related enzymes, exposure to
5 ppm decreased GRed and GST activity in the liver.
Exposure to 50 ppm did not affect liver GST, but decreased
liver GRed and GPx.
Adult rats exposed to air or SO2 were born to dams fed diets
with varying casein contents (180 [control], 120, 90 or 60
g/kg) during gestation. In the liver, SO2 exposure elevated
GSH level in the 120 g/kg dietary group but lowered it in the
60 g/kg dietary group. SO2 did not affect liver GST in any
group. SO2 increased GCS levels in the 180 and 90 g/kg
groups, GPx in the 60 g/kg group, and GRed in the 120 and
90 g/kg groups. This study provides information for an
extremely high concentration level but is being
acknowledged here with the unit corrected to verify that a
low-concentration level study was not missed.
May 2008
E-27
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Gunnison et al.
(1987)
Agar et al.
(2000)
KiJciJkatay et al.
(2003)
CONCENTRATION
10 or 30 ppm (26.2 or
78.6 mg/m3); whole
body
10 ppm (26.2 mg/m3);
whole body
10 ppm (26.2 mg/m3);
whole body
DURATION
6 h/day,
~5 days/wk for
21 wks (total of
99 days)
1 h/day, 7
days/wk for 6 wks
1 h/day, 7
days/wk for 6 wks
SPECIES
Sprague-Dawley CD rat,
male, 8 wks old, weight
NR, N: 70/group in
3 groups (inhalation
series)
Swiss Albino rat, male,
3 mos old, weight NR, N:
10/group
Rat, male, 3 mos old,
weight NR, N: 10/group
in 4 groups
EFFECTS
No effects on relative liver weight or histopathology were
found.
Effects were compared in non-diabetic rats, non-diabetic rats
exposed to SO2, alloxan-induced diabetic rats, and diabetic
rats exposed to SO2. SO2 increased blood glucose in all
groups, but did not affect total cholesterol, high density
lipoprotein cholesterol, low density lipoprotein cholesterol,
very low density lipoprotein cholesterol, ortriglyceride levels
in either normal or diabetic rats.
Effects compared in normal rats and rats with alloxan
induced diabetes. Among the significant effects observed,
SO2 exposure enhanced the body weight loss seen in the
diabetic group, but did not affect body weight gain in the
control group. SO2 elevated blood glucose levels in both
controls and diabetics, but lowered triglycerides only in
diabetics. Cholesterol parameters were not affected.
Table E-23.  Renal effects of SO2.
STUDY
Wu and Meng
(2003)
Langley-Evans
etal. (1996)
CONCENTRATION
22, 64, or 148 mg/m3
(8.4, 24.4, or
56.5 ppm)
5, 50, or 100 ppm
(13.1, 131,or262
mg/m3)
DURATION
6 h/day for 7 days
5 h/day for 7-28
days
SPECIES
Kunming-strain mice, male,
ageNR, 18-20g, N:
10/group
Wistar rat, male, 7 wks old,
weight NR, N: 4-57
treatment group, 8 controls
EFFECTS
GST was decreased in the kidney at 64 and 148 mg/m3
and glucose-6-phosphate dehydrogenase activity was
decreased at 148 mg/m3. Kidney GSH levels were
reduced at all exposure levels.
GSH was depleted in the kidney in the 5 and 100 ppm
groups but not in the 50 ppm group. No effects were
observed for other GSH-related enzymes.
Table E-24.  Lymphatic system effects of SO2 and SO2 mixtures.
STUDY
CONCENTRATION
DURATION
SPECIES
EFFECTS
Subchronic/Chronic Exposure
Smith
etal.
(1989)
Aranyi
etal.
(1983)
1 ppm (2.62 mg/m3);
whole body
13.2 mg/m3 (5.0 ppm)
SO2 + 1 .04 mg/m3
ammonium sulfate +
0.2 mg/m3 (0.10 ppm)
ozone; whole body
5 h/day, 5
days/wk for 4
mos.
5 h/day, 5
days/wk for
up to 103
days
Sprague- Dawley
rat, male, young
adult, initial
weight NR, N:
1 2-1 57 data point
CD1 mice,
female, 3-4 wks
old, weight NR,
N: 360/group total
(14-154/group in
each assay)
No significant effects were reported for spleen weight or mitogen-induced
activation of peripheral blood lymphocytes or spleen cells (data not shown by
authors).
Cytostasis of MBL-2 leukemia target cells by peritoneal macrophage was
increased in groups exposed to ozone alone or a mixture of the three
compounds but was significantly higher with the mixture than with ozone alone
at a macrophage:target cell ratio of 10:1 ; no significant effects were observed
with macrophage:target cell ratio of 20:1. Reduction in splenic lymphocyte
blastogenesis in response to phytohemagglutinin and concanavalin A occurred
after exposure to ozone alone, but increased response occurred after exposure
to the mixture; no response to alloantigen occurred after exposure to ozone
alone but increased response occurred after exposure to mixture; there were no
effects on S. typhosa lipopolysaccharide with either exposure scenario.
May 2008
E-28
DRAFT—DO NOT QUOTE OR CITE

-------
                   Annex  F.  Epidemiological  Studies
Table  F-1.    Associations of short-term exposure to SO2 with  respiratory morbidity in field/panel
                studies.
     STUDY
                             METHOD
                                                    POLLUTANTS
                                                                                           FINDINGS
                                                     UNITED STATES
Delfino et al. (2003)

Los Angeles, CA

Nov 1999-Jan 2000
Panel study of 22 Hispanic children
with asthma aged 10 to 16 yrs.
Participants performed twice-daily
PEF measurements and filled out
symptom diaries. Analyses of
symptoms conducted using GEE
with exchangeable correlation.
Linear mixed model used for PEF
analyses. GEE models controlled
for respiratory infections (data
available for 20 subjects) and
temperature.
Mean Levels:1-h max
SO2: 7.0 ppb (SD 4.0)
IQR: 4.0
8-h max SO2:4.6 ppb
(SD 3.0) IQR: 2.5

Co pollutants:
O3(r=-0.19)
NO2 (r = 0.89)
CO (r = 0.69)
PM10 (r = 0.73)
EC (r = 0.87)
OC (r = 0.83)
VOCs
None of the VOCs or gaseous pollutants associated with PEF.
Current-day, but not previous-day, SO2 concentrations
associated with symptom score > 1 and >2.

OR for symptom score > 1 per IQR increase in SO2:
    1-h maxSO2:LagO: 1.31 (1.10, 1.55)
    Lag 1: 1.11  (0.91, 1.36)
8-h max SO2: Lag 0: 1.23 (1.06, 1.41)
    Lag1: 1.11 (0.97, 1.28)
OR for symptom score >2 per IQR increase:
    1-h maxSO2: Lag 0: 1.37(0.87,2.18)
    Lag 1:0.76(0.35, 1.64)
8-h max SO2: Lag 0: 1.36 (1.08, 1.71)
    Lag 1:0.91  (0.51, 1.60)
Mortimer et al.
(2002)

Eight urban areas:
St. Louis, MO;
Chicago, IL; Detroit,
Ml; Cleveland, OH;
Washington, DC;
Baltimore, MD; East
Harlem, NY; Bronx,
NY

Jun-Aug 1993
Panel study of 846 asthmatic
children 4-9 yrs from the National
Cooperative Inner-City Asthma
Study (NCICAS). Study children
either had physician-diagnosed
asthma and symptoms in the past
12 mos or respiratory symptoms
consistent with asthma that lasted
more than 6 wks during the previous
yr. Respiratory symptoms recorded
in daily diary and included cough,
chest tightness, and wheeze. Mixed
effects models and GEE models
used to evaluate the effect of air
pollutants on PEF and respiratory
symptoms. Models adjusted for day
of study,  previous 12-h mean
temperature, urban area, diary
number,  rain in the past 24  h.
Mean Levels:
3-h avg SO2
(8 a.m.-11 a.m.) for all
8 areas (shown in
figure): 22 ppb

Avg intradiary Range:
53 ppb

Co pollutants:
03(r=0.29)
NO2
PM10
None of pollutants associated with evening PEF or evening
symptoms. Using single-pollutant model, SO2 had little effect on
morning PEF (data not shown). Significant associations between
moving avg of 1- to 2-day lag of SO2 and incidence of morning
asthma symptoms.

OR for morning symptoms associated with 20 ppb increase  in
    3-h avg  SO2 concentration (Lag 1-2 day):
8 urban areas: Single-pollutant model: 1.19 (1.06, 1.35)
    SO2 with O3 model: 1.18(1.05, 1.33)
7 urban areas: Single-pollutant model: 1.22 (1.07, 1.40)
    SO2 with O3 and NO2 model:1.19 (1.04, 1.37)
3 urban areas: Single-pollutant model: 1.32 (1.03, 1.70)
SO2with O3, NO2, and PM10 model:1.23 (0.94, 1.62)
Neasetal. (1995)

Uniontown, PA

Summer 1990
Panel study of 83 fourth-fifth
graders in Uniontown,
Pennsylvania. Participants reported
twice-daily PEF and presence of
cold, cough, or wheeze. During
summer of 1990, there were 3,582
child-days. PEF analyzed with
autoregressive linear regression
model that included a separate
intercept for evening measure-
ments, trend, temperature and 12-h
avg air pollutant concentration,
weighted by the number of hours
child spent outdoors during  the
previous 12 h.
12-h avg SO2: 10.2 ppb
Max: 44.9; IQR: 11.1

Daytime 12-h avg SO2
(8 am-8 pm): 14.5 ppb

Overnight 12-h avg
SO2 (8 pm-8 am): 5.9
ppb

Co pollutants:
PM10,PM2.5,O3
total sulfate particles
particle-strong acidity (r
= 0.44)
Incidence of new evening cough episodes significantly
associated with the preceding daytime 12-h avg SO2. Mean
deviation in PEF not associated with SO2.

Effects associated with 10 ppb increase in 12-h avg SO2:
Change in mean deviation in PEF:
-0.63 L/min (-1.33, 0.07)

OR for evening cough: 1.19 (1.00, 1.42)

Concentration weighted by proportion of hours spent outdoors
during prior 12-h: Change in mean deviation in PEF:
-1.25 L/min (-2.75, 0.25)

OR for evening cough: 1.53 (1.07, 2.20)
May 2008
                                    F-1
                               DRAFT—DO NOT QUOTE OR CITE

-------
      STUDY
                               METHOD
                                                        POLLUTANTS
                                                                                                   FINDINGS
Newhouse et al.
(2004)

Tulsa, OK

Sep-Oct 2000
Panel study of 24 patients 9-64 yrs
with physician-diagnosed asthma.
Subjects performed twice-daily PEF
(morning and evening)
measurements, and recorded
medications, symptoms. Simple
linear regression, forward stepwise
multiple regression, correlation
analysis performed. Multiple
regression analyses used to de-
velop predictive models for other
environmental factors. Analyses
produced complex models with diff-
erent  predictor variables for each
symptom.
Mean Levels:
24-h avg SO2: 0.01
ppm
Range: 0.00, 0.02

Co pollutants:
PM2.5
CO
03
pollen
fungal spores
Of the atmospheric pollutants, avg and max O3 were most
significant factors that influenced symptoms. Quantitative results
not provided for SO2.

Avg or max SO2 found to be negative predictors of asthma in
subgroup analyses of women and  nonsmokers and rhinitis in all
patients. Avg SO2 also negative predictor of evening PEF.

Quantitatively useful effect estimates not provided.
Ross et al. (2002)

East Moline, IL

May-Oct1994
Panel study of 59 asthmatics 5-49
yrs. Analysis based on 40 subjects,
due to withdrawal or failure to
provide requested health data.
Study assessed the effect of single
and combined exposures to air
pollutants and airborne allergens on
PEF, symptom scores and
medication use frequency. Multi-
variate linear-regression models
with 1st order autoregression used
for analysis of daily means of mean
-standardized PEF, symptom
scores and asthma medication use;
logistic regression used for di-
chotomized data for symptom score
and medication use, log-linear
models for log-transformed symp-
tom scores and medication use
frequency.
24-h avg SO2:
3.4ppb(SD3.1)

Median: 2.8
IQR: 2.4
Range: 0, 27.3

Co pollutants:
PM10
03
NO2
pollen
fungi
No associations observed with SO2.

No effect estimates provided.
Schildcrout et al.
(2006)

Albuquerque, NM;
Baltimore MD;
Boston MA; Denver,
CO; San Diego, CA;
Seattle, WA; St.
Louis, MO; Toronto,
Ontario, Canada

Nov 1993-Sept 1995
Meta-analysis of 8 panel studies
with 990 children of the Childhood
Asthma Management Program
(CAMP), during the 22-mo preran-
domization phase to investigate
effects of criteria pollutants on
asthma exacerbations (daily symp-
toms and use of rescue inhalers).
Poisson regression and logistic
regression models used in analy-
ses. Within city models controlled
for day of wk, ethnicity, annual
family income, flexible functions of
age and log-transformed sensitivity
to the methacholine challenge using
natural splines with knots fixed at
25th, 50th, and 75th percentiles.
Also controlled for confounding due
to seasonal factors. All city-specific
estimates included in calculations of
study-wide effects except Albuquer-
que where SO2 data were not
collected.
24-h avg SO2: Median
(10th, 25th, 75th, 90th
percentile):

Albuquerque: NA

Baltimore: 6.7 ppb (3.2,
4.7,9.8, 14.2)

Boston: 5.8 ppb (2.7,
3.7,9.1, 14.1)

Denver: 4.4 ppb (1.2,
2.5,6.7,9.5)

San Diego: 2.2 ppb
(1.2, 1.7,3.1,4.4)

Seattle: 6.0 ppb (3.7,
4.7,7.5,9.5)

St. Louis: 7.4 ppb (3.9,
5.3, 10.7, 13.6)

Toronto: 2.5 ppb (0.2,
1.0,4.8,8.8)

Co pollutants:
O3(-0.03< r<0.44)
NO2 (0.23 < r < 0.68)
PM10(0.31 
-------
     STUDY
                               METHOD
                                                       POLLUTANTS
                                                                                                FINDINGS
Schwartz et al.
(1994)

Watertown, MA (Apr-
Aug 1985);

Kingston-Harriman,
TN (Apr-Aug 1986);

St. Louis, MO; (Apr-
Aug 1986);

Steubenville, OH;
(Apr-Aug 1987);

Portage, Wl; (Apr-
Aug 1987);

Topeka, KS (Apr-
Aug 1988)
Longitudinal study of 1,844 children
in grades 2-5 from the Six Cities
Study to examine the effects of PM
and SOX on respiratory health. Daily
diaries completed by parents,
recording symptoms, such as
cough, chest pain, phlegm, wheeze,
sore throat, and fever. Logistic
regression models adjusting for
aurocorrelation were used for the
analysis. To examine possible non-
linearity in the relationship, smooth
functions of the air pollution vari-
ables were fit using GAM and the
significance of the deviation from
linearity was tested.
24-h mean SO2:
Median: 4.1 ppb
IQR: 1.4,8.2
Max: 81.9

Copolllutants:
O3 (r = -0.09)
NO2(r=0.51)
PM10 (r = 0.53)
PM25(r = 0.55)
PM2 5 sulfur (r= 0.50)
H+ (r = 0.23)
SO2 associated with incidence of cough and lower respiratory
symptoms. Local smooth showed increased cough incidence for
only above a 4-day avg of 20 ppb (less than 5% of data). Test for
nonlinearity was significant (p = 0.002). No increase in incidence
of lower respiratory symptoms was seen until 24-h avg SO2
concentrations exceeded 22 ppb.

ORs for cough and lower respiratory symptoms related to were
substantially  reduced after adjustment for PMio, suggesting the
SO2 associations might be confounded by particles.

OR for cough incidence associated with 10 ppb increase in 4-day
avg SO2 concentration:

Single-pollutant model: 1.15 (1.02, 1.31)
SO2with PM10 model: 1.08 (0.93, 1.25)
SO2 with O3 model: 1.15(1.01, 1.31)
SO2 with NO2 model: 1.09 (0.94, 1.30)

OR for lower respiratory symptoms associated  with 10 ppb
increase in 24-h avg SO2 concentration:

Single-pollutant model: 1.28 (1.13, 1.46)
SO2 with PMio model: Not presented. Stated as not statistically
significant.
                                                            EUROPE
Boezenetal. (1998)

Amsterdam and
Meppel, the
Netherlands

winter of 1993-1994
Panel study of 189 adults (48-
73 yrs) w/ and w/out chronic
respiratory symptoms in urban and
rural areas to investigate whether
bronchial hyperresponsiveness and
PEF variability can be used to iden-
tify subjects who are susceptible to
air pollution. Spirometry and metha-
choline challenge were  performed
and subjects with a fall  in FEVi of
20% or greater were considered
BHR. Subjects performed twice-
daily peak flow for 3 mos. A sub-
ject's basal PEF variability
calculated over an 8-day period with
low air pollution. PEF variability
expressed as (highest PEF-lowest
PEF/mean) or amplitude % mean
PEF. After calculation of daily PEF
variability, number of days where
the amplitude % mean was greater
than 5% was determined. This
resulted in 2 groups of subjects;
those with ampli%mean PEF of 5%
or less every day in the 8-day
period, and those with an Amplitude
% mean PEF greater than 5% on at
least 1 day. Effects of air pollutants
on prevalence of symptoms as-
sessed with logistic regression
models that adjusted for autocorre-
lation of the residuals, daily min
temp, time trend and week-
ends/holidays.
24-h avg SO2

Urban
Mean: 11.8 pg/m3
Range: 2.7, 33.5

Rural
Mean: 8.2
Range: 0.8, 41.5

Co pollutants:
PM10
BS
NO2
No association between SO2 and respiratory symptoms in
subjects with no BHR, BHR at < 2.0 mg of methacholine or BHR
at 2 cumulative dose < cum 1.0 mg methacholine. In subjects
with ampli% mean PEF > 5% and those with ampli%mean PEF >
5% for > 33% of days, SO2 was associated with the prevalence
of phlegm.

Odds ratio (per 40 pg/m3 SO2)
Subjects with no BHR
    URS: 0.86 (0.73, 1.03) LRS: 1.15 (0.90, 1.46)
    Cough: 1.01 (0.84, 1.21) Phlegm: 1.01 (0.86, 1.20)
BHR at cum 2.0
Methacholine: URS: 1.11 (0.78, 1.56). LRS: 1.03 (0.72, 1.47)
    Cough: 0.89 (0.66, 1.19). Phlegm: 1.03 (0.78, 1.37)
BHR at 1.0
Methacholine
    URS: 1.02 (0.65, 1.61). LRS: 0.96 (0.63, 1.47)
    Cough: 0.96 (0.64, 1.44). Phlegm: 1.00 (0.68, 1.46)
Ampl%mean PEF 5%
    URS: 0.82 (0.62, 1.08). LRS: 1.38 (0.93, 2.03)
    Cough: 0.72 (0.52, 0.98). Phlegm: 0.79 (0.59, 1.05)
Ampl%mean PEF > 5%
    URS: 1.04(0.88, 1.23). LRS: 1.14(0.96, 1.36)
    Cough: 1.07 (0.90, 1.26). Phlegm: 1.23 (1.05, 1.43)
Ampl%mean PEF > 5%, >33% of days
    URS: 1.10(0.85, 1.41). LRS: 1.14(0.91, 1.42)
    Cough: 1.14(0.89, 1.47). Phlegm: 1.36(1.14, 1.63)
May 2008
                                       F-3
                                 DRAFT—DO NOT QUOTE OR CITE

-------
     STUDY
                               METHOD
                                                       POLLUTANTS
                                                                                                FINDINGS
Boezenetal. (1999)
Bodeg raven,
Meppel, Nuspeet,
Rotterdam,
Amsterdam, The
Netherlands
3 winters of 1992-95
Panel study of 632 children (7 to 11
yrs) living in rural and urban areas
of the Netherlands, to investigate
whether children with bronchial
hyperresponsiveness (BHR) and
relatively high serum concentrations
of total IgE were susceptible to air
pollution. Methacholine challenge
performed to determine bronchial
hyperresponsiveness. Serum total
IgE higher than the median (60kU/L)
were defined as relatively high.
Peak flow was measured twice daily
and lower and upper respiratory
symptoms were recorded daily for
3 mos. Association between
symptoms and air pollutants
assessed using logistic regression
that adjusted for daily min temp,
linear, quadratic and cubic time
trend and weekends and holidays,
and incidence of influenza.
Examined 0,1,2 Lags and 5 day
mean of air pollutants.
1992-9:
Urban areas-
Mean: 22.5 pg/m3,
Range: (1.4,61.3)
Rural areas-
Mean: 9.8
Range: (1.3,34.2)
1993-4:
Urban areas-
Mean: 11.8,
Range: (2.7, 33.5)
Rural areas-
Mean: 8.2,
Range: (0.8,41.5)
1994-5:
Urban areas-
Mean: 8.3,
Range: (0.6, 24.4);
Rural areas-
Mean: 4.3,
Range: (0.5,17.0)
Co pollutants:
PM10
Black smoke
NO,
459 children had complete data. For children with BHR and
relatively high serum total IgE, the prevalence of LRS was
associated with increases in PM10, BS, SO2, and NO2. In the
group with no BHR and relatively low IgE, and the group with
BHR and low IgE, there was no consistent association between
air pollutants with symptoms or decreased PEF. In children with
no BHR but relatively high serum total IgE, there was a 28% to
149% increase in the prevalence of LRS per 40 pg/m3 SO2.

Odds ratio (per 40 pg/m3 SO2)
Children with BHR and relatively high IgE (N: 121)
    LRS LagO: 1.45(1.13, 1.85). Lag  1: 1.41 (1.09, 1.82)
    Lag 2: 1.40(1.10, 1.79). 5-day Mean: 2.25 (1.42, 3.55)
    URS Lag 0: 1.17 (0.99, 1.38). Lag 1: 1.06 (0.90, 1.25)
>10% morning PEF decrease
    Lag 0: 1.09 (0.89, 1.34). Lag 1: 1.00 (0.81, 1.23)
>10% evening PEF  decrease
    Lag 0: 1.06 0.86, 1.30). Lag 1: 0.83 (0.68,  1.02)
NO BHR and low IgE (N: 167)
    LRS Lag 0: 1.12 (0.76, 1.66). Lag  1: 0.61 (0.39, 0.94)
    URS Lag 0: 1.01 (0.89, 1.13). Lag 1: 1.08 (0.96, 1.22)
>10 morning PEF decrease
    LagO: 1.02(0.89, 1.16). Lag 1: 1.00(0.87, 1.15)
>10% evening PEF  decrease
    LagO: 1.10(0.97, 1.25). Lag 1: 1.06(0.93, 1.21)
With BHR and low IgE (N: 67)
    LRS Lag 0: 0.72 (0.41, 1.28). Lag  1: 1.03 (0.56, 1.91)
    URS Lag 0: 0.82 (0.62, 1.09). Lag 1: 0.84 (0.64, 1.12)
>10% morning PEF decrease
    Lag 0: 0.74 (0.51, 1.07). Lag 1: 0.96 (0.67, 1.37)
>10% evening PEF  decrease
    Lag 0: 1.23 (0.88, 1.73). Lag 1: 1.32 (0.96, 1.82)
With BHR and low IgE (N: 67)
    LRS. Lag 0: 0.72 (0.41, 1.28). Lag 1: 1.03  (0.56, 1.91)
    URS Lag 0: 0.82 (0.62, 1.09). Lag 1: 0.84 (0.64, 1.12)
>10% morning PEF decrease
    Lag 0: 0.74 (0.51, 1.07). Lag 1: 0.96 (0.67, 1.37)
>10% evening PEF  decrease
    Lag 0: 1.23 (0.88, 1.73). Lag 1: 1.32 (0.96, 1.82)
No BHR and high IgE (N: 104)
    LRS LagO: 1.44 (1.17,1.77). Lag 1: 1.28(1.00, 1.64)
    Lag 2: (1.38 (1.08, 1.77). 5-day Mean: 2.49 (1.54, 4.04)
URS Lag 0: 0.98 (0.84, 1.14). Lag 1: 1.01 0.87, 1.18)
>10% morning PEF decrease
    Lag 0: 0.92 (0.79, 1.08). Lag 1: 1.03(0.89, 1.21)
>10% evening PEF  decrease
    LagO: 1.00(0.85, 1.17) . Lag 1: 1.05(0.90, 1.23)
May 2008
                                       F-4
                                 DRAFT—DO NOT QUOTE OR CITE

-------
     STUDY
                               METHOD
                                                       POLLUTANTS
                                                                                                 FINDINGS
Boezen et al. (2005)
Meppel, Nunspeet,
Amsterdam, The
Netherlands
two winters 1993-
1995
Panel study of 327 elderly patients
(50 to 70 yrs) to determine suscep-
tibility to air pollution by airway
hyperresponsiveness (AHR), high
total immunoglobulin (IgE), and sex.
Methacholine challenges were
performed and  subjects with greater
than or equal to 20% fall in FE\/i
after inhalation  of up to 2.0 mg
methacholine were considered
AHR+. Subjects with total serum IgE
> 20 kU/L were defined as high total
IgE (lgE+). Twice daily PEF
measurements and daily symptoms
recorded for 3 mos. Data analysis
performed using logistic regression
with modeling of first-order autocor-
relation in the residuals that ad-
justed for daily  min temperature,
time trend, weekend/holidays and
influenza incident for the rural and
urban areas and the two winters
separately. Subjects were classified
as lgE+ AHR+, lgE+ AHR-, IgE-
AHR+ or IgE- AHR+. Examined
effects of pollutants on the same
day, Lag 1, Lag 2 and the 5-day
mean concentration of Lag 0 to Lag
4 preceding that day. Groups that
had effect estimates for PMio, BS,
SO2, and NO2 that were outside the
95% Cl of the effect estimates for
the AHR-/lgE- (control group) were
considered to have increased
susceptibility to air pollution.
24-h mean SO2 (pg/
in winter

Winter 1993/1994

Urban:
Mean: 11.8 pg/m3
MediaN: 10.2
Range: 2.7, 33.5

Rural:
Mean: 8.2
MediaN: 4.4
Range: 0.8, 41.5

Winter 1994/1995

Urban:
Mean: 8.3
MediaN: 7.4
Range: 0.6, 24.4

Rural:
Mean: 4.3
MediaN: .7
Range: 0.5, 17.0

Co pollutants:
PM10
BS
NO,
No consistent associations between the prevalence of LRS or
>10% fall in evening PEF and air pollution in any of the four
groups. In the AHR+/lgE group, the prevalence of URS was
associated with SO2 at 1 day Lag, and the prevalence of >10%
fall in morning PEF with SO2 at  Lag 1, Lag 2 and 5-day mean
(avg of Lag 0 to Lag 4). For females who were AHR+/lgE+, the
prevalence of >10% fall in PEF  was associated with SO2 Lag  1,
Lag 2 and 5-day mean. In subjects with AHR/lgE+ the prevalence
of URS was associated with SO2 the previous day and the mean
of Lag 0 to Lag 4. The effect estimate was outside the 95% Cl of
the estimate for the control group AHR/lgE. No consistent
positive associations found between prevalences of URS, cough
or >10% fall in morning PEF and air pollutants in subjects with
AHR+/lgE or AHR/lgE. Based on results of the study, authors
conclude  that subjects with AHR+/lgE+ were the most responsive
to air pollution.

Odds ratio (per 10 pg/m3 SO2).  AHR/lgE
URS Lag  0: 0.99 (0.93, 1.05). Lag 1: 1.02 (0.97, 1.08)fh

Cough: Lag 0: 1.03 (0.98, 1.08). Lag 1: 0.97 (0.93, 1.02)

>10% fall in morning PEF. Lag 1: 1.00 (0.92, 1.08)

AHR/lgE+
URS Lag  0: 0.98 (0.92, 1.03). Lag 1: 1.07(1.01, 1.12)
5-day mean 1.15(1.02, 1.29), OR outside 95% Cl of control
group

Cough: Lag 0: 1.01 (0.95, 1.07). Lag 1: 1.02 (0.96, 1.08)

>10 % fall in morning PEF. Lag  1:  1.00 (0.92, 1.08)
AHR+/lgE
LagO: 1.05(0.94, 1.17). Lag 1:  1.07(0.96, 1.19)

Cough: Lag 0: 1.03(0.95, 1.12). Lag 1: 1.01  (0.93, 1.09)

>10%fallin morning PEF. Lag  1:0.99(0.87, 1.12)
5-day Mean: 0.78 (0.61, 0.98),
OR  outside 95% Cl of control group
AHR+/lgE+
LagO: 1.06(0.97, 1.15). Lag 1:  1.13(1.05, 1.23)

Cough: Lag 0: 1.02 (0.94, 1.11). Lag 1: 1.02 (0.94, 1.10)
>10%fallin morning PEF. Lag  1:0.99(0.87, 1.12)
AHR+/lgE+

URS. LagO: 1.06(0.97, 1.15). Lag 1: 1.13(1.05, 1.23),
OR  outside 95% Cl of control group

Cough: Lag 0: 1.02 (0.94, 1.11). Lag 1: 1.02 (0.94, 1.10)

>10%fallin morning PEF. Lag  1:  1.15(1.04, 1.27),
OR  outside 95% Cl of control group
Lag 2 : 1.18(1.07, 1.30),
OR  outside 95% Cl of control group
5-day mean : 1.26(1.07, 1.49),
OR  outside 95% Cl of control group
Cuijpersetal. (1994)

Maastricht, the
Netherlands

Nov-Dec 1990
(baseline)
Aug8-16
(smog episode)
The effects of exposure to summer
smog on respiratory health were
studied in 535 children (age un-
specified). During a smog episode,
212 children were randomly chosen
to be reexamined for lung function
and symptomsOnly 112 of the had
adequately completed summer
questionnaires and were used for
the symptom analysis. Lung func-
tion measurements made with
forced oscillation technique were
available for 212 children  and valid
spirometry was available for 208
children. Corrected baseline lung
function compared using paired t
test and difference in the prevalence
in symptoms during baseline and
episode compared.
24-h avg SO2

Baseline 55 pg/m3
Summer episode
23 pg/m3

NO2
BS
03
PM10
Acid aerosol
H+
Small decrements in FEVi and FEF25.75 found in the 212 children
during the episode compared to baseline. However, there was
also a significant decrease in resistance parameters. No
increases observed in the prevalence of acute respiratory
symptoms.

Change in lung function and impedance between baseline and
smog episode:
FEV,: -0.032 L (SD 0.226), p < = 0.05
FEF2^75: -0.086 L/s (SD 0.415), p < = 0.01
Resistance at 8 Hz: -0.47 cm H2O (L/s)
(SD1.17), p< = 0.05
May 2008
                                       F-5
                                 DRAFT—DO NOT QUOTE OR CITE

-------
     STUDY
                               METHOD
                                                       POLLUTANTS
                                                                                                 FINDINGS
Desqueyroux et al.
(2002)

Paris, France

Nov 1995-Nov1996
Panel study of 60 patients with
moderate to severe physician-
diagnosed asthma (mean age 55
yrs). Asthma were noted by
physician at each consultation
(regular or emergency). Attacks
defined as need to increase twofold
the dose of beta2 agonist.
24-h avg SO2

Summer 7 (5) pg/m3
Range: 2, 27
Winter 19(12) pg/m3
Range: 3, 81

PM10, NO2, O3
         No association between asthma attacks and SO2 for any Lag or
         season. Mean 24-h SO2 (per 10 pg/m3)
         OR on incident of asthma attacks
         Lag 1: day 0.98 (0.76, 1.27); Lag 2: day 0.92 (0.72, 1.19)
         Lag 3: day 1.01 (0.82, 1.23); Lag 4: day 1.01 (0.86, 1.19)
         Lag 5: day 1.05 (0.85, 1.29)Cumulative exposure mean (-1 to -5
         days) 0.99 (0.76, 1.30)
Forsberg et al.
(1993)
Pitea, Northern
Sweden
Mar to Apr
Panel study of 31 asthmatic patients
(9 to 71 yrs) to assess relationship
between daily occurrence of asthma
symptoms and fluctuations in air
pollution and meteorological condi-
tions. Subjects recorded symptoms
(shortness of breath, wheezing,
cough, phlegm) for 14 consecutive
days.
24-h avg SO,
Mean: 5.7
Range: 1.3, 12.9
Correlations:
NO2 (r = 0.24)
BS  (r = 0.70)
(pg/m3)
No significant association observed with SO2. Positive asso-
ciation between severe shortness of breath and black smoke.
Regression coefficient and 90% Cl
Subjects with shortness of breath (N: 28):
0.0345 (-0.49, 0.118)
Subjects with 5 or more incident episodes of severe shortness of
breath (N:  10): -0.0266 (-0.140, 0.087)
Higginsetal. (1995)
United Kingdom
Panel study of 75 patients with
physician diagnosed asthma or
chronic bronchitis (mean age
50, range 18 to 82 yrs) to determine
if air pollution affects respiratory
function and symptoms. Subjects
asked to keep symptom records and
perform PEF for 28 days. PEF
values recorded every 2  h beginning
at 02.00 h each day. Methacholine
challenge performed on each
subjects. Those with PM20 FEVi of <
12.25 pmol were considered as
methacholine reactors. PEF
variability was calculated as the
amplitude % Mean: (highest-lowest
PEF value/mean) ><100. 75 patients
had PEF records, 65 completed
symptom  questionnaires.
Max 24-h SO2
                                                    Co pollutants:
                                                    03
                                                    NO2
         The amplitude % mean was significantly associated with
         increasing levels of SO2, on the same day for all subjects and
         among reactors. Mean daily PEF and min PEF associated with
         SO2 among reactors only. Significant associations also observed
         with wheeze and SO2 on the same day, at 24-h Lag, and 48-h
         Lag for all subjects and meta-choline reactors; and with
         bronchodilator use for all subjects at 24-h Lag.

         Regression coefficient per 10 pg/m3 SO2

         All subjects

            Mean PEF (L/min). Same day 0.021 (0.031)
               24-h Lag 0.003 (0.033). 8-h Lag 0.021 (0.032)
            Minimum PEF(L/min). Same day 0.062 (0.039)
               24-h Lag 0.048 (0.041). 48-h Lag 0.001 (0.040)
            Amplitude (% mean). Same day: 0.167 (0.072)
               24-h Lag 0.191 (0.76). 48-h Lag 0.022 (0.075)
            Wheeze. Same day: 1.14(1.03, 1.26)
               24-h Lag 1.22(1.09,  1.37). 48-h Lag 1.14(1.02, 1.27)
            Dyspnoea. Same day: 1.03 (0.94, 1.14)
               24-h Lag 1.07 (0.96,  1.18). 48-h Lag 0.94 (0.85, 1.05)
            Cough. Same day: 1.03 (0.95, 1.12)
               24-h Lag 1.04 (0.95,  1.13). 48-h Lag 1.02 (0.94, 1.12)
            Throat symptoms. Same day: 1.01 (0.92, 1.11)
               24-h Lag 1.00 (0.91,  1.10). 48-h Lag 0.96 (0.87, 1.06)
            Eye symptoms. Same day: 1.08 (0.97, 1.20)
               24-h Lag 1.11 (0.99,  1.24). 48-h Lag 1.10(0.99, 1.21)
            Bronchodilator use. Same day 1.11 (0.97, 1.26)
               24-h Lag 1.16(1.01,  1.34). 48-h Lag 1.12 (0.98, 1.27)

         Reactors

            Mean PEF (l/min). Same day 0.087 (0.054)
               24-h Lag 0.44 (0.058). 48-h Lag 0.012 (0.057)
            Minimum PEF(L/min). Same day 0.168 (0.071)
               24-h Lag 0.078 (0.076). 48-h Lag 0.026 (0.075)
            Amplitude (% mean). Same day: 0.157 (0.120)
               24-h Lag 0.083 (0.127). 48-h Lag 0.005 (0.126)
            Wheeze. Same day: 1.26 (1.08, 1.47)
               24-h Lag 1.57 (1.30,  1.89). 48-h Lag 1.24 (1.06, 1.45)
            Dyspnoea. Same day:1.04 (0.90, 1.20)
               24-h Lag 1.17 (1.00,  1.37). 48-h Lag 1.03 (0.89, 1.20)
            Cough. Same day:1.09 (0.96, 1.24)
               24-h Lag 1.05 (0.91,  1.20). 48-h Lag 1.00 (0.87, 1.15)
            Throat symptoms. Same day: 1.06 (0.92, 1.21)
               24-h Lag 1.06 (0.91,  1.23). 48-h Lag 1.01 (0.87, 1.17)
            Eye symptoms. Same day: 1.19 (1.01, 1.40)
               24-h Lag 1.21 (1.01,  1.45). 48-h Lag 1.08 (0.91, 1.28)
            Bronchodilator use. Same day 1.18 (0.99, 1.42)
               24-h Lag 1.23 (1.02,  1.50). 48-h Lag 1.31 (1.09, 1.58)
May 2008
                                       F-6
                                 DRAFT—DO NOT QUOTE OR CITE

-------
     STUDY
                               METHOD
                                                       POLLUTANTS
                                                                                                 FINDINGS
Hiltermann et al.
(1998)
Bilthoven, The
Netherlands
Jul-Oct 1995
Panel study of 60 adult (18 to
55 yrs) nonsmoking patients with
intermittent to severe persistent
asthma to examine the association
of summertime air pollution (ozone
and PMio) with  respiratory symp-
toms, medication use and PEF.
Subjects were followed over 96
days. Twice daily PEF, respiratory
symptoms, and medication use and
whether they were exposed to envi-
ronmental tobacco smoke were re-
corded daily. Analysis controlled for
time trends, aeroallergens, environ-
mental tobacco smoke exposures,
day of wk and temperature.
Examined Lag effects of 0 to 2
days.
24-h avg SO2 (pg/m3
Mean: 6.2
Range: 0.1, 16.2
Correlation with BS
r = 0.53
O3, PM10, NO2, BS
Correlation with
copollutants:
O3(r=0.30
)PM10(r=0.37)
NO2 (r = 0.49)
BS (r = 0.53)
SO2 not included in the analysis since levels were negligible
during the study period (< 17 pg/m3)

Effect estimates not provided.
Hoek and Brunekreff
(1993)
Wageningen,

The Netherlands
Panel study of 112 children (7 to
12 yrs, non-urban) to assess effects
of winter air pollution pulmonary
function and respiratory symptoms.
Parents filled out symptom diary
that was turned in every 2 wks.
Pulmonary function test performed
by technician every 3 wks. Addi-
tional pulmonary function tests
performed when SO2 was predicted
to be higher than 125 pg/m3 or NO2
> 90 pg/m3.
Daily concentrations
presented in graph;

Highest 24-h avg cone
SO2: 105 pg/m3(air
pollution episode)

PM10, BS, NO2
During the winter episode, pulmonary function of schoolchildren
was significantly lower than baseline. Significant negative
associations between SO2 and FVC, FEV, and MMEF. No
significant associations found with prevalence of respiratory
symptoms. Authors noted that it is not clear which components of
episode mix responsible for association and that the
concentrations of acid aerosol and SO2 were too low for direct
effects to be likely. SO2 moderately correlated with PMio (r =
0.69) and black smoke (r = 0.63) but not NO2 (r = 0.28).

Mean of individual  regression slopes and SE

FVC Same day -0.55 (0.10), p < 0.05
    Lag 1: -0.74 (0.15) p < 0.05. 1 wk -0.94 (0.20) p < 0.05
FEV, Same day -0.51 (0.09) p < 0.05
    Lag 1: -0.21 (-0.63) p < 0.05.  1 wk-0.78 (0.18) p < 0.05
PEF Same day -0.64 (-0.44)
    Lag 1: -0.21 (0.63). 1 wk-0.34 (0.81) p < 0.05
MMEF. Same day  -0.54 (0.20)
    Lag 1: -0.40 (0.29). 1 wk-0.61 (0.37)
Prevalence of acute respiratory symptoms regression coefficient
from time-series model and SE

Cough. Same day  0.02 (0.18)
    Lag 1:-0.14 (0.19). 1 wk 0.13 (0.76)
Upper respiratory symptoms. Same day 0.12  (0.16)
    Lag 1: -0.02 (0.17). 1 wk-0.24 (0.76)
Lower respiratory symptoms. Same day 0.06  (0.26)
    Lag 1: -0.11 (0.29). 1 wk-0.54 (0.92)
Any respiratory symptoms. Same day 0.01 (0.13)
    Lag 1: -0.03 (0.13). 1 wk-0.11 (0.60)
Hoek and Brunekreff
(1995)

Deurne and
Enkhuizen, The
Netherlands

Mar-JuM989
Panel study of 300 children
(7-11 yrs) to examine the effects of
photochemical air pollution on acute
respiratory symptoms. Occurrence
of respiratory symptoms recorded
by parents in daily diary. Symptoms
included cough, shortness of breath,
upper and lower respiratory
symptoms, throat and eye irritation,
headache and nausea. Association
of symptom prevalence and inci-
dence assessed using first order
autoregressive, logistic regression
model.
Daily concentration of
SO2 < 43 pg/m3

03
PM10
SO4
N03"
Same day concentrations of SO2 and NO2 not associated with
symptom prevalence.

No effect estimates for SO2 provided
Just et al. (Just et
al., 2002)

Paris, France

1996
Panel study consisting of 82 medi-
cally diagnosed asthmatic children,
7-15 yrs old, followed for 3 mos
(Jan-Mar). Examined the associa-
tion between air pollution and
asthma symptoms using regression
analyses based on generalized
estimating equations (GEE).
24-h avg (Mg/m3): 11.6
(5.7)

PM
BS
NO2
03
SO2 was not analyzed because it was only present at low
concentrations.
May 2008
                                       F-7
                                 DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Koppetal. (1999)
Two towns in Black
Forest, Germsany
Villingen and
Freudenstadt
Mar-Oct 1994








Lagorio et al. (2006)
May 24 to June 24,
1999 and Nov 18 to
Dec 22, 1999
Rome, Italy













METHOD
Panel study of 170 children (median
age 9.1 yrs) to investigate nasal
inflammation and subsequent
adaptation after ambient ozone
exposures. Nasal lavage was
sampled over 11 time points, and
skin prick tests performed. Nasal
lavage samples were analyzed for
eosinophil cationic protein, albu-
men, and leukocytes as markers of
nasal inflammation. To avoid
confounding with allergens, the
study population was restricted to
only children with no positive
reaction to any of the tested inhalant
allergens. GEE used in analysis.
Panel study of 29 patients with
either COPD (N: 11, mean age 67
yrs), asthma (N: 11, mean age 33
yrs) or ischemic heart disease (N: 7,
mean age 63 yrs) to evaluate
whether daily levels of air pollutants
have a measurable impact on lung
function in adults with preexisting
lung or heart disease.











POLLUTANTS
Mean SO2 (mg/m3)
Villingen
Mean: 3
5%: 0
95%: 9
Freudenstadt
Mean: 3
5%: 0
95%: 9
Copollutants: O3, NO2,
TSP, PM10




24-h mean SO2 (Mg/m3)
Spring mean 4.7
SD1.8
Winter mean 7.9
SD 2 2
Overall mean 6.4
SD: 2.6

Copollutants: PM25,
PM10.25, PMio, CD, Cr,
FE.NI.PB, PT, V, Zn,
NO2, CO, O3
Correlation with
Copollutants:
PM2.5 (r = 0.34)
PM10-2.5(r = -0.16)
PM10(r = 0.21)
NO2(r=0.01)
O3(r= -0.61)
CO (r = 0.65)
FINDINGS
Results for only O3. Authors noted that since there were very low
concentrations of NOX and SO2, the confounding effects of these
components in ambient air were negligible.
Eosinophil cationic protein and leukocyte levels peaked after the
first increase in ambient ozone levels.









Because avg 24-h concentrations of SO2 were low and showed
little variability, SO2 was not considered in the analysis














May 2008
F-8
DRAFT—DO NOT QUOTE OR CITE

-------
     STUDY
                               METHOD
                                                        POLLUTANTS
                                                                                                  FINDINGS
Neukirch et al.
(1998)
Paris, France
Nov 15, 1992to
May 9, 1993
Panel study of 40 nonsmoking, mild
to moderate asthmatics (16 to
70 yrs, mean 46) to examine the
short-term effects of winter air
pollution in asthma symptoms and
three daily peak flow measure-
ments. Patients were followed for
23 wks. Used GEE models that
controlled for autocorrelation of
responses, weather, and time
trends. Analysis conducted on entire
study  population and for subgroup
of subjects who took inhaled B2
agonists as needed. Assessed air
pollution effect on both incident and
prevalence of symptoms, Z-trans-
formed morning PEF and daily PEF
variability.
24-h avg SO2

Mean: 21.7
(13.5) pg/m3
Range: 4.4, 83.8

Co pollutants
NO2, PM13, Black
smoke

Correlation with
copollutants:

NO2 (r = 0.54)
PM13 (r = 0.83)
BS  (r = 0.89)
Significant effects on incidence and prevalence of symptoms.
Effects at Lag days 3-6 and weekly avg exposures. Based on
group avg PEF of 407 l/min, a 50 pg/m increase SO2 caused a
maximum decrease in morning PEF of 5.5%. 24-h avg SO2 (per
50 Mg/m3).

Odds ratio: all subjects - Incident episodes:
   Wheeze: Lag 5: 1.66 (1.01, 2.70)
   Nocturnal cough: Lag 3: 1.60 (0.98, 2.62)
      Lag 4: 1.71 (0.86, 3.40). Lag 6: 1.72(1.16,2.55)
   Respiratory infections: Lag 3: 3.14 (1.30, 7.59)
      Lag 4: 2.70 (1.36, 5.37). Lag 5: 2.79 (0.95, 8.21)
      Wk: 8.52 (1.20, 60.5)

Odds ratio: all subjects - Prevalent episodes:
   Wheeze: Lag 5: 1.35 (1.01, 1.81)
      Lag 6: 1.39 (1.04, 1.87). Wk: 1.64 (0.91, 2.94)
   Nocturnal cough: Lag 6: 1.34 (1.00, 1.79)
   Shortness of breath: Wk: 1.56 (1.06, 2.32)
   Respiratory infections: Lag 4: 2.40 (1.33, 4.33)
      Lag 5: 2.72 (1.67, 4.44). Lag 6: 2.94 (1.80, 4.79)
      Wk: 6.30 (1.31, 30.2)

Odds ratio: Subjects taking B2 agonists - Incident episodes:
   Asthma attacks: Lag 6: 2.19 (0.91, 5.29)
   Wheeze: Lag 5: 1.84 (1.13, 3.00)
   Nocturnal cough: Lag 3: 2.41 (1.47, 3.93
      Lag 4: 2.35 (0.88, 6.26). Lag 6: 1.86 (1.14, 3.04)

Odds ratio: Subjects taking B2 agonists - Prevalent episodes:
   Asthma attacks: Lag 5: 1.88 (0.95, 3.73)
      Lag 6: 2.82 (1.57, 5.07)
   Wheeze: Lag 5: 1.51 (1.02, 2.23)
      Lag 6: 1.57(1.06,2.32)
   Nocturnal cough: Lag 3: 1.73 (1.06, 2.82)
      Lag 4: 2.28 (1.27, 4.11). Lag 5: 1.91 (1.17,3.12)
      Lag 6: 1.91 (1.17,3.12)
   Shortness of breath: Lag 4: 1.81 (1.22,  2.67)
      Lag 5: 1.65 (1.11, 2.44). Lag 6: 1.61 (1.20,2.16)
      Wk: 3.03 (1.26, 7.33)

Regression coefficients of the effects and SE (per 1 pg/m3)
Z-transformed morning PEF
Lag 5: 0.450 (0.138) p = 0.001. Lag 6: 0.337 (0.164) p = 0.03
PEF daily variability. Lag 2: 0.025 (0.013) p = 0.05
Peacock et al.
(2003)

Southern England

Nov 1, 1996toFeb
14, 1997
Panel study of 177 children (mean
age 10.7 yrs, range 7 to 13) from
three schools (two urban and 1 rural
location) to investigate effects of
winter air pollution on respiratory
function. Children were followed for
13 wks. Used two sources of air
pollution in the rural area, one that
was "locally validated" and the other
"nationally validated."
24-h avg SO2 (ppb)

Rural

(nationally validated)
Mean: 5.1  (4.7)
Range: 0.0, 35.6

Rural

(locally validated)
Mean: 5.4 (5.1)
Range: 0.0, 39.1

Urban 1
Mean: 6.0 (6.0)
Range: 0.5, 32.5

03
NO2
PM10
SO4
No statistically significant association between winter SO2 and
PEFR, 0.70% decline in PEFR for a 10 ppb increase in the five-
day mean concentration of SO2 (community monitor)

24-h avg So2 - Change in PEF per 1 ppb SO2 - community
monitor - Regression Coefficient (95% Cl)
Lag 0: 0.05 (-0.05, 0.16). Lag 1: -0.04 (-0.13, 0.06)
Lag 2: -0.08 (-0.19, 0.04). Mean (0-4) -0.23 (-0.65, 0.18)
Change in PEF per 1 ppb SO2- local regression coefficient (95%
Cl)
                                                                           Lag 0:-0.01 (-0.10,0.07).
                                                                           Lag 2:-0.09 (-0.18, 0.01).
                                                                           Odds of 20% decrement in
                                                                           Lag 0 0.987 (0.958, 1.017),
                                                                           Lag 2 0.992 (0.963, 1.023)
                                                                           Odds of 20% decrement in
                                                                           children
                                                                           Lag 00.981 (0.925, 1.041),
                                                                           Lag 2 0.995 (0.939 1.054).
                                                                               Lag 1:0.02 (-0.05, 0.10)
                                                                               Mean (0-4) -0.09 (-0.25, 0.07)
                                                                               PEF below the median-all children
                                                                                Lag 1 1.007 (0.986, 1.030)
                                                                                Mean (0-4) 0.972 (0.887,  1.066)
                                                                               PEF below the median-wheezy

                                                                                Lag 1 0.999 (0.957, 1.042)
                                                                               Mean (0-4) 1.019 (0.890 to 1.167)
May 2008
                                       F-9
                                 DRAFT—DO  NOT QUOTE OR CITE

-------
     STUDY
                               METHOD
                                                       POLLUTANTS
                                                                                                FINDINGS
Peters etal. (1996)

Erfurt and Weimar,
former German
Democratic
Republic;
Sokolov, Czech
Republic

Sep 1990 to June
1992
Panel study of 102 adult (32 to
80 yrs) and 155 children (7 to
15 yrs) with asthma from the former
German Democratic Republic and
Czech Republic to investigate the
acute effects of winter type air
pollution on symptoms,  medication
intake and PEF. Used regression
analyses and distributed Lag
models.
Winter 1990/1991
Erfurt:
   Mean: 125 pg/m3
   Max: 564 pg/m3
   IQR: 113pg/m3
Weimar
   Mean: 236 pg/m3
   Max: 1018 pg/m3
   IQR: 207 pg/m3
Sokolov
   Mean: 90 pg/m3
   Max: 492 pg/m3
   IQR: 94 pg/m3

Winter 1991/1992
Erfurt
   Mean: 96 pg/m3
   Max  : 462 m/m3,
   IQR: 80 pg/m3
Weimar
   Mean: 153 pg/m3
   Max: 794 pg/m3
   IQR: 130pg/m3
Sokolov
   Mean: 71 pg/m3
   Max: 383 pg/m3
   IQR: 66 pg/m3
Copollutants: TSP,
PM10, SO4, PSA
(particle strong acidity)
5-day mean concentration of SO2 associated with PEF and
symptoms in children (combined analysis from former German
Democratic Republic and Czech Republic).

Correlation coefficient between SO2 and TSP in  Erfurt was r =
0.8, 0.9 during both winters and in Weimar during the first winter.

Correlation with TSP in Sokolov and in Weimar during the
second winter was r = 0.4, 0.5.

Combined analysis for children

% change in PEF
Concurrent day 0.18 (-0.44, 0.09) per 133 pg/m3
5-day mean -0.90 (-1.35, -0.46) per 128 pg/m3

% change in symptom score
Concurrent day -0.1 (-5.9, 5.7) per 133 pg/m3
5-day mean 14.7 (0.8, 28.6) per 128 pg/m3

Combined analysis for adults

% change in PEF
Concurrent day -0.20 (-0.53, 0.12) per 133 pg/m3
5-day mean -0.28 (-0.72, 0.16) per 128 pg/m
Pikhart et al. (2001)

Czech

1993-1994
SAVIAH study of 3045 children by
questionnaire to determine associa-
tion of SO2 to wheezing. Used
ecological and multilevel analysis
MediaN: 73.9 pg/nf
25th percentile: 63.5
75th percentile: 95.5

Copollutant: NO2
Positive association of SO2 with wheezing
Odds Ratio (95% Cl)
Logistic RegressioN: Individual outcome and area exposure: 1.08
(0.98, 1.20). Individual outcome and individual exposure: 1.08
(0.98, 1.19). Ecological analysis: 1.05(0.96, 1.16)
PonkaA. (1990)

Helsinki, Finland
1991
Survey study to compare weekly
changes in ambient SO2, NO2, and
temperature and the incidence of
respiratory diseases, and absen-
teeism for children in day-care
centers and schools and for adults
in the work place during a 1-yr
period (1987).
Mean weekly
concentration of SO2
(pg/m3)
Mean: 21.1
SD: 11.7
MediaN: 17.0
Range: 9, 61.5

Mean of daily max
Mean: 53
SD:20.8
MediaN: 48
Range: 25.9, 130.3

Copollutant: NO2
Mean SO2 concentration correlated with the incidences of URI
and tonsillitis reported from health centers. SO2 also correlated
with absenteeism due to febrile illness among children in day
care centers and adults. When comparing incidences during the
low and high levels of SO2, the number of cases of URI and
tonsillitis reported from health centers increased as well as
absenteeism. After standardization for temperature, the only
difference that was statistically significant was the occurrence of
URI diagnosed at health centers. Frequency of URI  was 15%
higher during high levels of SO2 compared to low.

Statistical significance of product moment correlation coefficients
(correlation  coefficient) between SO2 and respiratory disease and
absenteeism

Respiratory  tract infections diagnosed at health centers: URI SO2
arithmetic mean p < 0.001 (0.553)
SO2 mean of daily maximums: p = 0.0012 (0.437)

Tonsillitis: Arithmetic Mean: 0.0098 (0.355)
Mean of daily maximums: NS

Absenteeism due to febrile illness: Day care centers
SO2 arithmetic Mean: p = 0.012 (0.404)
Mean of daily maximums: p = 0.048 (0.323)

School children SO2 arithmetic Mean: NS
Mean of daily maximums: NS

Adults: SO2 arithmetic Mean: p < 0.0001 (0.644)
Mean of daily maximums: p < 0.0001  (0.604)
No significant correlation between SO2 and  URI, tonsillitis, otitis,
or LRI in day care center children

Statistical significance of weekly frequency of respiratory tract
disease and absenteeism during low and high levels of SO2:
Respiratory  infections diagnosed at health centers: URI
SO2 arithmetic Mean: p < 0.001. Mean of daily max: p = 0.0005

Tonsillitis
SO2 arithmetic Mean: 0.0351. SO mean of daily max: NS

Absenteeism due to febrile illness
Day care center childreN: p = 0.0256
School childreN: p = 0.0014. Adults: p = 0.0005
May 2008
                                      F-10
                                 DRAFT—DO NOT QUOTE OR CITE

-------
      STUDY
                               METHOD
                                                        POLLUTANTS
                                                                                                  FINDINGS
Pinter etal. (1996)

Tata Area, Hungary

winter mos between
Dec 1993-Mar 1994
Longitudinal (children < 14 yrs) and
cross-sectional study (9 to 11 yrs) to
examine air pollution and respiratory
morbidity in children. In the
longitudinal prospective study,
respiratory morbidity was evaluated
daily and on a weekly basis. In
cross-sectional study, anthropomet-
ric parameters, physical status,
pulse and blood pressure, lung
function parameters, eosinophils in
the nasal smear, hematological
characteristics and urinary excretion
of some metabolites were examine
and measured. Anova and linear
regression used in analysis.
Mean SO2 exceeded
the limit of yearly avg
150 pg/m3

Daily peaks reached as
high as 450 pg/m3

No specific values
given

Copollutant: NO2
Significant correlation between SO2 levels and acute daily
respiratory morbidity, but no correlation with weekly incidence.
Authors stated that in the cross-sectional study, almost all health
parameters were impaired  but no results were shown.

Results only provided in graph. No Cl provided
Roemeretal. (1993)

Wageningen and
Bennekom,
Netherlands
Panel of 73 children (mean age 9.3
yrs, range 6 to 12 yrs) with chronic
respiratory symptoms to investigate
effects of winter air pollution on lung
function, symptoms and medication
use. Subjects performed twice-daily
PEF measurements,  largest of three
PEF readings used in regression
analysis.  Both incidence and
prevalence  of symptoms analyzed,
using  logistic regression.
Daily concentrations of
SO2 shown in graph

Highest 24-h avg
concentration SO2: 105
pg/m3

Co pollutants;
NO2
PM10
BS

Correlation with
copollutants:
NO2 (r = 0.26)
PM10 (r = 0.65)
BS  (r = 0.63)
Positive association between incidence of phlegm and runny
nose with SO2 on the same day. Significant association also
found between evening PEF and SO2 on, the same day, previous
day and 1 wk (avg of same day and 6 days before). The use of
bronchodilators also associated with SO2.

Mean of individual regression coefficient
Morning PEF. Same day: -0.021 (0.024)
  Lag 1: -0.024 (0.031). Wk: -0.50 (0.069)
Evening PEF. Same day: -0.048 (0.018) p < 0.05
  Lag 1:-0.039 (0.021) p< 0.10.
  Wk:-0.110 (0.055) p< 0.05
Prevalence of symptoms (per 50 pg/m3 SO2)
Asthma attack. Same day: 0.008 (0.012)
  Lag 1: 0.016 (0.011).  1 wk: 0.058 (0.027) p < 0.05
Wheeze. Same day: 0.033 (0.17) p < 0.10
  Lag 1: 0.042 (0.016) p < 0.05. Wk: 0.069 (0.032) p < 0.05
Waken with symptoms. Same: day 0.033 (0.019) p < 0.10
  Lag 1: 0.032 (0.018) p < 0.10. Wk: 0.058 (0.045)
Shortness of breath. Same: day 0.029 (0.016) p < 0.10
  Lag 1: 0.016 (0.015). Wk: 0.044 (0.035)
Cough. Same day 0.018 (0.025)
  Lag 1: 0.012 (0.023). Wk 0.072 (0.066)
Runny nose.  Same day 0.070 (0.026) p < 0.05
  Lag 1 :-0.11 (0.025). Wk 0.153 (0.074) p< 0.05
Phlegm. Same day 0.011 (0.022)
  Lag 1: 0.014 (0.020). Wk -0.005 (0.056)
Roemeretal. (1998)

14 European
Centers: Umea,
Sweden; Malmo,
Sweden; Kuopi,
Finland; Oslo,
Norway; Amsterdam,
The Netherlands;
Berlin, Germany;
Katowice, Poland;
Cracow, Poland;
Teplice, Czech
Republic; Prague,
Czech Republic;
Budapest, Hungary;
Pisa, Italy; Athens,
Greece

Winter 1993-1994
Multicenter panel study of the acute
effects of air pollution on respiratory
health of 2010 children (aged 6 to
12 yrs) with chronic respiratory
symptoms. Results from individual
centers were reported by Kotesovec
et al. (1998), Kalandidi et al. (1998),
Haluszka et al. (1998), Forsberg
et al. (1998), Clench-Aas et al.
(1998), and Beyer et al. (1998).
Calculated effect estimates of air
pollution on PEF or the daily preva-
lence of respiratory symptoms and
bronchodilator use from the panel-
specific effect estimates
Range: -2.7 pg/m
(Umea, urban),
113.9pg/m3
(Prague, urban)

Copollutants:
PM10,
BS
NO2
No clear associations between PMio, BS, SO2 or NO2 and
morning PEF, evening PEF, prevalence of respiratory symptoms,
or bronchodilator use could be detected. Previous day PM10 was
negatively associated with evening PEF, but only in locations
where BS was high compared to PMio concentrations.

No consistent differences in effect estimates between subgroups
based on urban versus suburban, geographical location or mean
levels of PMio, BS, SO2, and NO2. Combined effect estimates
with 95% Cl of air pollution on PEF
Morning
Lag 0: 0.2 (-0.2, 0.6); Lag 1: 0.2 (-0.2, 0.6)
Lag 2: 0.6 (0.2, 1.0);  7-day mean 0.6 (-1.3, 2.5)
Afternoon
Lag 0: 0.1 (-0.3, 0.5); Lag 1: 0.0 (-0.4, 0.4)
Lag 2: 0.1 (-0.4, 0.6); 7-day mean 0.2 (-0.5, 0.9)
May 2008
                                      F-11
                                 DRAFT—DO NOT QUOTE OR CITE

-------
     STUDY
                               METHOD
                                                        POLLUTANTS
                                                                                                  FINDINGS
Segalaetal. (1998)

Paris, France

Nov 15, 1992to
May 9, 1993
Panel study of 84 children (7 to
15 yrs) with physician diagnosed
asthma to examine the effects of
winter air pollution on childhood
asthma. For 25 wks,  parents re-
corded the presence or absence of
asthma attacks, upper or lower
respiratory infections with fever, the
use of supplementary inhaled B2
agonist, the severity of symptoms
(wheeze, nocturnal cough and
shortness of breath). Children also
recorded PEF three times a day.
GEE models adjusted for age, sex,
weather and time trend.  Investi-
gated effects of SO2 at 0 to 6 day
Lags.
SO2 mean (SD):

21.7(13.5)pg/m3

Range:
(4.4, 83.8) pg/m3

Co pollutants:
NO2
PM13
BS

Correlation with
copollutants:
NO2 (r = 0.54)
PM13 (r = 0.43)
BS (r = 0.89)
SO2 associated with both incident and prevalent episodes of
asthma, use of supplementary beta 2 agonist, incident episodes
of nocturnal cough, prevalent episodes of shortness of breath
and respiratory infection.

OR per 50 pg/m3 SO2 (Only effects at 0 and 1-days lag shown
below unless statistically significant)
Incident episodes:  Mild asthmatics (N: 43)
Asthma: Lag 0: OR 2.86 (1.31, 6.27); Lag 1: 2.45 (1.01, 5.92)
Wheeze: Lag 0:  1.47 (0.90, 2.41); Lag1: 1.27 (0.48, 3.38)
Nocturnal cough:
Lag 3: 1.93 (1.18, 3.15) ; Lag 4: 2.12 (1.43, 3.13)
Respiratory infections: Lag 1: 1.52 (0.38, 5.98)
Prevalent episodes: Mild asthmatics (N: 43)
Asthma: Lag 0: 1.71 (1.15,2.53) ; Lag 1: 1.55(0.86,2.78)
Wheeze: Lag 4:  1.48 (0.90, 2.41)
Shortness of breath :
Lag 1: 1.36 (0.92, 2.01); Lag 2: 1.45(0.98,2.14)
Lag 3: 1.52 (1.03, 2.25); Lag 4: 1.51 (1.02, 2.24)
Respiratory infections:
Lag 0: 1.58 (0.72, 3.46); Lag 1: 1.91 (0.79, 4.62)
Lag 2: 2.13 (0.97, 4.67; Lag 3: 2.09 (1.05, 4.15)
Lag 4: 2.05 (1.14, 3.68)
Beta2 agonist: Lag 4: 1.63 (1.00, 2.66)
Beta2 agonist: Lag 4: 2.02 (1.02, 4.01)
Lag 5: 1.96(0.99,3.88)
Moderate asthmatics (N: 41)
Statistically significant (only) prevalent episodes: Beta2 agonist:
Lag 0: 3.67 (1.25,  10.8); Lag 1: 4.60 (2.10, 10.1)
Lag 2: 7.01 (3.53,  13.9); Lag 3: 4.74 (1.96, 11.5)
Taggartetal. (1996)

Runcorn and Widnes
in NW England

Jul-Sep 1993
Panel study of 38 nonsmoking
asthma subjects (18 to 70 yrs) to
investigate the relationship between
asthmatic bronchial hyperrespon-
siveness and pulmonary function
(PEF, FEV!, FVC) and summertime
ambient air pollution. Used univari-
ate nested (hierarchical) analysis of
variance to test hypothesis that
BHR or spirometry measurements
varied with air pollution levels.
Analysis was limited to within-
subject variation of (BHR, FEV,, or
FVC).
24-h avg SO2
Max: 103.7pg/m3

Copollutants: NO2, O3,
smoke
Correlation with
copollutants: O3 (r =
0.13)
NO2 (r = 0.65)
Smoke (r = 0.48)
No association between SO2 and FEVi or FVC.

Changes in BHR correlated significantly with changes in 24-h
mean SO2, NO2, and smoke.

Percentage change in BHR per 10 pg/m3 SO2
24-h mean SO2 -6.3 % (-13.6, 0.6)
48-h mean -2.9 % (-12.8, 8.2). 24-h Lag 7.4 % (-4.5, 20.8)
May 2008
                                      F-12
                                 DRAFT—DO NOT QUOTE OR CITE

-------
     STUDY
                              METHOD
                                                       POLLUTANTS
                                                                                                FINDINGS
Timonen and
Pekkanen (1997)
Kuopio, Finland
1994
Panel study of 169 children (7 to 12
yrs) with asthma or cough symp-
toms living in urban and suburban
areas of Kuopio, Finland to deter-
mine association between air
pollution and respiratory  health. In
the urban areas there were 39
asthmatics and 46 with cough only;
in the suburban areas there were 35
asthmatics and 49 with cough who
were included in the final analysis.
Twice daily PEF and  daily symp-
toms were recorded for 3 mos. First
order autoregressive models  used
to assess associations between air
pollutants and PEF and logistic
regression models used for symp-
tom prevalences and incidences.
Analysis conducted on daily mean
PEF deviations. Mean morning or
evening PEF calculated for each
child was subtracted from the daily
value of morning or evening PEF.
The daily deviations were then Avgd
to obtain daily mean PEF deviation
for morning or evening PEF.
Avg daily
Urban area:
Mean: 6.0
25th percentile: 2.6
50th percentile: 3.6
75th percentile: 7.1

Max: 32

Co pollutants:
PM10
BS
NO2
Correlation coefficient
with SO2
PM10(r = 0.21)
BS  (r = 0.20)
NO2(r=0.22)
Among children with cough only, morning and evening deviations
in PEF in the urban panel was negatively associated with SO2.
SO2 was also associated with an increase in the incidence of
URS in children with cough only in the urban area. When
excluding the three highest SO2 days, these effects were no
longer statistically significant. No associations found between
SO2 and morning or evening PEF or respiratory symptoms in
children with cough only in the suburban panel.

Asthmatic: Lag 0: 0.198 (0.804)
Lag 1: 0.382 (0.789). Lag 2: 0.648 (0.715)
4-day Mean: 1.39(1.14)
Odds ratio (per 10  pg/m3)
URS
Lag 1: 1.46 (1.07, 2.00). Lag 2: 1.46(1.14, 1.87)
4-day Mean: 1.55(1.08,2.24)
Odds ratio when excluded 3 highest SO2 days (no 95% Cl
provided, but effects were not significant)
Lag 1: 1.13. Lag 2: 1.46. 4-day Mean: 1.12
PM10 r = 0.21 BS r = 0.20 NO2 r = 0.22
Regression coefficient (SE) (per 10 pg/m3 SO2)

Morning PEF deviations
Children with cough alone
Lag 0: -0.229 (0.608). Lag 1: -1.38 (0.564)
Lag 2: -0.683 (0.523). 4-day Mean: -1.28 (0.633)
Evening PEF deviations
Children with cough alone
Lag 0:-1.84 (0.673). Lag 1:-0.144 (0.711)
Lag 2: -0.291 (0.613). 4-day Mean: -0.878 (0.868)
Asthmatics
Lag 0: 1.28 (0.711). Lag 1: 0.575 (0.727)
Lag 2: 0.819 (0.642). 4-day Mean:
1.34 (1.05)
May 2008
                                      F-13
                                 DRAFT—DO  NOT QUOTE OR CITE

-------
     STUDY
                               METHOD
                                                       POLLUTANTS
                                                                                                 FINDINGS
van der Zee et al.
(1999)

Netherlands
3 winters from 1992
to 1995

Rotterdam and
Bodegrven/Reeuwijk
(1992-1993)

Amsterdam and
Meppel (1993-1994)

Amsterdam and
Nunspeet
(1994-1995)
Panel study of 633 children (aged 7
to 11 yrs) with and without chronic
respiratory symptoms, living in
urban and nonurban areas in the
Netherlands. Volunteers measured
daily PEF and reported the occur-
rence of respiratory symptoms and
bronchodilator use in a diary.
Association between air pollution
and decrements in PEF, symptoms
and bronchodilator use evaluated
with logistic regression models that
adjusted for first order autocorrela-
tion, min daily temperature, day of
wk, time trend, incidence of influ-
enza and influenza-like illness.
Median and max 24-h
mean concentration
(pg/m3)

1992-1993
Urban 23 (152);
Nonurban 8.9 (43)

1993-1994
Urban 11 (34);
Nonurban 5.0 (42)

1994-1995
Urban 6.0 (24);
Nonurban 3.6 (17)

Co pollutants:
PM10
Black smoke
Sulfate
NO2
The correlation between SO2 and PM varied from 0.5 to 0.8
during first two winters. Correlation with NO2 about 0.50.

In the urban areas, SO2 was associated with > 10% decrements
in evening PEF, LRS and  use of bronchodilator in children with
symptoms. Most consistent associations found with PMio, BS,
and sulfate. No association found between SO2 and prevalence
of URS, cough, phlegm, and > 10% decrements in morning PEF.
In the nonurban areas, no associations found with SO2. In
children without symptoms, no consistent associations with SO2.
Authors concluded that children with symptoms are more
susceptible to particulate air pollution effects and that use of
medication for asthma did not prevent the  adverse effects of PM
in children with symptoms.

Odds ratio (per 40 pg/m3 SO2) Children with symptoms
Urban areas Evening PEF
LagO: 1.32(0.96, 1.80). Lag  1:0.83(0.60, 1.14)
Lag 2: 1.67(1.28,2.19)
Symptoms of lower respiratory tract
Lag 0: 1.35 (1.01, 1.79) . Lag 1: 1.23 (0.93, 1.64)
Symptoms of upper respiratory tract
Lag 0: 0.97 (0.82, 1.14). Lag  1: 1.10(0.94, 1.28)
Cough
Lag 0: 0.90 (0.77, 1.05). Lag  1: 1.12(0.96, 1.30)
Use of bronchodilator
Lag 0: 0.92 (0.72, 1.18). Lag  1: 1.45(1.13, 1.86)
Odds ratio (per 40 pg/m3 SO2)
Children without symptoms
Evening PEF
LagO: 1.13(0.88, 1.47). Lag  1: 1.16(0.90, 1.50)
URS
Lag 0:0.92 (0.76, 1.11)
Lag 1: 1.10(0.91, 1.34). Lag 2: 0.83 (0.70, 0.99)
Cough
Lag 0: 0.93 (0.78, 1.11). Lag  1: 1.02 (0.84, 1.23)
Nonurban areas Evening PEF
Lag 0: 1.20 (0.91, 1.58). Lag  1: 0.89 (0.68, 1.17)
Symptoms of lower respiratory tract
Lag 0: 0.91 (0.69, 1.19). Lag  1: 0.91 (0.69, 1.22)
Symptoms of upper respiratory
Lag 0:0.94 (0.81, 1.09). Lag  1:0.97(0.83, 1.13)
5-day Mean: 0.67 (0.47, 0.94)
Cough. Lag 0: 1.08 (0.94, 1.23).  Lag 1: 0.98 (0.85, 1.12)
Use of bronchodilator
Lag 0: 0.86 (0.59, 1.25). Lag  1: 1.18(0.80, 1.74)
Evening PEF
LagO: 1.10(0.87, 1.39). Lag  1: 1.07(0.85, 1.35)
URS. Lag 0: 1.07 (0.92, 1.25). Lag  1: 0.85 (0.72, 1.00)
Cough. Lag 0: 0.86 (0.76, 0.97).  Lag 1: 0.95 (0.83, 1.08)
May 2008
                                      F-14
                                 DRAFT—DO NOT QUOTE OR CITE

-------
     STUDY
                               METHOD
                                                       POLLUTANTS
                                                                                                 FINDINGS
van derZee (2000)
Netherlands,
3 winters from
1992to 1995
Rotterdam
1992-1993
Panel study of 489 adults (aged 50
to 70 yrs) with and without chronic
respiratory symptoms, living in
urban and nonurban areas in the
Netherlands. Volunteers measured
daily PEF and reported the occur-
rence of respiratory symptoms and
bronchodilator use in a diary.
Association between air pollution
and decrements in PEF, symptoms
and bronchodilator use evaluated
with logistic regression models that
adjusted for first order autocorrela-
tion, min daily temperature, day of
wk, time trend, incidence of influ-
enza and influenza-like illness.
Median (max) cone

1992/1993:
Urban 25 (61) pg/m3

1993/1994
Urban 11 (34) pg/m3
 Nonurban 5.0
(42) pg/m3

1994/1995
Urban 6.0 (24)
Nonurban 3.6
(17)Mg/m3

Co pollutants
PM10
BS
Sulfate
NO,
Among symptomatic adults living in urban areas, the prevalence
of >20% decrement in morning PEF was associated with SO2.
Moreover, there were no associations found with prevalence of
bronchodilator use, LRS, >10% decrement in morning PEF and
>10% and >20% decrement in evening PEF.

In the nonurban areas, there was no consistent association
between air pollution and respiratory health. In the
nonsymptomatic adults, no consistent associations observed
between health effects and air pollutants, but a significant and
positive association was observed with URS in the nonurban
area at 1 day Lag.

Range of Spearman correlation coefficients between 24-h avg
cone SO2 and copollutants:
PM10: 0.31, 0.78
BS:0.21,0.75
Sulfate: 0.29, 0.69
NO2: 0.47, 0.51
Odds ratio (per 40 pg/m3 SO2)
symptomatic adults
In urban areas >10% decline in PEF Morning
Lag 0: 0.86 (0.60, 1.23); Lag 1: 0.97 (0.68, 1.39)
>20% decline in PEF Morning
Lag 0: 1.33 (0.66, 2.71); Lag 1: 1.98 (1.03-3.79)
LRS
Lag 0: 1.01 (0.84, 1.20); Lag1: .97 (0.82, 1.16)
5-day mean: 0.71 (95% Cl: 0.53 to 0.95)
URS
LagO: 1.15(0.97, 1.37); Lag 1: 1.06(0.90, 1.26)
Bronchodilator use
Lag 0: 1.09 (0.93, 1.28); Lag 1: 1.05 (0.89, 1.24)
Lag 2: 0.85 (0.72, 0.99)
In nonurban areas >10 % decline in PEF
Morning
Lag 0: 79 (0.48, 1.29); Lag  1: 1.08 (0.68, 1.72)
>20% decline in PEF Morning
Lag 0: 0.79 (0.22, 2.88); Lag 1: 71 (0.13, 4.02)
LRS
LagO: 1.11 (0.94, 1.30); Lag 1: 1.04(0.88, 1.22)
URS
Lag 0: 0.97 (0.79, 1.20); Lag 1: 1.20 (0.98, 1.47)
Bronchodilator use
LagO: 1.04(0.91, 1.18); Lag 1: 1.08(0.95, 1.22)
Ward et al. (2002)
Birmingham and
Sandwell, England
1996
Children ages 9-yrs old in 5 different
schools were given a questionnaire
and administered PEF measure-
ment in summer and/or winter.
Study used bivariate correlation,
linear and logistic regressions for
analysis
Median, Range (ppb):
Winter: 5.4 (2-18)
Summer: 4.7 (2-10)
Copollutants: NO2; O3;
PM10;PM25; H+;Cf;
HCI; HNO3; NH3; NH4+
NO3-; SO42"
No consistent association was found between the 24-h avg SO2
and risk of wheezing bronchitis. However, after a 7-day lag, a
10 ppb increase in the 24-h avg SO2 was associated with a 21 %
increase in risk of wheezing bronchitis.
Wake at night with cough- Lag 0 day
Winter 1.00 (0.91, 1.10)
Summer 1.00 (0.87, 1.14)
May 2008
                                      F-15
                                 DRAFT—DO NOT QUOTE OR CITE

-------
     STUDY
                               METHOD
                                                       POLLUTANTS
                                                                                                FINDINGS
Ward et al. (2002)

Birmingham and
Sandwell,
England

Jan-Mar 1997
May-July 1997
Panel study of 162 children (9 yrs at
time of enrollment) from two inner
city locations to investigate the
association between ambient acid
species with PEF and symptoms.
Daily symptoms and twice-daily
peak flow measurements were
recorded over 8 wk periods in the
summer and winter. 39 of the
children reported wheezing in the
past 12 mos. Linear regression
used for PEF and logistic regression
for symptoms.
24-h avg SO2

Winter:
Jan 13-Mar10, 1997
MediaN: 5.4 ppb
Range: 2, 18 ppb

Summer:
May 19-July 14, 1997
MediaN: 4.7 ppb
Range: 2, 10 ppb

Co pollutants:
NO2, O3, PM10, H+ ,
Cf  , HCI , HNO3 , NH3 ,
NH4+ , NO3" , SO42"

SO2 concentrations
were not related to
changes in PEF or
respiratory symptoms
In the summer, changes in morning PEF were associated with
SO2 at 3-days lag and the 7-day mean SO2. Prevalence of cough
associated with SO2 on the same day. In the winter SO2 was only
associated with symptom of feeling ill on the same day. 24-h avg
SO2 (per 4.0 ppb in winter; per 2.2 ppb in summer) Data also
available for 3-,4-, and 7-day Lag

Change in PEF (L/min)

Morning- Lag 0-day
Winter-0.60 (-2.51, 1.32). Summer 0.91 (-0.95, 2.78)

Afternoon- Lag 0-day
Winter-0.32 (-2.71, 2.04). Summer-0.89 (-2.61, 0.83)

Morning- Lag 1-day
Winter 0.08 (-1.67,1.86). Summer 0.29 (-1.56, 2.14)

Afternoon- Lag 1-day
Winter-0.88 (-2.87, 1.10). Summer-0.02 (-1.68, 1.65)

Odds ratio for symptoms

Cough-Lag 0-day
Winter 0.92 (0.81, 1.05). Summer 1.08 (1.02, 1.15)

Ill-Lag 0-day
Winter 1.09 (1.01, 1.18). Summer 1.05 (0.96, 1.14)

Shortness of breath-Lag 0-day
Winter 1.02 (0.93, 1.13). Summer 0.98 (0.87,1.10)

Cough-Lag 1-day
Winter 1.00 (0.87, 1.15). Summer 1.04 (0.97, 1.11)

Ill-Lag 1-day
Winter 1.03 (0.95, 1.11). Summer 1.02 (0.94, 1.12)

Shortness of breath-Lag 1-day
Winter 1.00 (0.90, 1.09). Summer 1.00 (0.89, 1.13)

Wake at night with cough- Lag 0 day
Winter 1.00 (0.91, 1.10). Summer 1.00 (0.87, 1.14)

Wake at night with cough- Lag 1 day
Winter 1.05 (0.96, 1.15). Summer 1.02 (0.89, 1.16)

Wheeze- Lag 0 day
Winter 0.96, (0.85, 1.07). Summer 1.05 (0.92, 1.19)

Wheeze-Lag 1 day
Winter 0.96 (0.86, 1.07). Summer 1.00 (0.88, 1.13

Summer change in PEF 2.7 (1.03, 4.38) per 2.2 ppb SO2
Lag 3 days (p < 0.05)

Summer change in PEF 6.83 (0.98, 12.69) per 2.2 ppb SO2
Lag 0-6  days (p < 0.05)
                                                        LATIN AMERICA
Pino et al. (2004)

Santiago, Chile

1995-1997
Cohort study of 492 infants recruited
at 4 mos of age and followed
through the first yr of life to deter-
mine the association between air
pollution on wheezing bronchitis.
Mean concentration of
SO2 (ppb)

Mean: 11.6
SD:8.1
MediaN: 10.0
PM2.5
NO2
During summer, SO2 played a role in adverse health effects after
taking into account distance between community and health
providers. During winter, no relationship was found.

Study did not provide effect estimates.
Romieu et al. (1996)

Mexico City, Mexico

Apr-Jul 1991
Nov 1991-Feb 1992
Panel study of 71 mildly asthmatic
children (5 to 13 yrs) to assess the
relationship between air pollution
and childhood asthma exacerbation.
Children measured PEF three times
daily and recorded daily symptoms
and medication use. Examined both
incidence and prevalence of symp-
toms. Lower respiratory symptoms,
cough, phlegm, wheeze, and/or
difficulty breathing.
24-h avg SO2 (ppm)
Mean: 0.09
SD:0.05
Range: 0.02, 0.20

03
PM10
PM2.5
NO2
Found a short, marked decrease in FVC and FEV1 in smokers
after exposure to SO2 that lasted for up to 30 h.

Relative risk per IQR SO2 (5.68 ppb)
Total absences:
1.03(1.02, 1.05)
Non-illness related absences:
0.95 (0.92, 0.99)
Illness related absences:
1.09(1.07, 1.12)
2-pollutant model with O3:
1.10(1.08, 1.13)
May 2008
                                      F-16
                                 DRAFT—DO NOT QUOTE OR CITE

-------
     STUDY
                               METHOD
                                                       POLLUTANTS
                                                                                                 FINDINGS
                                                              ASIA
Chenetal. (1999)

Three towns in
Taiwan:
Sanchun, Taihsi,
Linyuan

May 1995-Jan 1996
Cross-sectional panel study of 895
children (8 to 13 yrs) to evaluate the
short-term effect of ambient air
pollution on pulmonary function.
Single and multipollutant models
adjusted for sex, height, BMI,
community, temperature, and
rainfall. Examined 1, 2, and 7-day
lag effects.
Peak concentrations of
SO2

Range: 0, 72.4 ppb
Day-time avg and
1-day lag

CO
N03
PM10 (r = 0.63)
NO2(r = 0.71)
During the dust days, SO2 levels were significantly lower
compared to control days. SO2 had no significant effect on PEF
variability or night symptoms.

Regression estimate and standard error
per In SO2 (pg/m3)

Height-adjusted FEV, (mL): -35.6 (17.3)

Height-adjusted FVC (mL): -131.4 (18.8)
Jadsri et al. (2006)

Thailand

1993-1996
Spatial regression analysis of
outpatient disease occurrence
(respiratory system diseases; ICD
chapter 10) in 25 communities in
Rayong Province.
TSP, NOX
                      An inverse linear association found between In outdoor SO2 and
                      FEVi and FVC after adjusting for age, height and sex.
Min et al. (2008)

Korea
Panel study consisting of 867
smokers, former smokers, and
never smokers 20-86 yrs old. Used
linear regression analysis, adjusting
for age, height, gender, and a
diagnosis of asthma to examine the
combined effects of cigarette
smoking and SO2 on lung function.
Lung function measurements used
in this analysis included forced vital
capacity (FVC), forced expiratory
volume in 1 sec (FEV1), percent
predicted value of FVC (FVC %
pred), and percent predicted value
ofFEVI  (FEV1 %pred).
24-h avg (ppm): 0.006
Found a short, marked decrease in FVC and FEV, in smokers
after exposure to SO2 that lasted for up to 30 h.

Study did not provide effect estimates.
Park etal. (2002)

Seoul, Korea

Mar 2, 1996 to
Dec 22, 1999
Time-series analysis of school
absenteeism due to illness and air
pollution in one elementary school
in Seoul. School located in area with
heavy traffic. Avg enrollment in
1996 was 1,264.
24-h avg SO2

Mean: 9.19 ppb
SD:4.61
Range: 2.68, 28.11
PM10
NO2
CO (r = 0.67)
03
SO2, PMio, and O3 associated with illness related school
absenteeism. SP2 and O3 are protective for non-illness related
absences.

Relative risk per IQR SO2 (5.68 ppb)

Total absences:
1.03(1.02, 1.05)
Non-illness related absences:
0.95 (0.92, 0.99)
Illness related absences:
1.09(1.07, 1.12)
2-pollutant model with O3:
1.10(1.08, 1.13)
Park etal. (2005a)

Korea

Mar to June 2002
Panel study of 69 patients (16 to 75
yrs) diagnosed with asthma by
bronchial challenge or by bron-
chodilator response. Patients
recorded twice-daily PE, symptoms
at the end of each day (cough,
wheeze, chest tightness, shortness
of breath, sputum changes and  the
next morning, night awakenings).
During the study period, 14 Asian
dust days were identified. GEE  and
generalized additive Poisson
regression model used in analysis.
Daily avg SO2

Control Days:
0.0069
(0.0019) ppm
Dust days:
0.0052 (0.0010) ppm
Co pollutants:
PM10,  NO2, CO, O3
During the dust days, SO2 levels were significantly lower
compared to control days. SO2 had no significant effect on PEF
variability or night symptoms.

Relative risk based on Poisson log-linear regression analysis

PEF variability (>20%) 0.76 (0.37, 1.56)
Night symptoms: 0.98 (0.59, 1.51)
May 2008
                                      F-17
                                 DRAFT—DO NOT QUOTE OR CITE

-------
     STUDY
                           METHOD
                                                 POLLUTANTS
                                                                                       FINDINGS
Xuetal. (1991)

Beijing, China
Three areas:
industrial,
residential, and
suburban (control)

Aug 1986
Cross sectional survey of 1140
adults
(40 to 69 yrs) who had never
smoked living in three areas of
Beijing, to determine respiratory
health effects of indoor and outdoor
air pollution. A trained interviewer
obtained pulmonary function meas-
urements and determined history of
chest illnesses, respiratory symp-
toms, cigarette smoking, occupa-
tional exposure, residential history,
education level, and type of fuel
used for cooking and heating.
Annual mean
concentration of SO2
(pg/m3)
Residential: 128
Industrial: 57
SuburbaN: 18

TSPM
An inverse linear association found between Ln outdoor SO2 and
FEV, and FVC after adjusting for age, height and sex.

Regression estimate and standard error
per Ln SO2 (pg/m3)

Height-adjusted FEV, (mL): -35.6 (17.3)

Height-adjusted FVC (mL): -131.4 (18.8)
Table F-2.    Associations of short-term exposure to SO2 with emergency department visits and
               hospital admissions for respiratory diseases
STUDY
METHODS
POLLUTANTS
FINDINGS
UNITED STATES
Gwynn'etal. (2000)
Buffalo and Rochester, NY
United States
Period of Study:
1988-1990
Days: 1 ,090










Hospital Admissions
Outcome(s) (ICD9):
Respiratory admissions:
Acute bronchitis/
bronchiolitis (466);
Pneumonia (480-4860);
COPD and Asthma (490-
493, 496)
Age groups analyzed: 6
Study design: Time-series
N:24,
Statistical analyses: Poisson
regression with GLM and
GAM
Covariates: season, day of
wk, holiday, temperature,
relative humidity
Lag: 0-3 days
24-h avg SO2
(ppb):
MiN: 1.63
25th: 8.4
Mean: 12.2
75th: 15.4
Max' 37 7
H+;r = 0.06
SO42"(r = 0.19)
PM10(r = 0.19)
03 (r = 0.02)
NO2 (r = 0.36)
CO (r = 0.11)
COH (r = 0.19)






Significant associations observed between several pollutants and
various health-effect outcomes make it difficult to discriminate the
influence of a single-pollutant. This is likely the a result of the relatively
high intercorrelations among the various pollutants, as well as the
possible interactive role of several pollutants in the reported
associations.
Increment: 25.5, 7.0 ppb
(Max -Mean; IQR)
SO2 alone:
Max-Mean RR 1 .096 (t = 3.05) lag 0
IQRRR 1.025(t = 3.05)lagO







May 2008
                                  F-18
                             DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Ito et al. (2007)
New York, NY
1999-2002





















Jaffe et al. (2003)
3 cities, Ohio, United States
(Cleveland, Columbus,
Cincinnati)
Period of Study:
7/91-6/96









Lin et al. (2004c)
New York (Bronx County),
United States
Period of Study: 6/1991-
12/1993












METHODS
ED Visits
Outcome(s): Asthma
Study design: Time-series
Statistical Analysis: Poisson
GLM
Age groups analyzed: All
ages
Covariates: Adjustment for
temperature (same day and
avg lag 1-3), dew point
(same day and avg lag 1-3)
# Hospitals: 11
Lag(s): Avg 0 and 1 day













ED Visits
Outcome(s) (ICD9): Asthma
(493)
Age groups analyzed: 5-34
Study design: Time-series;
N: 4,416
Statistical analyses: Poisson
regression using a standard
GAM approach
Covariates: City, day of wk,
wk, yr, min temperature,
overall trend, dispersion
parameter
Season: June to Aug only
Dose-response
investigated: Yes
Lag: 0-3 days

Hospital Admissions
Outcome(s) (ICD9):
Asthma (493)
Age groups analyzed: 0-14
Study design:
Case-control

N: 2,629 cases; 2,236
controls
Statistical analyses: logistic
regression
Covariates: Race and
ethnicity, age, gender,
season
Statistical package:

Lag: 0,1, 2,3, 0-3


POLLUTANTS
Allyr
24-h avg (ppb):
7.8 (4.6)
5th: 3
25th: 5
50th: 7
75th: 10
95th: 17
Warm mos (Apr-
Sep)
24-h avg (ppb):
5.4 (2.2)
5th: 3
25th' 4
50th: 5
75th: 7
95th: 10
Cold Mos (Oct-
Mar)
24-h avg (ppb):
10.2(5.1)
5th: 4
25th: 6
50th: 9
75th: 13
95th: 19
Copollutants:
PM25;N02;03;
CO
24-h avg:
Cincinnati: 35.9
(25.1) pg/m3
Range: 1.7, 132
Cleveland: 39.2
(25.3) pg/m3
Range: 2.6, 167
Columbus:
11.1 (8.5) pg/m3
Range: 0, 56.8
Cincinnati:
PM10(r = 0.31)
NO2 (r = 0.07)
O3(r = 0.14)
Cleveland:
PM10(r = 0.29)
NO2 (r = 0.28)
O3 (r = 0.26)
Columbus:
PM10(r = 0.22)
NO2 (r = NR)
O3 (r = 0.42)
Cases:
24-h avg: 16.78
ppb
50th: 13.72
Range: 2.88,
66.35
C I
24-h avg: 15.57
nnh
ppu
50th: 13.08
Range: 2.88,
66.35
Quartile
Concentrations
(Ppb) :
Q1 ' 2 88 8 37
Q2: 9.37! 13.38
Q3: 13.5, 20.91
04:20.21,66.35
FINDINGS
In single-pollutant models, NO2 was found to have the most significant
association with asthma ED visits for all-yr and warm mos. SO2 was
significantly associated with asthma ED visits for all single-pollutant
models for all-yr and both the warm and cold mos. In copollutant models
for the warm mos, NO2 eliminated the association between SO2 and
asthma ED visits.
This result is consistent with the monitor-to-monitor correlations, which
suggested that NO2 had less exposure error compared to SO2.
Warm Mos (Apr-Sep)
(Weather model including smoothing terms for same day temperature
and avg lag 1-3 day temperature.)
Relative Risk (95% Cl) (per 6 ppb SO2) 1 .20 (1 .13, 1 .28)















Wide confidence intervals for data from Cleveland and Columbus make
these data not significant and unstable. Only data for Cincinnati was
considered statistically significant and demonstrated a concentration
response function that was positive.
No multipollutant models were utilized.
Increment: 50 pg/m3
Cincinnati: 35% (9, 21) lag 2
Cleveland: 6% (-7, 21) lag 2
Columbus: 26% (-25, 213) lag 3
All cities: 12% (1,23)
Attributable risk from SO2
increment:
Cincinnati: 4.2%
Cleveland: 0.66%
Columbus: 2.94%


Odds ratios for risk of hospitalization for asthma increased with each
quartile of SO2 concentration.
Lag 1 , 2, or 3 all showed a concentration response that was positive for
odds ratio as each quartile was compared to the total exposure group
(trend p > 0.001).
Quartile (24-h avg)
Q2 OR 1.26 lag 3
Q3 OR 1 .45 lag 3
Q4 OR 2. 16 (1.77, 2.65) lag 3
Quartile (1-h max)
Q4 OR 1.86 (1.52, 2.27) lag 3
For a 4 ppb increase in SO2
(24-h avg)
RR 1.07 (1.04, 1.11)






May 2008
F-19
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Michaud et al. (2004)
Hilo, Hawaii
2/21/1997-5/31/2001

















Moolgavkar* et al. (1997)
United States: Minneapolis-
St. Paul; Birmingham
Period of Study:
1986-1991















METHODS
ED Visits
Outcome(s) (ICD9): COPD
(490-496); Asthma (493,
495); bronchitis (490, 491),
other COPD (492, 494, 496)
Age groups analyzed: All
Study design: Time-series
Statistical analyses:
Exponential regression
models
Covariates: temporal
variables, day of wk,
meteorology
Statistical package: Stata,
SAS
Lag: 0,1,2,3 days






Hospital Admissions
Outcome(s) (ICD9): COPD
including asthma
(490-496), Pneumonia
(480-487)
Age groups analyzed: 65+
Study design: Time-series
Statistical analyses: Semi-
parametric Poisson
regression, GAM
Covariates: day of wk,
season, temporal trends,
temperature
Statistical package:
SDh ic-
r IUS
Lag: 0-3 days






POLLUTANTS
1-h max:
1.92 (12.2) ppb
Range: 0.0, 447
24-h avg:
1.97 (7. 12) ppb
Range: 0.0, 108.5
PM1













SO2 24-h avg
(Ppb):
Minneapolis:
Mean: 4.82
10th: 1.9
25th: 2.66
50th: 4.02
75th: 6.0
90th: 8.5
Birminaham:
Mean: 6.58
10th' 2 2
25th: 3.7
50th: 6.0
75th: 8.6
90th' 116
Minneapolis:
PM10(r = 0.08)
NO2 (r = 0.09)
CO (r = 0.07)
O3(r = -0.12)
Birminqham:
PM10(r = 0.17)
CO (r = 0.16)
03 (r = 0.02)
FINDINGS
The lack of organic carbon shows the pure SO2 effect uncontaminated
by vehicle emissions.
Asthma is associated with Vog, but Vog is not a major cause of asthma
in Hawaii. The strongest association was with the mo of the yr.
Admission for asthma and respiratory conditions was higher in the winter
compared to the summer, based on admission per day (observational-
not statistical analysis).
Increment: 10 ppb
COPD
RR 1 .04 (0.99, 1 .09) lag 1
RR 1 .04 (1 .00, 1 .09) lag 2
RR 1.07 (1.03, 1.11) lag 3
RR1.01 (1.00, 1.10) lag 1
RR 1.02 (1.03, 1.1 2) lag 2
RR 1.02 (1.03, 1.1 2) lag 3
Bronchitis
RR1.01 (0.93, 1.1 3) lag 1
RR 0.99 (0.88, 1.05) lag 2
RR1.01 (1.00, 1.1 4) lag 3
Other COPD
RR 1.00 (0.78, 1.23) lag 1
RR 0.96 (0.62, 1.11) lag 2
RR 0.98 (0.75, 1.16) lag 3
SO2 with NO2 and PMio were associated with hospital admissions.
Evidence of mixture effects was found. No single-pollutant was more
important than the other for respiratory admissions. Each pollutant was
associated with admissions except CO.
Consideration of four pollutants together showed the strongest
association with ozone. No pollutant other than O3 was stable in its
association with hospital admissions.
No effects were reported for Birmingham. Positive results were only
observed in Minneapolis.
Increment: 3.5 ppb

Sum of Pneumonia and COPD
1.6% (-0.1, 3.3) lag 2
Pneumonia Only
Minneapolis:
65+ 0.9% (-1 .1 , 2.9) lag 2 20 df
0.5% (-1.5, 2.5)
lag 2 130dfS




May 2008
F-20
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Moolgavkar (2000)
Reanalysis (2003a)
Multicity, United States:
Chicago, Los Angeles,
Maricopa County, (Phoenix)
Period of Study:
1987-1995




























METHODS
Hospital Admissions
Outcome(s) (ICD9): COPD
including asthma (490-496)
Age groups analyzed:
0-19,20-64,65+
(LA only)
Study design: Time-series
Statistical analyses: Poisson
regression, GAM
Covariates: Day of wk,
temporal trends,
temperature, relative
humidity
Lag: 0-5 days
























POLLUTANTS
Chicago: MediaN:
6 ppb
25th: 4. 75th:
8
Range: 0.5,
36
Los Angeles:
MediaN: 2 ppb
25th: 1.75th:
4
Range: 0, 16
Maricopa:
MediaN: 2 ppb
25th: 0.5.
75th' 4
Range: 0, 14
Chicago:
PM10(r =
0.42)
CO (r = 0.35)
NO2 (r =
0.44)
03(r = 0.01)
Los Angeles:
PM2.5 (r =
0.42)
PM10(r =
0.41)
CO (r = 0.78)
NO2 (r =
0.74)
03(r =
-0.21)
Maricopa:
PM10(r =
0.11)
CO (r = 0.53)
NO2 (r =
0.02)
03(r =
-0.37)
FINDINGS
In Los Angeles there was a significant association with and hospital
admissions for COPD.
SO2 may be acting as a surrogate for other pollutants since
heterogeneous responses found in different cities are inconsistent with a
cause-effect model.
Increment: 10 ppb
COPD, >65 yrs
Chicago lag 0: 4.87 (t = 3.18) GAM-100
LA lag 0: 2.84 (t = 13.32) GAM-30
LA lag 0: 1 .80 (t = 9.60)
GAM-100
LA lag 0: 1 .78 (t = 7.72)
NS-100

























May 2008
F-21
DRAFT—DO NOT QUOTE OR CITE

-------
         STUDY
                                 METHODS
                                                    POLLUTANTS
                                                                                               FINDINGS
NY Department of Health
(2006)

Bronx and Manhattan, NY

1999-2000
ED Visits

Outcome(s) (ICD9): Asthma
(493), for infants (466.1 and
786.09)

Study design: Time-series

Statistical Analysis: Poisson
regression with GLM

Statistical package: S-Plus

Age groups analyzed: All
ages

Covariates: Season, day-of-
wk, temperature

# Hospitals: 22

Lag(s): Avg 0- to 4-day  lags
24-h avg (ppm):
0.011 (0.0072)

PM10
PM2.5
OC
EC
Cr
Fe
Pb
Mn
Ni
Zn
H+
Sulfate
03
NO2
SO2
In single-pollutant models, PM2.5, SO2, O3, and NO2 were all found to be
significantly associated with asthma ED visits in a community, Bronx,
with a high prevalence of asthma. This association was maintained in
both two- and three-pollutant models for O3 and SO2.

Single-Pollutant Models

Relative Risk (95% Cl)  (per 0.011 ppm SO2)
5-day moving avg
Manhattan: 0.99 (0.88,  1.12)
Bronx: 1.11 (1.06, 1.17)
Relative Risk (95% Cl)  (per 0.0072 ppm SO2)
Bronx: 1.07(1.04, 1.11)
Relative Risk based on Daily Max Hourly SO2 (95% Cl) (per 0.0227 ppm
S02)
Manhattan: 0.96 (0.86,  1.07)
Bronx: 1.07(1.03, 1.12)
Relative Risk (95% Cl)
(per 0.0072 ppm SO2)-model excludes temperature ManhattaN: 0.99
(0.88, 1.11)
Bronx: 1.11 (1.06, 1.17)
  Relative Risk (95% Cl) (per 0.0072 ppm SO2)-
  By Gender
Manhattan
  Male: 0.90 (0.75, 1.07); Female: 1.08 (0.91, 1.29)
Bronx
  Male: 1.08(1.00, 1.17); Female: 1.14(1.06, 1.23)
Relative Risk (95% Cl)  (per 0.0072 ppm SO2)-By Age
Manhattan
  0-4:0.82 (0.59, 1.15). 5-18: 1.03(0.77, 1.37)
  19-34: 1.01 (0.76, 1.35)
  35-64: 1.04 (0.86, 1.25). 65+: 0.88 (0.57, 1.37)
Bronx
  0-4: 1.13(1.01, 1.26). 5-18: 1.03(0.92, 1.16)
  19-34: 1.06(0.93, 1.21). 35-64: 1.18(1.07, 1.30)
  65+: 1.12(0.88, 1.42)
Two-Pollutant Models
Relative Risk (95% Cl)  (per 0.0072 ppm SO2)
5-day moving avg
Manhattan
  SO2 + Max 8-h O3: 0.99 (0.88, 1.12)
  SO2 + FRMPM25: 0.97 (0.85,  1.11)
  SO2 +Max PM25: 0.98 (0.85, 1.12)
  SO2 + NO2: 1.01 (0.87, 1.16)
Bronx
  SO2 + Max 8-h O3: 1.11 (1.05, 1.17)
  SO2 + FRM PM25: 1.11 (1.04, 1.18)
  SO2 + MaxPM25: 1.09(1.03, 1.16)
  SO2 + NO2: 1.11 (1.04, 1.17)
Norrisetal. (1999)
Seattle, Washington
1995-1996
ED Visits

Outcome(s) (ICD9): Asthma
(493)

Study design: Time-series

Statistical Analysis:
Semiparametric Poisson
regression model

Statistical package:  S-Plus

Age groups analyzed: < 18

Covariates: Adjustments for
day-of-wk indicator
variables, time trends,
temperature, dew point
temperature

# Hospitals: 6

Lag(s): 0, 2
24-h avg (ppb):
6.0 (3.0)
Range: 1.0, 21.0
1-h max (ppb):
16.0(14.0)
Range: 2.0, 84.0
Copollutants:
PM10, Dry light
scattering,
NO2,CO, O3
A significant association was found between asthma emergency
department visits in children and PM25 and CO. Estimates were not
found to be different between high and low hospital utilization areas. SO2
was negatively associated with asthma emergency department visits in
high utilization areas, and positively associated in low utilization areas.
Relative Rates (95% Cl) (per 3 ppb 24-h avg SO2; per 12 ppb 1-h max
S02)
High Utilization Areas
24-h avg: 0.92 (0.83, 1.03); 1-h max: 0.99 (0.89, 1.10)
Low Utilization Areas
24-h avg: 1.09(1.00, 1.19); 1-h max: 1.09(1.00, 1.19)
All Areas
24-h avg: 0.97 (0.91, 1.04) lag 0;  1-h max:  1.02 (0.95,  1.09) lag 2
May 2008
                                F-22
                                  DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Peel et al. (2005)
Atlanta, GA, United States
Period of Study:
1/93-8/2000
















Schwartz (1995)
New Haven, CT
Tacoma, WA
United States
Period of Study:
1988-1990










Schwartz et al. (1996)
Cleveland, OH
Period of Study:
1988-1990







METHODS
ED Visits
Outcome(s)(ICD9):AII
respiratory (460-6, 477,
480-6, 480-6, 490-3, 496);
Asthma (493); COPD (491-
2, 496); Pneumonia (480-
486); Upper Respiratory
Infection (460-6, 477)
Age groups analyzed: All
Study design: Time-series.
N: 484,830. # of Hospitals:
31
Statistical analyses: Poisson
Regression, GEE, GLM,
and GAM (data not shown
for GAM)
Covariates: Day of wk,
hospital entry/exit, holidays,
time trend; season,
temperature, dew point
temperature
Statistical package: SAS, S-
Plus
Lag: 0 to 7 days. 3 day
moving avgs.
Hospital Admissions
Outcome(s) (ICD9):AII
respiratory admissions
(460-519)
Age groups analyzed: >65
Study design: Time-series
N: 13,470
Statistical analyses: Poisson
regression, log linear
regression using GLM and
GAM
Covariates: dewpoint, temp,
long-term trends, days of wk
Statistical package: S- Plus
Lag: 0-1


Hospital Admissions
Outcome(s) (ICD9): All
respiratory disease
Age groups analyzed: > 65
Study design: Time-series
Statistical analyses: Poisson
regression
Covariates: Season,
temperature, day of wk
Statistical package:
Lag: 0-1
POLLUTANTS
1-h max: 16.5
(17.1)ppb
10th%:2.0
90th%: 39.0
03
NO2
CO
PM
rlVl2.5
Evaluated
multipollutant
models (data not
shown)











24-h avg
New Haven
Mean 78 pg/m3
(29.8 ppb)
10th: 23
25th: 35
50th: 78
75th: 100
90th' 159

Mean: 44 pg/m3
(16.8 ppb)
10th: 15
25th: 26
50th: 40
75th: 56
90th: 74
Copollutants:
PM2.5
03
24-h avg: 35 ppb
10th: 13
25th: 20
50th: 31
75th: 45
90th: 61

PM2.5
03


FINDINGS
Estimates from distributed lag models (0-13 days) tend to be higher than
for 3-day moving avg. Positive associations for URI and COPD with SO2
were noted for unconstrained lags (0-13 days) that covered the previous
two weeks of exposure.
Increment: 20 ppb
All respiratory
RR 1.008 (0.997, 1.019) lag 0-2, 3-day moving avg
Upper Respiratory Infection (URI)
RR 1.010 (0.998, 1.024) lag 0-2, 3-day moving avg
A th
All: 1.001 (0.984, 1.017) lag 0-2, 3-day moving avg
Pneumonia
RR 1 .003 (0.984, 1 .023) lag 0-2, 3-day moving avg
COPD
RR 1.016 (0.985, 1.049) lag 0-2, 3-day moving avg









In New Haven, risk associated with SO2was not affected by inclusion of
PM2 5 in the model and the effect of PM25 was not strongly affected by
inclusion of SO2. This suggests that in New Haven, SO2 and PM25 acted
independently.
In Tacoma, 2-pollutant model analysis showed risk associated with SO2
was attenuated by PM25. This suggested risks associated with SO2 and
PM2 5 were not independent. Possibly, SO2 acts as a surrogate for PM25
in this city.
Increment: 50 pg/m3 or 18.8 ppb
New Haven, CT
RR = 1 .03 (Cl 1 .02,1 .05), lag 0-1 . p < 0.001
2-pollutant model with PM25:
RR = 1 .04 (Cl 1 .02, 1 .06) p < 0.001
Tacoma, WA
RR= 1.06(CI 1.01, 1.12), lag 0-1. p > 0.02
2-pollutant model with PM25:
RR = 0.99 (Cl 0.93, 1 .06) p > 0.5


Significant associations were seen for PM25 and O3, with somewhat
weaker evidence for SO2.
Increment: 100 pg/m3
RR 1 .03 (0.99, 1 .06) lag 0-1







May 2008
F-23
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Sheppard etal. (1999)
Reanalysis (2003)
Seattle, WA, United States
Period of Study:
1987-1994








Sinclair and Tolsma (2004)

Atlanta, GA
1998-2000











Tolbert et al. (2007)
Atlanta, GA
1993-2004


















METHODS
Hospital Admissions
Outcome(s) (ICD9):
Asthma (493)
Age groups analyzed: < 65
Study design: Time-series
N: 7,837

# of Hospitals: 23
Statistical analyses: Poisson
regression with adjustment
for auto-correlation.
Covariates:
Statistical package: S-Plus
Lag: 0,1, 2,3
ED Visits

Outcome(s): asthma, upper,
and lower respiratory
infections.
Study design: Time-series
investigation
Statistical Analysis: Single
pollutant Poisson general
linear modeling
Statistical package: SAS v.
8.02
Age Groups Analyzed: All
# Hospitals: 10
Lag(s): 0-8 days
ED Visits
Outcome(s) (ICD9):
Cardiovascular (410-414,
427, 428, 433-437, 440,
443-445, 451-453);
Respiratory (493, 786.07,
786.09,491,492,496,460-
465,477,480-486,466.1,
466.11,466.19)
Study design: Time-series
Statistical Analysis: Poisson
Generalized Linear Model
(GLM).
Statistical package: SAS
Age groups analyzed: All
ages
Covariates: Adjustment for
day-of-wk, hospital entry,
holidays, time, temperature,
dew point temperature
# Hospitals: 41. N: 238,360
(Cardiovascular); 1,072,429
(Respiratory)
Lag(s): 3-day moving avg
POLLUTANTS
24-h avg: 8 ppb
IQR: 5 ppb
10th: 3.0
25th: 5.0
50th' 8 0

75th' 100
90th: 13.0
PM2.5(r = 0.31)
PM2.5(r = 0.22)
03 (r = 0.07)
CO (r = 0.24)

1 -hour Max
Mean' 19 28
nnhv
ppuv
SD:16.28
PM2.5
PM10
NO2

CO
03




1-h max (ppb):
14.9
Range: 1.0, 149.0
10th: 2.0. 25th:
4 0
75th: 20.0. 90th:
35.0

PM10
PM2.5
03
NO2
CO
Sulfate
Total Carbon
Organic Carbon
EC
Water-Soluble
Metals
Oxygenated
Hydrocarbons






FINDINGS
Sources of SO2 adjacent or near to monitoring site. Low concentrations.
No association with SO2 for asthma but positive association for
appendicitis.
Increment: 5 ppb (IQR)
GAM with stricter criteria:
1. 0% (-2.0, 3.0) lag 0

GLM with natural spline
smoothing:
0.0% (-3.0, 4.0) lag 0




No significant findings for child or adult asthma.

Significant negative associations with upper respiratory infections for 6-8
day lag (RR = 0.98).
Significant positive association with lower respiratory infections for 0-2
day lag
(RR= 1.067).
Not provided.







In single pollutant models, O3, PMio, CO, and NO2 significantly
associated with ED visits for respiratory outcomes.
Relative Risk (95% Cl)
(per 16.0 ppbSO2)
1.003(0.997, 1.009)

















May 2008
F-24
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Wilson et al. (2005)
Multicity, United States
(Portland, ME and
Manchester, NH)
Period of Study:
1996-2000 (Manchester)
1998-2000 (Portland)















CANADA
Bates etal. (1990)
Vancouver Region, BC,
Canada
Period of Study: 7/1/1984-
10/31/1986




















METHODS
ED Visits
Outcome(s) (I CD 9 codes):
All respiratory
(460-519); Asthma (493)
Age groups analyzed:
0-14 yrs; 15-64 yrs; >65 yrs
Study design: Time-series
Statistical analyses: Multiple
regression analysis
standard GAM with more
stringent criteria parameters
Covariates'
Time-trend season
influenza, temperature,
humidity, precipitation
Statistical package: S-Plus
Lag' 0-2






POLLUTANTS
SO2 1-h max:
Mean, (SD) (ppb)
Portland
Allyr: 11.1 (9.1)
Winter: 17.1
(12.0)
Spring: 10.0(7.1)
Summer: 9.1
(8.0)
Fall: 9.7 (7.1)

Manchester
Allyr: 16.5(14.7)
Winter: 25.7
(15.8)
Spring: 14.8
(12.0)
Summer: 10.6
(15.1)
Fall: 14.6(11.1)
Copollutants:
O3
PM-jc
nvi2.5


FINDINGS
Elevated levels of SO2 were positively associated with elevated
respiratory and asthmatic ER visits. The significance of these
relationships is not sensitive to analytic or smoothing techniques.
Increment: 6.3 ppb (IQR) for Portland; IQR for Manchester
Portland:
All respiratory
All ages RR 1 .05 (1 .02, 1 .07) lag 0
0-14 yrs RR 0.98 (0.93, 1.02) lag 0
1 5-64 yrs RR 1 .06 (1 .03, 1 .09) lag 0
>65yrsRR1.10(1.05, 1.15) lag 0
Asthma
All ages RR 1 .06 (1 .01 , 1 .1 2) lag 2
0-1 4 yrs RR 1.03 (0.93, 1.15) lag 2
15-64 yrs 1.07 (1.01, 1.15) lag 2
>65 yrs RR 1 .07 (0.90, 1 .26) lag 2
Manchester:
All respiratory
All ages RR 1.01 (0.99, 1.02) lag 0
0-14 yrs RR 1.00 (0.96, 1.04) lag 0
1 5-64 yrs RR 1 .00 (0.98, 1 .03) lag 0
>65yrsRR 1.04(0.97, 1.11) lag 0
Asthma
All ages RR 1 .03 (0.98, 1 .09) lag 2
0-14 yrs RR 1.11 (0.98, 1.25) lag 2
15-64 yrs 1 .02 (0.96, 1 .08) lag 2
>65 yrs RR 1 .06 (0.83, 1 .36) lag 2

ED Visits
Outcome(s) (ICD 9):
Asthma (493);
Pneumonia (480-486);
Chronic bronchitis
(491,492,496);
Other respiratory (466)
Age groups analyzed:
All; 15-60
Study design:
# of Hospitals: 9
Statistical analyses:
Pearson correlation
coefficients were calculated
between asthma visits and
environmental variables
Season:
Warm (May-Oct);
Cool (Nov-Apr)
Covariates' NR
Lag:0, 1,2







May-Oct
SO2 1-h max:

Range: 0.0137,
0.0151 ppm
Nov-Apr
Range: 0.012,
0.0164 ppm
Number of
stations: 11
May-Oct

03 (r = 0.23)
NO2 (r = 0.67)
CoH (r = 0.34)
SO4 (r = 0.46)
Nov-Apr
O3 (r = 0.47)

NO2(r = 0.61)
CoH (r = 0.64)
SO4 (r = 0.54)






SO2 effects depend on the season. In the summer a rise in ambient SO2
levels was seen to coincide with a rise in respiratory related hospital
•
Correlation Coefficients:
Warm Season (May-Oct)
Asthma (15-60 yrs)
r = 0.118lagOp<0.01
r = 0 139 lag 1
Respiratory (15-60 yrs)
r = 0.134lagOp<0.001
r = 0.164 lag 1 p < 0.001
Cool Season (Nov-Apr)
Respiratory
1-14 yrs
r = 0.205lagOp<0.001
r = 0.234 lag 1 p < 0.001
r = 0.234 lag 2 p < 0.001
15-60 yrs
r = 0.180lagOp<0.001
r = 0.214 lag 1 p < 0.001
r = 0.215 lag 2 p < 0.001
> 61 yrs
r = 0.257lagOp<0.001
r = 0.308 lag 1 p < 0.001
r = 0.307 lag 2 p < 0.001
Asthma (> 61 yrs)
r = 0.125lagOp<0.001
r = 0.149 lag 1 p < 0.001
r = 0.148 lag 2 p < 0.001
Total ER admissions (> 61 yrs)
r = 0.13 lag 1 p < 0.01
r = 0.13 lag 2 p < 0.01
May 2008
F-25
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Burnett et al. (1997a)
16 cities
Period of Study: 4/1981-
12/1991
Days: 3,927









Burnett etal. (1997b)
Toronto, Canada
Period of Study:
1992-1994









Burnett etal. (1999)
Metro Toronto, Canada
Period of Study:
1980-1994













METHODS
Hospital Admissions
Outcome(s) (ICD9): All
respiratory admissions (466,
480-6, 490-4, 496)
Study design: Time-series.
N: 720,519. # of Hospitals:
134

Statistical analyses: random
effects relative risk
regression model
Covariates: Long-term
trend, season, day of wk,
hospital
Lag: 0, 1, 2 day
Hospital Admissions
Outcomes (ICD 9 codes):
Respiratory
tracheobronchitis (480-6),
COPD(491-4, 496)
Study design: Time-series
Statistical analyses: Poisson
regression, GEE, GAM
Covariates: Temperature,
dew point temperature,
Long-term trend, season,
influenza, day of wk
Season: Summers only
Lag: 0,1,2,3,4 days



Hospital Admissions
Outcome(s) (ICD9):
Asthma (493); obstructive
lung disease (490-2, 496);
Respiratory infection (464,
466, 480-7, 494)
Study design: Time-series
Statistical analyses: Poisson
regression model with
stepwise analysis
Covariates: Long-term
trends, season, day of wk,
daily max temperature, daily
min temperature, daily avg
dew point temperature, daily
avg relative humidity
Statistical package: S-Plus,
SAS
Lag: 0,1,2 days, cumulative



POLLUTANTS
1-h maxSO2
(ppb)
Mean' 14 4 SD'
22.2
25th: 3
50th: 10
75th' 19
95th: 45
99th: 97
O3 r = 0.04

Copollutants: CO,
NO2,COH



Mean SO2: 7.9
ppb. CV: 64
Range: 0, 26
5th: 1
25th: 4
50th' 7
75th: 11
95th' 18

Number of
stations: 6-11
CO (r = 0.37)
H+ (r = 0.45)
SO4 (r = 0.42)
TP (r = 0.55)
FP (r = 0.49)
CP (r = 0.44)
COH (r = 0.50)
O3(r = 0.18)
NO2 (r = 0.46)
24-h Mean: 5.35
ppb
CV= 110;
5th: 0
25th: 1
50th: 4
75th' 8
95th: 17
100th: 57
Number of
stations' 4
PMi c (r = 0 461
rlv|2.5 V1 u.tuy
PM fr — fl 9JV\
r!Vlio-2.5 (r - u.zo;
PM25(r = 0.44)
CO (r = 0.37)
NO2 (r = 0.54)
03 (r = 0.02)






FINDINGS
Control of SO2 reduced but did not eliminate the ozone association with
respiratory hospital admissions.
Increment: 10 ppb
Single-pollutant
SO2 and respiratory admissions, p = 0.134
Multipollutant model (adjusted for CO, O3, NO2, COH, dew point): RR
1.0055(0.9982, 1.0128) lag 0







Risks of hospitalization for respiratory disease were summed for O3,
NO2, and SO2 at 11% increase in admissions. The proportion associated
with the single-pollutant SO2 was 3.6%. CoH was the strongest predictor
of hospitalization indicating particle associated pollutants are responsible
for effects and outcomes measured.
Increment: 4.00 ppb (IQR)
Respiratory-percent increase 4.0% (t = 4.14) lag 0
Copollutant and multipollutant models RR
(t-statistic):
SO2, COH: 1.012(1.10)
SO2, H+: 1.022 (1.96)
SO2, SO4: 1.021 (1.93)
SO2, TP: 1.021 (1.72)
SO2, FP: 1.022(1.92)
SO2, CP: 1.023(2.03)
SO2, O3, NO2: 1.019(1.64)



The percent hospital admissions associated with SO2 increased for:
asthma, COPD, and respiratory infection. However, in multipollutant
models significant increases were only seen in asthma and respiratory
infection. SO2 effects could be largely explained by other variables in the
pollution mix as demonstrated by the Multipollutant model. The greatest
contribution of SO2 is to respiratory infection. However, overall SO2 is a
small factor in total hospitalization response.
Increment: 5.35 ppb (Mean)
Single-pollutant model percent increase (t statistic)
Asthma: 1.01% (1.76) lag 0-2
OLD 0.03% (0.05) lag 0-1
Respiratory infectioN: 2.40% (5.04) lag 0-2
Multipollutant model percent increase (SE)
Asthma:
SO2 + CO + O3: 0.89% (SE < 2)
SO2 + CO + O3 + PM2.5: 0.69% (SE < 2)
SO2 + CO + O3 + PM10-25: 0.16% (SE < 2)
SO2 + CO + O3 + PM2.5: 0.76% (SE < 2)
Respiratory infection:
SO2 + NO2 + O3: 1 .85%
SO2 + NO2 + O3 + PM25: 0.67 (SE < 2)
SO2 + NO2+O3 + PM10-25: 1.71 (SE > 3)
SO2 + NO2 + O3 + PM25: 1 .00 (SE > 2)
May 2008
F-26
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Burnett* et al. (2001)
Toronto, Canada
Period of Study:
1980-1994











Cakmak et al. (2006)
Canada
(Calgary, Edmonton,
Halifax, London, Ottawa,
Saint John, Toronto,
Vancouver, Windsor,
Winnipeg)
1993-2000







Fung et al. (2006)
Vancouver, BC, Canada
Period of Study:
6/1/95-3/31/99










METHODS
Hospital Admissions
Outcome(s) (ICD9): Croup
(464.4), pneumonia (480-
486), asthma (493), acute
bronchitis/bronchiolitis (466)
Age groups analyzed: <
2 yrs
Study design: Time-series
Statistical analyses: Poisson
regression with GAM
Covariates: Temporal trend,
day of wk, temperature,
relative humidity
Statistical package:
S-Plus
Lag: 0-5 days

Hospital Admissions
Outcome(s) (ICD9):
Respiratory (466, 480-486,
490-494, 496)

Study design: Time-series
Statistical Analysis: Poisson
Statistical package: S-Plus
Age groups analyzed: All
ages
Covariates: Day-of-wk,
mean daily temperature,
max daily temperature, min
daily temperature, change
in barometric pressure,
mean relative humidity
N: 215,544
Lag(s): 2. 6 days


Hospital Admissions
Outcome(s) (ICD9): All
respiratory hospitalizations
(460-519)
Age groups analyzed: 65+
Study design: (1) Time-
series (2) Case-crossover,
(3) DM-models (Dewanji
and Moolgavkar, 2000,
2002)
N: 40,974
Statistical analyses: (1)
Poisson, (2) conditional
logistic regression, (3) DM
method-analyze recurrent
data in which the
occurrence of events at the
individual level overtime is

Covariates: Day of wk
Statistical package: S-Plus
and R
Lag: Current day, 3 and
5 day lag
POLLUTANTS
1-h maxSO2
(ppb)
Mean: 11.8
CV:93
5th: 0
25th: 5
50th: 10
75th: 15
95th: 32
99th' 55
100th: 110
Number of

'
03 (r = 0.39)
SO2
CO
PM2.5
PM10-2.5
24-h avg: 4.6 ppb
Range: 2.8 ppb to
10.2 ppb
03
NO2
CO







SO224-h avg:
Mean: 3.46 ppb
SD: 1.82
IQR: 2.50 ppb
Range: 0.00,
12.50
CO (r = 0.61)
COH (r = 0.65)
NO2 (r = 0.57)
PM10(r = 0.61)
PM25(r = 0.42)
PM10-2.5(r = 0.57)
Os (r = -30.35)







FINDINGS
SO2 had the smallest effect on respiratory admissions of all pollutants
considered.
Increment: NR
All respiratory admissions:
Single-pollutant:
Percent increase: 3.1% (t = 1.900)
Iag3
Multipollutant (adjusted for O3):
Percent increase: 1.21% (t = 0.67)
Iag3





SO2 associated with increased hospital admissions.
% increase
(per 4.6 ppb SO2)
Overall
Single-pollutant model
1.1% (0.5, 1.8)
Multi-pollutant model
0.5% (0.1, 0.9)
By Gender
Male: 0.4% (-0.2, 1.1); Female: 0.9% (-0.4, 2.1)
By Education
35,905: 0.7% (-0.4, 1.8)
No significant association was found between hospital admissions and
current day SO2 levels (lag 0). Significant associations were found with
SO2 using a 3, 5, and 7 day moving avg, with the strongest association
observed with a 7 day lag. The DM method produced slightly higher
relative risks compared to the Time-series and case crossover results.
Increment: 2.5 ppb (IQR)
SO2 Time-series
RR 1.013(0.997, 1.028) lag 0
RR 1.030(1.010, 1.051) lag 0-3
RR 1 .032 (1 .008, 1 .056) lag 0-5
RR 1 .031 (1 .003, 1 .060) lag 0-7
SO2 Case-crossover
RR 1.010(0.992, 1.027) lag 0
RR 1 .028 (1 .005, 1 .050) lag 0-3
RR 1 .030 (1 .004, 1 .057) lag 0-5
RR 1 .028 (0.998, 1 .058) lag 0-7
SO2 DM model
RR 1.013(0.998, 1.027) lag 0
RR 1.034 (1.015, 1.053) lag 0-3
RR 1.039(1.016, 1.061) lag 0-5
RR 1.044(1.018, 1.070) lag 0-7
DM method produced slightly higher RR estimates on O3, SO2 and PM25
compared to time-series and case-crossover, and slightly lower
RR estimates on COH, NO2, and PMio, though the results were not
significantly different from one another.

May 2008
F-27
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Kestenetal. (1995)
Toronto, ON, Canada
Period of Study:
1991-1992







Lin et al. (2003)
Toronto, ON
Period of Study:
1981-1993










Lin* et al. (2004b)
Vancouver, BC

Period of Study:
1987-1998














METHODS
ED Visits
Outcome(s) (ICD 9):
Asthma (493)
Age groups analyzed:
Study design: Time-series
N:854
# of Hospitals: 1
Statistical analyses: Auto
regression
Statistical package: SAS
Lag: 1 or 7
Hospital Admissions
Outcome(s) (ICD9):
Asthma (493)
Age groups analyzed:
6-12
Study design:
Bi-directional case-
crossover

N: 7,319
Statistical analyses:
Conditional logistic
regression
Covariates: Daily max and
min temperatures and avg
relative humidity
Lag: Cumulative lag of 1-7
ays.
Hospital Admissions
Outcome(s) (ICD9):
Asthma (493)
Age groups analyzed: 6-12
Study design: Time-series
N: 3,754 (2,331 male, 1,423
female)
Statistical analyses: Semi-
parametric Poisson
regression with GAM (with
default and more stringent
'
Covariates: Trend, day of
wk,
Statistical package:
S-Plus

Lag: Cumulative
1-7 day



POLLUTANTS
SO2 24-h avg
No data was
provided for
concentration or
for correlation
with other
pollutants.
NO2
03
API (TRS.CO,
TSP)


SO224-h avg:
0.36 ppbSD:
5 90
Range: 0, 57.00
25th' 1 00
50th: 4.00
75th: 8.00
Number of
stations: 4
CO (r = 0 37)
NO2 (r = 0.54)
PM10(r = 0.44)
O3 (r = -0.01)
PM2 5 (r = 0 46)
DM Ir — ft OR\
rlvlio-2.5 (r - U.zo)



24-h avg SO2
(Ppb)
Mean: 4.77
SD: 2.75
MiN:0
25th: 2.75
50th: 4.25
75th: 6.00
Max: 24.00
Number of
stations: 30
CO (r = 0.67)

NO2 (r = 0.67)
O3 (r = —Q 1 0)






FINDINGS
Fit of an auto-regression model with Covariates linked to same day gave
no evidence of association between asthma and SO2.
Despite multiple attempts to correlate individual or combinations of
pollutants with air quality indices, no association was found between ER
visits for asthma and ambient daily, weekly, or monthly levels of SO2,
NO2, or O3.
No relative risks were provided.





SO2 is positively associated with asthma hospitalizations, although the
relationship varies in boys and girls.
Increment: 7 ppb (IQR)
Boys 6-12 yrs; Girls 6-12 yrs
Lag 0: OR 1.00 (0.95, 1.05); 1.04(0.97, 1.11)
Lag 0-1: OR 0.99 (0.93, 1.06); 1.04(0.95, 1.13)
Lag 0-2: OR 0.98 (0.90, 1.06); 1.05 (0.95, 1.16)
Lag 0-3: OR 0.96 (0.87, 1 .05); 1 .09 (0.98, 1 .22)
Lag 0-4: OR 0.95 (0.86, 1.05); 1.13(1.00, 1.28)
Lag 0-5: OR 0.93 (0.83, 1.03); 1.17(1.02, 1.34)
Lag 0-6: OR 0.93 (0.83, 1 .04); 1 .20 (1 .04, 1 .39)
Multipollutant model with PMio-2.s and PM25
Boys 6-12 yrs; Girls 6-12 yrs
Lag 0: OR 0.98 (0.93, 1.04); 1.06(0.98, 1.14)
Lag 0-1: OR 0.99 (0.91, 1.06); 1.03(0.93, 1.14)
Lag 0-2: OR 0.96 (0.88, 1 .05); 1 .04 (0.92, 1.17)
Lag 0-3: OR 0.95 (0.85, 1 .05); 1 .08 (0.95, 1 .23)
Lag 0-4: OR 0.94 (0.84, 1.06); 1.12 (0.97, 1.29)
Lag 0-5: OR 0.91 (0.80, 1.04); 1.18(1.00, 1.38)
Lag 0-6: OR 0.91 (0.80, 1 .04); 1 .28 (1 .08, 1 .51)
Results presented are default GAM, but authors state that use of natural
cubic splines with a more stringent convergence rate produced similar
results

Increment: 3.3 ppb (IQR)
Boys 6-12 yrs by SES status: Low; High
LagORR 1.02(0.94, 1.10); 1.03(0.95, 1.12)
Lag 0-1 RR 1.03 (0.94, 1.13); 1.06(0.96, 1.17)
Lag 0-2 RR 1 .03 (0.93, 1.15); 1.06(0.95, 1.18)
Lag 0-3 RR 1.01 (0.90, 1.13); 1.04(0.92, 1.17)
Lag 0-4 RR 0.98 (0.88, 1.10); 1.02(0.90, 1.14)
Lag 0-5 RR 0.97 (0.86, 1.10); 1.02(0.89, 1.16)
Lag 0-6 RR 0.98 (0.86, 1.12); 1.05(0.91, 1.21)
Girls 6-12 yrs by SES status: Low; High
LagORR 1.05(0.95, 1.16); 1.07(0.96, 1.19)
Lag 0-1 RR 1.11 (0.99, 1.25); 1.07(0.94, 1.21)
Lag 0-2 RR 1.11 (0.97, 1.26); 1.07 (0.93, 1.23)
Lag 0-3 RR 1.18 (1.02, 1.36); 1.02(0.87, 1.19)
Lag 0-4 RR 1.18 (1.02, 1.35); 0.99 (0.85, 1.15)
Lag 0-5 RR 1.19 (1.01, 1.40); 0.95 (0.80, 1.13)
Lag 0-6 RR 1.15 (0.97, 1.36); 0.98 (0.81, 1.17)
Multipollutant model (adjusted for NO2)
Girls, Low SES:
1.17(1.00, 1.37) lag 0-3
1.19(1.00, 1.42) lag 0-5
May 2008
F-28
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Lin et al. (2005)
Toronto, ON
Period of Study:
1998-2001


























Luginaah et al. (2005)
Windsor, ON, Canada
Period of Study:
4/1/95-12/31/00













METHODS
Hospital Admissions
Outcome(s) (ICD9):
Respiratory infections
(464,466,
480-487)
Age groups analyzed:
0-14

Study design: Case-
crossover
N: 6,782

Statistical analyses:
Conditional logistic
regression
Covariates'
Statistical package:
SAS 8.2
Lag: 0-6 days













Hospital Admissions
Outcome(s) (ICD9):
Respiratory admissions
(460-519)
Age groups analyzed: 0-14,
15-64, 65+, all ages
Study design: (1) Time-
series and (2) case-
crossover
N:4,214.# of Hospitals: 4
Statistical analyses:
(1)Poisson regression, GAM
with natural splines (stricter
criteria), (2) conditional
logistic regression with Cox
proportional hazards model
Covariates: Temperature,
humidity, change in
barometric pressure, day of
wk
Statistical package: S-Plus
Lag: 1,2,3 days
POLLUTANTS
24-h avg:
Mean: 4.73 ppb
SD: 2.58 ppb
Range: 1.00,
19.67
25th: 3.00

50th: 4.00
75th' 6 00

Number of
monitors: 5
PM2.5(r = 0.47)
PM,0-2.5(r = 0.29)
PM10(r = 0.48)
CO (r = 0.12)
NO2 (r = 0.61 )













SO2 mean 1-h
Max: 27.5 ppb,
SD: 16.5;
Range: 0, 129
IQR:

Number of
stations: 4
NO2 (r = 0.22)
CO (r = 0.16)
PM10(r = 0.22)
COH (r = 0 14)

O3 (r = -0.02)
TRS (r = 0.13)





FINDINGS
Asthma hospitalization for boys was associated with SO2 before the
adjustment for fine and coarse PM. Asthma hospitalization for girls was
not associated with SO2 for any lag.
Increment: 3 ppb (IQR)
Unadjusted Model:
Boys only:
OR 1.06 (0.97, 1.16) lag 0-3
OR 1.02 (0.92, 1.13) lag 0-5
Girls only:
OR 1.05 (0.94, 1.16) lag 0-3
OR 1.07 (0.95, 1.21) lag 0-5
Boys and Girls:
OR 1.06 (0.99, 1.13) lag 0-3
OR 1.04 (0.96, 1.13) lag 0-5
Adjusted Boys only:
OR 1.11 (1.01, 1.21) lag 0-3
OR 1.08 (0.97, 1.21) lag 0-5
Girls only:
OR 1.07 (0.96, 1.19) lag 0-3
OR 1.1 2 (0.98, 1.28) lag 0-5
Boys and Girls:
OR 1.10 (1.02, 1.18) lag 0-3
OR 1.1 0(1 .01, 1.20) lag 0-5
Multipollutant model with PM25 and PM25
Boys only:
OR 1.02 (0.90, 1.15) lag 0-3
OR 0.99 (0.85, 1.16) lag 0-5
Girls only:
OR 1.09 (0.0.94, 1.26) lag 0-3
OR 1 .07 (0.90, 1 .28) lag 0-5
Boys and Girls:
OR 1.05 (0.95, 1.15) lag 4
OR 1.03 (0.91, 1.16) lag 6
The effect of SO2 on respiratory hospitalization varies considerably,
especially at low levels of exposurer. Increment: 19.25 ppb (IQR)
Time-series, females; males
All ages, 1.041 (0.987, 1.098)
0.953(0.900, 1.009) lag 1
0-14 yrs, 1.111 (1.011, 1.221)
0.952(0.874, 1.037) lag 1
15-65 yr, 1.031 (0.930, 1.144)
0.971 (0.845, 1.1 5) lag 1
65+ yr, 1.030(0.951, 1.115)
0.9409 (0.860, 1 .029) lag 1
Case-crossover, females; males
All ages, 1.047(0.978, 1.122)
0.939(0.874, 1.009) lag 1
0-14yrs, 1.119(0.995, 1.259)
0.923(0.831, 1.025) lag 1
15-65 yr, 1.002(0.879, 1.141)
0.944(0.798, 1.1 16) lag 1
65+ yr, 1.020(0.924, 1.126)
0.968(0.867, 1.082) lag 1




May 2008
F-29
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Stiebetal. (1996)
St. John, New Brunswick,
Canada
Period of Study:
1984-1992
(May-Sep only)








Stieb* et al. (2000)
Saint John, New
Brunswick, Canada
Period of Study:
Retrospective:
7/92-6/94
Prospective:
7/94-3/96



















METHODS
ED Visits. Outcome(s):
Asthma
ICD9 codes: NR
Age groups analyzed: 0-15,
>15
Study design: Time-series
N: 1,163. # of Hospitals: 2
Statistical analyses: SAS
NLIN (Equivalent to Poisson
GEE)
Covariates: Day of wk, long-
term trends
Season: Summers only
(May-Sep). Dose-response
investigated?: Yes
Statistical package: SAS
Lag: 0-3 days
ED Visits
Outcome(s): Asthma;
COPD; Respiratory infection
(bronchitis, bronchiolitis,
croup, pneumonia);
All respiratory ICD9 codes:
NR
Age groups analyzed: All
Study design: Time-series
N: 19,821
Statistical analyses: Poisson
regression, GAM
Covariates: Day of wk,
selected weather variables
in each model

Season: All yr, summer only
Dose-response
investigated: Yes
Statistical package: S-Plus
Lag: all yr = 0; summer
only = 0-3








POLLUTANTS
1-h maxSO2
(ppb)
Mean: 38.1
Range: 0, 390
95th 110

O3 (r = 0.04)
NO2 (r = -0.03)
SO42- (r = 0.23)
TSP(r = 0.16)







24-h avg:
Annual Mean: 6.7
(5.6) ppb
95th: 18.0. Max:
60.0
Warm season
Mean: 7.6 (5.2)
ppb
95th: 18.0 .Max:
29.0
Annual Mean:
23.8 (21.0) ppb
95th: 62.0. Max:
161.0
Warm season
Mean: 25.4 (17.8)
ppb
95th: 62.0. Max:
137.0
CO (r = 0.31)
O3(r = 0.10)
NO2(r = 0.41)
TRS (r = 0.08)
PM10(r = 0.36)
PM25(r = 0.31)
H+ (r = 0.24)
SO?" (r = 0.26)
COM (r = 0.3)
H2S(r = -0.01)
Assessed
multipollutant
models
FINDINGS
SO2 did not affect the rate of asthma ED visits when O3 was included in
the model.
Increment: NR
SO2 + O3: 3 = -0.0030 (0.0027) lag 0










Non-linear effect of SO2 on summertime respiratory visits observed and
log transformation strengthened the association.
Increment: 23.8 ppb (mean)
1-h max:
Respiratory visits: 3.9% lag 5
May to Sept: 3.9% lag 0-3
Multipollutant model (SO2, O3, NO2)
All yr: 3.7% (1.5, 6.0) lag 5
Multipollutant model (In (NO2), O3, SO2 COM)
May to Sept: 3.9% (1.1,6.7) lag 0-3
















May 2008
F-30
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Villeneuve et al., (2006)
Toronto, ON, Canada
Period of Study:
1995-2000
Days: 2, 190













Yang et al. (2003b)
Vancouver, Canada
Period of Study:
1986-1998
Days: 4748







Yang et al. (2005)
Vancouver, BC, Canada
Period of Study:
1994-1998
Days: 1826












METHODS
GP Visits
Outcome(s) (ICD9): Allergic
Rhinitis (177)
Age groups analyzed: >65
Study design: Time-series
N: 52,691
Statistical analyses: GLM,
using natural splines (more
stringent criteria than
default)
Covariates: Day of wk,
holiday, temperature,
relative humidity,
aero-allergens
Season: All Yr; Warm, May-
Oct; Cool, Nov-Apr
Statistical package: S-Plus
Lag: 0-6
Hospital Admissions
Outcome(s) (ICD9): All
respiratory admissions
(460-519)
Study design:
Case-crossover
Age groups analyzed: < 3, >
65
Statistical analyses:
conditional logistic
regression
Lag: 0-5 days



Hospital Admissions
Outcome(s) (ICD9): COPD
excluding asthma (490-2,
494 496)
Age groups analyzed: 65+
Study design: Time-series
N: 6,027
Statistical analyses: Poisson
regression with GAM (with
more stringent criteria)
Covariates: Temperature,
relative humidity, day of wk,
temporal trends, season
Statistical package:
S-Plus
Lag: 0-6 days, moving
averages



POLLUTANTS
24-h avg: 4.7 ppb
SD: 2.8
IQR: 3.2 ppb
Range: 0, 24.8
stations: 9
NO2
03
CO
PM10
PMlO-2.5
PM-jc
nvi2.5






24-h avg SO2
(Ppb):
Mean: 4.84
SD: 2.84
5th: 1.50
25th: 2.75
50th' 4 25
75th: 6.25
100th: 24.00
IQR: 3.50
Number of
stations: 30
CO
NO2
03 (r = -0.37)
COH
24-h avg: 3.79
ppb
SD:2.12;
IQR: 2.75 ppb;
Range: 0.75,
22.67
Winter: 4. 10
(7 R7\
\^ -° ' /
Spring: 3.40
(1 .58)
Summer: 4.10
07Q\
./y;
Fall: 3.56 (1.92)
Number of
stations: 31
PMio (r = 0 62)
NO2(r = 0.61)
CO (r = 0.67)
03 (r = -0.34)


FINDINGS
There were positive associations between allergic rhinitis and SO2 for
exposures occurring on the same day as physician visits, but only during
the wintertime.
Increment: 10.3 ppb (IQR)
All results estimated from Stick Graph:
All Yr: Mean increase: 1.7% (-0.4, 2.8) lag 0
Warm: Mean increase: 0.3% (-1.9, 2.5) lag 0
Cool: Mean increase: 1.9% (-0.2, 4.1) lag 0











SO2 showed the weakest effect among children and the second weakest
effect among older adults when compared to all other pollutants
considered in the study.
Increment: 3.50 ppb (IQR)
All respiratory admissions < 3 yrs:
SO2 alone: OR 1 .01 (0.98, 1 .05) lag 2
SO2 + O3: OR 1 .01 (0.97, 1 .04) lag 2
SO2 + O3 + CO + COH + NO2: OR 0.98 (0.94, 1 .03) lag 2
All respiratory admissions > 65 yrs:
SO2 alone: OR 1 .02 (1 .00, 1 .04) lag 0
SO2 + O3: OR 1 .02 (1 .00, 1 .04) lag 0
SO2 + O3 + CO + COH + NO2: OR 1 .01 (0.98, 1 .03) lag 0




This study produced a marginally significant association between COPD
hospitalization and 6-day SO2 exposure. Most previous studies have not
detected a significant effect of SO2 on respiratory ED visits or
hospitalizations.
Increment: 2.75 ppb (IQR)
COPD
>65 yrs, yr round
RR 1.00 (0.97, 1.04) lag 0
RR 1 .02 (0.98, 1 .06) lag 0-1
RR 1 .04 (0.99, 1 .08) lag 0-2
RR 1 .04 (0.99, 1 .09) lag 0-3
RR 1.05 (0.99, 1.11) lag 0-4
RR 1.06 (1.00, 1.1 3) lag 0-5
RR 1.06 (0.99, 1.1 3) lag 0-6
2-pollutant model
NO2: RR 0.99 (0.91, 1.08) lag 0
CO: RR 0.97 (0.87, 1 .07) lag 0-6
O3: RR 1.07 (1.00, 1.1 4) lag 0-6
PM10: 0.97 (0.88, 1.06) lag 0-6
Multipollutant models
SO2, CO, NO2, O3, PM10: RR 0.94 (0.85, 1.05)
SO2, CO, NO2, O3: RR 0.96 (0.86, 1.06)
May 2008
F-31
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
METHODS
POLLUTANTS
FINDINGS
AUSTRALIA/NEW ZEALAND
Barnett et al. (2005)
Multicity, Australia/New
Zealand; (Auckland,
Brisbane, Canberra,
Christchurch, Melbourne,
Perth, Sydney)
Period of Study:
1998-2001



















Lam (2007)
Australia (New South
Wales; Sydney)
2001-2002






Hospital Admissions
Outcomes (ICD 9/ICD 10):
All respiratory
(460-51 9/JOO-J99 excluding
J95.4-J95.9, RO9.1,
RO9.8), asthma (493/J45,
J46, J44.8), COPD (490-
492, 494-496/J40-J44, J47,
J67), pneumonia with
bronchitis (466, 480-
486/J12-17, J18.0J18.1
J18.8J18.9J20J21)
Age groups analyzed: 0,1-
4,5-14
Study design: Case-
crossover
Statistical analyses:
Conditional logistic
regression, random effects
meta-analysis
Covariates: Temperature,
current-previous day
temperature, relative
humidity, pressure,
extremes of hot and cold,
day of wk, holiday, day after
holiday
Season: Cool, May-Oct;
Warm, Nov-Apr
Statistical package: SAS
Lag: 0-1 days
ED Visits
Outcome(s): Fever,
gastroenteritis,
asthma/other respiratory
problems
Study design: Time-series
Statistical Analysis: Auto
Regression Integrated
Moving Average (ARIMA)
statistical modeling
Statistical package: SPSS
Age groups analyzed: < 6
Covariates: NR. Lag(s): NR
24-h avg (ppb)
(range):
Auckland: 4.3
(0, 24.3)

Brisbane: 1 .8
/r\ o o\
(U, Q.Z)
Canberra: NA
Christchurch' 2 8
/n 11 Q^
Vui ' ' •-3)
Melbourne: NA
Perth: NA
Sydney: 0.9
(0, 3.9)
Daily 1 -h max
(range):
Auckland' NA

Brisbane: 7.6
(0, 46.5)
Canberra: NA
Ch ' t h h'
10 1

(0.1, 42.1)
Melbourne: NA
Perth: NA
Sydney: 3.7
(0.1,20.2)

24-h avg (ppm):
0.35(0.19)
Range: 0.10, 0.90
1-h max (ppm):
0.38 (0.20)
Range: 0.10, 1.80
Copollutants:
PM.n
nviio
PMir
TIVI2.5
NO2
O3


Increased hospital admissions were significantly associated with SO2 for
acute bronchitis, pneumonia, and respiratory diseases. In multipollutant
models the impacts of particulate matter and NO2 were isolated.
There were seasonal impacts on pneumonia and acute bronchitis
admissions in the
1- to 4-yr-old age group for SO2.
Increment: 5.4 ppb
(1-h maxIQR)
Pneumonia and acute bronchitis
Oyrs 3.5% (-0.3, 7.3) lag 0-1
1-4 yrs 6.9% (2.3, 11. 7) lag 0-1
Respiratory
Oyrs 3.2% (0.3, 6.3) lag 0-1
1-4 yrs 2. 7% (0.6, 4.8) lag 0-1
5-14 yrs 2.0% (-5.5, 10.1) lag 0-1
Asthma
0 yrs No analysis (poor diagnosis)
1-4 yrs 3.4% (-4.3, 11.6)
lag 0-1

5-1 4 yrs 3.3% (-5.6, 13.0)

lag 0-1





Bivariate correlations resulted in ARIMA models for fever and NO2 max,
gastroenteritis and O3 avg and NO2 max; and respiratory problems and
O3 max. Neither NO2 nor O3 was significantly associated with any of the
childhood illnesses analyzed.
SO2 was not significantly correlated with fever, gastroenteritis, or
respiratory problems; therefore, SO2 was not included in the ARIMA
models.






May 2008
F-32
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Petroeschevsky et al.
(2001)
Brisbane, Australia
Period of Study:
1987-1994
Days: 2922
















METHODS
Hospital Admissions
Outcome(s)(ICD9):AII
respiratory (460-519);
Asthma (493)
Age groups analyzed: 0-4,
5-14, 15-64, 65+, all ages
Study design: Time-series
N: 33,710
(13,246 = asthma)
Statistical analyses: APHEA
protocol, Poisson
regression, GEE
Covariates: Temperature,
humidity, season, infectious
disease, day of wk, holiday
Season: Summer, Autumn,
Winter, Spring, All yr
Dose-response
investigated? Yes
Statistical package: SAS
Lag: Single: 1,2,3 day
Cumulative: 0-2, 0-4
POLLUTANTS
Mean: 24-h avg:
Overall: 4.1 ppb
Summer: 3.9 ppb
AutumN: 4.2 ppb
Winter: 4.8 ppb
Spring: 3.7 ppb
Mean: 1-h max
Overall: 9.2 ppb
Summer: 7.8 ppb
AutumN: 9.3 ppb
Winter: 11.3 ppb
Spring: 8.4 ppb
$ of stations' 3

Copollutants:
BSP
(~\
Us
MO
INW2






FINDINGS
SO2 was highly correlated with max daily ER admissions for respiratory
conditions. The highest association was observed in the winter followed
by autumn, spring, and summer. For asthma, the highest association
was observed in the winter and autumn.
No statistically significant contributions for respiratory admissions were
reported for the age group 5-1 4 yr olds for any pollutant.
Increment: 0 ppb
Respiratory:
0-4 yrs 24-h avg 1 .224 (1 .087, 1 .377) lag 0-4
5-14yrs 1-h max 1.049 (0.986, 1.116) lag 0-4
15-64 yrs 24-h avg 1.033 (0.895, 1.118) lag 1
65+ yrs 24-h avg 1.121 (1.019, 1.234) lag 0
All ages 24-h avg 1.080(1.030, 1.131) lag 1
Asthma:
0-1 4 yrs 24-h avg 1.080(0.971, 1.201) lag 0
15-64 yrs 1-h max 0.941 (0.900, 0.984) lag 0
All ages 24-h avg 0.941 (0.876, 1.011) lag 2






EUROPE
Anderson et al. (1997)
Multicity, Europe
(Amsterdam, Barcelona,
London, Paris, Rotterdam)
Period of Study:
1977-1989 for Amsterdam
and Rotterdam
1986-1992 for Barcelona
1987-1991 for London
1980-1989 for Milan
1987-1 992 for Paris








Hospital Admissions
Outcome(s) (I CD 9):
COPD-unspecified
bronchitis (490), chronic
bronchitis (491),
emphysema (492), chronic
airway obstruction (496)
Study design: Time-series
Statistical analyses: APHEA
protocol, Poisson
regression, meta-analysis
Covariates: Trend, season,
day of wk, holiday,
influenza, temperature,
humidity
Season:
Cool, Oct-Mar;
Warm, Apr-Sep
Lag: 0,1,2 days and
0-3 cumulative
24-h all yr avg
(Mg/m3):
Amsterdam: 21
Barcelona: 40
LondoN: 31
MilaN: 53
Paris: 23
Rotterdam: 32
1-h max
Amsterdam: 50
Barcelona: 60
LondoN: NR
MilaN' NR
Paris' 47
Rotterdam' 82

Copollutants:
NO2
BS
TSP
03

The effect of SO2 varied considerably across the cities; however, the
summer estimate was significantly associated with COPD for the 1-h
measure and borderline significant forthe daily mean. Both 24-h and 1-h
SO2 concentrations were significantly associated with COPD ER
admissions in the warm season. Only cumulative lags of SO2 showed
borderline significance.
Increment: 50 pg/m3
COPD-Warm season
24 h avg 1.05(1.01, 1.10) 1-h 1.02(1.00, 1.04)
COPD-Cool season
24 h avg 1.02 (0.98, 1.05) 1-h 1.01 (0.99, 1.03)
COPD-AII yr
24-h avg 1 .022 (0.981 , 1 .055) lag 1
24-h avg 1.021 (0.998, 1.054) lag 0-3, cumulative
1-h max 1.01 (0.994, 1.029) lag 1
1-h max 1.015 (1.003, 1.027) lag 0-3, cumulative




May 2008
F-33
DRAFT—DO NOT QUOTE OR CITE

-------
         STUDY
                                 METHODS
                                                    POLLUTANTS
                                                                                               FINDINGS
Anderson et al. (1998)

London, England

Period of Study:
Apr 1987-Feb 1992

Days: 1,782
Hospital Admissions
Outcome(s) (ICD 9):
Asthma (493)
Age groups analyzed: < 15,
15-64,65+
Study design: Time-series
Statistical analyses: APHEA
protocol, Poisson
regression
Covariates: Time trends,
seasonal cycles, day of wk,
public holidays, influenza
epidemics, temperature,
humidity
Season:
Cool (Oct- Mar); Warm (Apr-
Sep)
Lag: 0,1,2 days
24-h avg SO2
Mean: 32.0
SD: 11.7
Range: 9, 100
5th: 16
10th: 18
25th: 24
50th: 31
75th: 38
90th: 46
95th: 52

# of monitors: 2

Copollutants:
03
NO2
BS
The strongest association between SO2 and asthma admissions was for
those >65 yrs in the cool season. A weaker association was observed for
children in the warm season and all yr. The adult population showed no
association.

In 2-pollutant models ozone was overall the strongest pollutant
associated with hospital admission with weaker associations with NO2
and BS. The most consistent yr-round association for All ages was found
with BS. When looking at all ages combined, SO2 association remained
significant in all 2-pollutant models except with NO2, both for all yr and
the summer (warm) season.

Increment: 10 ppb in 24-h SO2

0-14 yrs Whole yr
1.64% (0.29, 3.01) lag 1 2.04% (0.29, 3.83) lag 0-3
+ O3 1.77% (0.22, 3.36) lag 1 + NO2 1.23% (-0.22, 2.69) lag 1
+ BS 1.66% (0.23, 3.12) lag 1

Warm season
3.33% (1.09, 5.63) lag 1 3.40% (0.41, 6.48) lag 0-3
+ O3 3.35% (0.89, 5.87) lag 1 + NO22.92% (0.58, 5.32)  lag 1
+ BS 3.66% (1.35, 6.02) lag 1

Cool season
0.56% (-1.16, 2.32) lag 1
1.24% (-0.95, 3.49) lag 0-2

15-64 yrs Whole yr
-0.69% (-2.28, 0.94) lag 2
-0.71% (-2.69, 1.30) lag 0-2
Warm season
-1.39% (-3.97, 1.27) lag 0
-2.2% (-5.46, 11.8) lag 0-2
Cool season
-0.24% (-2.28, 1.84) lag 0
0.20% (-2.28, 2.74) lag 0-2
Multipollutant model with PM25 and PM25
Boys only:
OR 1.02 (0.90, 1.15) lag 0-3
OR 0.99 (0.85, 1.16) lag 0-5
Girls only:
OR 1.09 (0.0.94,  1.26) lag 0-3
OR 1.07 (0.90, 1.28) lag 0-5
Boys and Girls:
OR 1.05 (0.95, 1.15) lag 4
OR 1.03 (0.91, 1.16) lag 6
65+ yrs
Whole yr
2.82% (-0.82, 5.96) lag 2
3.06% (-0.72, 6.98) lag 0-3
Warm season
-2.62% (-7.31, 2.31) lag 2
-4.27% (-9.89, 1.71) lag 0-3
Cool season
5.85% (1.81, 10.05) lag 2
7.28% (2.19, 12.62) lag 0-3
+ O3 7.84% (2.48, 13.48) lag 1
+ NO2 4.19% (-0.53, 9.13) lag 1
+ BS 5.29% (0.42, 10.40) lag 1
All Ages
Whole yr
1.64% (0.54, 2.75) lag 1
2.75% (1.22, 4.30) lag 0-3
+ O3 1.48% (0.24, 2.73) lag 1
+ SO2 1.14% (-0.04, 2.33) lag 1
+ BS 1.54%(0.36, 2.73) lag 1
Warm season
2.02% (0.22, 3.85) lag 1
2.60% (0.02, 5.25) lag 0-3
+ O3 1.91% (0.05, 3.81) lag 1
+ NO2 1.64% (-0.23, 3.56) lag 1
+ BS 2.18% (0.32, 4.07) lag 1
Cool season
1.41% (0.0, 2.83) lag 1
2.83% (0.89, 4.81) lag 0-3
+ O3 -0.09% (-1.61, 1.82) lag 1
+ NO2 0.83% (-0.67, 2.34) lag 1
+ BS 1.11% (-0.41, 2.66) lag 1
May 2008
                                F-34
                                  DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Anderson et al. (2001)
West Midlands
conurbation, United
Kingdom
Period of Study: 10/1994-
12/1996













Atkinson et al. (1999b)
London, England
Period of Study:
1992-1994











METHODS
Hospital Admissions
Outcome(s) (ICD9): All
respiratory (460-519),
Asthma (493), COPD (490-
496,
excluding 493)

Age groups analyzed: 0-14,
15-64,65+
Study design: Time-series
Statistical analyses:
followed APHEA 2 protocol,
GAM
Covariates: Season,
temperature, humidity,
epidemics, day of wk,
holidays
Statistical package:
S-Plus 4.5 Pro
Lag: 0,1, 2,3, 0-1, 0-2, 0-3
Hospital Admissions
Outcome(s) (ICD9): All
respiratory (460-519);
Asthma (493); Asthma and
COPD (490-496); LRD
(466,480-486)
Age groups analyzed: all
ages, 0-14 yr, 1 5-64 yr and
>65yr
Study design: Time-series
N: 165,032
Statistical analyses: Poisson
regression following APHEA
protocol
Covariates: Long-term
seasonal patterns, day of
wk, temperature, humidity,
influenza.
Statistical package: SAS
Investigated
Dose/Response: Yes
Lag: 0,1,2,3 days
POLLUTANTS
24-h avg:
7.2 ppb,
4.7 (SD)
MiN: 1.9 ppb
Max: 59.8 ppb
10th: 3.3 ppb
90th: 12.3 ppb

'
PM10(r = 0.55)
PM10-2.5(r = 0.31)
PM25(r = 0.52)
BS (r = 0.50)
NO2 (r = 0.52)
f\ /r — r\ oo\
\J3 (T — U.ZZj





SO2- 24-h
Mean: 21. 2 (7.8)
pg/m3
MiN: 7.4
10th' 13

50th: 19.8
90th: 31
Max: 82.2
$ of monitors' 5
O3,CO, PM10, BS,
NO2

Correlation
coefficients
ranged between
r= 0.5 and 0.6




FINDINGS
When admissions were analyzed by subgroups, respiratory and asthma
admissions were positively correlated with SO2. SO2 significantly
associated with asthma and respiratory admissions for the 0 to 14-yr-
age group; however, little evidence of a seasonal interaction was
observed
Increment: 9 ppb (90th-10th)

All respiratory
All ages 1 .3% (-0.7, 3.4) lag 0-1
0-14 yrs 4.6% (1.40, 7.8) lag 0-1
15-64 yrs -0.9% (-4.8, 3.3) lag 0-1
> 65 yrs -2.0% (-4.9, 1 .1) lag 0-1
COPD with asthma
0-14 yrs 10.9% (4.50, 17.8) lag 0-1
15-64 yrs 2.4% (-5.5, 10.9) lag 0-1
> 65 yrs -4.2% (-8.9, 0.8) lag 0-1





Asthma was closely linked with PM, CO, NO2, and traffic pollution. When
SO2 and PMio were included in the same model, the magnitude of the
individual associations was reduced, as were their statistical
significance. This reduction occurred in children, adults and the elderly.
The other pollutants all had the effect of reducing the magnitude of the
individual SO2 and PM25 associations, although their statistical
significance was unaffected. This indicates that both SO2 and PM25
were indicators of the same pollutant mixture.
Increment: 18 pg/m3
All respiratory
All ages 2.01% (0.29, 3.76) lag 1
0-14 yrs 5.14% (2.59, 7.76) lag 0
15-64 yrs 1 .90% (-0.79, 4.660 lag 3
> 65 yrs 2.25 (-0.09, 4.65) lag 3
Asthma
All ages 3.38 (0.42, 6.43) lag 1
0-14 yrs 6.74% (2.92, 10.69) lag 1
15-64 yrs 4.58% (-0.18, 9.57) lag 3
> 65 yrs 6.31% (-1.59, 14.83) lag 2
COPD and Asthma
> 65 yrs 1 .53% (-1.83, 5.00) lag 3
Lower Respiratory
> 65 yrs 5.16% (1 .19, 9.28) lag 3

May 2008
F-35
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Atkinson et al. (1999a)
London, United Kingdom
Period of Study:
1/92-1294
















Atkinson et al. (2001)

Multicity, Europe
(Barcelona, Birmingham,
London, Milan,
Netherlands, Paris, Rome,
Stockholm)
Period of Study:
1998-1997









Boutin-Forzano et al.
(2004)
Marseille, France
Period of Study:
4/97-3/98










METHODS
ED Visits
Outcome(s) (I CD 9):
Respiratory ailments
(490-496), including
asthma, wheezing, inhaler
request, chest infection,
COPD, difficulty in
breathing, cough, croup,
pleurisy, noisy breathing
Age groups analyzed:
0-14' 15-64' 2 65'
All ages
Study design: Time-series
N' 98 685

# of Hospitals: 12
Statistical analyses: Poisson
regression, APHEA protocol
Covariates: Long-term
trend, season, day of wk,
influenza, temperature,
humidity
Statistical package: SAS
Lag: 0,1,0-2 and
0-3 days
Hospital Admissions

Outcome(s) (ICD 9):
Asthma (493), COPD (490-
496), All respiratory (460-
519)
Study design: Time-series
Statistical analyses: APHEA
protocol, Poisson
regression, meta-analysis
Covariates: Season,
temperature, humidity,
holiday, influenza
Lag: NR





ED Visits
Outcome(s): Asthma
ICD 9 Code(s): NR
Age groups analyzed: 3-49
Study design: Case-
crossover
N:549
Statistical analyses: Logistic
regression
Covariates: Minimal daily
temperature, max daily
temperature, min daily
relative humidity, max daily
relative humidity, day of wk
Lag: 0-4 days
POLLUTANTS
24-havg:21.2
pg/m3, SD: 7.8
10th: 13.0
50th: 19.8
90th: 31.0
Range: 7.4, 82.2
# of Stations: 5
SO2
03 (8 h)
CO (24 h avg),
PM10 (24 h avg)
BS










1-h max of SO2
(un/m3!
\Hy/m /
Barcelona: NR
Birmingham: 24.3
LondoN: 23.6
MilaN: 29.1
Netherlands: 8.5
Paris: 17.7
Rome: 9.8
Stockholm: 3.8
NO2, O3, CO, BS,
PM10
BcircsloricT 0 32
B'gham: 0.77
LondoN: 0.72
MilaN: 0.64
Netherlands: 0.67
Paris: 0.63
Rome: 0.15
Stockholm: 0.36
Mean: SO2:
22.5 [ig/m
Range: 0.0, 94.0
NO2 (r = 0.56)
O3 (r = -0.25)









FINDINGS
SO2 was closely related to PMio, but 2-pollutant models showed that the
effect of SO2 was decreased by NO2 and PMio inclusion. Inclusion of
other pollutants did not significantly decrease the influence of SO2 on ER
admissions in
2-pollutant models.
Increment: 18 pg/m3 in 24-h
Single-pollutant model
Asthma only
0-14 yrs 9.92% (4.75, 15.34) lag 1
15-64 yrs 4.19% (-0.53, 9.13) lag 1
All ages 4.95% (1 .53, 8.48) lag 1
All respiratory
0-14 yrs 6.01% (2.98, 9.12) lag 2
15-64 yrs 2.72% (-0.18, 5.70) lag 3
65+ yrs -1 .82% (-5.72, 2.25) lag 3
All Ages 2.81% (0.72, 4.93) lag 1
Copollutant models for asthma among children:
SO2 + NO2: 5.42% (0.18, 10.93)
SO2 + O3: 8.39% (3.82, 13.17)
SO2 + CO: 8.05% (3.45, 12.86)
SO2 + PM10: 5.63 (0.53, 10.98)
SO2 + BS: 8.03 (3.32, 12.96)




The inclusion of SO2 in the models only modified PM10 associations in
the
0- to 14-yr age group.
Increment: 10 pg/m3 for PMio; change in SO2 not described.
Asthma, 0 to 14 yrs:
ForPM10: 1.2 (0.2,2.3)
For PM10 + SO2: 0.8 (-3.7, 5.6)
Asthma, 15 to 64 yrs:
ForPM10: 1.1 (0.3, 1.8)
ForPM10 + SO2: 1.6(0.6,2.6)
COPD + Asthma, > 65 yrs
ForPM10: 1.0(0.4, 1.5)
ForPM10 + SO2: 1.3(0.7, 1.8)
All respiratory, > 65 yrs of age
For PM10: 0.9 (0.6, 1.3)
ForPM10 + SO2: 1.1 (0.7, 1.4)




No association was observed between ER visits for asthma and SO2
levels.
Only single-pollutant models were utilized.
Increment: 10 pg/m3
Increased ER visits
OR 1.0023 (0.9946, 1.0101) lag 0
OR 0.9995 (0.9923, 1 .0067) lag 1
OR 0.9996 (0.9923, 1 .0069) lag 2
OR 0.9970 (0.9896, 1 .0045) lag 3
OR 0.9964 (0.9889, 1 .0040) lag 4




May 2008
F-36
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Buchdahletal. (1996)
London, United Kingdom
Period of Study:
3/1/92-2/28/93







Castellsague et al. (1995)
Bcircslorici Spciin

Period of Study:
1986-1989















Dabetal. (1996)
Paris, France
Period of Study: 1/1/87-
9/30/92




















METHODS
ED visits.
Outcomes: Daily acute
wheezy episodes.
Age groups analyzed: < 16
Study design: Case-control
N: 1,025 cases, 4,285
controls. # of Hospitals: 1
Statistical analyses: Poisson
regression
Covariates: Season,
temperature, wind speed
Season: Spring (Apr-Jun),
Summer (Jul-Sep)
Autumn (Oct-Dec),
Winter (Jan-Mar)
Statistical package: Stata.
Lag: 0-7 days
ED visits. Outcome(s):
Asthma
Age groups analyzed: 15-64
Study design: Time-series. #
of Hospitals: 4
Statistical analyses: Poisson
regression

Covariates: Long time trend,
day of wk, temperature,
relative humidity, dew point
temperature
Seasons : Winter : Jan-Mar;
Summer : Jul-Sep
Dose-response
investigated: Yes
Lag: 0, 1-5 days and
cumulative.
Summer: lag 2 daysWinter:
lag 1 day
Hospital Admissions
Outcome(s)(ICD9):AII
respiratory
(460-5 19), Asthma (493),
COPD
(490-496)
Age groups analyzed: All
ages
Study design: Time-series
Number of hospitals: 27
Statistical analyses: Poisson
regression, followed APHEA
protocol
Covariates: Temperature,
relative humidity, influenza,
Long-term trend, season,
holiday, medical worker
strike

Lag: 0,1, 2 days, 0-3
cumulative







POLLUTANTS
SO2 24-h yr
round Mean:
22pg/m3, SD: 14
IQR: pg/m3
Spring: 20(14)
Summer: 18 (22)
Fall: 24 (14)
Winter: 25 (14)
NO2 (r = 0.62)
O3 (r = -0.28)






Mean SO2
(Mg/m3)
Summer' 40 8
25th' 25 50th' 36
75th: 54. 95th: 82

Winter: 52.0
25th: 36. 50th: 49
75th: 67. 95th: 94

# of Stations: 15
manual, 3
automatic
NO2
03





All Yr:
24-h avg: 29.7
pg/m3
Median:23.0
5th' 7 0 99th'
125.0
1-h max' 59 9
Median'46 7
5th' 140 99th'
232.7
Warm season
24-h avg: 20.1
Median:18.3.5th:
6.0
99th: 49.3
1-h m8X'42 7
Msdi8rv37 0
5th: 13.0. 99th:
133.7
Cold season
24-h avg: 40.1
pg/m3
Median:31.3
5th: 8.7. 99th:
149.0
1-h max:78.3
Median:60.7
5th: 17.0.99th:
268.3
NO2,O3,PM13, BS
FINDINGS
Variations in SO2 could not explain the U-shaped relationship between
ozone and incidence of asthma.
Increment: 14 pg/m3 (Std. Dev.)
No adjustments to model
RR 1.16 (1.10, 1.23) lag not specified
Adjusted for temperature and season.
RR 1.12 (1.06, 1.19) lag not specified
Adjusted for temperature, season and wind speed.
RR 1 .08 (1 .00, 1 . 1 6) lag not specified







Interaction between pollutants and asthma emergency room visits was
influenced by soy-bean dust in the air. The daily mean of asthma visits
and level of SO2 were higher in the winter than in the summer. A positive
but not statistically significant increase in relative risk was found for SO2
in the summer. SO2 levels were higher in the winter, but the RR was
lower compared to the RR in the summer. SO2 was not significantly
associated with asthma related ER visits. Increment: 25 pg/m3

Seasonal differences
Summer: RR 1.052 (0.980, 1.129) lag 2
Winter: RR 1 .020 (0.960, 1 .084) lag 1









1-h max SO2 levels yielded lower relative risk when compared to 24-h
avg levels. COPD effects were only significantly associated with SO2
with no lag. The strongest association was observed with PMi3; 4.5%
increase in respiratory admission per 100 pg/m3 increment. SO2 was a
close second. Neither analysis by age or by season showed a significant
sensitivity for hospital admissions. The strongest association for asthma
admission for all pollutants was with SO2 24-h avg of 7% (0.14, 14.10),
but 1-hr max level was not significant. The strongest association for
admission with COPD diagnosis was also for 24-h avg of SO2 (9.9%
[2.3, 18]). Increment: 100 pg/m3
All respiratory (1987-1990)
24-h avg RR 1 .042 (1 .005, 1 .080) lag 0-2
1-h max RR 1.018 (0.988, 1.048) lag 0-2
Asthma (1987-1 992)
24-h avg RR 1.070 (1.004, 1.1 41) lag 2
1-h max RR 1.047 (0.998, 1.098) lag 2
COPD
24-h avg RR 1 .099 (1 .023, 1 .1 80) lag 0
1 -h max RR 1 .051 (1 .025, 1 .077) lag 0










May 2008
F-37
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
de Diego Damia et al.
(1999)
Valencia, Spain
Period of Study: 3/1994-
3/1995









Fusco et al. (2001)
Rome, Italy
Period of Study: 1/1995-
10/1997





















Galan et al. (2003)
Madrid, Spain
Period of Study:
1995-1998









METHODS
ED visits. Outcome(s) (ICD
9): Asthma (493)
Age groups analyzed: > 12
N:515.# of Hospitals: 1
Statistical analyses:
Stepwise regression and
ANOVA; Linear regression
Covariates: Season and
temperature
Statistical package: SPS





Hospital Admissions
Outcomes
(ICD 9 codes): All
Respiratory
(460-519, excluding
470-478); Acute respiratory
infections including
pneumonia (460-466, 480-
486), COPD
(490-492, 494-496), asthma
(4931
^t^oy
Age groups analyzed: All
ages, 0-14
Study design: Time-series
Statistical analyses: Poisson
regression with GAM
Covariates: Influenza
epidemics, day of study,
temperature, humidity, day
ofwk, holidays
Statistical package: S-Plus 4
Lag:0, 1,2,3,4








ED Visits
Outcome(s) (ICD9): Asthma
(493)
Age groups analyzed: All
Study design Time-series
N: 4,827
Statistical analyses: Poisson
regression, (1) classic
APHEA protocol and (2)
GAM with stringent criteria
Covariates: Trend, yr,
season, day ofwk, holidays,
temperature, humidity,
influenza, acute respiratory
infections, pollen
Lag: 0-4 days
POLLUTANTS
24-h avg SO2
(Mg/m3)
Winter
Mssrv 56
Range: 30, 86
Spring
Mean: 47.
Range: 34, 75

Mean: 40.
Range: 12, 62
Autumn
Mean: 50.
Range: 42, 59
Number of
monitors: 1
BS; r = 0.54
24-h avg: 9.1
(5.8) pg/m3
25th: 5.1
50th: 7.9
75th: 12.0
# of monitors: 5
O3 (r = -0.35)
CO (r = 0.56)
NO2 (r = 0.33)

Particles;
r=0.25
















24-h Mean:
23.6 pg/m3
SD: 15.4
10th: 9.2
25th: 12.3
50th: 18.7
75th: 31 .3
90th: 43.9
Range: 5, 121.2
# of Stations: 15
PM10(r = 0.581)
NO2(r = 0.717)
O3(r = -0.188)
FINDINGS
The SO2 concentration was averaged for each season and quartiles of
concentration determined. Asthma visits that occurred in each season
were examined. There were no significant associations with asthma ER
visits with any season or with any quartile of SO2 exposure.
Mean number of asthma-related ED visits based on quartile of SO2
All yr: < 41 ug/m3: 8.6
41-50 pg/m3: 9.1. 51-56 pg/m3: 11.6
>56pg/m3: 11.9








SO2 did not have an effect on respiratory hospitalizations.
Increment: 6.9 pg/m3 (IQR)
Respiratory conditions:
All ages:
0.4% (-1.3, 2.2) lag 0
0.8% (-0.9, 2. 4) lag 1
0.3% (-1.3, 1.8) lag 2
0-14 vrs'
-0.7% (-4.0, 2. 7) lag 0
-2.0 (-5.2, 1.3) lag 1
-0.8 (-3.8, 2.3) lag 2
Acute respiratory infections: All ages:
0.4% (-2. 1,3.0) lag 0
1.4% (-1.0, 3.9) lag 1
1.2% (-1.0, 3.5) lag 2
0-14yrs:
-0.1 % (-3.9, 3.8) lag 0
-2.7% (-6.3, 1.0) lag 1
-1.2% (-4.5, 2.2) lag 2
All ages:
-1. 5% (-6.6, 3.9) lag 0
-1.5% (-6.5, 3.7) lag 1
2.5% (-2.2, 7.4) lag 2
0-14yrs:
-2.6 (-10.4,6.0) lag 0
4.3% (-3.5, 12.7) lag 1
5.5% (-1.8, 13.2) lag 2
COPD:
All ages:
1. 0% (-1.9, 4.0) lag 0
-1.1% (-3.9, 1.8) lag 1
-0.5% (-3.1, 2.1) lag 2
SO2 registered a predominately winter based pattern, and was positively
correlated with PM25, NO2. The lag that described the strongest
association was 3 days.
Multipollutant models were fitted for cold season pollutants. SO2 was the
most affected when PM25 was included in the model.
Parametric estimates using APHEA protocol produced similar results as
GAM.
The SO2 association may be due to the concealing effects of other
pollutants. PM2 5 accounted for most of the observed effects.
Increment : 10 pg/m3
Asthma :
RR lag 01. 018 (0.984, 1.054)
RRlag 1 1.005(0.972, 1.039)
RRIag2 1.002(0.970, 1.036)
RRlag 3 1.029 (0.997, 1.062)
RRlag 4 1.025 (0.994, 1.058)
Multipollutant model:
SO2/PM10 0.966 (0.925, 1.009)
May 2008
F-38
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Garty etal. (1998)
Tel Aviv, Israel
Period of Study: 1993












Hagen et al. (2000)
Drammen, Norway
Period of Study:
1994-1997










METHODS
ED Visits
Outcome(s): Asthma
ICD 9 Code(s): NR
Age groups analyzed: 1-18
Study design: Descriptive
study with correlations
N: 1,076
Statistical analyses:
Pearson correlation and
partial correlation
coefficients
Covariates: Max and min
ambient temperatures,
relative humidity and
barometric pressure
Statistical package: Statistix
Hospital Admissions
Outcome(s)(ICD9):AII
respiratory admissions
(460-519)
Age groups analyzed: All
ages
Study design: Time-series
Number of hospitals: 1
Statistical analyses: Poisson
regression with GAM
(adhered to HEI phase 1.B
report)
Covariates: Time trends,
day of wk, holiday,
influenza, temperature,
humidity
Lag: 0,1,2,3 days
POLLUTANTS
24-h mean of
SO2 (estimated
from histogram):
27 pg/m3 Range:
11, 64

NOx
SO2

03








SO2 24-h avg
ftjg/m3): 3.64,
SD: 2.41
25th: 2.16
50th: 2.92
75th: 4.38
# of Stations' 2
PM10 (r = 0.42)
NO2 (r = 0.58)
benzene (r =
0.29)
NO (r = 0.47)
O3 (r = -0.24)
Formaldehyde (r
= 0.54)
Toluene (r = 0.48)


FINDINGS
Asthma morbidity was higher in the autumn and winter than the rest of
the yr. The number of ER visits is Sep was exceptionally high.
The percent of total variance showed positive correlation between
asthma ER visits in children and high levels of NOX, SO2, and increased
barometric pressure. NOX enhances the effects of SO2, whereas O3 had
a reverse relation to SO2.
Air borne pollen was not a significant contributor to ER visits.
Correlation between SO2 and ER visits for asthma:
All Yr:

Daily data r = 0.24
Running mean for 7 days r = 0.53
Excluding Sep:
Daily data r = 0.31
Running mean for 7 days r = 0.64

SO2 was significantly associated with respiratory hospital admissions.
This relationship was robust to the inclusion of PM25, but attenuated
when both PM25 and benzene were included in the model.
Increment: SO2: 2.22 pg/m3 (IQR)
Single-pollutant model
Respiratory disease only
1.056(1.013, 1.101)
All disease
0.990(0.974, 1.007)
2-pollutant model with PMio
1.051 (1.005, 1.099)
3-pollutant model with
PMio + Benzene
1 .040 (0.993, 1 .089)

May 2008
F-39
DRAFT—DO NOT QUOTE OR CITE

-------
        STUDY
                                METHODS
                                                    POLLUTANTS
                                                                                              FINDINGS
Hajatetal. (1999)

London, United Kingdom

Period of Study:
1992-1994
GP visits

Outcome(s) (ICD9): Asthma
(493); Lower respiratory
disease (464, 466, 476,
480-3, 490-2, 485-7, 4994-
6,500,503-5,510-5)

Age groups analyzed: 0-14;
15-64; 65+; all ages

Study design: Time-series
analysis

Statistical Analysis: Poisson
regression, APHEA protocol

Covariates: Long-term
trends, seasonality, day of
wk, temperature, humidity

Season:
Warm, Apr-Sep; Cool, Oct-
Mar;AII-yr

Dose-response
investigated? Yes

Statistical package: SAS

Lag: 0-3 days, cumulative
Allyr

24-havg:21.2
pg/m3, SD: 7.8

10th: 13.0

90th: 31.0

Warm:

24-h avg: 20.5
pg/m3, SD: 6.5

10th: 13.4

90th: 28.4

Cool:

24-h avg: 22.0
pg/m3, SD: 9.0

10th: 12.8

90th: 33.3

NO2(r = 0.61)

BS (r = 0.57)

CO (r = 0.51)

PM10(r = 0.63)

03(r = -0
This study showed weak, but consistent associations between SO2 and
consultations for asthma and other LRD, especially in children. Bubble
plot suggests a concentration-response relationship.

Increment: 18 pg/m3
(90th-10th percentile)

Asthma
All ages 3.6% (0.3, 6.9) lag 2; 4.4% (0.9, 7.9)
lag 0-2
0-14 yrs 4.9% (0.1, 9.8) lag 1; 4.4% (-0.7,9.7)
lag 0-2

Warm: 9.0% (2.2, 16.2) lag 1
Cool: 2.0% (4.5, 8.9) lag 1

15-64 yrs 3.6% (-0.6, 8.0) lag 2; 3.5% (-1.0, 8.2) lag 0-3

Warm: 2.5% (-3.3, 8.7) lag 2
Cool: 4.5% (-1.4, 10.7) lag 2

65 + yrs 4.5% (-3.5, 13.1) lag 1; 4.8% (-2.9, 13.2) lag 0-1

Warm: 7.5% (-4.0, 20.3) lag 1
Cool: 2.0% (-8.6, 13.9) lag 1

Lower respiratory disease

All ages 1.8% (0.2, 3.4) lag 2; 2.2% (0.4, 4.1)
lag 0-2
0-14 yrs 4.5% (1.4, 7.8) lag 2; 5.7% (1.7, 9.7)
lag 0-3

Warm: 2.4% (-2.6, 7.7) lag 2
Cool: 5.8% (1.6, 10.2) lag 2

15-64 yrs 1.5% (-0.7, 3.7) lag 1; 1.6% (-0.9, 4.1) lag 0-3

Warm: -0.5% (-3.8, 2.9) lag 1
Cool: 2.5% (-0.5, 5.5) lag 1

65 + -2.2% (-4.9, 0.6) lag 0; -1.4% (-4.4, 1.7)
lag 0-1

Warm: -3.1% (-6.9, 0.9) lag 0
Cool:-1.6% (-5.3, 2.3) lag 0

2-pollutant model -Asthma
SO2 alone 4.9% (0.1, 9.8)
SO2/O35.9% (1.1, 10.9)
SO2/NO22.7%(-2.7,8.4)
SO2/PM2.5 3.4% (-3.0, 10.2)
2-pollutant model-Lower respiratory disease
SO2 alone 4.5% (1.4, 7.8)
SO2/O34.8% (1.6,8.1)
SO2/NO23.1%(-0.6,6.9)
SO2/PM253.8% (0.4,7.2)
May 2008
                                F-40
                                  DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Hajat* et al. (2001)
London, United Kingdom
Period of Study:
1992-1994















Hajat* et al. (2002)
London, United Kingdom
Period of Study:
"1QQO "1QQ4
iyyz- lyy^f

















METHODS
GP visits
Outcome(s) (ICD9): Allergic
Rhinitis (477)
Age groups analyzed:
0-14; 15-64; 65+; all ages
Study design Time-series
analysis
N' 4 214

Statistical Analysis: Poisson
regression, GAM
Covariates: Long-term
trends, seasonality, day of
wk, temperature, humidity,
variation in practice
population, counts for
lagged allergic pollen
measures, daily number of
consultations for influenza
Dose-response
investigated? Yes
Statistical package:
S-Plus
Lag: 0-6 days, cumulative
GP visits
Outcome(s) (ICD9): Upper
respiratory disease,
excluding Rhinitis
(460-3, 465, 470-5, 478)
Age groups analyzed:
0-14; 15-64; 65+; all ages
Study design:
Time-series analysis
Statistical Analysis: Poisson
regression, GAM

Covariates: Long-term
trends, seasonality, day of
wk, holidays, temperature,
humidity, variation in
practice population, counts
for lagged allergic pollen
measures, daily number of
consultations for influenza
Season:
Warm, Apr-Sep;
Cool, Oct-Mar
Dose-response
investigated? Yes
Statistical package:
S-Plus

Lag: 0,1,2,3 days


POLLUTANTS
24-havg:21.2
Mg/m3,
SD: 7.8
10th' 130
90th: 31.0
NO2(r = 0.61)
BS (r = 0.57)
CO (r = 0.51)
PM10(r = 0.63)
O3 (r = -0.11)











Allyr
24-havg:21.2
Mg/m3,
SD: 7.8
10th: 13.0
90th: 31.0

Warm:
24-h avg: 20.5
Mg/m3,
SD: 6.5

10th: 13.4
90th: 28.4
Cool:
24-h avg: 22.0
Mg/m3,
SD: 9.0
10th: 12.8
90th: 33.3
# of Stations: 3
NO2(r = 0.61)
BS (r = 0.57)
CO (r = 0.51)
PMio (r = 0.63)

03(r = -0.11)

FINDINGS
The number of allergic rhinitis admissions peaked in Apr and June. After
2-pollutant model analysis, SO2 still remained highly significant in the
presences of other pollutants. For both children and adults exposure-
response associations showed that risk levels off at higher SO2 levels.
Increment: 18 pg/m3
(90th-10th percentile)
Single-pollutant model
< 1 to 14yrs
24.5% (14.6, 35.2) lag 4
24.9% (11. 9, 39.4) lag 0-4
15 to 64yrs
14.3% (6.2, 23.0) lag 3
15.5% (9.1, 22.3) lag 0-5
>64 yrs-too small for analysis
2-pollutant models
< ^ JQ 1 4 vrs
SO2&O3: 22.1% (12.0, 33.1)
SO2 & NO2: 28.5% (15.5, 42.9)
SO2 & PM10: 27.2% (15.3, 40.2)
15 to 64yrs
SO2 & O3: 8.5% (3.4, 13.9)
SO2&NO2:8.3%(1.7, 15.3)
SO2&PM10:6.7%(0.7, 13.0)
Increased consultations for URD were most strongly associated with
SO2 in children. For adults and the elderly the strongest associations
were for PMio and NO2. The most consistent lag in adults and the elderly
for development of URD was 2 days (one day after a pollution event).
Increment: 18 pg/m3
(90th-10th percentile)
Single-pollutant model
Allyr
0-1 4 yr 3.5% (1.4, 5.8) lag 0
1 5-64 yrs 3.5% (0.5, 6.5) lag 1
>65 yrs 4.6% (0.4, 9.0) lag 2


0-14 yrs 3.2% (-0.5, 7.0) lag 0
15-64 yrs 4.6% (1.5, 7.7) lag 1
> 65 yrs 1 .6% (-4.8, 8.5) lag 2 Cool
0-14 yrs 5.5% (2.4, 8.7) lag 0
15-64 yrs 2.7 (0.0, 5.4) lag 1
>65 yrs 5.7% (0.4, 11. 4) lag 2
2-pollutant models
0-1 4 yrs
SO2&O3: 1.0% (-2.2, 4.2)
SO2&NO2:4.7%(2.2,7.4)
SO2&PM10:4.6%(2.1,7.2)
For 15-64 yrs
SO2 & O3: 3.7% (0.6, 7.0)
SO2 & NO2: 2.6% (-0.0, 5.2)
SO2&PM10:2.4%(-0.1,5.0)
For >65 yrs
SO2&O3:9.0%(1.7, 16.9)
SO2&NO2:4.3% (-1.2, 10.2)
SO2&PM10:3.2%(-1.9, 8.7)
May 2008
F-41
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Llorca et al. (2005)
Torrelavega, Spain
Period of Study:
1992-1995
Days: 1,461









Oftedal et al. (2003)
Drammen, Norway
Period of Study:
1994-2000










Ponce de Leon et al.
(1996)
London, England
Period of Study:
04/1987-1988'
1991-02/1992'














METHODS
Hospital Admissions
Outcome(s)(ICD9):AII
respiratory admissions
(460-519)
Age groups analyzed: All
ages
Study design Time-series
Number of hospitals: 1
Statistical analyses: Poisson
regression
Covariates: Short and Long-
term trends
Statistical package: Stata
Lag: NR


Hospital Admissions
Outcomes (ICD 10): All
respiratory admissions (JOO-
J99)

Age groups analyzed: All
ages
Study design: Time-series
Statistical analyses: Semi-
parametric Poisson
regression, GAM with more
stringent criteria
Covariates: Temperature,
humidity, influenza
Lag: 2,3 days
Hospital Admissions
Outcome(s) (ICD 9): All
respiratory (460-519)
Age groups analyzed: 0-14,
15-64, 65+, all ages
Study design: Time-series
N: 19,901
Statistical analyses: APHEA
protocol, Poisson
regression GAM
Covariates: Long-term
trend, season, influenza,
day of wk, holiday,
temperature, humidity
Season'
Cool, Oct-Mar; Warm: Apr-
Sep

Dose-response
Investigated?: Yes
Statistical package: SAS
Lag: 0, 1, 2 days, 0-3
cumulative avg.
POLLUTANTS
24-h avg SO2:
13.3pg/m3,
SD: 16.7
# of Stations: 3
NO2 (r = 0.588)
NO (r = 0.544)
TSP(r = -0.40)
SH2 (r = 0.957)







Mean: 2.9 pg/m3,
SD:2.1
IQR: 2.03 pg/m3
PM.n
nvi^o
NO2

03
Benzene
Formaldehyde





SO224-h avg:
32.2 pg/m3,
SD: 12.6

5th: 15
10th: 18
25th: 24
50th: 31
75th: 39
90th: 47
95th: 54

# of stations: 2
NO2 (r = 0.44)
BS (r = 0.44)
O3 (r = -0.067)






FINDINGS
Associations between SO2 and admissions observed in the Single-
pollutant model disappear in a 5-pollutant model. Only NO2 was
significantly associated with admissions.
No relation was described for sulphur compounds including H2S or SO2.
The concentration of SO2 changes with temperature changes, which
may be responsible for cardiac stress.
SO2 was not significantly associated with cardiac respiratory or cardio-
respiratory admissions
Increment: 100 pg/m3
Single-pollutant model
All cardio-respiratory admissions: RR 0.98 (0.89, 1.07)
Respiratory admissions: 1.04
(0.90, 1.19)
5-pollutant model
All cardio-respiratory admissions: RR 0.98 (0.80, 1.21)
Respiratory admissions: 0.89 (0.64, 1.24)
The study found positive associations between daily number of hospital
admissions for acute respiratory diseases and concentrations of SO2;
associations did not change substantially from the first to the second
3-yr period.

Increment: 2.03 pg/m (IQR)
All respiratory disease
1.042(1.011, 1.073)







Though significant effects were observed with SO2 in some age groups,
they were not consistent or similar in magnitude to those of O3.
Increment: 90th-10th percentile
(24-h avg: 29 Mg/m3).
Allyr
All ages 1.0092 (0.9926, 1.0261) lag 1
0-1 4 yrs 1.0093 (0.9837, 1.0356)lag1
1 5-64 yr 1.0223 (0.9942, 1.0511) lag 1
> 65 yr 1 .0221 (0.9970, 1 .0478) lag 2
\Afarm S6cison
All ages 1.0111 (0.9864, 1.0364) lag 1
0-14 yrs 1 .0468 (1 .0066, 1 .0885) lag 1
15-64 yr 0.9996 (0.9596, 1.0411) lag 1
>65yr 1.0124 (0.9772, 1.0489)
lag 2
Cool season
All ages 1 .0079 (0.9857, 1 .0306) lag 1
0-14 yrs 0.9848 (0.951 5, 1 .01 92) lag 1
15-64 yr 1.0389 (1.0010, 1.0783) lag 1
>65 yr 1 .0280 (0.9945, 1 .0625)
lag 2




May 2008
F-42
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Ponka (1991)
Helsinki, Finland
Period of Study:
1987-1989




















Ponka and Virtanen (1994)
Helsinki, Finland
Period of Study:
1987-1989
Days: 1096










Ponka and Virtanen (1996)
Helsinki, Finland
Period of Study:
1987-1989







METHODS
Hospital Admissions
Outcome(s) (ICD9):
Asthma (493)
Age groups analyzed: 0-14;
15-64; > 65yrs
Study design: Time-series
N: 4,209
Statistical analyses:
Correlations and partial
correlations
Covariates: Min
temperature
Statistical package:
Lag: 0-1












Hospital Admissions
Outcome(s) (I CD 9):
Chronic bronchitis and
emphysema (493)
Age groups analyzed: < 65,
>65
Study design: Time-series
Statistical analyses: Poisson
regression
Covariates: Season, day of
wk, yr, influenza, humidity,
temperature
Season: Summer (Jun-
Auol
"uyy,
Autumn (Sep-Nov),
Winter (Dec-Feb),
Spring (Mar-May)
Lag: 0-7 days

Hospital Admissions
Outcome(s) (ICD9):
Asthma (493)
Age groups analyzed: 0-14,
15-64,65+
Study design: Time-series
Statistical analyses:
Covariates: Long-term
trend, season, epidemics,
day of wk, holidays,
temperature, relative
humidity
Statistical package:
Lag: 0-2
POLLUTANTS
24-havg: 19.2
(12.6)pg/m3
Range: 0.2, 94.6
Number of
monitors: 4
NO2(r = 0.4516)
NO (r = 0.4773)
O3(r = 0.1778)
TSP(r = 0.1919)
CO















24-h Mean:
19pg/m3, SD:
12.6;
Range: 0.2, 95
# of stations: 2
NO2
03
TSP








24-h avg (pg/m3):
Winter: 26
Spring: 22
Summer: 13
Fall' 15
NO2

03
TSP




FINDINGS
The frequency of all admissions for asthma was significantly correlated
to SO2.
Child asthma admissions were not significantly correlated with SO2, but
were correlated to O3 and NO. SO2 was also significantly correlated with
elderly admissions. Increased hospitalization correlated with SO2 was
also observed for adults.
Hospital admissions were more strongly correlated with SO2 than other
pollutants. ER visits were more strongly correlated with a mixture of
pollutants (TSP, SO2, O3, and temperature).
Multipollutant model
co-linear results of SO2, CO, NO2, and NO suggest a mixture of
pollutants is responsible for asthma admissions.
Correlations between hospital admissions (HA) for asthma and
pollutants and temperature by ages.
0-14yrs
HA: -0.01 391
Emergency HA: 0.0332
15-64yrs
HA: 0.1 039 p = 0.0006
Emergency HA: 0.1199 p < 0.0001
> 65 yrs
HA: 0.0796 p = 0.0085
Emergency HA: 0.1169 p < 0.0001
Partial correlations between admissions for asthma and SO2 were
standardized for temperature.
HA: 0.0770 p = 0.01 72
Emergency HA: 0.1050
p = 0.0011
SO2 was significantly associated with increased admissions for chronic
bronchitis and emphysema for patients < 65 yrs of age with a lag of 0
and 3 days.
In the steps leading to regression analysis no association was observed
between SO2 levels and the >65 population. Multipollutant models were
only used to examine NO2 and SO2.
SO2 had no significant association with morbidity caused by chronic
bronchitis and emphysema in the > 65 yr old population.
Increment: NR

Chronic bronchitis and emphysema
< 65 yrs
RR1.31 (1.01, 1.70) lag 0
RR 0.96 (0.73, 1.27) lag 1
RR 0.78 (0.59, 1.03) lag 2
RR 1 .39 (1 .05, 1 .86) lag 3
RR 0.89 (0.68, 1.16) lag 4
RR 1.28 (0.97, 1.70) lag 5
RR0.91 (0.69, 1.20) lag 6
RR 1 .09 (0.84, 1 .40) lag 7
65+ yrs: NR
Significant associations were observed between daily SO2
concentrations and daily counts of hospitalizations among
15- to 64-yr-old patients and among those over 64 yrs old, but not
among children. These effects were observed when mean daily SO2
values were lower than the max value recommended by WHO (1 25
pg/m3).
Parameter estimates (PE) and standard erro (SE) for a 1-unit increase:
Asthma 15-64 yrs :
PE 0.2176 (0.1081) p = 0.44 lag 2
PE 0.3086 (0.1545) p = 0.046 lag 0-3
Asthma 65+ yrs !
PE 0.2412 (0.0956) p = 0.012 lag 2



May 2008
F-43
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Prescott et al. (1998)
Edinburgh, United Kingdom
Period of Study: 10/92-6/95












Rossi etal. (1993)
Oulu, Finland
Period of Study: 10/1/1985-
9/30/1986













METHODS
Hospital Admissions
Outcome(s) (I CD 9):
Pneumonia
(480-7), COPD + Asthma
(490-496)

Age groups analyzed: < 65,
65+
Study design: Time-series
Statistical analyses: Poisson
log linear regression
Covariates: Trend, seasonal
and weekly variation,
temperature, wind speed,
day of wk
Lag: 0,1 , or 3 day rolling
avg
ED Visits
Outcome(s) (ICD 9):
Asthma (493)
Age groups analyzed: 15-85
Study design: Time-series
N:232
Statistical analyses:
Pearson's and partial
correlation coefficients and
multiple regression with
stepwise discriminate
analysis
Covariates: Temperature,
humidity

Statistical package: BMDP
software
Lag: 0,1, 2,3
POLLUTANTS
SO2: 14.5(9.0)
ppb
MiN: 0 ppb
Max: 153 ppb
# of Stations: 1

CO
PM10
NO2
03
DO
Do




24-h Mean:
10.0pg/m3
Range: 0, 56
1-h max:
31.0pg/m3
Range: 1, 24
# of monitoring
stations: 4
NO2; r = 0.48
JSP; r = 0.31
H2S







FINDINGS
No effect of SO2 on hospitalizations observed in either age category.
Increment: 10 ppb
Respiratory admissions
>65 yrs -2.5 (-11.0, 6.9) lag 0-2

< 65 yrs 0.0 (-8.3, 9.1) lag 0-2










Same day ER visits were correlated to daily SO2 levels, but the
significance was lost with longer lag periods.
When asthma visits were analyzed, SO2 was positively and significantly
correlated with asthma visits in the same wk and the wk after.
After regression analyses, SO2 became insignificant.
Pearson correlation coefficients
ED asthma visits and same day SO2:
r = 0.13 p < 0.01 lag 0
Weekly ED asthma visits and same wk SO2: r = 0.28 p < 0.05
Weekly ED asthma visits and previous wk SO2: 0.30 p < 0.05
Multipollutant (NO2; TSP; H2S)
Regression coefficient:
All yr: 3 = 0.037, p = 0.535
Winter: 3 = -0.024, p = 0.710

Summer: 3 = -0.003, p = 0.991



May 2008
F-44
DRAFT—DO NOT QUOTE OR CITE

-------
         STUDY
                                 METHODS
                                                    POLLUTANTS
                                                                                               FINDINGS
Schouten et al.

(1996)

Multicity, The Netherlands
(Amsterdam, Rotterdam)

Period of Study:
04/01/77-09/30/89
Hospital Admissions

Outcome(s)(ICD9):AII
respiratory (460-519),
COPD (490-2, 494, 496),
Asthma  (493)

Age groups analyzed:
15-64, 65+, all ages

Study design: Time-series

Statistical analyses: APHEA
protocol, Poisson
regression

Covariates: Long-term
trend, season, influenza,
day of wk, holiday,
temperature, humidity

Season:
Cool,  Nov-Apr;
Warm: May-Oct

Lag: 0,1,2 days; and
cumulative 0-1 and 0-3 day
lags
24-h avg SO2
Amsterdam
Mean/Med:
28/21 pg/m3
Rotterdam
Mean: 40/32
pg/m3
Daily 1 -h max
Amsterdam
Mean/Med:
65/50 pg/m3
Rotterdam
Mean/Med:
99/82 pg/m3
# of stations: 1
per city
NO2
BS
03
May 2008
                                F-45
The relationship between short-term air pollution and hospital
admissions was not always consistent at low levels of exposure. One
statistically significant association between hospital admissions and
asthma (all ages) occurred in Amsterdam after a cumulative lag of
1-3 days in the summer. Higher SO2 levels were reported for the winter;
therefore, this association was not a concentration response.

In Rotterdam neither
1 day nor cumulative lags in the summer or winter increased asthma
admissions to statistical significance. Rotterdam had much higher mean
SO2 concentrations. There were no significant associations to hospital
admissions when higher pollution levels were prevalent.

The analysis of all respiratory hospital admissions for all ages in the
entire country (Netherlands) produced a statistically significant
association for both 1-h and 24-h  periods (100 pg/m3).

Increment: 100 pg/m3 increment.

All respiratory, Amsterdam

24-h avg
15-64yrs
RR 0.944 (0.864, 1.032) lag 2
RR 0.915 (0.809, 1.035) lag 0-3
>65 yrs
RR 1.046(0.965, 1.134) lag 2
RR 1.008 (0.899, 1.131) lag 0-3
1-h max
15-64 yrs
RR 0.989(0.952, 1028) lag 2
RR 0.977 (0.927, 1.030) lag 0-3
>65 yrs
RR 1.022 (0.985, 1060) lag 2
RR 1.010 (0.955, 1.068) lag 0-3
RR 0.941 (0.863, 1.026) lag 0-3
COPD, Amsterdam
24-h avg-all ages
RR 0.907 (0.814, 1.011) lag 0
RR 0.948 (0.838, 1.072) lag 0-1
1-h max-all ages
RR 0.978 (0.933, 1.026) lag 0
RR 0.995 (0.940, 1.053) lag 0-1

Asthma, Amsterdam
24-h avg-all ages
RR 0.802 (0.696, 0.924) lag 1
RR 0.792 (0.654, 0.958) lag 0-3
1-h max-all ages
RR 0.995 (0.942, 1.051) lag 0

All respiratory, Rotterdam
24-h avg
15-64 yrs
RR 0.941 (0.855, 1.036) lag 1
RR 0.895 (0.787, 1.019) lag 0-2
>65 yrs 1977-1981
RR 1.027(0.904, 1.165) lag 2
RR 1.011  (0.834, 1.227) lag 0-3
>65 yrs 1982-1984
RR 1.087 (0.890, 1.328) lag 0
RR 1.258 (0.926, 1.710) lag 0-3
>65yrs 1985-1989
RR 1.045 (0.908, 1.204) lag 0
RR 0.968 (0.787, 1.190) lag 0-3
1-h max
15-64 yrs
RR 0.989(0.953, 1025) lag 1
RR 0.965 (0.915, 1.018) lag 0-2
>65 yrs 1977-1981
RR 0.892 (0.842, 0.945) lag 0
RR 0.987 (0.907, 1.074) lag 0-3
>65 yrs 1982-1984
RR 1.005(0.933, 1.081) lag 0
RR 1.062 (0.938, 1.202) lag 0-3
>65yrs 1985-1989
RR 1.010(0.955, 1.068) lag 0
RR 1.064 (0.992, 1.141) lag 0-1

COPD, Rotterdam
24-h avg-all ages
RR 0.963 (0.874, 1.059) lag 2
RR 1.019(0.887, 1.172) lag 0-3
1-h max-all ages
RR 0.991 (0.955, 1.029) lag 2
RR 1.013 (0.953, 1.076) lag 0-3 All respiratory, Rotterdam
24-h avg          DRAFT—DO NOT QUOTE  OR CITE
15-64 yrs
RR 0.941 (0.855, 1.036) lag 1
RR 0.895 (0.787, 1.019) lag 0-2
>65 yrs 1977-1981
DD -\ 007 /n artA 1 IRC.\ io« o

-------
STUDY
Spixetal. (1998)
Multicity (London,
Amsterdam, Rotterdam,
Paris, Milan), Europe
Period of Study: 1977 and
1991












Sunyeretal. (1997)
Multicity, Europe
(Barcelona, Helsinki, Paris,
London)
Period of Study:
1986-1992














Sunyeretal. (2003)
Multicity study (Birmingham
(B), London (L), Milan (M),
Netherlands (N), Paris (P),
Rome (R) and Stockholm
(S), Europe
Period of Study:
1992 and 1997







METHODS
Hospital Admissions
Outcome(s) (ICD9): All
respiratory (460-519);
Asthma (493)
Age groups analyzed:
15-64, 65+
Study design: Time-series
Statistical analyses: Poisson
regression following APHEA
protocol. Pooled meta-
analysis adjusted for
heterogeneity
Covariates: trend,
seasonality, day of wk,
holiday, temperature,
humidity, unusual events
(strikes, etc.)
Lag: 1 to 3 days
Hospital admissions/ED
Visits
Outcome(s) (ICD 9):
Asthma (493)
Age groups analyzed: < 15,
15-64
Study design: Time-series

Statistical analyses: APHEA
protocol, Poisson
regression, GEE; meta-
'
Covariates: Humidity,
temperature, influenza,
soybean, Long-term trend,
season, day of wk
Season'

Cool, Oct-Mar;
Warm: Apr-Sep
Lag: 0,1,2,3 and cumulative
1-3
Hospital admissions/ED
Visits
Outcome(s) (ICD 9):
Asthma (493); COPD and
Asthma
(490-496);
All respiratory (460-519)
Age groups analyzed: All, 0-
14 yrs; 16-64 yrs; > 65 yrs
Study design Time-series
Poisson regression with
GAM following APHEA 2
"

Covariates: temperature,
humidity, Long-term trend,
season
Lag:0, 1
POLLUTANTS
SO2 daily mean
(Mg/m3)
LondoN: 29
Amsterdam: 21
Rotterdam: 25

Paris: 23
MilaN' 66

NO2, O3, BS, TSP









24-h median
(range) (pg/m3)
Barcelona: 41
(2, 160)
Helsinki: 16
(3, 95)
LondoN: 31
(9, 100)

Paris: 23
(1,219)
# of stations:
Barcelona* 3

LondoN: 4
Paris' 4

Helsinki: 8
NO2
black smoke
03
SO2 24-h avg and
SD (Mg/m3)
624.3(12.7)
L 23.6 (23.7)
M 32 .5 (37.5)
N 8.5 (7.7)
P 17.7 (12. 5)
R 9 8 (9 9)

S 6.8 (6.2)
PM10(r = 0.64)
CO (r = 0.53)




FINDINGS
Daily counts of adult respiratory admissions were not consistently
associated with daily mean levels of SO2. Heterogeneity between cities
was likely due to the number of stations or temperature. Only hospital
admissions for > 65 yr olds were significantly associated with SO2 in the
warm season.
Increment: 50 pg/m3

All cities, yr round
15-64 yrs RR 1.009 (0.992, 1.025)
Warm RR 1.01 (0.98, 1.04)
Cold RR1.01 (0.97, 1.07)
> 65 yrs RR 1 .02 (1 .005, 1 .046)
WarmRR 1.06(1.01, 1.11)
Cold RR 1 .02 (0.99, 1 .04)
APHEA protocol pooled result from >65 yrs old from Europe
All respiratory
RR 1.02 (1.00, 1.05)



SO2 alone or as part of a mixture was a factor that exacerbated asthma
admissions.
In 2-pollutant models with SO2 and BS, the association of BS with SO2
was attenuated for < 15 yr olds, compared to single-pollutant model
associations. In addition, the association of NO2was also attenuated by
the inclusion of SO2.
Increment: 50 pg/m3 of 24-h avg for all cities combined.
Asthma
15-64 vrs
0.997(0.961, 1.034) lag 2
1.003 (0.959, 1.050) lag 0-3, cum
< 1 5 yrs
1.075(1.026, 1.1 26) lag 1
1.061 (0.996, 1.131) lag 2-3, cum
2-pollutant models:
SO2/Black smoke
< 15 yrs 1.092 (1.031, 1.156) lag 0-1
SO2/NO2
< 15yrs 1.075 (1.019, 1.135)


The magnitude of association with asthma across the seven cities was
comparable to earlier studies of London, Helsinki and Paris.
Exposure factors may be important. Children may spend greater time
outdoors compared with adults. Pneumonia requires chronic exposure to
produce inflammatory response and infection, whereas asthma is an
acute response.
Increment: 10 pg/m3
Asthma
0-14 yrs 1.3% (0.4, 2.2)
15-64 yrs 0.0% (-0.9, 1.00)
COPD and Asthma
> 65 yrs 0.6% (0.0, 1 .2)
All Respiratory
> 65 yrs 0.5% (0.1,0.9)
Asthma
0-1 4 yrs
SO2 + PM10:-3.7%(p>0.1)
SO2 + CO: -0.7% (p > 0.1)
May 2008
F-46
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Sunyeretal. (1991)
Barcelona, Spain
Period of Study:
1985-1986











Sunyeretal. (1993)
Barcelona, Spain
Period of Study:
1985-1989









Teniasetal. (1998)
Valencia, Spain
Period of Study:
1993-1995

Seasons:
Cold: Nov-Apr
Warm: May-Oct



















METHODS
ED Visits
Outcome(s) COPD
(ICD 9): 490-496
Age groups analyzed: > 14
Study design:
Time-series
# of Hospitals: 4

Statistical analyses:
multivariate linear

regression
Covariates: Meteorology,
season, day of wk
Statistical package:
Lag: Oto 2 days
ED Visits
Outcome(s) (ICD 9): COPD
(490-492;
494-496)

Study design:
"

Statistical analyses:
Autoregressive linear
regression
Statistical package:
Lag: 1,2

ED Visits
Outcome(s): Asthma
ICD 9 Code(s): NR
Age groups analyzed: > 14
Study design: Time-series
N: 734
Statistical analyses: Poisson
regression, APHEA protocol

Covariates: seasonality,
temperature, humidity, long-
term trend, day of wk,
holidays, influenza
Season: Cold: Nov-Apr;
Warm: May-Oct
Dose-response
investigated: Yes
Lag: 0-3 days









POLLUTANTS
24-h avg (SD):
56.5 (22.5) pg/m3
98th: 114.3
Range: 17, 160
1-h max (SD):
141.9(98.8)
pg/m3
98th' 461 3

Range: 17, 160

Number of
monitors: 14-720
BS, CO, NO2, O3



SO2, 24-h
Winter Tertiles
UJQ )
<40.4
40.4, 61


Winter Tertiles
(pg/m3)
<28.1
28.1,46.1
>46.1

BS
24 h avg: 26.6
pg/m3
25th: 17.9
50th: 26.2
75th: 34.3
95th: 42.6
Cold' 31 7
Warm: 21. 7
1-h max:
56.3 ug/m3

25th: 36.3
50th' 52 2
75th' 72 2
95th: 95.2
Cold' 64 6
Warm: 48.2
# of Stations' 2

24 h avg:
O3(r = -0.431)
NO2 (24 h av) (r =
0.265)
NO2(1-h)(r =
0.199)
1-h:
03 (r = -0.304)
NO2 (24 h avg) (r
= 0.261)
NO2(1-h)(r =
0.201)
FINDINGS
An incremental change of 25 pg/m3 in SO2 was correlated with an
adjusted increase of 0.5 daily visits due to COPD.
SO2 and ER visits were more strongly correlated in warm weather.
Even at 24-h avg levels less than
100 pg/m3, effects of SO2 were statistically significant for COPD
admissions.
Change in 24-h SO2 daily ER pg/m3 admissions P-value
150 0 55 < 0 01
100 0 7 < 0 01

72 0.7 0.04
52 0.41 > 0.05
39 -1. 27 > 0.05
0.5 excess daily admissions per 25 pg/m3 increment of SO2.


SO2 concentrations were associated with the number of COPD ER
admissions in the winter and summer. An increase of 25 pg/m3 in SO2
produced an adjusted change of ~6% and 9%, respectively, in the
number of COPD emergencies in the winter and summer. Controlling for
particulate matter resulted in a loss of significance. Co linearity of BS
with SO2 was observed.

Effects were expressed as adjusted changes in daily COPD ER
admissions based on an increment of 25 pg/m3.
Winter: 6%
Summer: 9%
Mean ER admissions for COPD (winter) were 15.8 (range 3, 34) and 8.3
(range 1 , 24) in the summer.

SO2 showed the strongest correlation to asthma admissions during the
warm mos.
Multipollutant models showed that O3 and black smoke had a small
effect on the association between SO2 and asthma ER visits while NO2
greatly depressed these effects. It is likely that NO2 was the dominant
pollutant for respiratory outcomes. SO2 was the "most vulnerable
pollutant" to the presence of other pollutants.
Increment: 10 pg/m3
SO2 24-h avg
All yr 1 050
(0.973, 1. 133) lag 0
Cold 1 .032
(0.937, 1. 138) lag 0
Warm 1 .070
(0.936, 1 .224) lag 0
SO2 1-h max
All yr 1.027 (0.998, 1.057) lag 0
Cold 1.018(0.980, 1.057) lag 0
Warm 1 .038 (0.990, 1 .090) lag 0








May 2008
F-47
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Tenias et al. (2002)
Valencia, Spain
Period of Study:
1994-1995













Thompson et al. (2001)
Belfast, Northern Ireland
Period of Study:
1993-1995














Tobias etal. (1999)
Barcelona, Spain
Period of Study:
1986-1989








METHODS
ED Visits
Outcome(s): COPD
ICD 9 Code(s): NR
Age groups analyzed: > 14
Study design: Time-series
N: 1,298
# of Hospitals: 1
Statistical analyses: Poisson
regression, APHEA
protocol; basal models and
GAM

Covariates: Seasonally,
annual cycles, temperature,
humidity, day of wk, feast
days
Season:
Cold, Nov-Apr;
Warm, May-Oct
Dose-response
investigated: Yes
Lag: 0-3 days
Hospital admissions/ED
Visits
Outcome(s): Asthma
ICD 9 Code(s): NR
Age groups analyzed:
Children
Study design: Time-series
N: 1,044
Statistical analyses:
Followed APHEA protocol,
Poisson regression analysis
Covariates: Season, long-
term trend, temperature,
day of wk, holiday
Season:
Warm (May-Oct); Cold
(Nov-Apr)
Statistical package: Stata
Lag: 0-3
ED Visits
Outcome(s): Asthma
ICD9: NR
Age groups analyzed: > 14
Study design:
Time-series
Statistical analyses: Poisson
regression, followed APHEA
protocol

Covariates: Temperature,
humidity, long-term trend,
season, day of wk
Lag: NR
POLLUTANTS
24 h avg: 26.6
pg/m3
25th: 17.9
50th: 26.2
75th: 34.3

95th: 42.6
Cold: 31 .7
Warm: 21. 7
1-h max: 56.3
pg/m3
25th: 36.3
50th: 52.2
75th: 72.2
95th: 95.2
Cold: 64.6
Warm: 48.2
BS (r = 0.687)
NO2(r = 0.194)
CO (r = 0.734)
O3(r = -0.431)
Warm Season
SO2 (ppb):
Mean: 12.60;
SD' 10 60'
IQR: 6.0, 16.0
Cold Sscison

SO2 (ppb):
Mean: 20.40;
SD: 17.90;
IQR: 11.0,24.0
PM10(r = 0.66)
NO2 (r = 0.82)
MOv (r — fl ftT\
IN^JX ^1 — U.OOJ
NO (r = 0.76)
03 (r = -0.58)
CO (r = 0.64)

Benzene (r =
0.80)
24-h avg SO2
pg/m3
Non-epidemic
Days: 85.8 (62.4)
Epidemic Days:
116.3(79.3)
BS
NO2
O3





FINDINGS
SO2 did not show any significant association with COPD ER visits for all
seasons analyzed.
SO2 did not affect O3 or CO association to ER admission for COPD
when assessed together in the Multipollutant model.
Possibility of a linear relationship between pollution and risk of
emergency cases could not be ruled out.
Increment: 10 pg/m3.
24-h avg SO2
All yrRR 0.971
(0.914, 1.031) lag 0
Cold, 24-h avg: RR 0.970
(0.905, 1 .038) lag 0
Warm, 24-h avg: RR 0.982
(0.885, 1 .090) lag 0
1-h max SO2
All yr RR 0.981 (0.958, 1 .027) lag 3
Cold, 24-h avg: RR 0.972
(0.945, 1 .000) lag 3
Warm, 24-h avg: RR 1.003
(0.979, 1 .056) lag 3
This study found weak, positive associations for SO2 and adverse
respiratory outcomes in asthmatic children.
SO2 Increment: Per doubling (ppb)
LagORR 1.07(1.03, 1.11)
Lag 0-1 RR 1 .09
(1.04, 1.15)
Lag 0-2 RR 1 .08
(1.02, 1.15)
Lag 0-3 RR 1 .08
(1 01 115)

Warm only Lag 0-1 RR 1.11
(1 04 119)

Cold only Lag 0-1 RR 1 .07
(1.00, 1.15)
Adjusted for Benzene Lag 0-1 RR 0.99 (0.90, 1 .09)



The study failed to find a significant association between SO2 and
asthma ED visits.
3 x 104 (SE D 104 ) using Std Poisson
Without modeling asthma epidemics: 3.99 (4.14)
Modeling epidemics with 1 dummy variable: 1.64 (2.76)
Modeling epidemics with 6 dummy variables: 1 .53 (2.75)
Modeling each epidemic with dummy variable: 2.20 (2.65)
3 D 104 (SE D 104) using Autoregressive Poisson
Without modeling asthma epidemics: 6.99 (14.37)
Modeling epidemics with 1 dummy variable: 1.68 (2.77)
Modeling epidemics with 6 dummy variables: 1 .72 (2.75)
Modeling each epidemic with dummy variable: 2.85 (2.89)



May 2008
F-48
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Vigottietal. (1996)
Milan, Italy
Period of Study:
1Qftn 1QftQ
I you- I sos



















Walters etal. (1994)
Birmingham, United
Kingdom
Period of Study:
1988-1990








METHODS
Hospital Admissions
Outcomes
(ICD 9 codes): Respiratory
disease
(460-519).
Age groups analyzed:
15-64yrs and
>64 yrs
Study design: Time-series
N: >73,000

Statistical analyses: APHEA
protocol
Covariates'

Season: Cold season (Oct.
to Mar) and Warm season
(Apr to Sep)
Lag: 0, cumulative 4 day
(0-3)




Hospital Admissions
Outcome(s) (ICD9):
Asthma (493) and acute
respiratory conditions
(466, 480-486, 490-496)
Study design Time-series
Statistical analyses:
Least squares regression
Covariates: Temperature,
pressure, humidity
Lag: 3 day moving avg.



POLLUTANTS
24-h avg:
117.7pg/m3
Range: 3.0, 827.8
5th: 15.0
25th: 34.0
50th: 65.5
75th: 162.5
95th' 376 3
Winter:248.6
Range: 30.6,
827.8
5th: 78.8
25th: 138.5
50th: 216.0
75th: 327.8
95th: 527.0
Summer:30.5
Range: 3.0, 113.8
5th: 9.1
25th: 18.5
50th: 27.8
75th: 39.2
95th: 62.7
# of monitors: 4;
r= 0.89, 0.91
TSP(r = 0.63)
SO2 24-h mean
(Mg/m3)
All yr: 39.06
Max: 126.3
Spring: 42.9
Summer: 37.8
AutumN: 40.9
Winter: 34.2
BS




FINDINGS
The effect of single day or cumulative day exposure to SO2 was more
pronounced during the cool mos. Interaction between seasons was not
significant. SO2 did not interact with TSP. No differences were noted
between age groups.
There were increased, but not significant (borderline), risks for increased
hospital admissions based on an increment change in SO2 of 125 pg/m3
in the winter.

Increment: 100 pg/m3
All respiratory
15-64 yrs
All yr round:
RR 1.05 (1.00, 1.10)lagO
Warm:
RR 1.04 (0.98, 1.11) lag 0
Cool:
RR 1.06 (1.00, 1.13)lagO
>64 yrs
All yr:
RR 1 .04 (1 .00, 1 .09) lag 0
Warm:
RR 1 .02 (0.96, 1 .08) lag 0
Cool:
RR 1.05 (1.00, 1.11) lag 0


In 2-pollutant models BS remained significant but SO2 was no longer
associated significantly with admission.
A 100 pg/m3 increment in SO2 might result in four (0-7) more asthma
admissions and 15.5
(6-25) move respiratory admissions/day. Spring and autumn did not
show associations with admissions for asthma or respiratory.
Increment of 100 pg/m3
Asthma
Summer: 1.4% (-10, 39) lag 0
Winter: 2.7% (-0.8, 6.1) lag 0
All respiratory
Summer: 5.9% (1.1, 10.6) lag 0
(p < 0.02)
Winter: 18% (8.8, 26.8) lag 0
(p < 0.0002)
LATIN AM ERICA
Braga'etal. (1999)
Sao Paulo, Brazil
Period of Study: 10/1992-
10/1993


















Hospital Admissions
Outcome(s) (ICD9): All
respiratory (466, 480-
486,491-492,496)
Age groups analyzed: <
13 yrs
Study design Time-series
N. CO QH O
. Do,:? I o
# of Hospitals:
112 Statistical analyses:
Multiple linear regression
models (least squares). Also
used Poisson regression
techniques. GLM and GAM
using LOESS for
smoothing.
Covariates: Season,
temperature, humidity, day
of wk,
Statistical package: SPSS,
S-Plus
Lag: 1 ,2,3,4,5,6,7 moving
avgs
24-h avg 22 .40
(9.90) pg/m3
MiN: 6.4
Max: 69.6
# of monitors: 13
PM10(r = 0.73)
CO (r = 0.62)

NO2 (r = 0.53)
03












SO2 did not show a correlation with respiratory hospital admissions with
any lag structure.
Increment: 22.4 pg/m3
0.1 2 (-0.04, 0.28) lag 0
0.1 8 (-0.00, 0.37) lag 0-1
0.1 9 (-0.01, 0.39) lag 0-2
0.1 8 (-0.04, 0.40) lag 0-3

0.1 8 (-0.05, 0.42) lag 0-4
0.12 (-0.13, 0.36) lag 0-5
0.08 (-0.18, 0.35) lag 0-6











May 2008
F-49
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Braga* et al. (2001)
Sao Paulo, Brazil
Period of Study:
1/93-11/97












Farhat* et al. (2005)
Sao Paulo, Brazil
Period of Study:
1996-1997














Gouveia and Fletcher
(2000)
Sao Paulo, Brazil
Period of Study:
11/92-9/94














METHODS
Hospital Admissions
Outcome(s)(ICD9):AII
respiratory admissions
(460-519)
Age groups analyzed: 0-19,
Z 2 3-5 6-13 14-19

Study design: Time-series
Statistical analyses: Poisson
regression with GAM

Covariates: Long-term
trend, season, temperature,
relative humidity, day of wk,
holiday
Statistical package: S-Plus
4.5
Lag: 0-6 moving avg
Hospital Admissions/ED
Visits
Outcome(s) (ICD9): Lower
Respiratory Disease (466,
480-5)
Age groups analyzed: < 13

Study design: Time-series
N' 4 534
# of Hospitals: 1
Statistical analyses:
1) Poisson regression and
2) GAM - no mention of
more stringent criteria
Covariates: Long-term
trends, seasonality,
temperature, humidity
Statistical package:
S-Plus

Lag: 0-7 days, 2,3,4 day
moving avg

Hospital Admissions
Outcome(s)(ICD9):AII
respiratory; Pneumonia
(480-486); asthma or
bronchitis (466, 490, 491,
493)

Age groups analyzed:
< 1 ; < 5 yrs
Study design: Time-series
Statistical analyses: Poisson
regression Covariates:
Long-term trend, season,
temperature, relative
humidity, day of wk, holiday,
strikes in public transport or
health services
Season:
Cool (May-Oct),
Warm (Nov-Apr)
Statistical package: SAS
Lag: 0, 1, 2 days
POLLUTANTS
SO2 Mean:
21.4pg/m3;
SD: 11.2
IQR: 14.4 pg/m3
Range: 1.6, 76.1

# of stations: 5-6
PMio (r = 0.61)
NO2 (r = 0.54)
CO (r = 0.47)

03(r = 0.17)




24-h avg:
Mean: 23.7 pg/m3
SD: 10.0
Range: 3.4, 75.2
IQR: 12.5
# of Stations: 6

PM10(r = 0.69)
NO2 (r = 0.66)
CO (r = 0.49)
O3 (r = 0.28)








24-h avg:
Mean: 18.3 pg/m3
SD: 9.0
Range: 3.2, 61.1
5th: 7.6

25th: 11.9
50th: 16.6
75th: 22.2
95th' 35 8

# of stations: 4
PMio (r = 0.72)

NO2 (r = 0.37)
CO (r = 0.65)
03 (r = 0.08)


FINDINGS
Children < 2 yrs were most susceptible to the effect of each pollutant.
Pneumonia and bronchopneumonia were the main cause of hospital
admissions (71%) in the < 2-yr-old group. Bronchitis/asthma were more
important for the intermediate age groups. However, in all age groups
the largest increase in admissions was caused by chronic disease in
tonsils and adenoids.

Multipollutant models rendered all pollutants except PMio and SO2 from
significance. The effect of PMio stayed relatively unchanged while SO2
was reduced; however, it remained significant.
Increment: pg/m3 (IQR)

All respiratory admissions
< 2 yrs 5.9% (4.5, 7.4)
3-5 yrs 1.6% (-1.3, 4.4)
6-13 yrs 0.6% (-2.2, 3.5)
14-1 9 yrs 1.3% (-3.2, 5.8)
All ages 4.5% (3.3, 5.8)
This study reports a significant effect of air pollution on respiratory
morbidity, though several pollutants were associated with increased
respiratory events, making it difficult to isolate a single agent as the main
atmospheric contaminant.
Increment: 12.5 pg/m3 (IQR)
Single-pollutant models (estimated from graphs):

Pneumonia -21% (4.8, 37)
Asthma -12% (-10, 38)
Pneumonia multipollutant models:
Adjusted for:
PM10 13.3 (-5.7, 32.3) 6-day avg
NO2 16.5 (-1.6, 34.6) 6-day avg
CO 18.4 (0.5, 36.2) 6-day avg
O3 18.4 (0.5, 36.2) 6-day avg
Multipollutant model
13.3 (-5.9, 32.6) 6-day avg
Asthma multipollutant models:
Adjusted for:
PM10 3.8 (-23.3, 31.0) 2-day avg
NO2 -1 .2 (-27.4, 25.0) 2-day avg
CO 6.2 (-18.8, 31. 2) 2-day avg
O3 9.4 (-14.6, 33.5) 2-day avg
Multipollutant model
-0.5 (-27.7, 26.6) 2-day avg
Current ambient air pollution concentrations have short-term adverse
effects on children's respiratory morbidity assessed through admissions
to hospitals.
Increment: 27.1 pg/m3
(90th -10th)

All Respiratory
< 5 yrs RR 1 .038
(0.983, 1 .096) lag 1
< 5 yrs Cool RR 1 .06
(0.99, 1.11) (estimated from graph)
< 5 yrs Warm RR 0.98
(0.89, 1.07) (estimated from graph)
Pneumonia
< 5 yrs RR 1 .024
(0.961, 1.091) lag 1
< 1 yrRR 1.071
(0.998, 1.149) lag 0
Asthma
< 5 yrs RR 1.106
(0.981, 1.247) lag 2


May 2008
F-50
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
llabacaetal. (1999)
Santiago, Chile
Period of Study:
2/1/95-8/31/96
Days: 578





























Linetal. (1999)
Sao Paulo, Brazil
Period of Study:
May 1991 -Apr 1993
QgyS' Q2 1









METHODS
ED Visits
Outcome(s) (ICD9): Upper
respiratory illness (460-465,
487)'
Lower respiratory illness
(466, 480-486, 490-494,
496,519.1,033.9);
Pneumonia
(480-486)
Age groups analyzed: < 15
Study design: Time-series
# of Hospitals: 1
Statistical analyses: Poisson
regression
Covariates: Long-term
trend, season, day of wk,
temperature, humidity,
influenza epidemic
Season:
Warm (Sep-Apr),
Cool (May-Aug)
Lag: 0-3 days

















ED Visits
Outcome(s): Respiratory
disease, Upper respiratory
illness, Lower respiratory
illness, Wheezing
ICD 9Code(s): NR
Age groups analyzed: < 13
Study design: Time-series
# of Hospitals: 1
Statistical analyses:
Gaussian and Poisson
regression
Covariates: Long-term
trend, seasonality, day of
wk, temperature, humidity
Lag: 5-day lagged moving
avgs
POLLUTANTS
24-h avg SO2
(pg/m3)
Warm:
Mean: 14.9
MediaN: 13.2
SD: 8.8
Range: 1.9, 60.2
5th: 5.6
95th: 32.0
Cool:
Mean: 31.8
MediaN: 28.2
SD: 18.4
Range: 5.6, 92.1
5th: 9.4
95th: 75.2

# of stations: 4
Warm:
NO2 (r = 0.6556)
O3(r = 0.1835)
PM10(r = 0.6687)
PM2.5(r = 0.5764)
Cool:

NO2 (r = 0.7440)
O3(r = 0.1252)
PM10(r = 0.7337)
PM2.5(r = 0.6874)







SO2 pg/m3:
Mean: 20
SD:8
Range: 4, 60
Number of
stations: 3
NO2 (r = 0.38)
CO (r = 0.56)
PMio (r = 0.73)

03(r = 0.21)




FINDINGS
SO2 was related to the number of respiratory ED visits, but because of
the high correlation between contaminants, it is difficult to establish
independent health effects. These results support the fact that exposure
to air pollution mixtures may decrease immune functions and increase
the risk for respiratory infections among children.
Increment: IQR
All respiratory
Cool
Lag 2 IQR: RR 1.0289
(1.0151, 1.0428)
Lag 3 IQR: RR 1.0374
(1.0236, 1.0513)
Lag avg 7 IQR: RR 1.0230 (1.0086, 1.0377)
Warm
Lag 2 IQR: RR 1.0029
(0.9860, 1 .0200)
Lag 3 IQR: RR 1.0108
(0.9937, 1.0282)
Lag avg 7 IQR: RR 1.0108 (0.9756, 1.0473)
Upper respiratory
Cool
Lag 2 IQR: RR 1.0584
(1.0394, 1.0778)
Lag 3 IQR: RR 1.051 3
(1.0324, 1.0706)
Lag avg 7 IQR: RR 1.0316 (1.0120, 1.0515)
Warm
Lag 2 IQR: RR 1.0061
(0.9850, 1.0277)
Lag 3 IQR: RR 1.0130
(0.9916, 1.0349)
Lag avg 7 IQR: RR 0.9815 (0.9390, 1.0260)

rllcUlllUllId
Cool
Lag 2 IQR: RR 1.0164
(0.9757, 1.0587)
Lag 3 IQR: RR 1.0342
(0.9938, 1 .0762)
Lag avg 7 IQR: RR 1.0291 (0.9850, 1.0751)
Warm
Lag 2 IQR: RR 1.1010
(1.0404, 1.1653)
Lag 3 IQR: RR 1.0248
(0.9669, 1 .0862)
Lag avg 7 IQR: RR 1 .2 1 51 (1 .0771 , 1 .3709)
The results of this study demonstrate a significant association between
the increase in emergency visits for all respiratory illness, especially
URI and SO2 levels
Increment: 10 pg/m3
All respiratory illness
SO2 alone RR 1.079
(1.052, 1.107) 5-day moving avg
SO2 + PM10 + O3 + NO2 + CO RR 0.938 (0.900, 0.977)
Lower respiratory illness
SO2 alone RR 1.052
(0.984, 1.125) 5-day moving avg
SO2 + PM10 + O3 + NO2 + CO RR 0.872 (0.783, 0.971)
Upper respiratory illness
SO2 alone RR 1.075
(1.044, 1.107) 5-day moving avg
SO2 + PM10 + O3 + NO2 + CO RR 0.951 (0.906, 0.999)
Wheezing
SO2 alone RR 1 .034 (0.975, 1 .096) 5-day moving avg
SO2 + PM10 + O3 + NO2 + CO RR 0.908 (0.824, 1 .002)

May 2008
F-51
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Martins* et al. (2002)
Sao Paulo, Brazil
Period of Study:
5/96-9/98




















METHODS
ED Visits
Outcome(s) (ICD10):
Chronic Lower Respiratory
Disease (CLRD) (J40-J47);
includes chronic bronchitis,
emphysema, other COPDs,
asthma, bronchiectasia
Age groups analyzed: >64
Study design: Time-series
N: 712
# of Hospitals: 1
Catchment area: 13,163
total ER visits
Statistical analyses: Poisson
regression and GAM - no
mention of more stringent
criteria
Covariates: Weekdays,
time, min temperature,
relative humidity, daily
number of non-respiratory
emergency room visits
made by elderly
Statistical package: S-Plus
Lag: 2-7 days and 3 day
moving avgs
POLLUTANTS
SO2 24-h avg
ftjg/m^lSJ,
SD' 106
Range: 2.0, 75.2
IQR: 15.1 pg/m3
# of Stations: 13
03 (r = 0.28)
NO2 (r = 0.67)
PM10(r = 0.72)
CO (r = 0.51)















FINDINGS
The results of the study show a significant association between SO2 and
CLRD among the elderly.
Increment: IQR of pg/m3
Percent increase: 17.5
(5.0, 23.0) lag 3-day moving avg (estimated from graph)
Single-pollutant model
3 = 0.0140(0.0056)
Multipollutant model (with ozone)
3 = 0.0104(0.0059)
















ASIA
Agarwal et al. (2006)
Safdarjung area of south
Deli
Period of Study: 2000-2003
















Hospital Admissions
Outcome(s) (ICD9): COPD,
asthma and emphysema
Study design: time-series
Statistical Analysis:
Performed Kruskal-Wallis
one way analysis of
variance by rank, chi-square
analysis.

Statistical package: SPSS

Age groups analyzed: all
Covariates: Temperature-
min and maximum, relative
humidity at 0830 and 1730
h and wind speed
N: NR
# Hospitals: 1
Lag: none
Mean, SD
Quarter 1:
16.7,5.5
Quarter 2:
136 26

Quarter 3:
12831


'
143 28

NO2
SPM
RSPM




SO2 was found to be in "low" category the entire time, so no analysis
could be performed


















May 2008
F-52
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Chewetal. (1999)
Singapore
Period of Study:
1990-1994












Hwang and Chan (2002)
Taiwan
Period of Study:
1998












Ko et al. (2007b)

Hong Kong
2000-2005









METHODS
Hospital Admissions/ED
Visits
Outcome(s) (ICD 9):
Asthma (493)
Age groups analyzed: 3-12,
13-21
Study design: Time-series
N: 23,000
# of Hospitals: 2
Statistical analyses: Linear
regression, GLM
Covariates: variables that
were significantly
associated with ER visits
were retained in the model

Statistical package:
SAS/STAT, SAS/ETS 6.08

Lag: 1, 2 days avgs
ED Visits
Outcome(s) (ICD 9): Lower
Respiratory Disease (LRD)
(466, 480-6) including acute
bronchitis, acute
bronchiolits, pneumonia
Age groups analyzed:
0-14

15-64, > 65, all ages
Study design: Time-series
Catchment area: Clinic
records from 50
communities
Statistical analyses: Linear
regression, GLM
Covariates: temperature,
dew point temperature,
season, day of wk, holiday
Lag: 0,1 ,2 days and avgs
Hospital Admissions

Outcome(s) (ICD9): Asthma
Study design: Retrospective
ecological study
Statistical Analysis:
Generalized additive
models with Poisson
'
Age groups analyzed: All
Covariates'

N: 69,716
# Hospitals: 15
Lag: 0-5 days
POLLUTANTS
24-h avg: 38.1
Mg/m3,
SD:21.8
Range: 3.0, 141.0
# of Stations: 15
NO2
03
JSP









24-h avg: 5.4
SD: 3.0
Range: 1.5, 16.9
NO2
PMio
03

CO
No correlations
for individual-
pollutants.






Mean, SD
fun/m3!
\Hy/m /
Whole yr: 18.8,
13 1

<20°C: 18.0,
10.0
>20°C: 19.1,
14 1
1 1. 1
NO2

10
PM2.5
O3

FINDINGS
SO2 was positively correlated to daily ER visits and hospitalization for
asthma in children (3-12 yrs), but not adolescents. The association of
ER visits with SO2 persisted after standardization for meteorological and
temporal variables. An adjusted increase in 2.9 ER visits for every
20 pg/m3 increase in ambient SO2 levels with a lag of 1 was observed.
The increased number of ER visits/day for each quartile are listed below:
Q1:<9
Q2: 10-12
Q3: 13-16
Q4: > 16
Categorical analysis (via ANOVA)
p-value and Pearson correlation coefficient (r) using continuous data
comparing daily air pollutant levels and daily number of ER visits
Age Group: 3-12 13-21
LagOr = 0.04r = 0.05
n tf- n nm n — n nsfi
p < U.UU I p — U.Uoo
Lag 1 r = 0.10 r = 0.06
p< 0.001 p = 0.016
Lag 2 r = 0.08 r = 0.07
p < 0 001 p = 0 019

Colinearity of pollutants prevented use of multipollutant models
Increment: 10% change in SO2 (natural avg) which is equivalent to
2.4 ppb. NOTE: The percent change isforthe rate of clinic use NOT for
relative risk for adverse effect.
Increased clinic visits for lower respiratory disease (LRD) by age group
0-1 4 yrs
Lag 0 0.5% (0.3, 0.6)
15-64 yrs
Lag 0 0.7% (0.5, 0.8)
>65 yrs
Lag 0 0.8% (0.6, 1.1)
All ages
Lag 0 0.5% (0.4, 0.7)






SO2 had a non-significant effect on respiratory admissions.

Relative Risk (95% Cl)
LagO: 1.004(0.998, 1.011)
Lag 1 : 1 .000 (0.994, 1 .007)
Lag 2: 0.999 (0.993, 1.006)
Lag 3: 1 .002 (0.998, 1 .008)
Lag 4: 1.004(0.997, 1.010)
Lag 5: 0.997 (0.990, 1.003)
Lag 0,1: 1.003(0.996, 1.011)
Lag 0,2: 1.003(0.994, 1.011)
Lag 0,3: 1.004(0.994, 1.014)
Lag 0-4: 1.007(0.996, 1.017)
Lag 0-5: 1.004(0.993, 1.016)


May 2008
F-53
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Ko et al. (2007a)
Hong Kong
2000-2004










Lee* et al. (2002)
Seoul, Korea
Period of Study:
12/1/97-12/31/99
Days: 822










Lee et al. (2006)
Hong Kong, China
Period of Study:
1997-2002
Days: 2, 191








METHODS
Hospital Admissions
Outcome(s) (ICD9): COPD
Study design: Retrospective
ecological study
Statistical Analysis: Poisson
distribution

Age groups analyzed: All
ages
Covariates: Autocorrelation
and overdispersion were

N: 119,225
# Hospitals: 15
Lag: 0-5 days
Hospital Admissions
Outcomes (ICD 10): Asthma
(J45-J46)
Age groups analyzed: < 15
Study design: Time-series
N: 6,436
Statistical analyses: Poisson
regression, log link with
GAM

Covariates: Time, day of wk,
temperature, humidity
Season: Spring (Mar-May),
Summer (Jun-Aug),
Fall (Sep-Nov),
Winter (Dec-Feb)
Lag: 0-2 days cumulative
Hospital Admissions
Outcome(s) (ICD 9):
Asthma (493)
Age groups analyzed: <18
Study design: Time-series
N: 26,663
Statistical analyses: Semi-
parametric Poisson
regression with GAM
(similar to APHEA 2)
Covariates: Long-term
trend, temperature, relative
humidity, influenza, day of
wk, holiday
Statistical package: SAS
8.02
Lag: 0-5 days
POLLUTANTS
15.0pg/m3
SD: 11.6
NO2
PMnn
r IVI1Q
03
PMic
nvi2.5






24-h SO2 (ppb)
Mean: 7.7
SD: 3.3
5th: 3.7
25th: 5.1
50th' 7 0
75th' 9 5

95th: 14.3

# of stations: 27
NO2 (r = 0.723)
03(r = -0.301)
CO (r = 0.81 2)
PM10(r = 0.585)


SO2 24-h Mean:
17.7pg/m3,
SD: 10.7
IQR: 11.1 pg/m3
25th: 10.6
50th' 152

75th: 21. 7
# of stations:
9-10
PM10(r = 0.37)
PM2.5(r = 0.47)
NO2 (r = 0.49)
03(r = -0.17)



FINDINGS
Positive association with hospital admission for acute exacerbations of
COPD.
Relative Risk (95% Cl)
LagO: 1.007(1.001, 1.014)
Lag 1:0.991 (0.981, 1.001)
Lag 2: 0.992 (0.985, 1.000)
Lag 3: 1.006(0.999, 1.013)
Lag 4: 1.004(0.998, 1.011)
Lag 5: 1.004(0.997, 1.010)
Lag 0-1: 0.998 (0.991, 1.006)
Lag 0-2: 0.993 (0.985, 1.001)
Lag 0-3: 0.998 (0.989, 1.007)
Lag 0-4: 1.001 (0.991, 1.010)
Lag 0-5: 1.004(0.994, 1.014)


This study reinforces the possible role of SO2 on asthma attacks,
although it should be interpreted with caution because the effect
estimates are close to the null and because results in the multipollutant
models are inconsistent.
Increment: 14.6 ppb (IQR)
Asthma
SO2RR1.11 (1.06, 1.17) lag 0-2
SO2 + PM10RR 1.08 (1.02, 1.14) lag 0-2
SO2 + NO2 RR 0.95 (0.88, 1 .03) lag 0-2

SO2 + O3RR 1.12 (1.06, 1.1 7) lag 0-2
SO2 + CO RR 0.99 (0.92, 1 .07) lag 0-2
SO2 + O3 + CO + PM10 + NO2 RR 0.949 (0.868, 1 .033)




Absence of an association of SO2 with asthma admissions was
attributed to low ambient SO2 levels during the study period due to
restrictions on sulfur content in fuel.
Increment: 11.1 pg/m3 (IQR)
Asthma
Single-pollutant model
Lag 0-1. 57% (-2.87, -0.26)
Lag 1 -1.77% (-3.06, -0.46)
Lag 2 -1.15% (-2.42, 0.14)
Lag 3 0.82% (-0.45, 2.11)
Lag 4 1.40% (0.13, 2.69)
Lag 5 1.46% (0.19, 2.74)
Multipollutant model-including PM, NO2, and O3
0.81% (-0.75, 2.4) lag 5
Other lags NR



May 2008
F-54
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Lee et al. (2007)
Kaohsiung, Taiwan
1996-2003















Tanakaetal. (1998)
Kushiro, Japan
Period of Study:
1992-1993











Tsai et al. (2006)
Kaohsiung, Taiwan
Period of Study:
1996-2003
Days: 2922

















METHODS
Hospital Admissions
Outcome(s) (ICD9): COPD
(490-492, 494, and 496)
identified by records from
the National Health
Insurance (NHI) program
Study design: Case-
crossover
Statistical Analysis:
Conditional logistic
regression
Statistical package: SAS
Age groups analyzed: All
ages
Covariates: Adjustment for
temperature and humidity
N: 25,108
# Hospitals: 63
Lag: Cumulative lag up to 2
days



ED Visits
Outcome(s): Asthma
Age groups analyzed:
15-79
Study design: Time-series
N' 102

# of Hospitals: 1
Statistical analyses: Poisson
regression
Covariates: temperature,
vapor pressure, barometric
pressure, relative humidity,
wind velocity, wind direction
at maximal velocity
Statistical package: NR
Hospital Admissions
Outcome(s) (ICD 9):
Asthma (493)
Study design:
Case-crossover
N: 17,682
Statistical analyses:
Conditional logistic
regression
Covariates: Temperature,
humidity
Season:
Warm (> 25 °C);
Cool (< 25 °C)

Statistical package: SAS
Lag: 0-2 days Cumulative







POLLUTANTS
24-h avg (ppb):
9.49
Range: 0.92,
31.33
PM10
NO2
CO
03












SO2 24-h avg
3.2 (2.4) ppb in
fog
3.7 (1.9) ppb in
fog free days
Max SO2 24-h
avg
< 11 ppb
NO2 (r = NR)
SPM (TSP);
r=O3;r=NR





SO2 24-h Mean:
9.49 ppb
Range: 0.92,
31.33

25th: 6.37
50th: 8.94
75th: 12.16
# of stations: 6
PMnn
r IVI1Q
NO2
03
CO










FINDINGS
All pollutants, except SO2, were significantly associated with COPD
hospital admissions on warm days, while on cold days all pollutants
were found to be significantly associated. In two pollutant models, CO
and O3 were significantly associated with each of the other pollutants on
warm days, and on cool days, only NO2 was significantly associated with
all pollutants.
Odds Ratio (95% Cl), Single-pollutant model
(per 5. 79 ppb SO2)
>25°C
1.024(0.973, 1.077)
<25°C
1.190(1.093, 1.295)
Odds Ratio (95% Cl),
Co-pollutant model
(per 5. 79 ppb SO2)
> 25 °c
SO2 + PM10: 1 .002 (0.951 , 1 .054)
SO2 + NO2: 0.979 (0.926, 1.034)
SO2 + CO: 0.929 (0.876, 0.985)
SO2 + O3: 1.057(1.004, 1.113)
<25°C
SO2 + PM10: 1.043(0.952, 1.143)
SO2 + NO2: 0.767 (0.689, 0.855)
SO2 + CO: 1.004(0.915, 1.103)
SO2 + O3: 1.198(1.100, 1.304)
The results reveal that ED visits by atopic subjects increased on low SO2
days. This observation is inconsistent with most air pollution
epidemiology, as high levels of air pollutants have conventionally been
linked with asthma exacerbation.
Increment: 5 ppb
Nonatopic
OR 1.1 8 (0.96, 1.46)
Atopic
OR 0.78 (0.66, 0.93)







Positive associations were observed between air pollutants and hospital
admissions for stroke. In single-pollutant models SO2was not
associated with either PIH or IS. The season did not affect these
associations. SO2 was also not significant in
2-pollutant models.
Increment: 5.79 ppb (IQR)
Seasonality
Single-pollutant model
>25°C 1.018 (0.956, 1.083)
lag 0-2
< 25 °C 1.187 (1.073, 1.314)
lag 0-2
Dual-pollutant model
Adjusted for PM10
>25 °C 0.993 (0.932, 1.058) lag 0-2
< 25 °C 1.027 (0.921, 1.146) lag 0-2
Adjusted for CO
>25 °C) 0.910 (0.847, 0.978) lag 0-2
< 25 °C 1 .036 (1 .027, 1 .046) lag 0-2
Adjusted for NO2
>25 °C 0.967 (0.903, 1 .035) lag 0-2
< 25 °C 0.735 (0.646, 0.835) lag 0-2
Adjusted for O3
>25 °C 1.055 (0.990, 1.123) lag 0-2
< 25 °C 1.195 (1.080, 1.323) lag 0-2
May 2008
F-55
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Wongetal. (1999)
Hong Kong, China
Period of Study:
-\CtCtA 1QQ^
i yy^f- 1 yyo














Wong et al. (Wong et al.,
2001)
Hong Kong, China
Period of Study:
1993-1994














METHODS
Hospital Admissions
Outcome(s)(ICD9):AII
respiratory admissions
(460-6,471-8,480-7,
490-6); Asthma (493),
COPD (490-496),
Pneumonia (480-7)
Age groups analyzed:
0-4, 5-64, > 65, all ages
# of Hospitals: 12
Study design Time-series
Statistical analyses: Poisson
regression (followed APHEA
protocol)
Covariates: Trend, season,
day of wk, holiday,
temperature, humidity
Statistical package: SAS
8.02
Lag: days 0-3 cumulative
Hospital Admissions
Outcome(s) (ICD 9):
Asthma (493)
Age groups analyzed: < 15
N: 1,217
# of Hospitals: 1
Study design: Time-series
Statistical analyses: Poisson
regression (followed APHEA
protocol)
Covariates: Season,
temperature, humidity
Season:
Summer (Jun-Aug), Autumn
(Sep-Nov), Winter (Dec-
Feb),
Spring (Mar-May)
Lag: 0,1 ,2,3,4,5 days; and
cumulative 0-2 and 0-3
days.
POLLUTANTS
Median 24-h SO2:
17.05pg/m3
Range: 2.74,
68.49
25th' 12 45

75th: 25.01
# of stations:
7, r =
03
SO2
PM10








24-h avg SO2
Mean: 12.2 pg/m3
SD: 12.9
Range' 0 98
pg/m3
AutumN: 10.6
(9.6)
Winter: 10.0(7.5)
Spring: 9.6 (8.8)
Summer: 18.5
(19.5)
# of stations: 9

PM10
NO2





FINDINGS
Adverse respiratory effects of SO2 were noted at low concentrations.
Results for respiratory outcomes were attributed to the elderly
population. This was also true for the other pollutants. Therefore, it is
difficult to be certain that the effects were due mainly to SO2.
Pair-wise comparisons in multipollutant models showed significant
interactions of PM25, NO2, and O3.
Increment =10 pg/m3
Overall increase in admissions:
1.013(1.004, 1.021) lag 0
Respiratory relative risks (RR)
0-4 yrs: 1.005(0.991, 1.018) lag 0
5-64 yrs: 1.008 (0.996, 1.021) lag 0
>65yrs: 1.023(1.012, 1.036) lag 0
Asthma: 1.017 (0.998, 1.036) lag 0
COPD: 1.023(1.011, 1.035) lag 0
Pneumonia: 0.990
(0.977, 1 .004) lag 4




SO2 levels were found to be the highest during the summer. There were
consistent and statistically significant associations between asthma
admission and increased daily levels of SO2. No associations were
noted in the spring or winter. No significant associations were found
between hospital admissions and day of the wk, humidity, temperature
or atmospheric pressure.
Total admissions were limited to one hospital.
Increment: 10 pg/m3
Asthma
AllyrRR 1. 06 p = 0.004
AutumN: NR
Winter: NR
Spring: NR
Summer: NR








May 2008
F-56
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Wong et al. (2002a)*
London England and Hong
Kong
Period of Study:
LondoN:
1992-1994
Hong Kong:
1995-1997
Days: 1 ,096




























Yang and Chen (2007)
Taipei, Taiwan
Period of Study: 1996-2003














METHODS
Hospital Admissions
Outcome(s)(ICD9):AII
respiratory admissions
(460-519); asthma (493)
Age groups analyzed: 15-
64, 65+, all ages
Study design: Time-series
Statistical analyses: APHEA
protocol, Poisson
regression with GAM
Covariates: Long-term
trend, season, influenza,
day of wk, holiday,
temperature, humidity,
thunderstorms

Cool Oct-Mar'
Warm: Apr-Sep
Dose-Response
Investigated?: Yes
Statistical package: S-Plus
Lag:0, 1, 2, 3, 4 days, 0-1
cum. avg.














Hospital Admissions
Outcome(s) (ICD9): COPD
(490-492, 494, and 496)
identified by records from
the National Health
Insurance (NHI) program
Study design: Case-
crossover
Statistical Analysis:
Conditional logistic
regression
Statistical package: SAS
Age groups analyzed: All
ages
Covariates: Adjustments for
weather variables, day of
the wk, seasonality, and
long-term time trends
N: 46,491
# Hospitals: 47
Lag: Cumulative lag up to 2
days
POLLUTANTS
24-h SO2 pg/m3
Hong Kong
Mean: 17.7
Warm: 18.3
Cool: 17.2
SD: 12.3
Range: 1.1, 90.0
10th: 6.2
50th: 14.5
90th: 32.8

Mean: 23.7
Warm: 22.2
Cool: 25.3
SD: 12.3
Range: 6.2, 113.6
10th: 13.2
50th: 20.6
90th: 38.1
Uli-ini-i lfr\nn
nong Kong
PM2.5(r = 0.30)
NO2 (r = 0.37)
O3 (r = -0.18)


PM2.5(r = 0.64)
NO2 (r = 0.71)

03 (r = -0.25)









24-h avg (ppb):
4.33
Range: 0.15,
17.82
25th: 2.67
50th: 3.90
75th: 5.46
PM10
NO2

CO
3








FINDINGS
Similar non-statistically significant associations between asthma hospital
admissions and SO2 were found in both cities. The association between
respiratory hospital admissions and SO2 showed significance in the cold
season in Hong Kong and on an all yr basis. Respiratory hospital
admissions were not significantly associated with SO2 in Britain.
In the 2-pollutant model the association between respiratory hospital
admission and SO2 in London was insignificant, and remained
insignificant after adjusted for the second pollutants.
In Hong Kong, the positive association of SO2 was most affected by
NO2, losing statistical significance. The positive association remained
robust when adjusted for O3, and a slight decrease in association after
adjusted for PM25.
Increment: 10 pg/m3
Asthma, 15-64 yrs
Hong Kong
ER -0.1 (-2.4, 2.2) lag 0-1
ER -1.5 (-3.4, 0.5) lag
Warm: ER 1 .5 (-1 .5, 4.6) lag 0-1
Cool: ER -2.0 (-5.4, 1 .4) lag 0-1
London
ER 0.7 (-1 .0, 2.5) lag 0-1 ER 2.1 (0.7, 3.6) lag 3
Warm: ER -1 .4 (-4.7, 1 .9) lag 0-1
Cool: ER 1.6 (-0.5, 3.8) lag 0-1
Respiratory 65+ yrs
Hong Kong
ER1.8
(0.9, 2.6) lag 0-1 ER 1.7 (1.0, 2.4) lag 0
Warm:ER 1.1 (0.0, 2.2) lag 0-1
Cool: ER 2. 7 (1.4, 4.0) lag 0-1
+O3ER1.9(1.1,2.8) lag 0-1
+PM25ER1.2 (0.3, 2.2) lag 0-1
+NO2 ER 0.3 (-0.7, lag 1.4) lag 0-1
London
ER0.2 (-0.6, 1.1)
lag 0-1 ER 1.2 (0.5, 2.0) lag 3
Warm: ER 1.3 (-0.5, 3.1) lag 0-1
Cool: ER -0.3 (-1.3, 0.8) lag 0-
+O3 ER 0.5 (-0.4, 1.5) lag 0-1
+PM25ER1.2 (0.3, 2.2) lag 0-1
+NO2ER0.5(-0.7, 1.7)lagO-1
In single-pollutant models, all pollutants, except SO2, significantly
associated with COPD hospital admissions on warm days (>20 °C). On
cold days (< 20 °C), only SO2 was significantly associated with COPD
hospital admissions. In multi-pollutant models, NO2 and O3 were
significantly associated with each pollutant on warm days.
Odds Ratio (95% Cl),
Single-pollutant model
(per 2. 79 ppb SO2)
>20°C: 1.006(0.970, 1.043)
<20°C: 1.071 (1.015, 1.129)
Odds Ratio (95% Cl),
Co-pollutant model
(per 2. 79 ppb)
>20°C
SO2 + PM10: 0.909 (0.872, 0.949
SO2 + NO2: 0.835(0.798, 0.873)
SO2 + CO: 0.920(0.884, 0.958)
SO2 + O3: 0.978(0.943, 1 .015)
SO2 + PM10: 1.067(0.997, 1.141)
SO2 + NO2.1. 147 (1.072, 1.227)
SO2 + CO: 1.140(1.066, 1.219)


'Default GAM
+Did not report correction for
over-dispersion
APHEA: Air Pollution and
Health: a European Approach
May 2008
F-57
DRAFT—DO NOT QUOTE OR CITE

-------
Table F-3.    Associations of short-term exposure to SO2 with cardiovascular morbidity in
                 field/panel studies.
    STUDY
                              METHODS
                                                      POLLUTANTS
                                                                                               FINDINGS
                                                         UNITED STATES
Dockery et al.
(2005)

Boston, MA

Period of Study:
JuM995-Jul2002
Cohort study of 203 cardiac patients with
implanted cardioverter defibrillators.
Patients were followed for an avg of 3.1
yrs from 1995-2002 to assess the role of
air pollution on the incidence of
ventricular arrhythmias. The association
of arrhythmic episode-days and air
pollutions analyzed with logistic
regression using GEE with random
effects. Model adjusted for patient,
season, min temperature, mean  humidity,
day of the wk, and previous arrhythmia
within 3 days. Only effects of 2-day
running mean of air pollution
concentration reported.
48-h avg SO2;

MediaN: 4.9 ppb
25th%: 3.3 ppb
75%: 7.4 ppb

95%: 12.8 ppb

Copollutants:
PM2.5
BC
so42-
PN
NO2
CO
03
No statistically significant association between any of the air pollutant
and ventricular arrhythmias when all events were considered.
However, ventricular arrhythmias within 3 days of a prior event were
statistically significant with  SO2, PM25, BC, NO2, CO, and marginally
with SO42", but not with O3 or PN. CO, NO2, BC, and PM2.5
correlated, thus it was impossible to differentiate the independent
effects. Since the increased risk of ventricular tachyarrhythmia was
associated with air pollution observed among patients with a recent
tachyarrhythmia, it was suggested that air pollution acts in
combination with cardiac electrical instability to increase risk of
arrhythmia.

For IQR (4.0 ppb) increase in 48-h mean SO2: All events: OR = 1.04
(0.94, 1.14), p = 0.28. Prior arrhythmia event < 3 Days: 1.30 (95%
Cl: 1.06, 1.61), p = 0.013. Prior arrhythmia event >3 Days: 0.98
(0.87, 1.11)p = 0.78
Gold et al. (2000)

Boston, MA
Jun-Sep

Period of Study:
1997
Panel study on 21 active Boston
residents aged 53-87 yrs to investigate
the association between short-term
changes in ambient air pollution and
short-term changes in cardiovascular
function. Participants observed up to 12
times from June to Sep 1997 (163
observations made in total). Protocol
involved 25 mins perwk of continuous
ECG monitoring, that included 5 mins of
rest, 5 mins of standing, 5 mins of
exercise outdoors, 5 mins of recovery,
and 20 cycles of slow breathing. Fixed
effects models adjusted for time-varying
covariates and individuals traits.
24-h avg mean
3.2 ppb
Range: 0, 12.6
ppb
IQR: 3.0 ppb

Copollutants:
PM2.5
PM10.2.5
03
NO2
CO
In single-pollutant models, 24-h mean SO2 associated with reduced
heart rate in the first rest period but not overall. Associations weaker
for shorter averaging periods. Association between SO2 and heart
rate not significant with the multipollutant model (SO2 and PM25).
SO2 not associated with r-MSSD.

Heart rate, first rest period, mean 66.3 bpm, single-pollutant model,
estimated effect (SE) -1.0 (0.5); % mean 1.5, p = 0.03. Heart rate,
first rest period, mean 66.3 bpm
Multipollutant model (PM25 and SO2): SO2 estimated effect (SE) -0.8
(0.5); % mean 1.2, p = 0.09. PM2.5 estimated effect (SE) -1.6 (0.7);
% mean 2.5, p = 0.03. Overall heart rate, mean 74.9 bpm, Single-
pollutant model estimated effect (SE) -0.5 (0.5), p = 0.30.
Multipollutant model
SO2 estimated effect (SE) -0.2 (0.5), p = 0.6 PM25 estimated effect
(SE) -1.9 (0.7) p = 01% mean 2.6
Liao et al. (2004)

Three locations in
United States:
Minneapolis, MN;
Jackson, MS;
Forsyth County,
NC

1996-1998
Cross-sectional study of 6,784 cohort
members of the Atherosclerosis Risk in
Communities Study. Participants were
45-64 yrs of age; baseline clinical
examinations conducted from 1987-1989.
HRV data collected from 1996-1998. Air
pollutants obtained form EPA AIRS for
this same period. Resting, supine, 5-min
beat-to-beat RR interval data were
collected over a 4-h period. Multivariable
linear regression models used to assess
associations between pollutants
measured 1-3 days prior to HRV
measurements. Models controlled for
age, ethnicity-center, sex, education,
current smoking, BMI, heart rate, use of
cardiovascular medication, hypertension,
prevalent coronary heart disease, and
diabetes.
Mean (SD) SO2
measured 1 day
prior to HRV
measurement
was 4 (4) ppb

Copollutants:
PM10
03
CO
NO2
Significant interaction between SO2 and prevalence of coronary
heart disease for low-frequency power analyses. SO2 inversely
associated with SD of normal R-R intervals and low-frequency power
and positively associated with heart rate. SO2 association with low-
frequency power stronger among those with history of coronary heart
disease. Effect size of PMio larger than for gaseous pollutants.

Log-transformed low-frequency power effect estimate and SE per 1
SD increment (4 ppb) SO2 lag 1 day:
Log transformed high-frequency power -0.024 (SE 0.016)
Standard deviation of normal
R-R intervals -0.532 (SE 0.270), p < 0.05=Heart rate: 0.295 (SE
0.130), p < 0.05. Prevalent CHD: -0.122 (SE 0.056), p < 0.01. No
prevalent CHD -0.012 (SE 0.016)
May 2008
                                         F-58
                               DRAFT—DO NOT QUOTE OR CITE

-------
    STUDY
                              METHODS
                                                       POLLUTANTS
                                                                                                FINDINGS
Liao et al. (2005)

United States

1996-1998
Cross-sectional survey
10,208 participants (avg age 54 yrs) from
Atherosclerosis Risk in Communities
(ARIC) study cohort to assess the associ-
ation between criteria air pollutants and
hemostatic and inflammatory markers.
57% of participants were female and 66%
male. Used hemostatis/ inflammation
variables collected during the baseline
examination and air pollution data 1-3
days prior to the event. Used multiple
linear regression models that controlled
for age, sex, ethnicity-center, education,
smoking, drinking status, BMI, history of
chronic respiratory disease, humidity,
seasons, cloud cover, and temperature.
Also history of CVD and diabetes if not
effect modifier in a particular model.
SO2 mean (SD)
0.0005 (0.004)
ppm

Q1-3: 0.005
(0.003) ppm

Q4: 0.006
(0.005) ppm

Copollutants:
PM10
CO
NO2
03
Significant curvilinear association between SO2 witfactor VIII-C,
WBC, and serum albumin. Curvilinear association indicated
threshold effect

Results shown in graph.
Luttmann-Gibson
et al. (2006)

Steubenville, OH

2000
Conducted a panel study during the
summer and fall of 2000, which consisted
of 32 subjects 54-90 yrs old living in
Steubenville, OH. Used linear mixed
models, fixed effects of pollution, age,
gender, race, obesity, season, time of
day, apparent temperature, and a first
order autoregressive process for within-
subject residuals to examine the relation
between air pollution and log-transformed
HRV parameters and heart rate.
24-h avg (ppb):
4.1

Copollutants:
PM2.5
SO42"
EC
NO2
03
Increasing concentrations of PM25 and SO4 " in the previous day
were both found to be associated with reduced HRV. No association
was observed between increasing SO2 concentrations in the
previous day and HRV.

% Change (95% Cl) (per 4.3 ppb SO2). Standard Deviation of Normal
RR Intervals (SSDN) 0.7 (-1.0, 2.5). Differences Between Adjacent
RR Intervals (r-MSSD) 0.5 (-2.8, 4.0). High-Frequency Power (HF)
1.7 (-4.9, 8.7). Low-Frequency Power (LF) 4.9 (-1.4, 11.5). Heart
Rate (HR) 0.3 (-0.2, 0.8)
Metzger et al.
(2007)

Atlanta, GA

1993-2002
Collected information on 518 patients
(6287 event-days) for ventricular
tachyarrhythemic events over 10-yr
period. Used GEE analysis, a case-
crossover analysis, and a sensitivity
analysis stratified on subject
15.5 ppb (±16.4)

Copollutants:
PM10
03
NO2
CO
PM2.5
Little evidence of associations between ambient air quality
measurements and ventricular tachyarrhythmic events.

Odds ratio (95% Cl) All events: 1.002 (0.968-1.037)
Events resulting in cardiac pacing or defibrillatioN: 0.988 (0.936-
1.042). Events resulting in defibrillatioN: 1.004 (0.911-1.105).
Primary GEE model: 1.002 (0.968-1.037)
Controlling for min temperature: 1.010 (0.976-1.046)
Using an unconstrained distributed Lag: 0.996 (0.952-1.083). Warm
Season: 1.029 (0.989-1.116). Cold Season: 0.986 (0.956-1.023)
Park etal. (2005)

Greater Boston
area, MA

Nov 2000-Oct
2003
Cross-sectional study of effect of ambient
air pollutants on heart rate variability
(HRV) in 497 men who were in the
Normative Aging Study and examined
from Nov 2000 and Oct 2003. HRV mea-
sured between 0600 and 1300 h after
resting for 5 mins. 4-h, 24-h, and 48-h
moving avgs of air pollution matched to
time of ECG measurement. Linear
regression models included: age, BMI,
fasting blood glucose, cigarette smoking,
use of cardiac medications, room temp,
season, and the lagged moving avg of
apparent temp corresponding to the
moving avg period for the air pollutant.
Mean arterial blood pressure (MAP) and
apparent temperature also included.
Assessed modifying effects of
hypertension, IHD, diabetes or use of
cardiac/antihypertensive meds.
24-h avg SO2
4.9 ppb
SD: 3.4

Range: 0.95,
24.7 ppb

Copollutants:
PM2.5
PNC
BC
NO2
03
CO
No significant association between HRV and SO2for any of the
averaging periods, but positive relationship.

4-h moving avg SO2: (per 1 SD, 3.4 ppb SO2)

Log10SDNN:2.3(-1.7, 6.4)

Log10HF:5.6(-4.9, 17.3)

Log10LF:2.2(-5.9, 11.1)

LoglO (LF:HF): -3.2 (-10.1, 4.2)
May 2008
                                         F-59
                               DRAFT—DO  NOT QUOTE OR CITE

-------
    STUDY
                              METHODS
                                                        POLLUTANTS
                                                                                                  FINDINGS
Peters et al.
(2000a)

Eastern Massa-
chusetts, U.S.

1995-1997
Pilot study to test hypothesis that patients
with implanted cardioverter defibrillators
would experience potentially life-
threatening arrhythmias associated with
air pollution episodes. Records detected
arrhythmias and therapeutic interventions
downloaded from the implanted
defibrillator. Mean age of patients
62.2 yrs. 100 patients followed for over
3 yrs for 63,628 person-days. 33 patients
with any discharges and 6 patients with
10 or more events.  Data analyzed by
logistic regression models using fixed
effects models with individual intercepts
for each patient. Model controlled for
trend, season, meteorologic conditions,
and day of week. Evaluated air pollutants
on same day, lags 1,2, and 3 days, and
5-day mean.
24-h avg SO2:
7ppb

MediaN: 5 ppb
Max: 87 ppb

Copollutants:
PM10
PM2.5
BC
CO
03
NO2
No association between increased defibrillator discharges and SO2.
33 patients with at least 1 defibrillator discharge
Odds Ratio (95% Cl)

Lag 00.76 (0.48, 1.21); Lag 1 0.91 (0.60, 1.37)
Lag 2 0.89 (0.59, 1.34); Lag 3 1.09 (0.78, 1.52)
5-day mean 0.85 (0.50, 1.43)

6 patients with at least 10 discharges
Odds Ratio (95% Cl)

Lag 0 0.72 (0.40, 1.31); Lag 1 0.77 (0.44, 1.37)
Lag 2 1.01 (0.63, 1.61); Lag 3 1.08(0.72, 1.62)
5-day mean 0.75 (0.38, 1.47)
Peters et al.
(2001)

Greater Boston
area, MA

Jan 1995-May
1996
Case cross over Study design used to
investigate association between air
pollution and risk of acute myocardial
infarctions in 772 patients (mean age
61.6 yrs) with Ml as part of the
Determinants of Myocardial Infarction
Onset Study. For each subject, one case
period was matched to 3 control periods,
24 h apart. Used conditional logistic
regression models that controlled for
season, day of wk, temperature, and
relative humidity.
24-h avg SO2:
7 ppb
SD: 7 ppb
1-h avg SO2:
7 ppb
SD: 10 ppb
Copollutants:
PM2.5, PM10,
PM10-2s, BC
O3, CO, NO2
SO2 not statistically associated with risk of onset of Ml. Limitation of
study is only 1 air pollution monitoring site available.

OR for 2-h avg SO2 and 24-h avg SO2 estimated jointly:
2 h per 2 ppb increase SO2
Unadjusted: 1.00(0.87, 1.14)
Adjusted: 0.96 (0.83, 1.12)
24 h per 2 ppb increase
Unadjusted: 0.92 (0.71, 1.20)
Adjusted: 0.91 (0.67, 1.23)
Rich et al. (2005)

Boston, MA

JuM995-Jul2002
Case cross-over design used to evaluate
association between ventricular
arrhythmias detected by implantable
cardioverter defibrillators and air
pollution. Same study population as
Dockery et al. (2005): 203 patients with
ICD and residential zip codes within
40 km of central particle monitoring site.
Analyses conducted on 84 subjects with
confirmed ventricular arrhythmias during
the follow-up. Case periods defined by
time of each confirmed arrhythmic event.
Control periods (3-4 per case) selected
by matching on weekday and hour of the
day within the same calendar mo. Used
conditional logistic regression that
controlled for temperature, dew point,
barometric pressure, and a frailty term for
each subject. ORs presented for IQR
increase in mean concentration and
averaging time. Moving avg of
concentrations considered: lags 0-2, 0-6,
0-23, and  0-47 h.
1-h avg SO2:
MediaN: 4.3 ppb
25th  %: 2.6
75th  %: 7.5
Max: 71.6

24-h  avg  SO2:
MediaN: 4.8
25th  %: 3.2
75th  %: 7.3
Max: 31.4

Copollutants:
PM2.5
BC
NO2
CO
03
An IQR increase in the 24-h moving avg SO2 (4.1 ppb) marginally
associated with a 9% increased risk of ventricular arrhythmia and an
increased risk with 48-h moving avg. There was no risk associated
with 24-h moving avg after controlling for PM25 cases that had a prior
ventricular arrhythmia within 72-h had greater risk associated with
SO2 compared to those without a recent event, suggesting that risk is
greater among cases with more irritable or unstable myocardium.

Odds ratios- single-pollutant model
0-2-h lag (per 4.7 ppb) 1.07 (0.97, 1.18)
0-6-h lag (per 4.5 ppb) 1.09 (0.98, 1.20)
0-23-h lag (per 4.1 ppb) 1.09 (0.97, 1.22)
0-47-h lag (per4.0ppb) 1.17(1.02, 1.34)
Odds ratios- 2-pollutant model: SO2 and PM2 5 Per 4.1 ppb SO2: 1.00
(0.84, 1.20). SO2 and O3 Per 4.1  ppb SO2: 1.12 (0.99, 1.27). Per
4.1 ppb increase SO2 Prior arrhythmia event  < 3 Days: 1.20
(1.01, 1.44). Prior arrhythmia event >3 Days:  0.96 (0.83, 1.10)
Rich et al.
(2006b)

Boston, MA

1995-2002
Case-crossover study consisting of 203
individuals with implantable cardioverter
defibrillators (ICDs) implanted between
Jun 1995 and Dec 1999. Used
conditional logistic regression, which
included variables for mean pollutant
concentration in the hour of the
arrhythmia, and natural splines for mean
temperature, dew point, and barometric
pressure in the 24 h before the
arrhythmia. The regression analyses
were run for each pollutant individually to
examine the association between
increasing pollutant levels and
paroxysmal atrial fibrillation episodes
(PAF).
Max 24-h avg
(ppb): 31.4

Max 1-h max
(ppb): 71.6

Copollutants:
PM2.5
BC
NO2
CO
03
Ozone was significantly associated with PAF in the hour preceding
the arrhythmia, but the effect was not significant when analyzing the
preceding 24-h. Increasing levels of PM25, NO2, and BC resulted in
non-significant positive associations with PAF. SO2 was not
associated with PAF.

Odds Ratio (per 4.9 ppb SO2): 1.02 (0.81, 1.28) lag 0 h
Odds Ratio (per 4.1 ppb SO2): 0.99 (0.71, 1.39) lag 0-23-h
May 2008
                                          F-60
                                DRAFT—DO NOT QUOTE OR CITE

-------
    STUDY
                              METHODS
                                                       POLLUTANTS
                                                                                                FINDINGS
Rich et al.
(2006a)

St. Louis,
Missouri

May 2001-Dec
2002
Case-crossover design study of
56 patients with implantable cardioverter
defibrillators. Subjects ranged from 20 to
88 yrs (mean 63). Case period defined by
time of confirmed ventricular arrhythmia.
Control periods matched on weekday
and hour of the day within the same
calendar mo. Used conditional logistic
regression model that included  mean of
the previous 24-h temperature,  relative
humidity, barometric pressure, mean
pollutant concentration in the 24 h before
the arrhythmia. Model also included a
frailty term for each subject.
599 days

25th %: 2 ppb
50th %: 4 ppb
75th %: 7 ppb
Daily IQR: 5 ppb
Case/control
IQR: 5 ppb

Copollutants:
PM25, EC, OC,
NO2 , CO,  O3
Statistically significant increase in risk of ventricular arrhythmias
associated with each 5 ppm increase in 24-h moving avg SO2.
OR for ventricular arrhythmia associated with IQR increase
6-h moving avg SO2per4 ppb: 1.04 (95% Cl: 0.96, 1.12)
12-h moving avg SO2 perS ppb: 1.17 (95% Cl: 1.04, 1.30)

24-h moving avg SO2 per 5 ppb: 1.24 (95% Cl: 1.07, 1.44)
48-h moving avg SO2 per 4 ppb: 1.15 (95% Cl: 1.00, 1.34)
Sarnat et al.
(2006)
Steubenville, OH
2000
Panel study consisting of 32, non-
smoking older adults approximately 53-
90 yrs old. Electrocardiograms (ECGs)
and questionnaires regarding symptoms
were administered on a weekly basis.
Used a logistic mixed effects regression
to examine the association between
increasing air pollutant concentrations
and supraventricular ectopy (SVE) and
ventricular ectopy (VE).
1-day avg SO2
24-h avg (ppb):
10.4(8.3)
Range: 1.8, 58.3
5-day moving
avg

24-h avg (ppb):
10.7(5.5)
Range: 2.4, 31.3

Copollutants:
PM25, EC, 03,
NO2, SO42", CO
PM25 was significantly associated with SVE, whereas, SO4  and O3
were marginally associated in models including 5-day moving avg
pollutant concentrations. However, no pollutants were found to be
associated with VE in similar models. Overall, subjects that reported
previous cardiovascular conditions (e.g., myocardial infarction and
hypertension) were found to be more susceptible to SVE due to
increasing air pollutant concentrations.

Odds Ratio (per 5.4 ppb SOJ

5-day moving avg
SVE: 1.04(0.78, 1.39)
VE: 1.28(0.85, 1.92)
Schwartz et al.
(Schwartz et al.,
2005)
Boston, MA
12 wks during the
summer of 1999
Panel study of 28 subjects (aged 61-89
yrs) to examine association between
summertime air pollution and HRV.
Subjects examined once a wk up to 12
wks and HRV measured for
approximately 30 mins. Analyses used
hierarchical models that controlled for
baseline medical condition, smoking
history, day of wk and hour of day,
indicator variable for whether subjects
had taken their medication before they
came, temperature and time trend.
24-h avg SO2:

25th %:
0.017 ppm
50th %:
0.020 ppm
75th %:
0.54 ppm

Copollutants:
03
NO2
CO
PM2.5
BC
No significant association with SO2

Percentage change in HRV associated with IQR (0.523 ppm)
increase in SO2
SDNN (ms)0.4 (-1.3 to 2.1)
RMSSD (ms) 1.4 (-2.6 to 5.5)
PNN50 (ms) 3.8 (-12.1 to 22.5) for
1-h avg SO2
SDNN (ms) 0.4 (-4.2 to 5.1) for 24-h avg SO2
RMSSD (ms) -0.3 (-1.3 to 0.8)
PNN50 (%) -0.2 (20.9 to 17.6)
LFHFR2.9(-4.9to11.4)
Sullivan et al.
(2004)
King County,
Washington
1988-1994
Case-crossover study of 5,793 confirmed
cases of acute Ml. Data was analyzed
using simple descriptive analyses and
Pearson's correlation coefficient.
9 ppb

Range: 0-38 ppb

Copollutants:
PM2.5
PM10
CO
Increases in SO2 were not associated with Ml after adjusting for
relative humidity and temperature

Averaging time, Odds Ratio (95% Cl) (per 10 ppb SO2)
1-h: 0.97 (0.94, 1.01)
2 h: 0.98 (0.95, 1.01)
4 h: 0.99 (0.96, 1.03)
24 h: 1.0(0.95, 1.06)
Tolbert et al.
(2007)

Atlanta, GA

1993-2004
ED Visits

Outcome(s) (ICD9): Cardiovascular (410-
414, 427, 428, 433-437, 440, 443-445,
451-453); Respiratory (493, 786.07,
786.09, 491, 492, 496, 460-465, 477,
480-486, 466.1, 466.11, 466.19)

Study design: Time-series
Statistical Analysis: Poisson Generalized
Linear Model (GLM). Statistical package:
SAS
Age groups analyzed: All ages
Covariates: Adjustment for day-of-wk,
hospital entry, holidays, time,
temperature, dew point temperature

# Hospitals: 41
N: 238,360 (Cardiovascular)
1,072,429 (Respiratory)
Lag(s): 3-day moving avg
1-h max (ppb):
14.9
Range: 1.0,
149.0
10th: 2.0
25th: 4.0
75th: 20.0
90th: 35.0

Copollutants:
PM10,PM25, O,
NO2, CO, SO4
Total Carbon
OC, EC
Water-Soluble
Metals
Oxygenated
Hydrocarbons
In single pollutant models, O3, PMio, CO, and NO2 significantly
associated with ED visits for respiratory outcomes.

Relative Risk (95% Cl) (per 16.0 ppb SO2)

1.003(0.997, 1.009)
May 2008
                                         F-61
                               DRAFT—DO NOT QUOTE OR CITE

-------
    STUDY
                               METHODS
                                                        POLLUTANTS
                                                                                                  FINDINGS
Wheeler et al.
(2006)

Atlanta, Georgia

1999-2000
30 individuals with myocardial infarction
or COPD were administered a
questionnaire and an HRV protocol.
Linear mixed-effect models were used to
analyze the data.
Mean: 1.9 ppb

Copollutants:
PM25, EC, 03
CO, NO2
No association with SO2
                                                              CANADA
Rich et al. (2004)

Vancouver,
British Columbia,
Canada

Feb-Dec 2000
Case-crossover analysis used to
investigate association between air
pollution and cardiac arrhythmia in 34
patients (aged 15-85 yrs, mean 62) with
implantable cardioverter defibrillators.
Study included only patients who
experienced at least 1 ICD discharge
during the study period. Control days
were 7 days before and 7 days after day
of ICD discharge. Conditional logistic
regression analyses were stratified by
individual.
24-h avg:
2.6 ppb
SD: 1.3 ppb
IQR: 1.6 ppb
Copollutants:
PM25, EC, OC
SO42", PM10, CO
N02, 03
No statistically significant association between SO2 and implantable
cardioverter defibrillator discharges. However, when an analysis was
stratified by season, OR for SO2 were higher in the summer
compared to winter.

No quantitative results provided. Results shown in graph.
Vedal et al.
(2004)

Vancouver,
British Columbia,
Canada

1997-2000
Retrospective, longitudinal panel study of
50 patients, aged 12-77 yrs with
implantable cardioverter defibrillators.
Total of 40,328 person-days over 4-yr
period. GEE used to assess associations
between short term increases in air
pollutants and implantable cardioverter
defibrillator discharges. Models controlled
for temporal trends, meteorology, and
serial autocorrelation.
24-h mean (SD)
SO2:2.4 (1.2)
ppb

Range: 0.3,
8.1 ppb
MediaN: 2.2 ppb
25th percentile:
1.5
75th percentile:
3.1

Copollutants:
PM10
03
NO2
CO
Concluded that in general no consistent effect of air pollution on
cardiac arrhythmias in this population. There were no statistically
significant associations between SO2 and cardiac arrhythmias at any
lag day, but positive associations at lag 2. When analysis was
restricted to only patients who had at least 2 arrhythmias per yr over
their period of observation (N: 16), a positive and significant
association was seen with SO2 at 2 days lag. When analysis was
restricted to patients averaging 3 or more arrhythmias per yr (N: 13),
there was no significant association, but a positive association was
seen at 2 days lag.

When stratified by season, SO2 effects were in the  in the positive
direction in the winter, but in the negative direction  in the summer.
Authors noted results may be due to chance because of multiple
comparisons or SO2 may be surrogate for some other factor.

Summer analysis: significant negative association with SO2 at lag
days 2 and 3 (data not shown). When stratified to patients with 2 or
more arrhythmia  event-days peryr, significant negative associations
observed with SO2 at lag of 3 days. Winter analysis: significant
positive effect of  SO2 at 3 days lag  (data  not shown). If restricted to
patients with at least 2 arrhythmias per yr, a significant positive
association was seen at lags 2 and 3 days. When restricted to
patients with 3 or more arrhythmia event days per yr, positive
associations observed for SO2 at lags of 2 and 3 days.

No quantitative results, but % change in arrhythmia event-day rate
for each SD increase in  pollution concentration on log scale provided
in figures.
                                                              EUROPE
Beger et al.
(2000)

Erfurt, Germany

Oct 2000-Apr
2001
Prospective panel study of 57 non-
smoking men, of which 74% are ex-
smokers, with coronary heart disease
aged 52-76 yrs old. Subjects underwent
24-h electrocardiogram (ECG) recordings
and analysis once every 4 wks.
Associations analyzed using Poisson and
linear regression modeling, for
supraventricular and ventricular
tachycardia, respectively, adjusting for
trend, weekday, and meteorologic data.
24-h avg (SD)
Ojg/m3):4.1(1.8)

Range: 3.0, 11.7

Copollutants:
Ultrafine
Particles
Accumulation
Mode Particles
PM2.5
PM10
NO2
CO
NO
UFP, ACP, PM25, and NO2 associated with increased risk for
supraventricular tachycardia and ventricular tachycardia at almost all
lags. The majority of statistically significant associations was
observed in the previous 24-71-h and with the 5-day moving avg.
Associations were not observed for increasing concentrations of
SO2. Relative Risk (per 1.5 pg/m3 SOJ

Supraventricular Extrasystoles
0.92 (0.77, 1.09) lag 0. 0.98 (0.86, 1.12) lag 0-23-h
1.04 (0.93, 1.16) lag 24-47 h. 1.14 (0.98, 1.34) lag 48-71-h
0.95 (0.83, 1.09) lag 72-95-h. 1.01 (0.80, 1.27) lag 5-d avg

Ventricular Extrasystoles
-2.1 (-6.1, 2.1) lag 0. -1.8 (-6.1, 2.7) lag 0-23-h
-0.1 (-4.4, 4.4) lag 24-47 h. 4.5 (-0.4, 9.5)  lag 48-71-h
-2.2 (-6.4, 2.3) lag 72-95-h. -1.2 (-7.5, 5.5) lag  5-d avg
Henrotin et al.
(2007)

Dijon, France

1994-2004
Bi-directional case-crossover design to
examine association between air
pollutant and ischaemic stroke onset
(2078 cases).
Mean: 6.9 pg/m
SD: 7.5
MiN:0
Max: 65

Copollutants:
SOX O3, CO
PM,n
SO2 not significantly associated with occurrence of strokes Odds
Ratio (95% Cl)

Ischaemic stroke:
DO: 0.978 (0.868, 1.103). D-1: 0.978 (0.863, 1.108)
D-2: 1.015(0.902, 1.143). D-3: 1.003(0.892, 1.127)

Hemorrhagic stroke:
DO: 1.099(0.815, 1.483). D-1: 1.014(0.747, 1.376)
D-2: 0.961 (0.712, 1.297). D-3: 0.954 (0.729, 1.248)
May 2008
                                          F-62
                                DRAFT—DO NOT QUOTE OR CITE

-------
    STUDY
                             METHODS
                                                      POLLUTANTS
                                                                                               FINDINGS
Ibald-Mulli et al.
(2001)

Augsburg,
Germany

1984-85, 1987-88
Retrospective analysis of 2607 subjects
(25-64 yrs, subset of the participants of
first and second MONICA survey who
had valid electrocardiograms recordings
in both surveys and blood pressure
measurements). Used regression models
for repeated measures that controlled for
age, current smoking, and cardiovascular
medication, BMI, total and high density
lipoprotein cholesterol, temp, RH, and
barometric pressure.
24-h avg SO2
(pg/m3)

1984-1985:
Mean: 60.2
SD:47.4
Range: 13.0,
238.2

follow up 1987-
1988
Mean: 23.8
SD: 12.3
Range: 5.6, 71.1

Copollutants:
TSP, CO
SO2 and TSP associated with increases in systolic blood pressure. In
the multipollutant model with TSP, the effect of TPS remained
significant, but the SO2 effect was substantially reduced. No clear
association between SO2 and CO and diastolic blood pressure was
observed. Same day concentrations: mean change in systolic blood
pressure per 5th to 95th percentile increase in SO2 (per 80 pg/m3)

Same day concentrations (per 80 pg/m3):
Men (N: 1339): 0.96 (0.07,  1.85)
Women (N: 1268): 0.96 (-0.46,  1.49)
Men and womeN: 0.74 (0.08, 1.40)
5-day  avgs: Mean change in systolic blood  pressure per 5th to 95th
percentile increase in SO2 (per 75 pg/m3)
MeN:  0.97 (0.09, 1.85). WomeN: 1.23 (0.23, 2.22)
Men and womeN: 1.07 (0.41, 1.73)
2-pollutant model. Men and womeN: 0.23 (-0.50, 0.96)
Peters et al.
(1999)

Augsburg,
Germany

Winter 1984-1985
Winter 1987-1988
Retrospective analysis on subsample of
2,681 subjects (25-64 yrs) of the
MONICA cohort who had valid
electrocardiogram readings from both
surveys and no acute infections. GEE for
clusters used to assess association
between heart rate and air pollution.
Analyses adjusted for temperature,
relative humidity, and air pressure.
24-h avg SO2
(pg/m3)

Winter 1984-85
Outside episode:
Mean: 48.1
SD:23.1
Range: 13, 103

Winter 1984-85
During episode:
Mean: 200.3
SD:26.6
Range: 160, 238

Winter: 1987-88
Mean: 23.6
SD: 12.2
Range: 6, 71

Copollutants:
CO, TSP
Increases in SO2 concentrations associated with increases in heart
rate

Mean change in heart rate per 5th to 95th percentile SO2
Same day concentrations
(per 80 pg/m3 SO2)
MeN: 1.02 (0.41, 1.63)
WomeN: 1.07(0.41, 1.73)
Men  and womeN: 1.04 (0.60, 1.49)

5-day avg (per 75 pg/m3 SO2)
MeN: 1.29(0.68, 1.90)
WomeN: 1.26(0.57, 1.95)
Men  and womeN: 1.28 (0.82, 1.74)
Ruidavets et al.
(2005)

Toulouse, France

1995-1997
Cross-sectional survey of 863 randomly
chosen adults (35-65 yrs) living in
Toulouse (MONICA center) to examine
the relationship between resting heart
rate and air pollution. Resting heart rate
was measured twice in a sitting position
after a five minute rest. Used polytomous
logistic regression models with quintiles
of RHR. Final model controlled for sex,
physical activity, systolic blood pressure,
cardiovascular drug use, CRP, relative
humidity, and season mos.
Mean SO2: 13.3
(7.5) Mg/m3

Range: 1.3,
47.7 pg/m3

Copollutants:
N02, 03
Marginally significant association between SO2 and RHR in Q5
compared with Q1. No association with SO2 at 1, 2, or 3 days lag.
OR based on daily levels of So2- OR for resting heart rate = 1.19
(95% Cl:  1.02, 1.39) in 5th quintile (>75 bpm) compared to first
quintile (< 60 bpm) for 5 pg/m3 increase in SO2 same day 0 am-12
pm. OR for resting heart rate 1.14 (95% Cl: 1.01 to 1.30) in 5th
quintile (>75 bpm) compared to first quintile (< 60 bpm) for 5pg/m3
increase in SO2 same day 12 am-12 pm
Not-significant associations not listed
                                                         LATIN AMERICA
Holguin et al.
(2003)

Mexico City,
Mexico

Feb 8 to Apr 30,
2000
Panel study of 34 nursing home residents
(60-96 yrs) to assess association
between heart rate variability and air
pollution. Heart rate variability measured
every alternate day for 3 mos. Thirteen of
the subjects had hypertension. Used
GEE models that controlled for age and
avg heart rate during HRV measurement.
24-h mean SO2
(ppb). Mean: 24
SD: 12
Range: 6, 85

Copollutants:
Indoor PM25
Outdoor PM25
O3, NO2, CO
SO2 not related to heart rate variability on the same day or lag 1 day

Change in HRV per 10 ppb
Beta Coefficient (95% Cl)
HRV-HF -0.003 (-0.035, 0.035)
HRV-LF -0.004 (-0.004, 0.003)
HRV-LF/HF 0.012 (-0.060, 0.082)
May 2008
                                         F-63
                               DRAFT—DO NOT QUOTE OR CITE

-------
Table F-4.   Associations of short-term exposure to SO2 with emergency department visits and
           hospital admissions for cardiovascular diseases.
STUDY METHODS
POLLUTANTS
EFFECTS
UNITED STATES
Gwynn et al.
(2000)
New York
(Buffalo;
Rochester)
1988-1990





Koken et al.
(2003)
Denver, United
States
Period of Study:
Jul and Aug,
1993-1997,
N: 310 days









Low et al.
(2006)
New York City,
NY
Period of Study:
1995-2003,
3287 days



Hospital Admissions
Outcome(s) (ICD9): Respiratory (466, 480-486),
Circulatory (401-405, 410-417), Total (minus 800)
from Statewide Planning and Research
Cooperative System (SPARCS)
Study design: Time-series
Statistical Analysis: Loess fits of temperature and
relative humidity
Age groups analyzed: All ages
Covariates: Adjustments for weather
Lag(s): 0, 3
Outcome(s) (ICD9):
Acute Ml 410.00-410.92;
Atherosclerosis 14.00-414.05;
Pulmonary Heart Failure 416.0-416.9;
Dysrhythmia 427.0-427.9; CHF 428.0.
Discharge data from Agency for Healthcare
Research and Quality (AHRQ) database.
Age group analyzed: 65+ yrs
Study populatioN: 60,000
Covariates : Seasonal adjustment not needed.
Adjustment for temperature, dew point
temperature made.
Study design: Time-series
Statistical Analysis: GLMs to analyze frequency of
admissions as a function of exposure. GEEs to
estimate parameters in Poisson regression
models, adjusting for overdispersion.
Lag(s): 0-4 day
Outcome(s) (ICD): Ischemic stroke 433-434;
Undetermined stroke 436; monitored intake in 1 1-
hospitals (ER or clinic visits). Excluded stroke
patients admitted for rehabilitation.
Study design: Time-series
Statistical Analysis: Autoregressive integrated
moving avg (ARIMA) models
Software package: SAS


24-h avg (ppb): 12.2
Range: 1.63, 37.7
H+
SO42"
PM10
Filled PM10
03
CO
NO2
CoH
SO2 24-h avg (ppb)
Mean (SD): 5.7 (2.94)
MiN: 0.4
25th: 3.8
50th: 5.3
75th: 7.2
Max: 18.9

O3(r = -0.10)
CO (r = 0.21)
PMio (r = 0.36)
NO2 (r = 0.46)






SO2 24-h avg (ppm)
Mean (SD): 0.009124
MiN:0
25th: 0.005
MediaN: 0.009
75th: 0.01 4
Max: 0.096
PM10 (0.042)
NO2 (0.33)
CO (0.303)
Pollen (0.085)
3=0.000245(0.000917)
t = 0.27
Relative Risk (per 7 ppb SO2)
1.002






Effects were reported as percent change in
hospitalizations based on an increment of
3.4 ppb.
Single-pollutant model
Dysrhythmia
8.9% (-0.34, 18.93) lag 0, adjusted for gender
but not temperature
SO2 was found to be associated with cardiac
dysrhythmia but not other outcomes. No
association was observed for PM or NO2 with the
outcomes.







At the highest concentration of SO2 (96 ppb) in
New York city over the study period the expected
increase in strokes would be 0.857 visits on the
day of the event.
Each 1000 ppb (1 ppm) SO2 would produce an
additional 8.878 visits (SE 4.471)
(p = 0.0471) for stroke.



May 2008
F-64
DRAFT—DO NOT QUOTE OR CITE

-------
    STUDY
                                 METHODS
                                                                   POLLUTANTS
                                                                                                         EFFECTS
Metzger et al.
(2004)
Atlanta, GA
Period of Study:
Jan 1993-Aug
31 2000, 4 yrs
Outcome(s): IHD 410-414; AMI 410;
Dysrhythmias 427; cardiac arrest 427.5;
congestive heart failure 428; peripheral and
cerebrovascular disease 433-437, 440,
443-444, 451-453; atherosclerosis 440; stroke
436. ED visits from billing records.
N: 4,407,535 visits, 37 CVD visits/days
# Hospitals: 31
Age groups analyzed: adults > 19, elderly 56+
Statistical Analysis: Poisson regression, GLM.
Sensitivity analyses using GEE and GAM (strict
convergence criteria)
Covariates: long-term trends, mean and dew
point temp,  relative humidity (cubic splines)
Statistical Software: SAS
Season: Warm, Apr 15-Oct 14,
Cool.Oct 15-Apr 14.
Lag(s): 0-3 days
SO21-h max (ppb)

MediaN: 11.0
10th-90th Range: 2.0 to
39 ppb

PM10 (0.20)
03(0.19)
NO2 (0.34)
CO (0.26)
PM25(0.17)
PM10.2.6(0.21)
Ultrafine (0.24)

Multipollutant models used.
All models specified a priori.
Results presented for RR of an incremental
increase in SO2 of 20 ppb (a priori lag 3 day
moving avg).

All CVD: 1.007(0.993, 1.022)

Dysrhythmia: 1.001  (0.975, 1.028)

CHF: 0.992 (0.961,  1.025)

IHD: 1.007(0.981, 1.033)

PERI: 1.028(0.999, 1.059)

Finger wounds 1.007 (0.998, 1.026)

Single day lag models presented graphically.

No multipollutant models run for SO2 since
association was not observed in single-pollutant
models.
Michaud et al.
(2004)
Hilo, Hawaii

Period of Study:
1997-2001,
N: 1385 days
Outcome(s) (ICD9):
Cardiac 410-414, 425-429,
Emergency visits, primary diagnosis.

Study design: Time-series

Statistical Analysis: Exponential regression,
autocorrelation assessed by regressing square
root of number of ED visits on covariates (Durbin-
Watson statistic). Newey-West procedure also
conducted for assessment of autocorrelation.

Covariates: Temperature, humidity, interaction
between SO2 and PM

Lag(s): 1-3 days
SO2 (all hourly
measurements) (ppb)
Mean (SD): 1.92(12.2)
MiN:0
Max: 447
Daily SO2
(12am-6am) (ppb)
Mean (SD): 1.97(7.12)
MiN:0
Max: 108.5
PM
Effects were presented as relative risk based on
an increment of 10 ppb and the 24-h avg SO2
concentration.

Cardiac
0.92(0.85, 1.00) lag 3

No associations of cardiac ER visits with VOG
(SO2-acidic aerosols) observed.
Moolgavkar
(2000)*

Cook County IL,
Los Angeles
County, CA,
Maricopa
County, AZ

1987-1995
Outcome(s) (ICD9): CVD
390-429; Cerebrovascular disease 430-448.
Hospital admissions from CA department of
health database.
Age groups analyzed: 20-64, 65+ yrs
Study design: Time-series
N: 118 CVD admissions/days
# Hospitals: NR
Statistical Analysis: Poisson regression, GAM
Covariates: Adjustment for day of wk, long-term
temporal trends, relative humidity, temperature
Statistical package: SPLUS
Lag: 0-5 days
SO2 24-h avg (ppb)

Cook County:
MiN:0.5;Q1:4
MediaN: 6; Q3: 8
Max: 36

LA County:
MiN:0;Q1:  1
MediaN: 2; Q3: 4
Max: 16

Maricopa County:
MiN:0;Q1:0.5
MediaN: 2; Q3: 4
Max: 14

PM10 (0.11,0.42)
PM25 (0.42) (LA only)
CO (0.35, 0.78)
NO2 (0.02, 0.74)
O3 (-0.37, 0.01)
Results reported for percent change in hospital
admissions per 10 ppb increase in SO2. T
statistic in parentheses.

CVD, 65+:
Cook County
4.0 (6.1), lag 0
3.1 (4.5), lag 0, 2-pollutant model (CO)
1.0 (1.4), lag 0, 2-pollutant model (NO2)
LA County
14.4(15.2), lag 0
-2.5 (-1.6), lag 0, 2-pollutant model (CO)
7.7 (5.7), lag 0, 2-pollutant model (NO2)
Maricopa County
7.4 (4.5), lag 0
3.0 (1.8), lag 0, 2-pollutant model (CO)
3.9 (1.5), lag 0, 2-pollutant model (SO2)
Cerebrovascular Disease, 65+:
Cook County
3.1 (3.3)
LA County
6.5 (4.9)

Lags 1-5 also presented. Effect size generally
diminished with increasing lag time. Increase in
hospital admissions (10.3 for CVD and 9.0 for
cerebrovascular) also observed for the
20-64 age group.
May 2008
                                          F-65
                        DRAFT—DO NOT QUOTE OR CITE

-------
    STUDY
                                 METHODS
                                                                   POLLUTANTS
                                                                                                         EFFECTS
Moolgavkar
(2003a)

Cook County IL,
Los Angeles
County, CA,
Maricopa
County, AZ

1987-1995
Outcome(s) (ICD9): CVD
390-429; Cerebrovascular disease 430-448 was
not considered in the reanalysis. Hospital
admissions from CA department of health
database.
Age groups analyzed: 20-64,
65+yrs
Study design: Time-series
N: 118 CVD admissions/day
# Hospitals: NR
Statistical Analysis: Poisson regression, GAM
with strict convergence criteria (10-8), GLM using
natural splines
Covariates: Adjustment for day of wk, long-term
temporal trends, relative humidity, temperature
Statistical package: SPLUS
Lag: 0-5 days
See original analysis
(Moolgavkar, 2000) above.
See original analysis
(Moolgavkar, 2000) above.
Use of stringent criteria in GAM did not alter
results substantially. However, increased
smoothing of temporal trends attenuated results
for all gases and effect size diminished with
increasing lag.

Results reported for incremental increase of 10
ppb SO2. Estimated coefficient and T statistic in
parentheses.

GLM with 100 df (LA County)

13.67(11.82), lagO

6.44 (5.23), lag 1

0.23(0.18), lag 2
Morris et al.
(1995)

U.S. (Chicago,
Detroit, LA,
Milwaukee,
NYC,
Philadelphia)

Period of Study:

1986-1989,4
yrs
Outcome(s) (ICD9): CHF 428. Daily Medicare
hospital admission records.
Study design: Time-series
Statistical analyses: GLM, negative binomial
distribution
Age groups analyzed: > 65 yrs
Covariates: Temperature, indicator variables for
mo to adjust for weather effects and seasonal
trends, day of wk, yr
Statistical software: S-PLUS
Lag(s): 0-7 days
SO21-h max (ppm)

Mean (SD)

LA: 0.010 (0.005)
Chicago: 0.025 (0.011)
Philadelphia: 0.029 (0.015)
New York: 0.032 (0.015)
Detroit: 0.025 (0.013)
HoustoN: 0.018 (0.009)
Milwaukee: 0.017(0.013)

NO2
03
CO

Correlations of SO2 with
other pollutants strong.
Results reported for RR of admission for CHF
associated with an incremental increase in SO2
of 0.05 ppm.

CHF:
LA: 1.60(1.41, 1.82)
Chicago: 1.05(1.00, 1.10)
Philadelphia: 1.01 (0.96, 1.06)
New York: 1.04 (1.01, 1.08)
Detroit: 1.00(0.95, 1.06)
HoustoN: 1.07(0.97, 1.17)
Milwaukee: 1.07(0.99, 1.15)

RR diminished in multipollutant (4 copollutants)
models for all cities.
Peel et al. (Peel
et al., 2007)

Atlanta, GA

Period  of Study:

Jan 1993-Aug
2000
Outcome(s) (ICD9): IHD 410-414; dysrhythmia
427; CHF 428; peripheral vascular and
cerebrovascular disease
433-437, 440, 443, 444, 451-453. Computerized
billing records for ED visits.

Comorbid conditions: Hypertension
401-405; diabetes 250; dysrhythmia 427, CHF
428; atherosclerosis 440; COPD 491, 492, 496;
pneumonia 480-486; upper respiratory infection
460-465, 466.0; asthma 493, 786.09.

# Hospitals: 31

N: 4,407,535 visits

Study design: Case-crossover and time-series.
CVD outcomes among susceptible groups with
comorbid conditions.

Statistical analyses: Conditional logistic
regression and Poisson GLM.

Covariates: Cubic splines for temperature and
humidity included in models. Time  independent
variables controlled through design.

Statistical Software: SAS

Lag(s): 3 day avg, lagged 0-2 day
SO21-h max (ppb)

Mean (SD): 16.5(17.1)

10th: 2

90th: 39

PM10

03
NO2

CO
Results expressed as OR for association of CVD
admissions with a 20 ppb incremental increase
in SO2.

Case-Crossover:
All CVD
1.009(0.995,1.024), 0-2 avg
IHD
1.013 (0.988, 1.039), 0-2 avg
Dvsrhvthmia
1.003(0.975, 1.031), 0-2 avg
Peripheral and  Cerebrovascular
1.024 (0.993, 1.055), 0-2 avg
CHF
0.993 (0.961, 1.026), 0-2 avg
Time-series:
Odds Ratio (95% Cl) (per 20 ppb SO2)
All cardiovascular disease:
1.007(0.993, 1.022)
Ischemic heart  disease:
1.007 (0.981, 1.003)
Dvsrhvthmia:
1.001 (0.975, 1.028)
Peripheral and  cerebrovascular disease:
1.028 (0.999, 1.059)
Congestive heart failure:
0.992 (0.961, 1.025)
Effect modification by comorbid conditions was
not observed.
May 2008
                                          F-66
                        DRAFT—DO NOT QUOTE OR CITE

-------
    STUDY
                                 METHODS
                                                                   POLLUTANTS
                                                                                                         EFFECTS
Schwartz and
Morris (1995)*

Detroit, Ml

Period of Study:

1986-1989
Outcome(s) (ICD9): IHD 410-414; CHF 428;
Dysrhythmia 427. Medicare data, diagnosis at
discharge.
Study design: Time-series
Statistical Analysis: Poisson regression, GAM
Age groups analyzed: 65+ yrs
Covariates: Adjustments for long-term patterns,
temperature, humidity, days of the wk, holidays,
viral infections, etc.
Lag(s): 0-3, cumulative up to 3 days
SO2 24-h avg (ppb):

Mean: 25.4
IQR: 18 ppb
Q2: 15
Q3:33

# Stations: 6

PM10 (0.42)
CO (0.23)
03(0.15)
Effects were expressed as relative risk based on
an increment of 18 ppb.
IHD
1.014 (1.003, 1.026) lag 0, single-pollutant
1.009 (0.994, 1.023), 2-pollutant model with
PM10
CHF
1.002 (0.978, 1.017), single-pollutant model
Risks for dysrhythmia were NR for SO2.
Schwartz
(1997)*

Tuscon, AZ

Period of Study:

Jan1988-Dec
1990
Outcome(s) (ICD9): CVD 390-429. Ascertained
from hospital discharge records.
Study design: Time-series
Statistical Analysis: Poisson regression, GAM
Age groups analyzed: 65+
Covariates: Long-term and seasonal trends, day
of the wk, temperature, dew point,
Statistical software: S-PLUS
SO2 24-h avg (ppb)

Mean: 4.6 ppb
IQR: 3.9 ppb
10th: 0.7
Q2: 2.0
MediaN: 3.4
Q3: 5.9
90th: 10.1

PM10 (0.095)
NO2 (0.482)
CO (0.395)
03 (-0.271)
Results were expressed as percent change
based on an increment of 3.9 ppb.
0.14% (-1.3%, 1.6)
No other statistically significant associations for
cardiovascular outcomes were observed.
Tolbert et al.
(2007)

Atlanta, GA

1993-2004
ED Visits. Outcome(s) (ICD9): Cardiovascular
(410-414, 427, 428, 433-437, 440, 443-445, 451-
453); Respiratory (493, 786.07, 786.09, 491, 492,
496, 460-465, 477, 480-486, 466.1, 466.11,
466.19). Study design: Time-series
Statistical Analysis:  Poisson Generalized Linear
Model (GLM). Statistical package: SAS

Age groups analyzed: All ages. Covariates:
Adjustment for day-of-wk, hospital entry, holidays,
time, temperature, dew point temperature

# Hospitals: 41. N: 238,360 (Cardiovascular);
1,072,429 (Respiratory). Lag(s): 3-day moving
avg
1-h max (ppb): 14.9

Range:

1.0, 149.0

PM10
PM2.5
03
NO2
CO
Sulfate
Total Carbon
Organic Carbon
EC
Water-Soluble Metals
Oxygenated Hydrocarbons
Relative Risk (95% Cl)
(per 16.0 ppbSOJ

1.003(0.994, 1.011)
Wellenius et al.
(2005b)
Birmingham,
Chicago,
Cleveland,
Detroit,
Minneapolis,
New Haven,
Pittsburgh,
Seattle
Period of Study:
Jan 1986-Nov
1999 (varies
slightly
depending on
city)
Outcome(s) IS, primary diagnosis of acute but ill-
defined cerebrovascular disease or occlusion of
the cerebral arteries; HS, primary diagnosis of
intracerebral hemorrhage. ICD codes not
provided. Hospital admissions ascertained from
the Centers for Medicare and Medicaid Services.
Cases determined from discharge data were
admitted from the ER to the hospital.

N IS: 155,503

N HS: 19,314

Study design: Time-stratified Case-crossover.
Control days chosen such that they fell in same
mo and same day of wk. Design controls for
seasonality, time trends, chronic and other slowly
varying potential confounders.

Statistical Analysis:  2-stage hierarchical model
(random effects), conditional logistic regression
for city effects in the first stage

Software package: SAS

Covariates:

Lag(s): 0-2, unconstrained distributed lags
SO2 24 h avg (ppb)

10th: 2.17

25th: 3.57

MediaN: 6.22

75th: 10.26

90th: 16.17

SO2 data not available for
Birmingham, AL

PM10 (0.39)
CO, NO2
Results reported for percent increase in stroke
admissions for an incremental increase in SO2
equivalent to one IQR (6.69).
Ischemic Stroke: 1.35 (0.43, 2.29), lag 0
Hemorrhagic Stroke: 0.68 (-1.77, 3.19)
Multipollutant models not run.
May 2008
                                          F-67
                        DRAFT—DO NOT QUOTE OR CITE

-------
    STUDY
                                  METHODS
                                                                    POLLUTANTS
                                                                                                          EFFECTS
Wellenius et al.
(2005a)
Allegheny
County, PA
(near
Pittsburgh)
Period of Study:
Jan 1987-Nov
1999
Outcome(s): CHF 428. Cases are Medicare
patients admitted from ER with discharge of CHF
Study design: Case-crossover, control exposures
same mo and day of wk, controlling for season by
design.
Statistical Analysis: Conditional logistic regression
N: 55,019 admissions, including repeat
admissions, 86% admitted < 5 times
Age groups analyzed: 65+ yrs (Medicare
recipients)
Covariates: Temperature and pressure. Effect
modification by age, gender, secondary diagnosis
arrhythmias, atrial fibrillation, COPD,
hypertension, type 2 diabetes, AMI within 30
days, angina pectoris, IHD, acute respiratory
infection.
Statistical software: SAS
Lag(s): 0-3
SO2 24-h avg (ppb):

Mean (SD): 14.78(9.88)

5th: 3.98

25th: 7.70

MediaN: 12.24

75th: 18.98

95th: 33.93

# Stations: 10

PM10 (0.51)
CO (0.54)
NO2 (0.52)
O3(-0.19)
Effects were reported as percent change based
on an increment of 11 ppb.

CHF, single-pollutant models:
2.36(1.05, 3.69) lag 0, or
2.14 (0.95, 3.35) lag 0 after adjusted to an
increment of 10 ppb.

CHF, 2-pollutant models:
1.35 (-0.27, 2.99), SO2with PM10
0.10 (-1.35, 1.57), SO2with CO

0.68 (-0.82, 2.21), SO2 with NO2

2.02 (0.68, 3.37), SO2 with O3
                                                              CANADA
Burnett et al.
(1997a)*

Metropolitan
Toronto
(Toronto, North
York, East
York,
Etobicoke,
Scarborough,
York), Canada

Period of Study:

1992-1994,388
days, summers
only
Outcome(s) (ICD9):
IHD 410-414;
Cardiac Dysrhythmias 427;
Heart failure 428. All Cardiac
410-414, 427, 428. Obtained from hospital
discharge data.

PopulatioN: 2.6 Million residents

Study design: Time-series

Age groups analyzed: All

# Hospitals: NR

Statistical Analysis: Relative risk regression
models, GAMs.

Covariates: Adjusted for long- term trends,
seasonal and subseasonal variation, day of the
wk, temperature, dew point

Season: Summer only

Dose response:  Figures presented

Lag:  1-4 days
SO2 daily 1-h max (ppb):
Mean: 7.9
CV:64
MiN:0
25th percentile: 4
50th percentile: 7
75th percentile: 11
Max: 26
# of Stations: 4-6

(Results are reported for
additional metrics including
24-h avg and daytime avg
(day)

H+ (0.45)
SO4 (0.42)
TP  (0.55)
FP  (0.49)
CP  (0.44)
COH (0.50)
03(0.18)
NO2 (0.46)
CO (0.37)
Effects were expressed as relative risk based on
an increment of 7.00 ppb (IQR). T ratio in
parentheses.
All cardiac disease
Single-pollutant model
1.041 (2.66), daily max over 4 days, lag 0
Multipollutant model w/ SO2, O3, NO2
Of 7.72 excess hospital admissions, 2.8%
attributed to SO2.
Objective of study was to evaluate the role of
particle size and chemistry on cardiac and
respiratory diseases.
Burnett et al.
(1999)*

Metropolitan
Toronto
(Toronto, North
York, East
York,
Etobicoke,
Scarborough,
York), Canada

Period of Study:

1980-1995, 15
yrs
Outcome(s) (ICD9): IHD 410-414; Cardiac
Dysrhythmias 427; Heart failure 428; All cardiac
410-414, 427, 428; Cerebrovascular Disease
obtained from hospital discharge  data 430-438;
Peripheral Circulation Disease 440-459.
PopulatioN: 2.13-2.42 million residents
Study design: Time-series
Statistical Analysis: GAMs to estimate log RR per
unit changes, stepwise regression used to select
min number of air pollutants in multipollutant
models.
Covariates: Long-term trends, seasonal variation,
day of wk, temperature, and humidity.
Statistical package: S-PLUS
Lag(s): 0-2 days
SO2 daily avg (ppb)

Mean: 5.35

5th percentile: 0

25th percentile: 1

50th percentile: 4

75th percentile: 8

95th percentile: 17

Max: 57

Multiple day avgs used in
models

PM2.5 (0.50)

PM10.2.5 (0.38)
PM10 (0.52)
CO (0.55)

SO2 (0.55)

O3 (-0.04)
Effects were reported as % change based on an
increment of 5.35 ppb.

Single-pollutant model
Dysrhythmias 0.8% (-0.3,  1.9)
Cerebrovascular 0.04% (-0.7, 0.8)
CHF 1.93% (0.9, 2.9)
IHD 2.32% (1.6, 3.1)

Attributed percent increase in admissions for
SO2 were determined from multipollutant models.
IHD
Attributed percent increase: 0.95%

Authors note SO2 effects could be largely
explained by other variables  in the pollution mix
as demonstrated by the multipollutant model.
May 2008
                                          F-68
                        DRAFT—DO NOT QUOTE OR CITE

-------
    STUDY
                                 METHODS
                                                                   POLLUTANTS
                                                                                                         EFFECTS
Fung et al.
(2005)
Windsor,
Ontario,
Canada
Period of Study:
Apr 1995-Jan
2000
Outcome(s) (ICD9): CHF 428; IHD 410-414;
dysrhythmias 427 and all cardiac. Hospital
admissions from Ontario Health Insurance Plan
records.
Study design: Time-series
Statistical Analysis: GLM
N: 11,632 cardiac admission, 4.4/day for 65+ age
group
Age groups analyzed: 65+, < 65 yrs
Statistical Software: SPLUS
Lag(s): lag 0, 2, 3 day avg
SO21-h max (ppb)

Mean (SD): 27.5 (16.5)
MiN:0
Max: 129

IQR: 19.3 ppb

CO (0.16)

03 (-0.02)

PM10 (0.22)
NO2 (0.22)
Effects were expressed as percent change of
cardiac disease hospital admissions based on an
increment of 19.3 ppb.
Single-pollutant model:
< 65 yrs
2.3% (-1.8, 6.6) lag 0
3.9% (-1.5, 9.6) lag 0-1
3.4% (-3.0, 10.1) lag 0-2
> 65 yrs
2.6% (0.0, 5.3) lag 0
4.0% (0.6, 1.6) lag 0-1
5.6% (1.5, 9.9) lag 0-2

Inclusion of particulate matter and adjustment for
meteorological variables did not change the
association  between SO2 and cardiac
hospitalization.
Stieb et al.
(2000) *
Saint John,
New Brunswick
Canada
Period of Study:
July 1992-Mar
1996
Outcome(s): Angina pectoris; Ml;
dysrhythmia/conduction disturbance; CHF; All
Cardiac. ED Visits collected prospectively.

Study design: Time-series

Statistical analyses: Poisson regression, GAM

N: 19,821 ER visits

# Hospitals: 2

Lag(s): 1-8 days
SO2 24-h avg (ppb)
Mean (SD): 6.7 (5.6)
95th: 18
Max: 60
SO2 max (ppb)
Mean (SD): 23.8 (21.0)
95th: 62
Max: 161
CO (0.31)
H2S (-0.01)
O3 (-0.02)
NO2(0.41)
PM10 (0.36)
PM25(0.31)
H+ (-0.24)
SO4 (0.26)
COH (0.31)
Results reported for percent change in
admissions based on a single-pollutant model for
incremental increase in NO2 equivalent to one
IQR (8.9 ppb)
Cardiac visits (p-value in parentheses):
4.9 (0.002), 1 day avg, lag 8, all yr
2.8 (0.067), 5 day avg, lag 6, May-Sept
Multi-pollutant models:
4.9, (1.7,8.2), 1 day avg, lag 8, allyr(O3)
Lags 0-10 presented graphically. All but lag 8 in
single-pollutant model approximately null.
Villeneuve et al.
(2006)
Edmonton,
Canada
Period of Study:
Apr 1992-Mar
2002
Outcome(s) (ICD9): Acute ischemic stroke 434,
436; hemorrhagic stroke 430, 432; transient
ischemic attach (TIA) 435; Other 433, 437, 438.
ED visits supplied by Capital Health.

N: 12,422
Stroke Visits Catchment area: 1.5 million people

Study design: Case-crossover, exposure index
time compared to referent time. Time
independent variables controlled in the design.
Index and referent day matched by day of wk.

Statistical Analysis: Conditional logistic
regression, stratified by season and gender.

Covariates: Temperature and humidity

Statistical software: SAS

Season:
Warm: Apr-Sep;
Cool: Oct-Mar.

Lag(s): 0, 1, 3 day avg
SO2 24 h avg ppb:
Allyr
Mean (SD): 2.6 (1.9)
MediaN: 2.0
25th: 1.0
75th: 4.0
IQR: 3.0

Summer
Mean (SD): 2.1 (1.6)
MediaN: 2.0
25th: 1.0
75th: 3.0
IQR: 2

Winter
Mean (SD): 3.1 (2.0)
MediaN: 3.0
25th: 2.0
75th: 4.0
IQR: 2.0

Correlation  between SO2
and other pollutants (all yr):
NO2 (0.42)
CO (0.41)
03 (-0.25)
PM25(0.22)
PM10(0.19)
Effects were reported as odds ratios based on
an increment of 3 ppb.
Acute Ischemic stroke, > 65 yrs
Allyr OR 1.05(0.99,1.11)
lag 0
Warm OR 1.11 (1.01, 1.22)
lag 0
Cold OR 1.00 (0.93, 1.09)
lagO
Effect stronger among males
Hemorrhagic stroke, > 65 yrs
All yr:  0.98 (0.90, 1.06), lag 0
Cold: 0.94 (0.84, 1.05), lag 0
Warm: 1.03(0.90, 1.17)
Effect stronger among males
Transient Cerebral Ischemic Attack, > 65 yrs
Allyr:  1.06(1.00, 1.12), lag 0
Cold: 1.03(0.95, 1.11), lag 0
Warm OR 1.11 (1.02, 1.22)
lagO
2-pollutant models presented graphically.
Association of SO2 with Acute Ischemic stroke
diminished with inclusion of CO and NO2.
May 2008
                                          F-69
                        DRAFT—DO NOT QUOTE OR CITE

-------
STUDY METHODS
POLLUTANTS
EFFECTS
EUROPE
Anderson et al.
(Anderson et
al., 2001)*
West Midlands
conurbation, UK
Period of Study:
1994-1996,
N: 832 days










Atkinson et al.
(Atkinson et al.,
1999a)
England
Period of Study:
1992-1994,
N: 1,096 days















Outcome(s) (ICD9): All CVD 390-459; cardiac
disease 390-429; IHD 410-414; stroke 430-438.
Emergency admissions counted.
Catchment area: 2.3 million
Age groups analyzed: 0-14, 15-64, > 65.
Study design: Time-series, APHEA 2 methods
Statistical analyses: GAMs for modeling non-liner
dependence of some variables.
Covariates: Adjusted for effects of seasonal
patterns, temperature and humidity, influenza
episodes, day of wk and holidays.
Software package: S-PLUS
Season: Interaction by warm and cool season
investigated.
Lag(s): 0-3 days


Outcome(s) (ICD9):
All CVD 390-459; IHD 410-414. Emergency
admissions obtained from the Hospital Episode
Statistics (HES) database (complaints).
Ages groups analyzed: 0-14 yrs, 15-64 yrs, 0-64
yrs, 65+ yrs, 65-74 yrs, 75+ yrs
Study design: Time-series, hospital admission
counts
N: 189,109 CVD admissions
Catchment area: 7 million residing in 1,600 Km2
area of Thames basin.
Statistical analyses: APHEA protocol, Poisson
regression
Covariates: Adjusted long-term seasonal
patterns, day of wk, influenza, temperature,
humidity (compared alternative methods for
modeling meteorological including linear,
quadradic, piece-wise, spline)
Season: Warm season Apr-Sep, cool season
remaining mos, interactions between season
investigated
Dose response investigated: Yes, bubble charts
presented
Statistical package: SAS
Lag: 0-3
Dose response: Bubble plots presented
SO2 24-h avg (ppb)
Mean (SD): 7.2 (4.7)
MiN: 1.9
10th: 3.3
MediaN: 5.8
90th' 123

Max: 59.8
# of Stations: 5 sites
p[y] /Q 55)
PM-, c (0 52)
' IVI2.5 \*J-'J£-/
PM2.^io (0.31)
BS (0.50)
SO4(0.19)
NO2 (0.52)
O3 (-0.22)
SO2 24 h avg (ppb):
Mean: 21.2
SD: 7.8
MiN: 7.4
10th: 13
MediaN: 19.8

90th: 31
Max' 82 2
10th-90th percentile: 11.2
# of Stations: 3, results
averaged across stations
PM10
CO
SO2
03
BS

Correlations of SO2 with
CO, NO2, O3, BS ranged
from 0.5-0.6

Correlation of SO2 with O3
negative


Results reported for % change in admissions,
increment = 9 ppb (10th-90th).
All CVD all ages
-0.4 (-2.2, 1.5), mean lags 0 + 1
Cardiac all ages:
0.7 (-1 .3, 2.8), mean lags 0 + 1
IHD > 65 yrs

1 .5 (-2.5, 5.6), mean lags 0 + 1
Stroke > 65 yrs
-5.1 (-9.6, -0.4), mean lags 0 + 1






Results reported for % change in admissions,
increment 10th-90th percentile (11.2 ppb).
All CVD, all ages
1.57(0.22,2.93), lag 0
All CVD, 0-64 yrs
2.44 (0.3, 4.63), lag 0
All CVD, 65+
1.72(0.15, 3.32), lag 0
IHD, 0-64 yrs
-2.03 (-5.35, 0.91), lag 2
IHD, 65+
3.10(0.61,5.65), lag 0
Effect size and significance diminished in models
containing SO2 and BS.









May 2008
F-70
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Ballester et al.
(2001) *
Valencia, Spain
Period of Study:
1992-1996















Ballester et al.
(2006)
Multicity, SpaiN:
Barcelona,
DilKo^
mirjao,
Castellon,
Gijon, Huelva,
Madrid,
Granada,
Oviedo, Seville,
Valencia,
Zaragoza
Period of Study:
1995/1996-
1999, N: 1,096
days













METHODS
Outcome(s) (ICD9):
All CVD 390-459;
heart diseases 410-414, 427, 428;
cerebrovascular diseases 430-438. Admissions
from city registry - discharge codes used.

Study design: Time-series
N: 1080 CVD admissions
# of Hospitals: 2
Catchment area: 376,681 inhabitants of Urban
Valencia
Statistical analyses: Poisson regression, GAM,
APHEA/ Spanish EMECAM protocol. Both Linear
and nonparametric model, including a loess term
was fitted, departure from linearity assess by
comparing deviance of both models.
Covariates: Long-term trend and seasonality,
temperature and humidity, weekdays, flu, special
events, air pollution.
Season:
Hot season May to Oct;
Cold season Nov to Apr
Statistical package: SAS
Lag: 0-4
Outcome(s) (ICD9):
All CVD 390-459;
Heart diseases 410-414,427,428. Emergency
admission from hospital records. Discharge data
used.
Study design: Time-series, meta-analysis to pool
cities
N: Daily mean admissions reported by city.
Statistical analyses: Poisson regression and
GAM, with stringent convergence criteria, meta-
analysis with random effect model. Tested
linearity by modeling pollutant in linear and non-
linear way (spline smoothing). Linear model
provided best results 55% of time but used in all
cases to facilitate comparability.
Covariates: Temperature, humidity and influenza,
day of wk unusual events, seasonal variation and
trend of the series
Season'
Hot: May to Oct;
Cold: Nov to Apr

Statistical package: S-PLUS
Lag: 0-3






POLLUTANTS
24 h avg (pg/m3):
Mean: 25.6
SD:NR
MiN' 4 4

Max: 68.4
mediaN: 25
# of Stations: 14 manual, 5
automatic
Correlation among stations:
0.3-0.62 for BS, 0.46-0.78
for gaseous pollutants
CO (0.74)
NO2 (0.22)
O3 (-0.35)
BS (0 63)






SO2 24-h avg (pg/m2)
Mean 10th 90th
Barcelona: 15.5, 6.6, 27.9
Bilbao: 18.6, 10.2,29.3
Cartagena: 27.1, 14.6,40.8
CastelloN: 7.7, 3.8, 12.7
GijoN:29.4, 10.3,52.4
Granada: 19.1, 8.8, 31.5
Huelva: 11.9,4.5,22.6
Madrid: 21. 8, 8.7, 41.8
Oviedo: 40.9, 16.3,75.5
Pamplona: 7.6, 1.8, 17.0
Seville: 9.6, 5.6, 14.6
Valencia: 16.6, 9.4, 24.4
Vigo:9.3, 2.6, 18.2
Zaragoza: 9.3, 2.0, 19.9
# of Stations: Depends on
the city
Correlation among stations:
Correlations between SO2
stations within cities poor.
CO (0.58)

O3 (-0.03)
NO2 (0.46)
BS (0.24)
TSP (0.31)
PM10 (0.46)
Correlations reported are
the median for all cities.
EFFECTS
Results expressed as relative risk, increment of
10 pg/m3.
All CVD
1.0302(1.0042, 1.0568), lag 2

Heart disease
1.0357(1.0012, 1.0714), lag 2
Cerebrovascular disease
1 .0378 (0.9844 to 1 .0940), lag 5
Digestive diseases
1.0234(0.9958, 1.0518), lag 1
All CVD, hottest semester
1.050(1.010, 1.092), lag 2
Effect size for all CVD and cerebrovascular
disease diminished in 2-pollutant models.






Results reported for % change in admissions,
increment 10 (pg/m3).
All cardiovascular
1.33% (0.21, 2.46) lag 0-1
Heart diseases
1.72% (0.50, 2.95) lag 0-1
Single day lags presented graphically. Effect size
decreased with increasing lag.
Multi-pollutant results presented graphically.
Control for CO and particulates diminished SO2
effects.













May 2008
F-71
DRAFT—DO NOT QUOTE OR CITE

-------
    STUDY
                                 METHODS
                                                                   POLLUTANTS
                                                                                                         EFFECTS
D'lppoliti et al.
(2003)
Rome, Italy
Period of Study:
Jan 1995-June
1997
Outcome(s) (ICD): AMI 410 (first episode).
Computerized hospital admission data.
Study design: Case-crossover, time stratified,
control days within same mo falling on the same
day.
Statistical analyses: Conditional logistic
regression, examined homogeneity across co-
morbidity categories
N:6531 cases
Age groups analyzed: 18-64 yrs,
65-74 yrd,  > 75
Season: Cool: Oct-Mar;
Warm: Apr-Sep
Lag(s): 0-4 day, 0-2 day cum avg
Dose Response: OR for increasing quartiles
presented  and p-value for trend.
SO2 24 h avg (pg/m3)
All yr:
Mean (SD): 9.5 (6.0)
25th: 5.4
50th: 8.2
75th: 12.6
IQR: 7.2
Cold season:
Mean (SD): 12.7(6.5)
Warm Season:
Mean (SD): 88.3 (15.4)
# Stations: 5
TSP (0.29)
NO2 (0.37)
CO (0.56)
Results reported as odds ratios for increment
equal to one IQR (7.2 pg/m3).
AMI
Quartile I  (referent)
Quartile II
0.987 (0.894, 1.089), lag 0-2
Quartile III
1.008(0.892, 1.140), lag 0-2
Quartile IV
1.144(0.991, 1.321), lag 0-2
Results at various lags NR for SO2.
Llorca et al.
(2005)
Torrelavega,
Spain
Period of Study:
1992-1995
Outcome(s) (ICD): CVD (called cardiac in paper)
390-459. Emergency admissions, excluding
nonresidents. Obtained admissions records from
hospital admin office.
Study design: Time-series
Statistical analyses: Poisson regression, APHEA
protocol
Covariates: Rainfall, temperature, wind speed
direction
N: 18,137 admissions
Statistical software: STATA
Lag(s): NR
SO2 24 h avg pg/m3:
Mean (SD): 13.3(16.7)
TSP  (-0.40)
NO2 (0.588)
SH2 (0.957)
NO (0.544)
Results expressed as rate ratios. Increment =
100 pg/m3.
Cardiac admissions, single-pollutant model
0.94(0.84, 1.05)
Five-pollutant model
1.09(0.83, 1.42)
All cardiorespiratory admissions, single-pollutant
model
RR 0.98 (0.89, 1.07)
Five-pollutant model
0.98(0.80, 1.21)
Poloniecki et al.
(1997)
London, UK
Period of Study:
Apr, 1987-Mar
1994, 7 yrs
Outcome(s): All CVD 390-459; Ml 410; Angina
pectoris 413; other IHD 414; ARR 427;
congestive heart failure 428; cerebrovascular
disease 430-438. Hospital Episode Statistics
(HES) data on emergency hospital admissions.
Study design: Time-series
N: 373, 556 CVD admissions
Statistical analyses: Poisson regression with
GAM, APHEA protocol
Covariates: Long-term trends, seasonal variation,
day of wk, influenza, temperature and humidity.
Season: Warm, Apr-Sep;
Cool, Oct-Mar
Lag: 0-1
SO224 h avg ppb:
MiN:0
10%: 2
MediaN: 6
90%: 21
Max: 1
Black Smoke
CO 24 h avg
NO2 24 h avg
038h
Correlations between
pollutants high but not
specified
Effects were expressed as relative risk based on
an increment of 19 ppb
(10th-90th percentile).
Single-pollutant models (lag 0-1)
Ml: 1.0326(1.0133, 1.0511)
Angina: 1.0133 (0.9907, 1.0383)
IHD: 0.9944 (0.9651, 1.0239)
ARR: 1.0181 (1.0000, 1.0448)
CHF: 1.0057 (0.9846, 1.0258)
Cerebrovascular: 1.0019 (0.9837, 1.0189)
All circulatory: 1.0248 (1.0062, 1.0444)
Ml, 2-pollutant models, cool season
1.0399(1.0171, 1.0628), SO2 only
1.0285 (1.0019, 1.0571), SO2with NO2
1.0380 (1.0057, 1.0704), SO2/CO
1.0285 (1.0019, 1.0552), SO2/BS
1.0476 (1.0209, 1.0742), SO2 with O3
In the warm season no significant associations
were observed in 2-pollutant models..
Prescott et al.
(1998)*
Edinburgh, UK
Period of Study:
Oct 1992-Jun
1995
Outcome(s) (ICD9): Cardiac and cerebral
ischemia 410-414, 426-429,
434-440. Extracted from Scottish record linkage
system.
Study design: Time-series
Statistical Analysis: Poisson, log linear regression
models
Age groups analyzed: < 65, 65+ yrs
Covariates: Seasonal and weekday variation,
temperature, and wind speed.
Lag(s): 0,1,3 day moving avg
NO2 24 h avg ppb
Mean (SD): 8.3 (5.6)
Range: 1-50
90th-10th
Percentile =12 ppb
O3, 24 h avg
PM, 24 h avg
NO2, 24 h avg
CO, 24 h avg
Correlations NR.
Results reported as % increase in admissions,
increment 10 ppb.
All CVD, < 65 yrs
4.9 (-1.0, 11.1), 3 day moving avg
All CVD, > 65 yrs
-3.7 (-12.4, 5.9), 3 day moving avg
May 2008
                                          F-72
                        DRAFT—DO NOT QUOTE OR CITE

-------
    STUDY
                                 METHODS
                                                                   POLLUTANTS
                                                                                                         EFFECTS
Sunyer et al.
(2003)
Europe
(Birmingham,
London, Milan,
Paris, Rome,
Stockholm, the
Netherlands)
1990-1996
Hospital Admissions
Outcome(s) (ICD9): Cardiovascular diseases
(390-429); IHD (410-413); stroke (430-438)
Study design: Time-series
Statistical Analysis: Poisson autoregression with
GAM
Age groups analyzed: All ages
Covariates: Trend, seasonal patterns,
meteorological factors
Lag(s): 0 + 1
24-h median (pg/m3):
Birmingham: 19
LondoN:21
MilaN: 18
Netherlands: 9
Paris: 15
Rome: 9
Stockholm: 5
PM10
BS
NO2
03
% increase in Hospital Admissions (95% Cl)
(per 10 pg/m3SO2)
Cardiovascular
All ages: 0.7  (0.3, 1.1)
>65:0.7(0.3, 1.2)
IHD
< 65: 0.6 (0.2, 1.1)
>65: 1.2(0.8, 1.6)
IHD after adjustment for CO, NO2, BS < PM10
< 65: 0.7 (0.1, 1.3)
>65:-1.4 (-8.0, 6.0)
Stroke
>65: 0.0 (-0.5, 0.5)
Yallop et al.
(2007)
London,
England
Period of Study:
Jan 1998-Oct
2001,>  1400
days
Outcome(s): Acute pain in Sickle Cell Disease
(HbSS, HbSC, HbS/30, thalassaemia, HbS/3+).
Admitted to hospital for at least one night.
Study design: Time-series
Statistical analyses: Cross-correlation function
N: 1047 admissions
Covariates: No adjustment made in analysis,
discussion includes statement that the effects of
weather variables and copollutants are inter-
related.
Statistical package: SPSS
Lag(s): 0-2 days
Dose response: quartile analysis, graphs
presented, ANOVA comparing means across
quartiles.
NR
03
CO
NO
NO2
PM10
No association for SO2
                                                           AUSTRALIA
Jalaludin et al.
(2006)
Sydney,
Australia
Period of Study:
Jan 1997-Dec
2001
Outcome(s) (ICD9): All CVD
390-459; cardiac disease
390-429; IHD 410-413; and cerebrovascular
disease or stroke 430-438; Emergency room
attendances obtained from health department
data.
Age groups included: 65+
Study design: Time-series, multicity APHEA2
Protocol.
Statistical Analysis: GAM (with appropriate
convergence criteria) and GLM Models. Only
GLM presented.
Lag: 0-3
Covariates: Daily avg temperature and daily
relative, humidity, long-term trends, seasonality,
weather, day of wk,  public school holidays,
outliers and influenza epidemics.
Dose response:  quartile analysis
Season: Separate analyses for warm (Nov-Apr)
and cool periods (May-Oct).
SO2 24 h avg avg (ppb)
Mean (SD): 1.07(0.58)
MiN: 0.09
25th: 0.64
MediaN: 1.01
75th: 1.39
Max: 3.94
IQR: 0.75
# of Stations: 14
BS (0.21)
PM10 (0.37)
03 (0.454)
NO2 (0.52)
CO (0.46)
Effects were presented as percent change based
on an increment of 0.75 ppb.
Single-pollutant model:
All CVD, all yr
1.33% (0.24, 2.43) lag 0
Cardiac: 1.62% (0.33, 2.93) lag 0
IHD: 1.12% (-0.84, 3.12) lag 0
Stroke: -1.41% (-3.67, 0.90) lag 0
Cool Season
All cardiovascular: 2.15% (0.84, 3.46) lag 0
Cardiac: 2.48% (0.94, 4.04) lag 0
IHD: 2.49% (0.13, 4.91) lag 0
Stroke: -0.19% (-2.90, 2.60) lag 0
Warm Season
All cardiovascular: 0.06% (-1.48, 1.62) lag 0
Cardiac: 0.38% (-1.37, 2.16) lag 0
IHD: -0.47% (-3.08,  2.22)  lag 0
Stroke: -2.74% (-5.92, 0.55) lag 0
Results for lags 0-3 presented. In general, effect
size diminished with increasing lag.
Effects of SO2 on all CVD were diminished with
inclusion of PM and CO (graphically presented.)
May 2008
                                          F-73
                        DRAFT—DO NOT QUOTE OR CITE

-------
    STUDY
                                 METHODS
                                                                   POLLUTANTS
                                                                                                         EFFECTS
Petroeschevsky
et al. (2001)

Brisbane,
Australia

Period of Study:
Jan 1987-Dec
1994,
2,922 days
Outcome(s) (ICD9): CVD 390-459. Hospital
admissions, non-residents excluded.
Study design: Time-series
Statistical analyses: Poisson regression, APHEA
protocol, linear regression and GEEs
Age groups analyzed: 15-64, 65+
Covariates: Temperature, humidity, rainfall. Long-
term trends, season, flu, day of wk, holidays.
Dose  response: Quintile analysis.
Statistical software: SAS
Lag(s): lag 0-4, 3 day avg,
5 day avg
SO2 24-h avg (pphm)

Summer: Mean, min, max
0.39,0.0, 1.63
Fall: Mean, min, max
0.42,0.01,3.55
Winter: Mean, min, max
0.48, 0.0, 2.08
Spring: Mean, min, max
0.37, 0.0, 6.02
Overall: Mean,  min, max
0.41,0.0,3.55
SO21-h max (pphm)
Summer: Mean, min, max
0.78, 0.0, 5.5
Fall: Mean, min, max
0.93, 0.05, 5.95
Winter: Mean, min, max
1.13,0.0,6.68
Spring: Mean, min, max
0.84, 0.0, 6.01
Overall: Mean,  min, max
0.92, 0.0, 6.68
BSP
03
NO,
Effects were expressed as relative risk based on
an increment of 10 ppb and the 24-h avg SO2
concentrations.

All CVD
15 to > 65 yrs
1.028 (0.987, 1.070) lag 0
15 to 64yrs
1.081  (1.010, 1.158) lag 0
> 65 yrs

1.038(0.988, 1.091) lag 1

Non-significant increasing risk for CVD in those
15-64 by quintile of SO2 concentration observed.
                                                               ASIA
Chan et al.
(2006) *

Taipai, Taiwan

Period of Study:
Apr 1997-Dec
2002, 2090
days
Outcome(s) (ICD9): Cerebrovascular disease
430-437; stroke 430-434; hemorrhagic stroke
430-432; ischemic stroke 433-434. Emergency
admission data collected from National Taiwan
University Hospital.
Ages groups analyzed: age >50 included in study
Study design: Time-series
N: 7341 Cerebrovascular admissions among
those >50 yrs old
Catchment area:
Statistical analyses: Poisson regression, GAMs
used to adjust for non-linear relation between
confounders and ER admissions.
Covariates: Time trend variables: yr, mo, and day
of wk, daily temperature difference, and dew point
temperature.
Linearity: Investigated graphically by using the
LOESS smoother.
Lag: 0-3, cumulative lag up to 3 days
SO2 24-h avg (ppb):

Mean: 4.3

SD: 2.4

MiN: 0.4

Max: 17.1

IQR: 3.1 ppb

# of Stations: 16

Correlation among stations:
NR

PM10 (0.59)

PM2.5 (0.51)

CO (0.63)

NO2 (0.64)

03(0.51)
Results reported for OR for association of
emergency department admissions with an IQR
increase in SO2 (3.1 ppb)

Cerebrovascular:

1.008 (0.969, 1.047), lag 0

Stroke:

0.991 (0.916, 1.066), lag 0

Ischemic stroke:

1.044(0.966, 1.125), lag 0

Hemorrhagic stroke:

0.918(0.815, 1.021), lagO

No significant associations for SO2 reported. Lag
0 shown but similar null results were obtained for
lags 0-3.

2-pollutant models to adjust for copollutants but
not for SO2, which was not associated with
health outcomes.
Chang et al.
(2005)

Taipei, Taiwan

Period of Study:
1997-2001,5
yrs
Outcome(s) (ICD9): CVD 410-429.
Daily clinic visits or hospital admission from
computerized records of National Health
Insurance. Discharge data.
Source populatioN: 2.64 Million
N: 40.8 admissions/day, 74,509/5 yrs
# Hospitals: 41
Study design: Case-crossover, referent day 1 wk
before or after index day
Statistical analyses: Conditional logistic
regression.
Covariates: Same day temperature and humidity.
Season: warm/cool (stratified by temperature
cutpoint of 20 °C)
Lag(s): 0-2 days
SO2 24-h avg (ppb)

Mean: 4.32
MiN: 0.15

25th: 2.74

MediaN: 3.95

75th: 5.49
Max: 14.57
IQR: 2.75

# of Stations: 6

CO

03
NO2

PM10 Correlations NR.
Effects were expressed as odds ratios based on
an increment of 2.75 ppb.

Warm (> 20 °C) 0.967 (0.940, 0.995)
Cool (< 20 °C)  1.015 (0.965, 1.069)

In 2-pollutant models with (PM10, NO2, CO, or
O3) the effect of SO2 was attenuated for both
temperature ranges such that it was negatively
associated with CVD.

> 20 °C: 0.874  (0.77, 0.880), w/ PM10
< 20 °C: 0.986  (0.928, 1.048), w/ PM10
> 20 °C: 0.826  (0.798, 0.854), w/ NO2
< 20 °C: 0.922  (0.865, 0.984), w/ NO2
> 20 °C: 0.903  (0.876, 0.931), w/ CO
< 20 °C: 0.960  (0.901, 1.022), w/ CO
> 20 °C: 0.953  (0.926, 0.981), w/ O3
<20°C: 1.014(0.963, 1.067),w/O3
May 2008
                                          F-74
                        DRAFT—DO NOT QUOTE OR CITE

-------
    STUDY
                                 METHODS
                                                                   POLLUTANTS
                                                                                                         EFFECTS
Lee et al.
(2003)
Seoul, Korea
Period of Study:
Dec 1997-Dec
1999, 822-days,
184 days in
summer
Outcome(s) (ICD10): IHD: Angina pectoris 120;
Acute or subsequent Ml 121-123; other acute IHD
124. Electronic medical insurance data used.
Study design: Time-series
Statistical methods: Poisson regression, GAM
with strict convergence criteria.
Age groups analyzed: all ages, 64+
Covariates: Long-term trends LOESS smooth,
temperature, humidity, day of wk.
Season: Presented results for summer (Jun, Jul,
Aug) and entire period.
Lag(s): 0-6
SO2 24 h avg (ppb):
5th: 3.7
10th: 5.1
MediaN: 7.0
75th: 9.5
95th: 14.3
Mean (SD): 7.7 (3.3)
IQR: 4.4
Allyr
NO2 (0.72)
03 (-0.30)
CO (0.81)
PM10 (0.59)
Warm season
NO2 (0.79)
O3 (-0.56)
CO (0.41)
PM10 (0.61)
Results reported for RR of IHD hospital
admission for an incremental increase in SO2
equivalent to one IQR (4.4 ppb).
Single-pollutant model:
Entire season- IHD
All ages 0.96 (0.92, 0.99) lag 3
> 64 yrs 0.95 (0.90, 1.01) lag 3
Summer-IHD
All ages 1.09 (0.96, 1.24) lag 3
> 64 yrs 1.32 (1.08, 1.62) lag 3
2-pollutant model:
Entire season; SO2 and PMio
> 64 yrs 0.98 (0.94, 1.03) lag 3
Tsai et al.
(2003)
Kaohsiung,
Taiwan
Period of Study:
1997-2000
Outcome(s) (ICD9): All cerebrovascular
430-438; SHS 430;
PIH 431-432; IS 433-435; Other
436-438. Ascertained from National Health
Insurance Program computerized admissions
records.
Study design: Case-crossover
Statistical Analysis: Conditional logistic
regression.
N: 23,179 stroke admissions
# Hospitals: 63
Statistical software: SAS
Season: Warm (> 20 °C); Cool (< 20 °C).
Lag(s): 0-2, cumulative lag up to 2 previous days
S02 (ppb)
MiN: 1.25
25th: 6.83
MediaN: 9.76
75th: 13.00
Max: 26.80
Mean: 10.08
# StatioN: 6
PM10
SO2
CO
03
Results reported as OR for the association of
admissions with an incremental increase of SO2
equivalent to the IQR of 6.2 ppb
PIH admissions
Warm: 1.06 (0.95, 1.18), lag 0-2
Cool: 0.85 (0.58, 1.26), lag 0-2
IS admissions:
Warm: 1.06(1.00, 1.13), lag 0-2
Cool: 1.11  (0.83, 1.48), lag 0-2
2-pollutant models:
PIH 0.91 (0.80, 1.03)w/NO2
130.93(0.87, 1.00)w/NO2
PIH 0.94 (0.83, 1.06),w/CO
130.94(0.88, 1.02), w/CO
PIH 1.08(0.96, 1.20)w/O3
IS 1.08(1.01, 1.15)w/O3
PIH 0.99 (0.88, 1.11)w/PM
IS 1.01 (0.95-1.08) w/PM
Wong et al.
(1999)
Hong Kong,
China
Period of Study:
1994-1995
Outcome(s) (ICD9): CVD:
410-417, 420-438, 440-444; CHF 428; IHD 410-
414; Cerebrovascular Disease 430-438. Hospital
admissions through ER departments via Hospital
Authority (discharge data).
Study design: Time-series
Statistical analyses: Poisson regression, APHEA
protocol
# Hospitals: 12
Covariates: Daily temperature, relative humidity
day of wk, holidays, influenza, long-term trends
(yr and seasonality variables). Interaction of
pollutants with cold season examined.
Season: Cold (Dec-Mar)
Lag(s): 0-3 days
SO2 24-h avg (
Mean: 20.2
IQR: 10
PM10
SO2
03
Results reported for RR associated with
incremental increase in NO2 equal to 10 pg/m3.
All CVD, All ages
1.016(1.006, 1.026) lag 0-1
All CVD, 5-65 yrs
1.004 (0.989, 1.020) lag 0-1
All CVD, >65 yrs
1.021 (1.010, 1.032) lag 0-1
CHF
1.036(1.013, 1.059) lag 0
IHD
1.010(0.995, 1.025) lag 0-1
Cerebrovascular
0.990 (0.978, 1.002) lag 3
2-pollutant model results not presented for SO2
May 2008
                                          F-75
                        DRAFT—DO NOT QUOTE OR CITE

-------
STUDY METHODS
Wong et al.
(2002a)*
Hong Kong,
London
Period of Study:
1995-1997
(Hong Kong),
1992-1994
(London)






















Yang et al.
(2004)
Kaohsiung,
Taiwan
Period of Study:
1997-2000








Outcome(s) (ICD9): Cardiac disease 390-429;
IHD 410-414. Patients admitted to hospitals from
emergency departments, out patient departments
or directly to inpatient wards.
Statistical Analysis: Poisson regression, GAMs
Covariates: Smooth functions of time,
temperature, humidity (up to 3 days before
admission) day of wk, holidays and unusual
events.
Statistical software:
S-PLUS
Season: Warm/cold
Lag(s): 0-3, cumulative 0-1


















Outcome(s) (ICD9): All CVD: 410-429 * (All CVD
typically defined to include ICD9 codes 390-459)
N: 29,661
Study design: Case-crossover
Statistical Analysis: Poisson Time-series
regression models, APHEA protocol
# of Hospitals: 63
Season: Authors indicate not considered because
the Taiwanese climate is tropical with no apparent
seasonal cycle
Covariates: Stratified by warm (> 25°) and cold (<
25°) days, temperature, and humidity
measurements included in the model
Statistical package: SAS
Lag: 0-2 days




POLLUTANTS
SO2 24-h avg (pg/m3)
Mono Kono
Mean, all yr: 17.7(12.3)
Mean, warm: 18.3
Mean, cold: 17.2
MiN: 1.1
10th' 6 2

50th: 14.5
90th: 32.8
Max: 90
London
Mean, all yr: 23.7 (12. 3)
Mean, warm: 22.2
Mean cold* 25 3

Min 6.2
10th: 13.2
50th: 20.6

90th: 38.1
Max: 113.6

Hong Kong
NO2 (0.37)
PM10 (0.30)
O3(-0.18)
London
NO2(0.71)
PM10 (0.64)
03 (-0.25)
SO2 24-h avg (ppb)
MiN: 1.25
25%' 6 83
50%' 9 76
75%: 13.00
Max: 26.80
Mean: 10.08
# of Stations: 6
Correlation among stations:
NR
PM10
CO
SO2

038
2-pollutant models used to
adjust for copollutants
Correlations NR
EFFECTS
Effects expressed as % change, increment was
10 pg/m3 Cardiac (all ages)
Hong Kong
All yr: 2.1% (1.3, 2.8) lag 0-1
Warm: 1.0% (0.0, 2.0) lag 0-1
Cold: 1.9% (1.2, 2.7) lag 0-1
London
Allyr: 1.6% (1.0, 2.2) lag 0-1
Warm: 0.6% (-0.6, 1 .7) lag 0-1
Cold: 1.9% (1.2, 2.7) lag 0-1
IHD (all ages)
Hong Kong
Allyr: 0.1% (-1.1, 1.2) lag 0-1
Warm: -0.6% (-2.0, 0.8) lag 0-1
Cold: 1.0% (-0.8, 2.8) lag 0-1

London
Allyr: 1.7% (0.8, 2.6) lag 0-1
Warm: 1.0% (-0.6, 2.6) lag 0-1
Cold: 2.0% (0.9, 3.1) lag 0-1
Multipollutant model
Cardiac (all ages)
Hong Kong
SO2 alone 2.1% (1.3,2.8)
SO2 with NO2 1.4% (0.4, 2.3)
SO2 with O3 2.1% (1.4, 2.9)
SO2 with PM10 2.0% (1.1,2.8)

London
SO2 1.6% (1.0, 2.2)
SO2with NO2 1.4% (0.6, 2.3)
SO2 with O3 1.6% (0.9, 2.2)
SO2 with PM10 2.2% (1.2,3.2)

OR's for the association of one IQR (1 7.08 ppb)
increase in SO2 with daily counts of CVD hospital
admissions are reported
All CVD (ICD9: 410-429), one-pollutant model
> 25°: 0.999 (0.954, 1.047)
<25°: 1.187(1.092, 1.291)
All CVD (ICD9: 410-429), 2-pollutant models
Adjusted for PMi0:
> 25°: 0.961 (0.917, 1.008)
<25°: 1.048(0.960, 1.145)
Adjusted for NO2:
> 25°: 0.921 (0.875, 0.969)
< 25°: 0.711 (0.641,0.789)
Adjusted for CO:
> 25°: 0.831 (0.785, 0.879)
< 25°: 0.996 (0.910, 1.089)
Adjusted for O3
>25°: 1.034(0.987, 1.084)
<25°: 1.194(1.098, 1.299)

MIDDLE EAST
Hosseinpoor et
al. (2005)
Tehran, Iran
Period of Study:
Mar1996-Mar
2001, 5 yrs




Outcome(s) (ICD9): Angina pectoris 413. Primary
discharge diagnosis from registry databases or
records

Study design: Time-series
Statistical methods: Poisson regression
# Hospitals: 25
Covariates: Long-term trends, seasonality,
temperature, humidity, holiday, post-holiday, day
of wk.
Lag(s): 0-3
SO2 24-h avg (pg/m3)
Mean (SD): 73.74 (33.30)
MiN: 0.30
25th: 48.23
MediaN: 74.05
75th' 98 64
Max: 499.26

NO2 CO O3 PM10
Correlations NR
Results reported for relative risk in hospital
admissions per increment of 10 pg/m3 SO2.
Angina
0.99995 (0.99397, 1 .00507), lag 1
In a multipollutant model only CO (lag 1) was
significantly associated with angina pectoris
related hospital admissions.



May 2008
F-76
DRAFT—DO NOT QUOTE OR CITE

-------
Table F-5.   Associations of short-term exposure to SO2 with mortality.
STUDY
METHODS
POLLUTANTS
OUTCOME
FINDINGS
META ANALYSIS
Stieb et al. (2002;
reanalysis 2003)
meta-analysis of estimates
from various countries.
The lags and multiday
averaging used varied
Meta-analysis of time-series
study results.
24-h avg ranged from 0.7
ppb (San Bernardino) to 75
ppb (Shenyang)
"Representative"
concentratioN: 9.4 ppb
Copollutants: PMio, O3,
NO2, CO
All cause
Single-pollutant model
(29 estimates): 1.0% (0.6,
1.3)
Multipollutant model
estimates (10 estimates):
0.9% (0.3, 1 .4)
UNITED STATES
Chock et al. (2000)
Pittsburgh, PA
1989-1991
De Leon et al. (2003)
New York City
1985-1994
Dockeryetal. (1992)
St. Louis, MO and
Eastern Tennessee
1985-1986
Gamble (1998)
Dallas, TX
1990-1994
Gwynn et al. (2000)
Buffalo, NY
Kelsalletal. (1997)
Philadelphia, PA
1974-1988
Kinney and Ozkaynak
(1991)
Los Angeles County, CA
1970-1979
Klemm and Mason (2000);
Klemm et al. (2004)
Atlanta, GA
Aug 1998-Jul2000
Lags:0, 1,2,3
Poisson GLM. Time-series
study. Numerous results
Lags: 0 or 1
Poisson GAM with
Stringent convergence
Criteria; Poisson GLM.
Time-series study.
Lag: 1
Poisson with GEE.
Time-series study.
Lag:0
Poisson GLM.
Time-series study.
Lag:0, 1,2,3
Poisson GAM with
Default convergence
criteria. Time-series study.
Lag: 0
(AIC presented for 0
through 5)
Poisson GAM.
Lag:1
OLS (ordinary least
squares) on high-pass
filtered variables. Time-
series study.
Lag: 0-1
Poisson GLM using
quarterly, monthly, or
biweekly knots for temporal
smoothing. Time-series
study.
Mean NR
Copollutants: PM10, O3,
NO2, CO; 2-, 5-, and 6-
pollutant models
24-h avg: 15 ppb
Copollutants: PMio, O3,
NO2, CO; 2-pollutant
models
24-h avg: St. Louis: 9 ppb
Eastern Tennessee:
5 ppb
Copollutants: PM10, PM25,
SO42", H+, O3, NO2
24-h avg: 3 ppb
Copollutants: PMio, O3,
NO2, CO; 2-pollutant
models
24-h avg: 12 ppb
Copollutants: PM10, CoH,
SO42- , O3, NO2, CO, H+
24-h avg: 17 ppb
Copollutants: TSP, CO,
NO2, O3
24-h avg: 15 ppb
Copollutants: KM (particle
optical reflectance), Ox,
NO2, CO; multipollutant
models
1-h max: 19 ppb
Copollutants: PM25, PM10.
25, EC, OC, SO42' , NO3-,
O3, NO2, CO
All cause; age < 74 yrs; age
75+ yrs
Circulatory and cancer with
and without contributing
respiratory causes
All cause
All cause; respiratory;
cardiovascular
All cause; respiratory;
circulatory
All cause; respiratory;
cardiovascular
All cause; respiratory;
circulatory
All cause; respiratory;
cardiovascular; cancer;
other; age < 65 yrs; age
65+ yrs
All cause:
Age 0-74 yrs:
Lag 1:0.7% (-0.7, 2.2)
Age 75+ yrs:
Lag 1: -0.2% (-1.6, 1.3)
Gaseous pollutants results
were given only in figures.
Circulatory:
Age < 75 yrs: -2%
Age 75+ yrs: -2%
All cause:
St. Louis, MO: 0.8% (-1.7,
3.2)
Eastern Tennessee: 0.4%
(-0.4, 1.1)
All cause: -0.8% (-3.8, 2.4)
Respiratory: -1.0% (-5.8,
4.1)
Cardiovascular: -0.5%
(-11.4, 11.8)
All cause:
Lag 0: -0.1% (-1.8, 1.7)
Circulatory: Lag 3: 1 .3%
(-2.9,5.6)
Respiratory: Lag 0: 6.4%
(-2.5, 16.2)
All cause:
Single-pollutant:
0.8% (0.3, 1 .4)
With all other pollutants:
0.8% (0.1, 1.6)
All cause:
Exhaustive multipollutant
model: 0.0% (-1.1, 1.2)
All cause -Age 65+ yrs:
Quarterly knots:
4.7% (-2.6, 12.5)
Monthly knots:
3.4% (-4.1, 11.5)
Bi-weekly knots:
1.0% (-6.7, 9.3)
May 2008
F-77
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Klemm et al. (2004)
Georgia (Fulton; DeKalb
counties)
1998-2000





Lipfert et al. (2000a)
Seven counties in
Philadelphia, PA area
1991-1995
Lippmann et al. (2000);
reanalysis Ito, (2003)
Detroit, Ml
1985-1990
1992-1994

Mar et al.(2000; reanalysis
in 2003)
Phoenix, AZ.
1995-1997



Moolgavkar et al. (1995)
Philadelphia, PA
1973-1988.



Moolgavkar (2000;
reanalysis 2003a)
Cook County, IL; Los
Angeles County, CA; and
Maricopa County, AZ
1987-1995




Moolgavkar (2003b)
Cook County, IL and Los
Angeles County, CA
1987-1995





METHODS
Lags: 2-day avg (avg of lag
0 and lag 1)
Poisson with GLM. Time-
series study.





Lag: 0-1
Linear with 19-day weighted
avg Shumway filters. Time-
series study. Numerous
results.
Lags:0, 1,2,3,0-1,0-2,0-
3
Poisson GAM, reanalyzed
with stringent convergence
criteria; Poisson GLM.
Numerical SO2 risk
estimates were not
presented in the re-
analysis. Time-series study.
Lags: 0 for all cause; 0, 1 ,
2, 3, 4 for cardiovascular
Poisson GAM with default
convergence criteria (only
cardiovascular deaths were
reanalyzed in 2003). Time-
series study.
Lag:1
Poisson GLM. Time-series
study.



0, 1,2,3,4,5
Poisson GAM with default
convergence criteria in the
original Moolgavkar (2000);
GAM with stringent
convergence criteria and
GLM with natural splines in
the 2003 re-analysis. The
2000 analysis presented
total death risk estimates
only in figures.
Lags:0, 1,2,3,4,5
Poisson GAM with default
convergence criteria. Time-
series study.





POLLUTANTS
1-h max(ppb): 19.4(13.42)
Copollutants: PM25, Coarse
mass, O3, NO2, CO, Acid,
Ultrafine surface area,
Ultrafine count, EC, Organic
carbon, SO4, Oxygenated
hydrocarbons, Nonmethane
hydrocarbons, NO3



24-h avg: 8 ppb
1-h max: 18 ppb
Copollutants: PM10, PM25,
PM10-2.5, SO42-, other PM
indices, O3, NO2, CO; 2-
pollutant models
24-h avg:
1985-1990: 10 ppb
1992-1 994: 7 ppb
Copollutants: PM10, PM25,
PM10-2.5, SO42-, H+, O3,
NO2, CO;
2-pollutant models

24-h avg: 3.1 ppb
Copollutants: PM25, PM10,
PM10-2.5, CO, NO2, O3, and
selected trace elements,
ions, EC, OC, TOC, and
factor analysis components

24-h avg:
Spring: 17 ppb
Summer: 16 ppb
Fall: 18 ppb
Winter: 25 ppb
Copollutants: TSP, O3;
2-pollutant models
24-h avg median:
Cook County: 6 ppb Los
Angeles: 2 ppb Maricopa
County:
2 ppb
PM2.5, PM10, O3, NO2, CO;
2- and 3-pollutant models



24-h avg median:
Cook County: 6 ppb Los
Angeles: 2 ppb
Copollutants: PM2.5, PM10,
O3, NO2, CO; 2-pollutant
models




OUTCOME
All-cause








All cause; respiratory;
cardiovascular; all ages;
age 65+ yrs; age
< 65 yrs; various
subregional boundaries
All cause; respiratory;
circulatory; cause-specific



All cause, cardiovascular





All cause




Cardiovascular;
cerebrovascular;
COPD






All cause; cardiovascular







FINDINGS
> 65 yrs old
Quarterly knots (SE)
3 = 0.00115(0.00092)
t= 1.24
Monthly knots (SE)
3 = 0.00084 (0.00096)
t = 0.87
Biweekly knots (SE)
3 = 0.00024(0.00101)
t = 0.24
All-cause:
Philadelphia:
0.7% (p > 0.05)

Poisson GAM:
All cause:
1985-1990:
Lag 1:0.5% (-1.5, 2.4)
1992-1994:
Lag 1: 1.1% (-1.4, 3.6)

Poisson GAM:
All cause:
LagO: 11.2% (-1.5, 25.6)
Poisson GLM:
Cardiovascular:
Lag 1:7.4% (-13.1, 32.6)

All yr: 1 .3% (0.8, 1 .8)
Spring: 1.7% (0.6,2.9)
Summer: 0.9% (-0.7, 2.5)
Fall: 1.3% (0.0, 2.6)
Winter: 2.0% (0.9, 3.0)

GLM (re-analysis):
Cook County:
All-cause:
Lag 1:2.6% (1.4, 3.8)
Cardiovascular:
Lag 1:2.9% (1.0, 4.8)
Los Angeles:
Cardiovascular:
Lag 1:5.9% (3.0, 9.0)

All cause: Cook County:
Single-pollutant:
Lag 1:2.6% (1.5, 3.7)
With PM10:
Lag 1: 1.9% (0.6, 3.2)
Los Angeles:
Single-pollutant:
Lag 1:6.9% (5.4, 8.4)
WithPM25:
Lag 1:7.6% (3.4, 12.0)
May 2008
F-78
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Samet et al. (2000a;
2000b); reanalysis Dominici
et al. (2003)
90 U.S. cities (58 U.S.cities
with SO2 data)
1987-1994
Schwartz (1991)
Detroit, Ml
1973-1982
Schwartz (2000)
Philadelphia, PA
1974-1988
Schwartz (2004)
14 U.S. cities that had daily
PM10 data
METHODS
Lags:0, 1,2
Poisson GAM, reanalyzed
with stringent convergence
criteria; Poisson GLM.
Time-series study.
Lags:0, 1,0-1
Poisson GEE. Time-series
study.
Lag: 0
Poisson GAM model in 15
winter and 15 summer
periods. The second stage
regressed the TSP and SO2
risk estimates on SO2/TSP
relationships.
Lag: 1
Case-crossover design,
estimating PMio risks by
matching by the levels of
gaseous pollutants.
POLLUTANTS
24-h avg ranged from
0.4 ppb (Riverside) to
1 4.2 ppb (Pittsburgh)
Copollutants: PM10, O3,
NO2, CO; multipollutant
models
24-h avg: 12 ppb
Copollutants: TSP
(predicted from extinction
coefficient); 2-pollutant
models
24-h avg summer mean
declined from 20 ppb in
1974 to 9 ppb in 1988;
winter mean declined from
35 ppb in 1974 to 17 in
1988
Copollutants: TSP,
extinction coefficient
24-h avg median ranged
from 2.2 ppb (Spokane,
WA) to 39.4 ppb
(Pittsburgh, PA)
Copollutants: PMio risk
estimates computed,
matched by the levels of
SO2, CO, NO2, and O3
OUTCOME
All cause; cardiopulmonary
All cause
All cause
All cause
FINDINGS
Posterior means:
All cause:
Single-pollutant:
Lag 1:0.6% (0.3, 1.0)
With PM10 and NO2:
Lag 1 : 0.4% (-0.6, 1 .4)
Poisson regression
coefficient
Single-pollutant:
Lag 1 : 0.863 (SE = 0.323)
With TSP:
Lag 1 : 0.230 (SE = 0.489)
(Though SO2 levels were
reported in ppb, these
coefficients must have been
for SO2 in ppm.)
Single-pollutant:
2.3% (1.6, 3.0)
With TSP:
0.4% (-2.2, 3.1)
SO2 risk estimates not
computed. PMio risk
estimates showed the
largest risk estimate when
matched for SO2.
CANADA
Burnett et al. (2004)
12 Canadian cities
1981-1999
Burnett etal. (1998a)
11 Canadian cities
1980-1991
Burnett etal. (1998)
Toronto
1980-1994
Goldberg et al. (2003)
Montreal, Quebec
1984-1993
Vedal et al. (2003)
Vancouver, British
Columbia
1994-1996
Lag: 1
Poisson GLM. Time-series
study.
Lags:0, 1,2,0-1,0-2
examined but the best
lag/averaging for each city
chosen
Poisson GAM with default
convergence criteria. Time-
series study.
Lags:0, 1,0-1
Poisson GAM with default
convergence criteria. Time-
series study.
Lags:0, 1,0-2
Poisson GLM with natural
splines. Time-series study.
Lags:0, 1,2
Poisson GAM with stringent
convergence criteria. Time-
series study. By season.
24-h avg ranged from 1 ppb
(Winnipeg) to 10 ppb
(Halifax)
Copollutants: PM25, PM10-
2.5, O3, NO2, CO
24-h avg ranged from 1 ppb
(Winnipeg) to 11 ppb
(Hamilton)
Copollutants: O3, NO2, CO
24-h avg: 5 ppb
Copollutants: O3, NO2, CO,
TSP, COH, estimated PM10,
estimated PM25
24-h avg: 6 ppb
Copollutants: PM25,
coefficient of haze, SO42-,
O3, NO2, CO
24-h avg: 3 ppb
Copollutants: PMio, O3,
NO2, CO
All cause
All cause
All cause
Congestive heart failure
(CHF) as underlying cause
of death versus those
classified as having CHF 1
yr prior to death
All cause; respiratory;
cardiovascular
Single-pollutant:
0.7% (0.3, 1 .2)
With NO2:
0.4% (0.0, 0.8)
Single-pollutant:
3.4% (2.0, 4.7)
With all gaseous pollutants:
2.6% (1.3, 3.9)
Single-pollutant:
LagO: 1.0% (0.3, 1.8)
With CO:
Lag 0: 0.6% (-0.4, 1 .5)
CHF as underlying cause of
death: Lag 1:-0.1% (-8.9,
9.6)
Having CHF 1 yr prior to
death: Lag 1:5.4% (1.3,
9.5)
Results presented in figures
only.
All cause:
Summer: Lag 0: -3%
Winter: Lag 1:~1%
May 2008
F-79
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Villeneuve et al. (2003)
Vancouver, British
Columbia
1986-1999
METHODS
Lags:0, 1,0-2
Poisson GLM with natural
splines. Time-series study.
POLLUTANTS
24-h avg: 5 ppb
Copollutants: PM2.5, PM10,
PM10-2.5, TSP, coefficient of
haze, SO42-, O3, NO2, CO
OUTCOME
All cause; respiratory;
cardiovascular; cancer;
socioeconomic status
FINDINGS
All yr: All cause:
Lag 1: 1.7% (-1.1, 4.5)
Cardiovascular:
Lag 1: 1.1% (-3.1, 5.4)
Respiratory:
Lag 1:8.3% (0.6, 16.6)
EUROPE
Anderson et al. (1996)
London, England
1987-1992
Anderson et al. (2001)
West Midlands region,
England
1994-1996
Ballester et al. (2002)
13 Spanish cities
1990-1996
Biggeri et al. (2005)
8 Italian cities
Period variable between
1990-1999
Bremneret al. (1999)
London, England
1992-1994
Clancy et al. (2002)
Dublin, Ireland
1984-1996
Dab etal. (1996)
Paris, France
1987-1992
Diaz etal. (1999)
Madrid, Spain
1990-1992
Fischer etal. (2003)
The Netherlands
1986-1994
Lag:1
Poisson GLM. Time-series
study.
Lag: 0-1
Poisson GAM with default
convergence criteria. Time-
series study.
Lags: 0-1 for 24-h avg SO2;
Ofor 1-h max SO2
Poisson GAM with default
convergence criteria. Time-
series study.
Lag: 0-1
Poisson GLM. Time-series
study.
Lags: Selected best from 0,
1,2,3, (all cause); 0, 1, 2,
3, 0-1, 0-2, 0-3 (respiratory,
cardiovascular)
Poisson GLM. Time-series
study.
NA
Comparing standardized
mortality rates for 72 mos
before and after the ban on
coal sales in Sep 1990.
Lag:1
Poisson autoregressive.
Time-series study.
Lag:1
Autoregressive OLS
regression. Time-series
study.
Lags: 0-6
Poisson GAM with default
convergence criteria. Time-
series study.
24-h avg: 11 ppb
Copollutants: BS, O3, NO2;
2-pollutant models
24-h avg: 7 ppb
Copollutants: PM10, PM25,
PM10-2.5, BS, SO42-, O3,
NO2, CO
24-h avg SO2 ranged from
2.8 ppb (Sevilla) to 15.6 ppb
(Oviedo)
Copollutants: TSP, BS, PM10
24-h avg ranged from 2 ppb
(Verona) to 14 ppb (Milan)
Copollutants: O3, NO2, CO,
PM10
24-h avg: 7 ppb
Copollutants: BS, PM10, O3,
NO2, CO; 2-pollutant
models
24-h avg:
1984-1990: 11. 7 ppb
1990-1996: 7.7 ppb
Copollutants: BS
24-h avg: 10 ppb
1-h max: 21 ppb
Copollutants: BS, PM13, O3,
NO2, CO
24-h avg: Levels NR.
Copollutants: TSP, O3, NO2,
CO
24-h avg mediaN: 3.5 ppb
Copollutants: PM10, BS, O3,
NO2, CO
All cause; respiratory;
cardiovascular
All cause; respiratory;
cardiovascular
All cause, cardiovascular,
respiratory
All cause; respiratory;
cardiovascular
All cause; respiratory;
cardiovascular; all cancer;
all others; all ages; age
specific (0-64, 65+, 65-74,
75+ yrs)
All cause, cardiovascular,
and respiratory
Respiratory
All cause; respiratory;
cardiovascular
All-cause, cardiovascular,
COPD, and pneumonia in
age groups < 45, 45-64, 65-
74, 75+
All cause: 1.0% (0.0, 2.0)
Respiratory: 1.7% (-1.3,
4.9)
Cardiovascular: 0.2% (-1.4,
1.8)
All cause: -0.2% (-2.5, 2.1)
Respiratory: -2.2% (-7.4,
3.2)
Cardiovascular: -0.2%
(-3.5,3.1)
All cause:
Lag 0-1: 1.4% (0.2,2.7)
Cardiovascular:
Lag 0-1 : 1 .4% (-0.4, 3.3)
Respiratory:
Lag 0-1 : 3.5% (1 .0, 6.0)
All cause: 4.1% (1.1, 7.3)
Respiratory: 7.4% (-3.6,
19.6)
Cardiovascular: 4.9% (0.4,
9.7)
All cause: Lag 1 : 1 .6%
(-0.5, 3.7)
Respiratory: Lag 2: 4.8%
(-0.2, 10.0)
Cardiovascular Lag 1:
1.3% (-1.7, 4.3)
BS mean declined by a
larger percentage (70%)
than SO2 (34%) between
the two periods.
All cause death rates
reduced by 5.7% (4, 7);
respiratory deaths by 15.5%
(12, 19); cardiovascular
deaths by 10.3% (8, 13).
Lag 1:2.3% (-0.9, 5.5)
Only significant regression
coefficients were shown, but
description of the table was
not clear enough to derive
risk estimates.
Cardiovascular:
Age < 45 yrs: 4.3% (-4.6,
13.9)
Age 45-64 yrs: -0.5% (-3.6,
2.7)
Age 65-74 yrs: 1 .6% (-0.8,
4.2);
Age 75+ yrs: 2.8% (1.3, 4.3)
May 2008
F-80
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Garcia-Aymerich et al.
(2000)
Barcelona, Spain
1985-1989











Hoek et al. (2002)
Rotterdam, the Netherlands
1983-1991
Hoek et al. (2000;
reanalysis Hoek, 2003)
The Netherlands: Entire
country, four urban areas
1986-1994





Hoeketal. (2001;
reanalysis Hoek, 2003)
The Netherlands
1986-1994






Katsouyanni et al. (1997)
12 European cities
Period of Studys vary by
city, ranging from 1977 to
1992
METHODS
Selected best averaged lag
Poisson GLM. Time-series
study.












Lag:1
Poisson GAM with default
convergence criteria. Time-
series study.
Lag: 1,0-6
Poisson GAM, reanalyzed
with stringent convergence
criteria; Poisson GLM.
Time-series study.





Lag: 0-6
Poisson GAM, reanalyzed
with stringent convergence
criteria; Poisson GLM.
Time-series study.






"Best" lag variable across
cities from 0 to 3
Poisson autoregressive.
Time-series study.

POLLUTANTS
Levels NR.
Copollutants: BS, O3, NO2












24-h avg median:
7.7 ppb
Copollutants: TSP, BS, Fe,
03, CO
24-h avg mediaN: 3.5 ppb
in the Netherlands; 5.6 ppb
in the four major cities
Copollutants: PM10, BS,
SO42', NO3-, O3, NO2, CO;
2-pollutant models




24-h avg mediaN: 3.5 ppb
in the Netherlands; 5.6 ppb
in the four major cities
Copollutants: PMio, O3,
NO2, CO






24-h avg median of the
median across the cities
was 14 ppb, ranging from
5 ppb (Bratislava) to 26 ppb
(Cracow)
Copollutants: BS, PM10
OUTCOME
All cause; respiratory;
cardiovascular; general
population; patients with
COPD












All cause


All cause; COPD;
pneumonia; cardiovascular






Total cardiovascular;
myocardial infarction;
arrhythmia; heart failure;
cerebrovascular;
thrombosis-related






All cause


FINDINGS
All cause: General
population:
Lag 0-3: 4.4% (2.3, 6.5)
COPD patients:
Lag 0-2: 2.6% (-5.0, 10.7)
Respiratory: General
populatioN: Lag 0-1: 3.5%
(-0.6, 7.8)
COPD patients:
Lag 0-2: 2.3% (-8.9, 15.0)
Cardiovascular:
General populatioN: Lag 0-
3: 5.1% (2.3, 8.0)
COPD patients: Lag 0-2:
2.0% (-11.5, 17.5)
Single-pollutant:!. 5% (0.0,
3.0)
With TSP and O3: 0.5%
(-1.2,2.3)
Poisson GLM: All cause:
Lag 1: 1.3% (0.7, 1.9)
Lag 0-6: 1.8% (0.9,2.7)
WithBS: 1.1% (-0.3, 2.4)
Cardiovascular:
Lag 0-6: 2.7% (1.3,4.1)
COPD:
Lag 0-6: 3.6% (-0.3, 7.7)
Pneumonia:
Lag 0-6: 6.6% (1.2, 12.2)
Poisson GLM:
Total cardiovascular: 2.7%
(1.3,4.1)
Myocardial infarctioN: 0.8%
(-1.2,2.8)
Arrhythmia: 2.3% (-3.9, 8.8)
Heart failure: 7.1% (2.6,
11.7)
Cerebrovascular: 4.4%
(1.4,7.5)
Thrombosis-related: 9.6%
(3.1, 16.6)
All cities: 1.1% (0.9, 1.4)
Western cities: 2.0% (1.2,
2.8)
Central eastern cities: 0.5%
(-0.4, 1.4)
May 2008
F-81
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Keatinge and Donaldson
(2006)
London
1991-2002























Kotesovec et al. (2000)
Northern Bohemia, Czech
Republic
1982-1994
Le Tertre et al. (2002)
Bordeaux , Le Havre, Lille,
Lyon, Marseille, Paris,
Rouen, Strasbourg, France
Period of Study varies by
city, ranging from 1990-
1995
Michelozzi et al. (1998)
Rome, Italy
1992-1995

Peters etal. (2000)
NE Bavaria, Germany
1982-1994
Coal basin in Czech
Republic 1993-1994

Ponka etal. (1998)
Helsinki, Finland
1987-1993

Prescott et al. (1998)
Edinburgh, Scotland
1992-1995

Rahlenbeckand Kahl
(1996)
East Berlin, Germany
1981-1989
METHODS
Lags: Mean of 0, -1, -2
Graphic analysis and GAM.
Time-series study.
























Lags:0, 1,2,3,4,5,6,0-6
Poisson GLM, time-series
study

Lags: 0-1
Poisson GAM with default
convergence criteria. Time-
series study.

Lags:0, 1,2,3,4
Poisson GAM with default
convergence criteria. Time-
series study.
Lags:0, 1,2,3
Poisson GLM. Time-series
study.



Lags:0, 1,2,3,4,5,6,7
Poisson GLM. Time-series
study.

Lag:0
Poisson GLM. Time-series


Lags:0, 1,2,3,4,5
OLS, with log of SO2, Time-
series study.

POLLUTANTS
24-h avg: Levels NR
Copollutants: O3, PMio
























24-h avg: 34.9 ppb
Copollutants: TSP

24-h avg ranged from 3 ppb
(Bordeaux) to 9 ppb
(Rouen)
Copollutants: BS, O3, NO2

24-h avg: 5.7 ppb
Copollutants: PM13, NO2,
O3, CO

24-h avg:
Czech Republic: 35 ppb
Bavaria, Germany: 14 ppb
Copollutants: TSP, PM10,
O3, NO2, CO

24-h avg median:
3.5 ppb
Copollutants: TSP, PM10,
O3, NO2
24-h avg: 1981-1995: 15
ppb
1992-1 995: 8 ppb
Copollutants: BS, PM10, O3,
NO2, CO; 2-pollutant
models
24-h avg: 61.9 ppb
"SP" (beta absorption)


OUTCOME
All-cause

























All cause, cardiovascular
(only age = < 65
presented), cancer

All cause; respiratory;
cardiovascular



All-cause


All cause; respiratory;
cardiovascular; cancer




All cause; cardiovascular;
age < 65 yrs, age 65+ yrs


All cause; respiratory;
cardiovascular; all ages;
age < 65 yrs; age 65+ yrs


All cause



FINDINGS
Relative Risk for a 1 06
Increase in Mortality (per 10
ppb SO2)
SO2 + Temp: 3.1 (0.6, 5.5)
SO2 + Temp + Acclim.: 2.2
(-0.1,4.6)
SO2 + Temp + Acclim. +
Acclim. D Temp:
2.5(0.2,4.8)
SO2 + Temp + Acclim. +
Acclim. D Temp + SuN:
2. 3 (-0.03, 4.5)
SO2 + Temp + Acclim. +
Acclim. D Temp + Sun +
Wind: 1.6 (-0.7, 3.8)
SO2 + Temp + Acclim. +
Acclim. D Temp + Sun +
Wind +
Abs. Humidity: 1.7 (-0.6,
3.9)
SO2+ Temp + Acclim. +
Acclim. D Temp + Sun +
Wind+
Abs. Humidity + RaiN: 1.8
(-0.4,4.1)
SO2 + Temp. + Abs.
Humidity: 2.5 (0.03, 4.9)
All cause:
Lag 1:0.1% (-0.1, 0.4)

8-city pooled estimates:
All cause:
2.0% (1.2, 2.9)
Respiratory: 3.2% (0.1 , 6.3)
Cardiovascular: 3.0% (1.5,
4.5)
Lag 1: -2.0% (-4.4, 0.5);
(negative estimates at all
lags examined)

Czech Republic:
All cause:
Lag 1:0.8% (-0.2, 1.8)
Bavaria, Germany:
All cause:
Lag 1 : 0.3% (-0.3, 0.9)
No risk estimate presented
for SO2. PMio and O3 were
reported to have stronger
associations.

Results presented as
figures only. Essentially no
associations in all
categories. Very wide
confidence intervals.

Single-pollutant:
Lag 1:4.4% (0,8.7);
With SP:
Lag 1:2.9% (-2. 7, 8.5)

May 2008
F-82
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Roemer and van Wijinen
(2001)
Amsterdam, the
Netherlands
1987-1998
Saezetal. (1999)
Barcelona, Spain
1986-1989
Saez et al. (2002)
Seven Spanish cities
Variable periods of study
between 1991 and 1996
Spix and Wichman (1996)
Koln, Germany
1977-1985
Sunyeretal. (2002)
Barcelona, Spain
1986-1995
Sunyeretal. (1996)
Barcelona, Spain
1985-1991
Verhoeff et al. (1996)
Amsterdam, the
Netherlands
1986-1992
Zeghnoun et al. (2001)
Rouen and Le Havre,
France
1990-1995
Zmirou et al. (1996)
Lyon, France
1985-1990
METHODS
Lags: 1,2,0-6
Poisson GAM with default
convergence criteria (only
one smoother). Time-series
study.
Lags: 0-1
Poisson with GEE. Time-
series study.
Lags: 0-3
Poisson GAM with default
convergence criteria. Time-
series study.
Lags: 0, 1, 0-3
Poisson GLM. Time-series
study.
Lags: 0-2
Conditional logistic (case-
crossover)
Selected best single-day lag
Autoregressive Poisson.
Time-series study.
Lags:0, 1,2
Poisson GLM. Time-series
study.
Lags:0, 1,2,3,0-3
Poisson GAM with default
convergence criteria. Time-
series study.
Lags: Selected best from 0,
1,2,3
Poisson GLM. Time-series
study.
POLLUTANTS
24-h avg:
Background sites: 3.1 ppb
Traffic sites: 4.2 ppb
Copollutants: BS, PM10, O3,
NO2, CO
Levels NR.
Copollutants: BS, O3, NO2,
Values for SO2 NR.
Copollutants: O3, PM, NO2,
CO
24-h avg: 15 ppb
1-h max: 32 ppb
Copollutants: TSP, PM7,
NO2
24-h avg median:
6.6 ppb
Copollutants: PM10, BS,
NO2, O3, CO, pollen
24-h avg median:
Summer: 13 ppb
Winter: 16 ppb
Copollutants: BS, NO2, O3
24-h avg: 4.5 ppb
Copollutants: BS, PM10, O3,
CO; multipollutant models
24-h avg: RoueN: 10 ppb
Le Havre: 12 ppb
Copollutants: NO2, BS,
PM,3, 03
24-h avg: 16 ppb
Copollutants: PM13, NO2, O3
OUTCOME
All cause
Asthma mortality; age 2-45
yrs
All cause; respiratory;
cardiovascular
All-cause
All cause, respiratory, and
cardiovascular mortality in a
cohort of patients with
severe asthma
All cause; respiratory;
cardiovascular; all ages;
age 70+ yrs
All cause; all ages; age 65+
yrs
All cause; respiratory;
cardiovascular
All cause; respiratory;
cardiovascular; digestive
FINDINGS
Total population using
background sites:
Lag 1:2.6% (-0.6, 5.8)
Traffic population using
background sites:
Lag 1 : 0.6% (-6.9, 8.6)
Total population using traffic
sites: Lag 1:2.4% (-0.3,
5.1)
RR = 1.9(0.7,4.4)
Risk estimates for SO2 was
NR. Including SO2 in
regression model did not
appear to reduce NO2 risk
estimates.
Lag 1:0.8% (0.2, 1.4)
Odds ratio: Patients with 1
asthma admission:
All cause: 14.8% (-19.8,
64.4)
Patients with more than 1
asthma adm: All cause:
50.4% (-48.6, 340.4)
Patients with more than 1
asthma or COPDadm: All
cause: 20.2% (-17.5, 75.0)
NO2 and O3 were more
strongly associated with
outcomes than SO2.
All yr, all ages: All cause:
Lag 1:3.5% (1.9, 5.1)
Respiratory:
Lag 0: 3.5% (-0.2, 5.0)
Cardiovascular:
Lag 1:2.2% (0.5, 3.9)
Single-pollutant:
Lag 1: 1.4% (-1.4, 4.2)
With BS:
-3.7% (-8. 1,0.9)
All cause:
RoueN: Lag 1:2.3% (-1.1,
5.9)
Le Havre: Lag 1: 1.1%
(-0.3, 2.5)
All cause: Lag 0: 3.4% (1.4,
5.4)
Respiratory: Lag 3: 2.8%
(0.9, 4.8)
CardiovascularLag 0-3:
4.5% (2.0, 7.0)
May 2008
F-83
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Zmirou et al. (1998)
10 European cities
Period of Studys vary by
city, ranging from 1985-
1992





METHODS
Lags:0, 1,2,3,0-1,
0-2, 0-3 (best lag selected
for each city)
Poisson GLM. Time-series
study.





POLLUTANTS
24-h avg: Cold Season:
Ranged from 12 ppb
(London) to 87 ppb (Milan)
ppb
Warm Season: Ranged
from 5 ppb (Bratislava) to
21 ppb (Cracow) in warm
season
Copollutants: BS, TSP, NO2,
03
OUTCOME
Respiratory; cardiovascular







FINDINGS
Western cities:
Respiratory: 2.8% (1.7, 4.0)
Cardiovascular: 2.3% (0.9,
3.7)
Central eastern cities:
Respiratory: 0.6% (-1.1,
2.3)
Cardiovascular: 0.6% (0.0,
1.1)

LATIN AMERICA
Borja-Aburto et al. (1998)
SW Mexico City
1993-1995



Borja-Aburto et al. (1997)
Mexico City
1990-1992


Cakmak et al. (2007)
7 Chilean urban centers
1997-2003














Cifuentes et al. (2000)
Santiago, Chile
1988-1966


Conceifao et al. (2001)
Sao Paulo, Brazil
1994-1997

Loomisetal. (1999)
Mexico City
1993-1995

Ostroetal. (1996)
Santiago, Chile

1989-1991

Lags:0, 1,2, 3, 4, 5, and
multiday avg.
Poisson GAM with default
convergence criteria (only
one smoother). Time-series
study.
Lags:0, 1,2
Poisson iteratively weighted
and filtered least-squares
method. Time-series study.


Lags:0, 1,2,3,4,5,0-5
Poisson GLM with random
effects between cities.
Time-series study.













Lags: 1-2
Poisson GAM with default
convergence criteria;
Poisson GLM. Time-series
study.
Lag: 2
Poisson GAM with default
convergence criteria. Time-
series study.
Lags:0, 1,2,3,4,5,3-5
Poisson GAM with default
convergence criteria. Time-
series study.
Lag:0
OLS, Poisson. Time-series
study


24-h avg: 5.6 ppb
Copollutants: PM25, O3,
NO2; 2-pollutant models



24-h avg median:
5.3 ppb
TSP, O3CO;
2-pollutant models


24-h avg ranged from 9.12
ppb (Las Condes) to 64.06
ppb (Independencia)
Population-weighted avg
concentratioN: 14.08 ppb
Copollutants: PM10, O3, CO











24-h avg: 18.1 ppb
Copollutants: PM25, PM10-
2.5, CO, NO2, O3


24-h avg: 7.4 ppb
Copollutants: PM10, CO, O3


24-h avg: 5.6 ppb
Copollutants: PM25, O3

1-h max: 60 ppb
Copollutants: PMio, O3,
NO2;

2-pollutant models
All cause; respiratory;
cardiovascular; other; all
ages; age >65 yrs



All cause; respiratory;
cardiovascular; all ages;
age < 5 yrs; age >65 yrs


All cause; respiratory;
cardiovascular; all ages;
age
< 65 yrs; age
65-74 yrs; age 75-84 yrs;
age 85+ yrs











All cause



Child mortality (age under 5
yrs)


Infant mortality


All cause




SO2 risk estimates NR.
PM2 5 and O3 were
associated with mortality.



All-cause: Lag 0: 0.2%
(-1.1,1.5)
Cardiovascular: Lag 0: 0.7%
(-1.6,3.0)
Respiratory: Lag 0: -1.0%
(-5.0, 3.2)
All cause: All ages:
Single-pollutant:
Lag 1:4.0% (2.4, 5.6)
Lag 0-5: 6.5% (4.5, 8.5)
Multipollutant:
Lag 1:3.2% (1.3, 5.1)
< 65 yrs:
Lag 0-5: 3.0% (0.6, 5.5)
65-74 yrs:
Lag 0-5: 5.1% (1.2,9.1
75-84 yrs:
Lag 0-5: 7.8% (4.1, 11.6)
85+ yrs: 7.8% (4.2, 11.5)
Warm Season: Lag 0-5:
7.2% (4.1, 10.3)
Cool Season: Lag 0-5: 3.0%
(-0.4, 6.5)
Poisson GLM: Single-
pollutant:
Lag 1-2:0.2% (-0.9, 1.3)
Wth other pollutants: Lag 1-
2: -0.6% (-1.7, 0.5)
Single-pollutant:
Lag 2: 17.0% (7.0, 28.0);
Wth all other pollutants:
Lag 2: 13.7% (-1.1, 30.8)
SO2 risk estimates NR.
PM2 5 and O3 were
associated with mortality.

Lag 0: 0.7% (-0.3, 1.7)




May 2008
F-84
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Pereiraetal. (1998)
Sao Paulo, Brazil
1991-1992
Saldivaetal. (1994)
Sao Paulo, Brazil
1990-1991
Saldivaetal. (1995)
Sao Paulo, Brazil
1990-1991
METHODS
Lag:0
Poisson GLM. Time-series
study.
Lags: 0-2
OLS of raw or transformed
data. Time-series study.
Lag: 0-1
OLS; Poisson with GEE.
Time-series study.
POLLUTANTS
24-h avg: 6.6 ppb
Copollutants: PM10, O3,
NO2, CO
24-h avg: 6.0 ppb
Copollutants: PMio, O3,
NO2, CO; multipollutant
models
24-h avg: 6.5 ppb
Copollutants: PMio, O3,
NO2, CO; 2-pollutant
models
OUTCOME
Intrauterine mortality
Respiratory; age < 5 yrs
All cause; age 65+ yrs
FINDINGS
Single-pollutant model:
11. 5% (-0.3, 24.7)
With other pollutants: 8.6%
(-8.7, 29.3)
-1.0% (-47. 1,45.1)
Single-pollutant: 8.5%
(1.3, 15.6)
With other pollutants:
-3.1% (-13.0, 6.9)
ASIA
Ha et al. (2003)
Seoul, Korea
1995-1999
Hong et al. (2002)
Seoul, Korea
1995-1998
Hong et al. (2002)
Seoul, Korea
1995-1998
Kwon et al. (2001)
Seoul, Korea
1994-1998
Lee et al. (2000)
Seoul, Korea
2000-2004
Lee etal. (1999)
Seoul and Ulsan, Korea
1991-1995
Lee and Schwartz (1999)
Seoul, Korea
1991-1995
Lee et al. (2000)
7 Korean cities
1991-1997
Lag:0
Poisson GAM with default
convergence criteria. Time-
series study.
Lag: 2
GAM with default
convergence criteria. Time-
series study.
Lag: 2
Poisson GAM with default
convergence criteria. Time-
series study.
Lag:0
Poisson GAM with default
convergence criteria; case-
crossover analysis using
conditional logistic
regression.
Lag: 1
GAM with stringent
convergence criteria. Time-
series study.
Lags: 0-2
Poisson with GEE. Time-
series study.
Lags: 0-2
Conditional logistic
regression. Case-crossover
with bidirectional control
sampling.
Lags: 0-1
Poisson GAM with default
convergence criteria. Time-
series study.
24-h avg: 11.1 ppb
Copollutants: PMio, O3,
NO2, CO
24-h avg (ppb): 12.1 (7.4)
Copollutants: PM10 NO2
CO, O3
24-h avg: 12.1 ppb
Copollutants: PMio, O3,
NO2, CO
24-h avg: 13.4 ppb
Copollutants: PMio, O3,
NO2, CO
24-h avg (ppb): 5.20 (2. 17)
Copollutants: PM10, CO,
NO2, O3
1-h max: Seoul: 26 ppb
UlsaN: 31 ppb
Copollutants: TSP, O3
1-h max: 26 ppb
Copollutants: TSP, O3
24-h avg SO2 ranged from
12.1 ppb (Kwangju)to31.4
ppb (Taegu)
Copollutants: TSP, NO2, O3,
CO
All cause; respiratory;
postneonatal (1 mo to 1 yr);
age
2-64 yrs; age 65+
Stroke
Acute stroke mortality
Mortality in a cohort of
patients with congestive
heart failure
Non-accidental
All cause
All cause
All cause
All cause:
Postneonates: 11.3% (4.0,
19.1)
Age 65+ yrs: 3.2% (3.1, 3.3)
% increase (per 5.7 ppb
SO,)
2.9% (0.8, 5.0) lag 2
Stratified by PM10
(MediaN: 47.4 pg/m3)
Med:3.8%
5.2% (1 .4, 9.0)
Odds ratio in general
populatioN: 1.0% (-0.1, 2.1)
Congestive heart failure
cohort: 6.9% (-3.4, 18.3)
% Increase (per 3.06 ppb
S02)
2.7(1.8, 3.5) lag 1
Seoul: 1.5% (1.1, 1.9)
UlsaN: 1.0% (-0.2, 2.2)
Two controls, ± 1 wk:
0.3% (-0.5, 1.0)
Four controls, ± 2 wks:
1 .0% (0.3, 1 .6)
Single-pollutant :
Lag 0-1 : 0.6% (0.3, 0.8)
Multipollutant :
Lag 0-1 : 0.6% (0.2, 0.9)
May 2008
F-85
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Qian et al. (2007)
Wuhan, China
2000-2004






















HEI International Scientific
Oversight Committee
(2004)
East Asian cities.





Tsai et al. (2003)
Kaohsiung, Taiwan
1994-2000



Vennersetal. (2003)
Chonqing, China
1995


METHODS
Lag:0
Poisson GAM with stringent
convergence criteria. Time-
series study.






















The lags and multi-day
averaging used in varied
Meta-analysis of time-series
study results





Lags: 0-2
Conditional logistic
regression. Case-crossover
analysis.


Lags:0, 1,2,3,4,5
Poisson GLM, time-series
study



POLLUTANTS
24-h avg (Mg/m3): 44.1
(25.3)
Copollutants:
PM10
NO2
03




















The levels of SO2 in these
Asian cities were generally
higher than those in the
U.S. or Canadian cities,
with more than half of these
studies reporting the mean
SO2 levels higher than 10
ppb.
Copollutants considered
varied across studies.
24-h avg: 11.2 ppb
Copollutants: PM10, NO2,
O3, CO



24-h avg: 74.5 ppb
Copollutants:PM25



OUTCOME
Non-accidental,
cardiovascular, stroke,
cardiac, respiratory,
cardiopulmonary






















All-cause







All cause; respiratory;
cardiovascular; tropical area




All cause, cardiovascular,
respiratory, cancer, and
other



FINDINGS
Mean % change (per 10
pg/m3 SOJ Non-accidental
All Ages: 0.01 (-0.46,0.47)
< 65: -0.55 (-1.33, 0.23)
> 65: 0.22 (-0.32, 0.76)
Cardiovascular
All Ages: 0.20 (-0.45, 0.86)
< 65: -0.63 (-1.96, 0.72)
> 65: 0.41 (-0.31, 1.14)
Stroke
All Ages: -0.27 (-1.04,
0.51)
< 65: -1.35 (-3.01, 0.33)
> 65: 0.01 (-0.87, 0.88)
Cardiac
All Ages: 0.88 (-0.22, 1.99)
< 65: 0.29 (-2. 11, 2.75)
>65: 1.01 (-0.18,2.21)
Respiratory
All Ages: 1.13 (-0.28, 2.56)
< 65: -0.59 (-4.24, 3.19)
>65: 1.36 (-0.05, 2.80)
Cardiopulmonary
All Ages: 0.29 (-0.33, 0.92)
< 65: -0.80 (-2.07, 0.49)
> 65: 0.53 (-0.1 5, 1.20)
The estimates were found
to be heterogeneous across
11 studies. Random-Effects
Estimate: 1.49% (95% Cl:
0.86, 2.13); Fixed-Effects
Estimate: 1.01%(95%CI:
0.73, 1.28).



Odds ratios:
All cause: 1.1% (-4.4, 6.8)
Respiratory: 3.5% (-17.6,
29.9)
Cardiovascular: 2.4% (-9.1,
15.4)
All cause: Lag 2: 1.1%
(-0.1,2.4)
Cardiovascular: Lag 2: 2.8%
(0.4, 5.2)
Respiratory: Lag 2: 3.0%
(0.4, 5.7)
May 2008
F-86
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Vennersetal. (2003)
Chongqing, China 1995




























































Wong et al. (2001)
Hong Kong
1995-1997
METHODS
Lags:0, 1,2,3,4,5
Robust Poisson regression.
Time-series study.




























































Lags:0, 1,2
Poisson GAM with default
convergence criteria. Time-
series study.
POLLUTANTS
24-h avg (Mg/m3): 213.0
Copollutants: PM2.5




























































24-h avg:
Warm Season: 6.4 ppb
Cool Season: 6.0 ppb
Copollutants: PM10, O3,
NO2; 2-pollutant models
OUTCOME
Total
(Non-accidental),
Cardiovascular, Respiratory,
Cancer, Other




























































All cause; respiratory;
cardiovascular
FINDINGS
Relative Risk (95% Cl) Total
(per 100 [ig/m SO2)
1.01 (0.96, 1.06) lag 0
1 .03 (0.98, 1 .08) lag 1
1 .04 (1 .00, 1 .09) lag 2
1 .04 (0.99, 1 .08) lag 3
1.01 (0.96, 1.05) lag 4
1.01 (0.97, 1.06) lag 5
AIIYr-Lag2 (per100pg/m3
S02)
Total: 1.04(1.00, 1.09)
Respiratory: 1.11 (1.02,
1.22)
Cardiovascular: 1.10(1.02,
1.20)
Cancer: 1.02 (0.93, 1.28)
Other: 1.03(0.97, 1.10)
6 mos (Jan-Jun)-Lag 2 (per
100pg/m3SO2)
Total: 1.08(1.02, 1.14)
Respiratory: 1.16 (1.04,
1.29)
Cardiovascular: 1.23(1.11,
1.17)
Cancer: 0.95 (0.70, 1.29)
Other: 1.08(0.99, 1.14)
All Yr (Excluding high-
mortality days)-Lag 2 (per
100pg/m3SO2)
Total: 1.02(0.97, 1.07)
Respiratory: 1.07 (0.98,
1.18)
Cardiovascular: 1 .05 (0.96,
1.14)
Cancer: 1.06(0.82, 1.35)
Other: 1.00(0.93, 1.07)
All Yr-Lag 3 (per 100 pg/m3
S02)
Total: 1.04(0.99, 1.08)
Respiratory: 1.00 (0.91,
1.10)
Cardiovascular: 1.20(1.11,
1.30)
Cancer: 0.94 (0.74, 1.18)
Other: 0.99 (0.85, 1.06)
6 mos (Jan-Jun)-Lag 3 (per
100 ug/m3SO2)
Total: 1.01 (0.96, 1.07)
Respiratory: 0.97 (0.87,
1.09)
Cardiovascular: 1.18(1.07,
1.30)
Cancer: 1.02 (0.76, 1.37)
Other: 0.96 (0.88, 1.04)
All Yr (Excluding high-
mortality days)-Lag 3
(per 100 pg/n^SOJ
Total: 1.03(0.99, 1.08)
Respiratory: 1.01 (0.92,
1.12)
Cardiovascular: 1.20(1.10,
1.30)
Cancer: 0.90 (0.70, 1.17)
Other: 0.97 (0.90, 1.04)
All cause: Lag 1:3.2% (1.1,
5.3)
Respiratory: Lag 0: 5.3%
(2.2,8.6)
Cardiovascular: Lag 1 : 4.3%
(1.1,7.5)
May 2008
F-87
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Wong et al. (2002b)
Hong Kong
1995-1998
Yang et al. (2004)
Taipei, Taiwan
1994-1998
METHODS
Lags:0, 1,2,0-1,0-2
Poisson GLM. Time-series
study.
Lags: 0-2
Conditional logistic
regression. Case-crossover
analysis.
POLLUTANTS
24-h avg: 29 ppb
Copollutants: PM10, O3,
NO2; 2-pollutant models
24-h avg: 5.5 ppb
Copollutants: PM10, NO2,
O3, CO
OUTCOME
Respiratory; cardiovascular;
COPD; pneumonia and
influenza; ischemic heart
disease; cerebrovascular
All cause; respiratory;
cardiovascular; subtropical
area
FINDINGS
Respiratory: Lag 0-1: 2.6%
(0.2,5.1)
Cardiovascular: Lag 0-1:
1.2% (-1.0, 3.5)
Odds ratios:
All cause: -0.5% (-7.0, 6.6)
Respiratory: -1.8% (-23.1,
25.3);
Cardiovascular: -3.4%
(-15.2, 10.0)
AUSTRALIA
Simpson et al. (1997)
Brisbane, Australia
1987-1993
Lag: 0
Autoregressive Poisson
with GEE. Time-series
study.
24-h avg: 4.2 ppb
1-h max: 9.6 ppb
Copollutants: PM10, bsp,
O3, NO2.CO
All cause; respiratory;
cardiovascular
All cause:
All yr:
Lag 0: -2.8% (-2.7, 8.6)
Summer:
Lag 0: 2.8% (-8.3, 15.2)
Winter:
Lag 0: 2.8% (-3.9, 9.8)
Table F-6.   Associations of long-term exposure to SO2 with respiratory morbidity.
STUDY
METHODS
POLLUTANT
RESULTS
UNITED STATES AND CANADA
Dockery et al.
(1996)
18 sites in U.S.
6 sites in Canada












Study of the respiratory health effects of acid
aerosols in 13,369 white children aged 8 to
12 yrs old from 24 communities in the United
States and Canada between 1988 and 1991.
Information was gathered by questionnaire and
a pulmonary function.











SO2 mean 4.8 ppm
SD 3 5
Range 0.2, 12.9













With the exception of the gaseous acids (nitrous
and nitric acid), none of the particulate or
gaseous pollutants, including SO2, were
associated with increased asthma or any
asthmatic symptoms. Stronger associations with
particulate pollutants were observed for bronchitis
and bronchitic symptoms.
Odds Ratio (95% Cl) for 12.7 ppb range of SO2
pollution
Asthma 1 .05 (057, 1 .93)
Attacks of Wheeze 1 .07 (0.75, 1 .55)
Persistent Wheeze 1 .19 (0.80, 1 .79)
Any asthmatic symptoms 1.16 (0.80, 1.68)
Bronchitis 1.56 (0.95, 2.56)
Chronic cough 1.02 (0.66, 1.58)
Chronic phlegm 1 .55 (1 .01 , 2.37)
Any Bronchitic symptoms 1.29 (0.98, 1.71)
May 2008
F-88
DRAFT—DO NOT QUOTE OR CITE

-------
     STUDY
                                   METHODS
                                                                     POLLUTANT
                                                                                                          RESULTS
Dockery et al.
(1989)

Watertown, MA;
St. Louis, MO;
Portage, Wl;
Kingston-Harriman,
TN; Steubenville,
OH; Topeka, KS

1980-1981 school
yr
Cross-sectional assessment of the association
between air pollution and chronic respiratory
health of 5,422 (10-12 yrs) white children
examined in the 1980-1981 school yr. Children
were part of the cohort of children in the Six
Cities Study of Air pollution and Health.
Symptoms were analyzed using logistic
regression that included sex, age, indicators of
parental education,  maternal smoking, indicator
for gas stove, and an indicator for city.
Respiratory symptoms investigated were
bronchitis, chronic cough, chest illness,
persistent wheeze, asthma. The logarithm of
pulmonary function was fitted to a multiple
linear regression  model that included sex, sex-
specific log of height, age, indicators of parental
education, maternal smoking, a gas stove
indicator, and city indicator. Annual means of
the 24 h avg  air pollutant concentration for the
12 mos preceding the examination of each child
was calculated for each city.
Daily mean
concentrations,
averaging hourly
concentrations for each
day with at least
18 hourly values

Portage: 4.2 ppb
Topeka: 3.5
Watertown: 10.5
Kingston: 6.5
St. Louis: 13.5
Steubenville: 27.8
No significant associations between SO2 and any
pulmonary function measurements. No significant
association between SO2 and symptoms.

Relative odds and 95% Cl between most/least
polluted cities:
Bronchitis: 1.5(0.4,5.8)
Chronic cough: 1.8(0.3, 12.5)
Chest illness: 1.5 (0.4,5.9)
Persistent wheeze: 0.9 (0.4, 1.9)
Asthma: 0.6(0.3, 1.2)

Reference symptoms:
Hay fever: 0.6 (0.2, 1.7)
Earache: 1.2(0.3,5.3)
Nonrespiratory illness: 1.0 (0.6, 1.5)

Analysis stratified by asthma or persistent wheeze
bronchitis
No wheeze or asthma 1.5 (0.5, 4.3)
Yes wheeze or asthma 2.0 (0.3,  14.3)

Chronic cough
No wheeze or asthma 2.4 (0.5, 11.7)
Yes wheeze or asthma 1.9 (0.1,  44.1)

Chest illness
No wheeze or asthma 1.5 (0.4, 5.6)
Yes wheeze or asthma 1.9 (0.3,  13.0)
Euleretal. (1987)

California, USA
Cross-sectional study of 7,445 (25 yrs or older)
Seventh-Day Adventists who lived in their 1977
residential areas (Los Angeles and it border
counties, San Francisco, and San Diego) for at
least 10 yrs to determine the effect of long-term
cumulative exposure to ambient levels of TSP
and SO2 on COPD symptoms. Study population
is subgroup of NCI-funded ASHMOG study that
enrolled 36,805 Seventh-Day Adventists in
1974. Each participant's cumulative exposure to
the pollutant exceeding 4 different threshold
levels were estimated using moly residence ZIP
code histories and interpolated dosages from
state monitoring stations. Participants
completed a questionnaire on respiratory
symptoms, smoking history, occupational
history, and residence history.
None provided
Study reported that SO2 exposure was not
associated with symptoms of COPD until
concentrations exceeded 4 ppm. The correlation
coefficient of SO2 (above 4 ppm) with TSP (above
200 pg/m3) the highest exposure levels for these
two pollutants was 0.30; thus, the authors
believed that it was possible to separate the
effects of SO2 from TSP. Multiple regressions
used in the analysis. No significant effect at
exposures levels below 4 ppm or above 8 ppm.

Relative risk estimate (based on 1,003 cases)
SO2 exposure above 2 ppm during 11  yrs of study
 2000 h/yr: 1.09 1000h/yr: 1.04
 500 h/yr: 1.03

SO2 exposure above 4 ppm
 500 h/yr: 1.18 250 h/yr: 1.09
 100 h/yr: 1.03

SO2 above 8 ppm
 60 h/yr: 1.07 30 h/yr: 1.03
 15 h/yr: 1.02

SO2 above 14 ppm
 10 h/yr: 1.03 5 h/yr:  1.01
 1 h/yr: 1.00
Goss et al. (2004)

U.S. nationwide

1999-2000
Cohort study of 18,491 cystic fibrosis patients
over 6 yrs of age who were enrolled in the
Cystic Fibrosis Foundation National Patient
Registry in 1999 and 2000. Mean age of
patients was 18.4 yrs; 92% had pancreatic
insufficiency. Air pollution from the Aerometric
Information Retrieval System linked with
patient's home ZIP code. Air pollutants studied
included O3, NO2, SO2, CO, PM10, and PM2.5.
Health endpoints of interest were pulmonary
exacerbations, lung function, and mortality.
However, study did  not have enough power to
assess the outcome of mortality. Logistic
regression and polytomous regression models
that adjusted for sex, age, weight, race, airway
colonization, pancreatic function, and insurance
status were used.
Mean (SD): 4.91
(2.6) ppb
MediaN: 4.3 ppb
IQR: 2.7-5.9 ppb
With the single-pollutant model, no significant
association between SO2 and pulmonary
exacerbations.

Odds ratio per 10 ppb increase in SO2:
0.83 (95% Cl: 0.71, 1.01), p = 0.068

No clear association between pulmonary function
and SO2. No effect estimates provided.
May 2008
                                       F-89
                      DRAFT—DO NOT QUOTE OR CITE

-------
     STUDY
                                   METHODS
                                                                     POLLUTANT
                                                                                                          RESULTS
McDonnell et al.
(1999)

California, U.S.

1973-1992
Prospective study (over 15 yrs) of 3,091
nonsmokers aged 27-87 yrs that evaluated the
association between long-term ambient O3
exposure and the development of adult-onset
asthma. Cohort consisted of nonsmoking, non-
Hispanic white, California Seventh Day
Adventists who were enrolled  in 1977 in the
AHSMOG study. Logistic regression used to
assess the association between the 1973-1992
mean 8-h avg ambient O3 concentration and the
1977-1992 incidence of doctor-told asthma.
Levels of PMio, NO2, and SO4 were measured
but no effect estimates were given.
Mean: SO26.8 pg/m
Range: 0.0-10.2 pg/m3

Correlation coefficient
r = 0.25 with O3
No significant positive association between SO2
and asthma for males or females. Addition of a
second pollutant to the O3 model for the male
subjects, did not result in a decrease of more than
10% in the magnitude of the regression
coefficient for O3, and for the females addition did
not cause the coefficient for O3 to become
significantly positive
Schwartz (1989)

United States

1976-1980
Cross-sectional study using data from the
Second National Health and Nutrition
Examination Survey (NHANES II) to examine
the relation between air pollution and lung
function growth in 4,300 children and youths 6-
24 yrs old. A two-staged analysis was
performed that consisted of (1) regression
equations including factors known to affect lung
function and (2) a regression of the  residuals of
the first regression on air pollution.
Annual percentiles
(ppm):

10th: 0.0060
25th: 0.0106
50th: 0.0131
75th: 0.0159
90th: 0.0193
The study did not find an association between
SO2 and any of the lung function growth
measurements (i.e., FVC, FEV1, and Peak flow).
                                                              EUROPE
Ackermann-Liebrich
etal. (1997)

8 communities in
Switzerland
Aarau, Basel,
Davos, Geneva,
Lugano, Montana,
Payerne, and Wald

1991-1993
Cross-sectional population based study of
9,651 adults (18-60 yrs) in 8 areas in
Switzerland (SAPALDIA), to evaluate the effect
of long-term exposure of air pollutants on lung
function. Examined the effects of SO2, NO2, O3,
TSP, and PMio. Participants were given a
medical exam that included questionnaire data,
lung function tests, skin prick testing, and end-
expiratory CO concentration. Subjects had to
reside in the area for at least 3 yrs to be in the
study.
Mean SO2 in 1991
(pg/m3)
Mean: 11.7
SD:7.1
Range: 2.5, 25.5
Mean values of SO2, PMio, and NO2were
significantly associated with reduction in
pulmonary function. SO2was correlated with Pm30
(r = 0.78), PM10 (r = 0.93) and NO2 (r = 0.86).
Authors stated that the association with SO2
disappeared after controlling for PMio but no data
was shown.

Regression coefficients and 95% Cl in healthy
never smokers (per 10 pg/m3 increase in annual
avg SOJ

FVC: -0.0325 (-0.0390, -0.0260)
FEV,: -0.0125 (-0.0192,-0.0058)
Braun-Fahrlander
etal. (1997)

10 communities in
Switzerland
Anieres, Bern, Biel,
Geneva, Langnau,
Lugano, Montana,
Payerne, Rheintal,
Zurich

1992-1993
Cross-sectional study of 4,470 children (6-15
yrs) living in 10 different communities in
Switzerland to determine the effects of long
term exposure to PMio, NO2, SO2, and O3 on
respiratory and allergic symptoms and
illnesses. Part of the Swiss Study on Childhood
Allergy and Respiratory Symptoms with
Respect to Air Pollution (SCARPOL).
Annual mean SO2
(pg/m3)

Lugano:23
Geneva: 13
Zurich: 16
BerN: 11
Anieres: 4
Biel: 15
Rheintal: 8
Langnau: NA
Payerne: 3
MontaN: 2
This study reported that the annual mean SO2,
PMio, and NO2 were positively and significantly
associated with prevalence rates of chronic
cough, nocturnal dry cough, and bronchitis and
conjunctivitis symptoms. Strongest association
found with PMio. However, there was no
significant association between SO2 and asthma
or allergic rhinitis.

Adjusted relative odds between the most/least
polluted community 2-23 pg/m2 (0.8, 8.8 ppb)
Chronic cough: 1.57 (1.02, 2.42)
Nocturnal dry cough: 1.66 (1.16, 2.38)
Bronchitis: 1.48 (0.98, 2.24)
Wheeze: 0.88 (0.54, 1.44)
Asthma (ever): 0.74 (0.45, 1.21)
Sneezing during pollen Season: 1.07 (0.67, 1.70)
Hay fever: 0.84 (0.55, 1.29)
Conjunctivitis symptoms: 1.74 (1.22, 2.46)
Diarrhea: 1.02(0.75, 1.39)
Charpin et al.
(1999)

Etang de Berre
area of France:
Aries, Istres, Port
de Bouc, Rognac-
Velaux, Salon de
Provence, Sausset,
Vitrolles

Jan-Feb1993
Cross-sectional cohort study of 2,073 children
(10-11 yrs) from 7 communities in France (some
with the highest photochemical exposures in
France) to test the hypothesis that atopy is
greater in towns with higher photochemical
pollution levels. Mean levels of SO2, NO2, and
O3 were measured for 2 mos in 1993. Children
tested for atopy based on skin prick test (house
dust mite, cat dander, grass pollen, cypress
pollen, and Alternaria). To be eligible for the
study, subjects must have resided  in current
town for at least 3 yrs. Questionnaire filled out
by parents that included questions on
socioeconomic status and passive smoking at
home. Two-mo mean level of air pollutants used
in logistic regression analysis.
24-h mean (SD)
S02 (pg/m3)

Aries: 29.7 (15.5)
Istres: 23.8 (12.7)
Port de Bouc: 32.3 (24.5)
Rognanc-Velaux:
39.5(21.8)
Salon de Provence:
17.3(11.6)
Sausset: 29.0 (28.7)
Vitrolles: 57.4 (32.0)
Study did not demonstrate any association
between air pollution and atopic status of the
children living in the seven communities, some
with high  photochemical exposures. A limitation of
study is that authors did not consider short-term
variation in air pollution and did not have any
indoor air pollution measurements.
May 2008
                                       F-90
                      DRAFT—DO NOT QUOTE OR CITE

-------
     STUDY
                                   METHODS
                                                                    POLLUTANT
                                                                                                         RESULTS
Frischer et al.
(2001)

Nine communities
in Austria

Sep-Oct1997
Cross-sectional cohort study of 877 children
(mean age 11.2 yrs) living in 9 sites with
different O3 exposures. Urinary eosinophil
protein U-EPX) measured as a marker of
eosinophil activation. U-EPX determined from a
single spot urine sample analyzed with  linear
regression models.
1/2-h avg SO2:

30-day mean 2.70 ppb
IQR2.1 ppb
No significant association between SO2 and U-
EPX

Regression coefficient and SE -10.57 (0.25) per
ppb SO2
Frischer et al.
(1999)

Nine communities
in Austria

1994-1996
Longitudinal cohort study of 1150 children
(mean age 7.8 yrs) to investigate the long-term
effects of O3 on lung  growth. Children were
followed for 3 yrs and lung function was
recorded biannually,  before and after
summertime. The dependant variables were
change in FVC, FEV,, and MEF50. The 9 sites
were selected to represent a broad range of O3
exposures. GEE models adjusted for baseline
function, atopy, gender, site, environmental
tobacco smoke exposure, season, and change
in height. Other pollutants studied included
PM10, SO2, and NO2.
Annual mean SO2 (ppb)
in 1994

AmstetteN: 3.75
St. ValentiN: 3.00
Krems: 3.75
HeidenreichsteiN: 4.13
Ganserndorf: 5.63
Mistelbach: 5.25
Wiesmath: 6.00
Bruck: 4.88
Pollau: 2.25
No consistent association observed between lung
function and SO2, NO2 and PM,o. A negative
effect estimate was observed during the summer
and a positive estimate during the winter.

Change in lung function (per ppb SO2):
FEV, (mL/day):
Summer: -0.018 (0.004), p < 0.001
Winter: 0.003 (0.001), p < 0.001

FVC (mL/day):
Summer: -0.009 (0.004), p = 0.02
Winter: 0.002 (0.001), p = 0.03

MEF50 (mL/s/day):
Summer: -0.059 (0.010), p < 0.001
Winter: 0.003 (0.003), p = 0.26
Frye et al. (2003)

Zerbst, Hettstedt,
Bitterfeld.East
Germany

1992-93, 1995-
1996, 1998-1999
Three consecutive cross-sectional surveys of
children (11-14 yrs) from three communities in
East Germany. Parents of 3,155 children
completed a questionnaire on symptoms.  Lung
function tests performed on 2,493 children.
Study excluded children if they lived for less
than 2 yrs in current home and if their previous
home was more than 2 km away. The log-
transformed lung function parameters were
used as the response variables in a linear
regression analysis that controlled for sex,
height, season of examination, lung function
equipment, parental education, parental atopy,
and environmental tobacco smoke. Used  avg of
annual means of pollutants 2 yrs preceding
each survey.
Used avg of annual
means of pollutants 2 yrs
preceding health
measurement

High of 113 pg/m3 (in
Bitterfeld) to a low of
6 pg/m3. (Pollution values
only described in figure)
The annual mean TSP declined from 79 to
25 pg/m3 and SO2 from 113 to 6 pg/m3 and the
mean FVC and FEV, increased from 1992-1993
to 1998-1999. Study concluded that reduction of
air pollution in a short time period may improve
children's lung function.

Percent change of lung function for a 100-pg/m3
decrease in SO2 2 yrs before the investigation
(N: 1,911)
FVC: 4.9 (0.7, 9.3)
FEV,: 3.0 (-1.1, 7.2)
FEV,/FVC:-1.5(-3.0, 0.1)
Garcia-Marcos
etal. (1999)

Cartagena, Spain

winter 1992
A total of 340 children (10-11 yrs) living in and
attending schools within a polluted and a
relatively nonpolluted area were included in this
study which aimed to establish the relative
contribution socioeconomic status, parental
smoking, and air pollution on asthma
symptoms, spirometry, and bronchodilator
response. Parents completed questionnaire on
respiratory symptoms and risk factors including,
living in polluted area, maternal smoking,
paternal smoking, number of people living in the
house, proximity to heavy traffic roads.
Spirometry was performed before and after an
inhaled 0.2 mg fenoterol was delivered to
determine bronchodilator response.
Bronchodilator response was considered
positive if the FVC after fenoterol was increased
by at least 10% or PEF by 12%. Logistic
regression included as independent variables
all the risk factors.
Annual mean SO2
(pg/m3) Polluted areas
75 pg/m3
Nonpolluted areas:
20 pg/m3
This study found that living in the polluted areas
reduced the risk of a positive bronchodilator
response (RR = 0.61,
p = 004).
May 2008
                                      F-91
                      DRAFT—DO NOT QUOTE OR CITE

-------
     STUDY
                                    METHODS
                                                                     POLLUTANT
                                                                                                          RESULTS
Gokirmak et al.
(2003)
Malatya, Turkey
Study on occupational exposure to SO2 in
apricot sulfurization workers that investigated
the role of oxidative stress resulting exposure to
high concentrations of SO2 on
bronchoconstriction. Forty workers (mean age:
28 yrs, range 16-60 yrs) who have been
working in apricot sulfurization for 20-25 days
each yr and 20 controls (mean age: 29 yrs,
range 17-42) who had no SO2 exposure
participated in the study. Activities of
antioxidant enzymes (glutathione peroxidase
(GSH-Px), superoxide dismutase (SOD) and
catalase) malondialdehyde (MDA)
concentrations (marker of lipid peroxidation),
and pulmonary function test measured in
subjects.
SO2 cone ranged from
106.6 to 639.2 ppm in
9 apricot farms.

Mean cone around
sulfurization chamber:
324.1 (35.1) ppm
SOD, GSH-Px, and catalase activities were lower
and malondialdehyde concentrations were higher
in the apricot sulfurization workers compared to
controls. Pulmonary function decreased after SO2
exposure among the apricot sulfurization workers.
Authors concluded that occupational exposure to
high concentrations of SO2 enhances oxidative
stress and that lipid peroxidation may be a
mechanism of SO2 induced bronchoconstriction.

Apricot sulfurization workers vs. controls
Mean (SD)
SOD (U/mL): 2.2 (0.6) vs. 3.2 (0.7) U/m , p <
0.0001
Glutathione peroxidase (U/mL): 0.6 (0.3) vs. 1.1
(0.3),
p < 0.0001
Catalase (L/L): 107.6 (27.4) vs. 152.6 (14.3), p <
0.0001
MDA (nmol/L): 4.1 (0.9) vs. 1.9 (5.3) , p < 0.0001

Before vs. after SO2 exposure among apricot
sulfurization workers
Mean (SD)
FVC (% predicted) 88 (17) vs. 84 (16) , p < 0.001
FEV, (% predicted) 98 (14) vs. 87 (14), p < 0.001
FEW1/FVC: 92 (7) vs. 86 (9), p < 0.001
FEF25-75% (% predicted) 108 (19) vs. 87 (23) , p
< 0.001
Heinrich et al.
(2002)

Reunified Germany
Bitterfeld, Hettstedt,
Zerbst

1992-1993, 1995-
1996, 1998-1999
Three cross-sectional surveys of children (5-14
yrs) from 3 areas that were formerly part of East
Germany to investigate the impact of declines
in TSP and SO2 on prevalence of nonallergic
respiratory disorders in children. Study
excluded children if they lived for less than 2 yrs
in current home and if their previous home was
more than 2 km away. GEE used for analysis.
SO2 concentration in
pg/m3

Yr/Zerbst/Bitterf/ Hettst
1991/78/113/84
1992/58/75/46
1993/42/60/49
1994/29/35/38
1995/21/30/26
1996252425
1997/13/13/13
1998/8/9/6
Study found that SO2 exposure was significantly
associated with prevalence of bronchitis, frequent
colds, and febrile infections. While results are
reported as risk for an increase in air pollutant,
the respiratory health of children improved with
declines in TSP and  SO2. Authors concluded that
exposure to combustion-derived air pollution is
causally related to nonallergic respiratory health
in children.

Odds ratio and 95%  Cl: (per 100 pg/m3 in 2 yr
mean SO2)

All children:
Bronchitis: 2.72 (1.74, 4.23)
Otitis media:  1.42 (0.94, 2.15)
Sinusitis: 2.26 (0.85, 6.04)
Frequent colds: 1.81 (1.23, 2.68)
Febrile infections: 1.76 (1.02, 3.03)
Cough in morning: 1.10 (0.73, 1.64)
Shortness of breath: 1.31 (0.84, 2.03)

Children without indoor exposures (living in damp
houses with visible molds, ETS in the home, gas
cooking emissions, and contact with cats)
Bronchitis: 4.26 (2.15, 8.46)
Otitis media:  1.43 (0.73, 2.81)
Sinusitis: 2.95 (0.52, 16.6)
Frequent colds: 2.29 (1.15, 4.54)
Febrile infections: 1.75 (0.78, 3.91)
Cough in morning: 1.00 (0.38, 2.64)
Shortness of breath: 2.07 (0.90, 4.75)
May 2008
                                       F-92
                       DRAFT—DO NOT QUOTE OR CITE

-------
     STUDY
                                   METHODS
                                                                     POLLUTANT
                                                                                                          RESULTS
Herbarth et al.
(2001)

East Germany

1993-1997
Meta-analysis of three cross-sectional studies:
(1) Study on Airway Diseases and Allergies
among Kindergarten Children (KIGA), (2) the
Leipzig Infection, Airway Disease and Allergy
Study on School starters (LISS), and (3) KIGA-
IND, which was based on the KIGA design but
conducted in 3 differentially polluted industrial
areas. A total of 3,816 children participated in
the three studies. Analysis of data from parent-
completed questionnaires to determine the
effect of life time exposure to SO2 and TSP on
the occurrence of acute bronchitis. Total lifetime
exposure burden corresponds to the exposure
duration from birth to time of the study. The
LISS study was divided in to LISS-U for the
urban area and LISS-R for the rural area.
Logistic regression analysis used that adjusted
for predisposition in the family (mother or father
with bronchitis), ETS, smoking during
pregnancy or in the presence of the pregnant
Avg lifetime exposure
burden of SO2 (pg/m3)

KIGA: 142
LISS: 48
LIISS:R47
KIGA-IND: 59
This study found the highest bronchitis
prevalence in the KIGA cohort and the lowest in
the LISS cohort, which is consistent with the SO2
concentrations in these cohorts. Study found a
correlative link  between SO2 and bronchitis (R =
0.96, p < 0.001) but not TSP (R = 0.59). Results
of study suggest that SO2 may  be a more
important factor than TSP in the occurrence of
bronchitis in these study areas.

Odds ratio for bronchitis adjusted for parental
predisposition,  smoking, and lifetime exposure to
SO2 and TSP (2-pollutant model).

SO2: 3.51 (2.56, 4.82)
TSP: 0.72 (0.49, 1.04)
Hirschetal. (1999)
Dresden, Germany
Cross sectional study to relate the prevalence
of respiratory and allergic diseases in childhood
to measurements of outdoor air pollutants.
5,421 children ages 5-7 yrs and 9-11 yrs were
evaluated by questionnaires, skin-prick testing,
venipuncture for (lg)E, lung function, and
bronchial challenge test.
Mean (pg/m3): 48.3

Range: 29.0-69.3

25-75 percentile 42.7-
54.3
Sox was positively associated with current
morning cough but not with bronchitis.

Prevalence odds ratio (95% Cl) for symptoms
within past 12 mos, +10 pg/m3:

Wheeze:
Atopic 103 (0.79, 1.35)pg/m3
Nonatopic 1.36 (1.01, 1.84)

Morning Cough:
Atopic 1.22 (0.92, 1.61)
Nonatopic 1.32 (1.07, 1.63)

Prevalence odds ratio (95% Cl) for doctor's
diagnosis, +10 pg/m3:
Asthma
Atopic 1.07 (0.79, 1.45)
Nonatopic 1.35 (1.00, 1.82)

Bronchitis
Atopic 1.04 (0.87, 1.25)
Nonatopic 0.99 (0.88, 1.12)
Horak et al. (2002)
Eight communities
in Austria
1994-1997
Longitudinal cohort study that continued the
work of Frischer et al. (1999) by adding one
more yr of data and analyzing the effects of
PM10 in addition to SO2, NO2, and O3. At the
beginning of the study 975 children (mean age
8.11 yrs) were recruited for the study, but only
80.6% of the children performed all 6 lung
function tests (twice a yr). The difference for
each lung function parameter between two
subsequent measures was divided by the days
between measurements and presents as
difference  per day (dpd) for that parameter.
860 children were included in the GEE analysis
that controlled for sex, atopy, passive smoking,
initial height, height difference, site, and initial
lung function.
Seasonal mean SO2
pg/m3:
Winter:
Mean: 16.8
Range: 7.5, 37.4
Summer:
Mean : 6.9 pg/m3
Range: 3.1, 11.7
Moderate correlation between PMio and SO2 in
the winter (r = 0.52). In a one-pollutant model for
SO2, long term seasonal mean concentration of
SO2 was had a positive association with FVC dpd
and FEV, dpd in the winter, but no effect on
MEF25_75 dpd. In a two-pollutant model with PMio,
wintertime SO2 had a positive association with
FEV, dpd.

Single-pollutant model
FVC dpd:
Summer: 0.009, p = .336
Winter: 0.006, p = .009

FEV, dpd:
Summer: 0.005,  p = 0.576
Winter: 0.005, p = 0.013

MEF25.75:
Summer: 0.015, p = 0.483
Winter: 0.003, p = 0.637

Two-pollutant model:
SO2 + PM10
FVC dpd:
Summer: 0.008, p = 0.395
Winter: 0.004, p = 0.225

FEV, dpd:
Summer: 0.010 (0.271)
Winter: 0.007 (0.025)

MEF25.75dpd:
Summer: 0.037,  p = 0.086
Winter: 0.007, p = 0.429
May 2008
                                       F-93
                      DRAFT—DO NOT QUOTE OR CITE

-------
     STUDY
                                   METHODS
                                                                    POLLUTANT
                                                                                                         RESULTS
Jedrychowski et al.
(1999)
Krakow, Poland
1995 (Mar-Jun) and
1997(Mar-Jun)
Cohort prospective study consisting of 1,001
preadolescent children (9 yrs old) from two
areas of Krakow, Poland. The study examined
lung function growth using  FVC and FEV1
measurements taken in 1995 and then again
two yrs later, 1997. Used a two-stage analysis
that consisted of (1) multivariate linear
regression analyses to determine body
variables that are significant predictors of lung
function growth, and then (2) multivariate
logistic regression to examine the relation
between air pollution and lung function growth.
Annual avg:

City Center (pg/m3):
43.87 (32.69)

Control Area (pg/m3):
31.77(21.93)
The study did not provide individual estimates for
SO2.
Koksal et al. (2003)
Malatya, Turkey
Study on occupational exposure to high
concentrations of SO2 on respiratory symptoms
and pulmonary function on apricot sulfurization
workers. Apricot sulfurization workers (N: 69)
from 15 apricot farms who have been working
in sulfurization of apricots for 20-25 days a yr
during each summer were recruited for the
study. Subjects rated symptoms (itchy eyes,
runny nose, stuffy nose, itchy or scratchy throat,
cough, shortness of breath, phlegm, chest  pain,
and fever) before during and 1-h after each
exposure.
SO2 cone ranged from
106.6to721.0ppm
SO2 exposure at high concentrations increased
symptoms of itchy eyes, shortness of breath,
cough, running and/or stuffy nose, and itchy or
scratchy throat during exposure
(p < 0.05). Inhalation of high concentrations of
SO2for 1-h caused significant decreases in
pulmonary function.
Difference in pulmonary function measured
before and after exposure:
FVC (L) 0.16 (0.42), p< 0.05
FEV, (L) 0.39 (0.36), p < 0.001
FEV,/FVC: 5.22 (6.75), p < 0.001
PEF(L/s) 1.39(1.06), p< 0.001
FEF25.75*, (L/s) 0.82 (0.70), p < 0.001
Kopp et al. (2000)
Ten communities in
Austria and
SW Germany
Longitudinal cohort study of 797 children (mean
age 8.2 yrs) from 2nd and 3rd grades of 10
schools in Austria and SW Germany to assess
the effects of ambient O3 on lung function in
children over a
2-summer period. Study also examined the
association between avg daily lung growth and
SO2, NO2, and PM10. Each child performed 4
lung function tests during spring 1994 and
summer 1995. ISAAC questionnaire used for
respiratory history. Linear regression models
used to assess effect of air pollutants on FVC
and FEV,, which were surrogates of lung
growth.
Mean SO2 (95% Cl) ppb
Apr-Sep 1994
AmstetteN: 3.7 (0.7, 3.9)
StValentiN:2.6(1.5, 5.2)
Krems: 3.7 (0.7, 7.5)
VillingeN: 0.7 (0 , 3.0)
HeindenreichsteiN: 3.7,
(0.7, 7.5)
Ganserndorf: 3.7 (0.7,
11.2)
Mistelbach: 3.7 (0.7, 7.5)
Wiesmath: 6.3 (3.4, 9.4)
Bruck: 1.5(0.7,4.1)
Freudenstadt: 0.7 (0, 3.0)
Oct1994-Mar1995
AmstetteN: 3.7 (0.7, 7.5)
StValentiN:3.0(1.1,9.4)
Krems: 3.7 (0.7, 11.0)
VillingeN: 1.9(0,3.0)
HeindenreichsteiN: 3.7
(0.7, 15.0)
Ganserndorf: 3.7 (0.7,
22.5)
Mistelbach: 3.7 (0.7,
22.5)
Wiesmath: 2.23 (0.7,
10.1)
Bruck: 15(1.1,7.9)
Freudenstadt: 1.57 (0.4,
5.3)
Apr-Sep 1995
AmstetteN: 3.7 (0.7, 3.8)
StValentiN:2.6(1.1,6.8)
Krems: 3.7 (0.5, 3.8)
VillingeN: 0.7 (0 , 2.6)
HeindenreichsteiN: 0.7
(0.5, 0.9)
Ganserndorf: 3.7 (0.7,
7.5)
Mistelbach: 3.7 (0.7, 7.5)
Wiesmath: 7.5 (0.7, 14.9)
Bruck: 3.7 (0.4, 4.9)
Freudenstadt: 0.7 (0, 3.4)
Lower FVC and FEVi increases observed in
children exposed to high ambient O3 levels vs.
those exposed to lower levels in the summer. This
study found no effect of SO2 and PMio on FVC
increase during the summer of 1995 and winter
1994/1995, however, SO2 was negatively
associated with FVC during the summer of 1994.

Change in FVC (per ppb SO2)
Summer 1994: -0.044, p = 0.006
Winter  1994/95: 0.007, p = 0.243
Summer 1995: 0.045, p = 0.028
May 2008
                                       F-94
                      DRAFT—DO  NOT QUOTE OR CITE

-------
     STUDY
                                    METHODS
                                                                     POLLUTANT
                                                                                                           RESULTS
Kramer etal. (1999)

East and West
Germany

1991 to 1995
Repeated cross-sectional studies between
1991 and 1995 on 7-yr-old children in East
Germany and between 1991 and 1994 in West
Germany. Comparison of prevalence of airway
diseases and allergies in East and West
Germany during the first five yrs after
reunification. A total of 19,090 children
participated in the study. Logistic regression
used to assess the  effect of SO2 and TSP on
airway diseases and allergies. Analysis
performed on 14,144 children with information
on all covariates of interest.
East Germany 2-yr avg
concentration ranged
from 45 to 240 pg/m3

West Germany 2-yr avg
concentration ranged
from 18-33
All infectious airway diseases and irritation of the
airway was associated with either SO2 or TSP in
East Germany in 1991. The decrease of pollution
between 1991 and 1995 had a favorable effect on
the prevalence of these illnesses. SO2 was
significantly associated with more than 5 colds in
the last 12 mos, tonsillitis, dry cough in the last 12
mos, and frequent cough in 1991-1995.

Odds ratio and 95% Cl: (per 200 pg/m3 SO2) in
East Germany areas, 1991-1995 for children
living at least 2 yrs in the areas, adjusted for time
trend:

Infectious airway diseases
Pneumonia ever diagnosed: 1.17 (0.85, 1.62)
Bronchitis ever diagnosed: 0.85 (0.68, 1.05)
$5 colds in last 12 mos: 1.55 (1.18, 2.04)
Tonsillitis in the  last 12 mos: 1.89 (1.49, 2.39)
Dry cough in the last 12 mos:  1.46 (1.12, 1.91)
Frequent cough  ever: 2.51 (1.79. 3.53)

Allergic diseases and symptoms:
Irritated eyes in  the last 12 mos: 1.06 (0.66, 1.70)
Irritated nose in  the last 12 mos:  1.26 (0.96, 1.66)
Wheezing ever diagnosed: 0.68 (0.46, 1.01)
Bronchial asthma ever diagnosed: 2.73  (1.24,
6.04)
Hay fever ever diagnosed: 0.60 (0.24, 1.52)
Eczema ever diagnosed: 0.87 (0.65, 1.18)
Allergy ever diagnosed: 0.93 (0.67, 1.29)
Liebhart et al.
(2007)

Poland (Bialystok,
Bydgoszcz,
Gdansk, Krakow,
Lublin, Lodz,
Poznan, Rabka,
Warszawa,
Wroclaw, Zabrze)

1998-1999
The Polish Multicentre Study of Epidemiology of
Allergic Diseases (PMSEAD), which consisted
of a cohort of 16,238 individuals aged 3-80 yrs
old from 33 areas in 11 regions of Poland.
Asthma diagnosis was determined through
household questionnaires. Conducted
multivariate and univariate logistic regression
analyses to examine the prevalence of and risk
factors for asthma.
Range (pg/m3):

4.0-35.0
In multivariate logistic regression models, black
smoke was found to be a significant risk factor for
asthma for both children and adults. SO2 was
found to be a significant risk factor for asthma in
both children and adults, but only in a univariate
logistic regression.

Adjusted Odds Ratio (95% Cl)
Univariate logistic regression
ChildreN: 1.34(1.04, 1.72)
Adults: 1.19(1.02, 1.38)
Multivariate logistic regression
ChildreN: 1.20(0.91, 1.59)
Adults: 1.01 (0.85, 1.20)
Kohlhammer et al.
(2007)

Hettstedt, Germany
1992-1999
Three repeated cross-sectional studies of 5,360
children ages 5-14 examining health impacts
(lifetime pneumonia) of social and
environmental factors
                         No relationship between SO2 and pneumonia was
                         observed.
Penard-Morand
et al. (2006)

Six communities in
France: Bordeaux,
Clermont-Ferrand,
Creteil, Marseille,
Strasbourg and
Reims

Mar 1999-Oct 2000
Cross-sectional study of 4,901 children
(9-11 yrs) form 108 randomly selected schools
in 6 cities to assess the association between
long-term exposure to background air pollution
(NO2, SO2, PMio, O3) and atopy and respiratory
outcomes. Analysis restricted to children who
had lived at least the last 3 yrs in their house at
the time of the examination. Analysis used three
yr avgd air pollutant concentrations at the
children's schools. Parents completed
questionnaire on  respiratory  and allergic
disorders (asthma, allergic rhinitis (AR), and
atopic dermatitis) and children underwent
examination that  included a skin prick test to
assess allergic sensitization, evidence of visible
flexural dermatitis and measure of exercise-
induced bronchial reactivity (EIB).
Estimated 3-yr avg
concentrations at 108
schools

Low cone: 4.6 pg/m3
(Range: 1.3, 7.4),
High cone: 9.6 pg/m3
(range 7.7, 13.7)
Increased concentrations of SO2 were
significantly associated with an increased risk of
EIB, lifetime asthma and lifetime AR. Past yr
wheeze and asthma were also associated with
SO2. In a two-pollutant model with PMio,
significant associations were observed between
SO2 and EIB and past yr wheeze.

Odds ratio and 95% Cl (per 5 pg/m3 SO2)
EIB: 1.39(1.15, 1.66), p < 0.001
Flexural dermatitis: 0.86 (0.73,  1.02), p < 0.10
Past yr wheeze: 1.23(1.0, 1.51), p < 0.05
Past yr asthma: 1.28 (1.00, 1.65),  p < 0.01
Past yr rhinoconjunctivitis: 1.05 (0.89, 1.24)
Past yr atopic dermatitis: 1.01 (0.86, 1.18)
Lifetime asthma: 1.19(1.00, 1.41), p<0.10
Lifetime allergic rhinitis:  1.16(1.01, 1.32), p<
0.05
Lifetime atopic dermatitis: 0.93 (0.82, 1.05)
Two-pollutant model with PM10
EIB: 1.46(1.12, 1.90)
Past yr wheeze: 1.45 (1.09, 1.93)
May 2008
                                       F-95
                       DRAFT—DO NOT QUOTE OR CITE

-------
     STUDY
                                   METHODS
                                                                    POLLUTANT
                                                                                                         RESULTS
Pikhart et al. (2001)

Czech Republic,
Poland,

1993-1994
Part of the small-area variation in air pollution
and health (SAVIAH) study to assess long-term
effects of air pollution on respiratory outcomes.
Analysis on data from two centers of the
multicenter study: Prague, Czech Republic, and
Poznan, Poland. Both cities had wide variation
in air pollution levels. Parents/guardians of
6,959 children (7-10 yrs) completed a
questionnaire about the socioeconomic
situation of the family, type of housing, family
history of atopy, parental smoking, family
composition, and health of the child. SO2 was
measured at 80 sites in Poznan and 50 sites in
Prague during 2-wk campaigns. From these
data GIS was used to estimate pollutant
concentrations at a small area level. Logistic
regression used to assess effect of air pollution
on the prevalence of respiratory outcomes.
Mean SO2 (pg/m )

Prague: 83.9
Range: 65.8-96.6

PoznaN: 79.7
Range: 44.2-140.2
SO2 levels (mean of home and school) were
associated with the prevalence of
wheezing/whistling in the past 12 mos. There was
a marginal association between SO2 and lifetime
prevalence of wheezing and physician diagnosed
asthma. Fully adjusted model controlled for age,
gender, maternal education, number of siblings,
dampness at home, heating and cooking on gas,
maternal smoking, and family history of atopy and
center. Authors noted SO2 is strongly spatially
correlated with particles in the Czech Republic
and probably Poland, so SO2 may be proxy for
exposure to other pollutants. Not other pollutants
measured in study.

Odds ratio (per 50 pg/m3) SO2
Wheezing/whistling in past 12 mos: 1.32 (1.10,
1.57)
Wheezing/whistling ever: 1.13 (0.99, 1.30)
Asthma ever diagnosed by doctor: 1.39 (1.01,
1.92)
Dry cough at night: 1.06 (0.89, 1.27)
Ramadour et al.
(2000)

Seven towns in SE
France

Jan-Feb1993
Cross-sectional cohort study of 2,445 children
(age 13-14 yrs) who had lived for at least 3 yrs
in their current residence to compare the levels
of O3, SO2, and NO2 to the prevalence rates of
rhinitis, asthma, and asthmatic symptoms.
Some of the communities had the heaviest
photochemical exposure in France. Subjects
completed ISAAC survey of asthma and
respiratory symptoms. Analysis conducted with
logistic regression models that controlled for
family history of asthma, personal history of
early -life respiratory diseases, and SES. Also
performed simple univariate linear regressions.
Mean (SD) pg/m of SO2
during 2-mo period

Port de Bouc: 32.3 (24.5)
lstres:23.8(12.7)
Sausset: 29.0 (28.7)
Rognanc-Velaux: 39.5
(21.8)
Vitrolles: 57.4 (32.0)
Aries: 29.7 (15.5)
SaloN: 17.3(11.6)
Study found no relationship between mean levels
of SO2, NO3, orO3and rhinitis ever, 12-mo
rhinitis, rhinoconjunctivitis, and hay fever or
asthmatic symptoms. Simple regression analyses
of respiratory outcomes vs. mean SO2 levels in
the 7 towns indicated that nocturnal dry cough
was associated with mean SO2 levels (r = 0.891).
Potential confounding across towns.
Soyseth et al.
(1995)

Ardal and Laerdal,
Norway

winter seasons
1989-92
Cross-sectional study of 529 children
(aged 7-13 yrs) to determine whether exposure
to SO2 during infancy is related to the
prevalence of bronchial hyperresponsiveness
(BHR). A sulfur dioxide emitting aluminum
smelter is present in Ardal, but there is no air
polluting industry in Laerdal. Parents filled out
questionnaire regarding family history of
asthma, type of housing, respiratory symptoms
and parent's smoking habits. Spirometry was
performed on each child and bronchial
hyperactivity was determined by methacholine
challenge or reversibility test. Skin prick test
done to assess atopy. Also examined, the
effects of fluoride.
Median SO2

37.1 pg/m3at ages
0-12 mos
37.9 pg/m at
ages 13-36 mos
This study found that the risk of BHR was
associated with SO2 exposure at 0-12 mos

Odds ratio for BHR (per 10 pg/m3 SO2) for various
ages at exposure
0-12 mos: 1.62 (1.11,2.35)
13-36 mos: 1.40 (0.90, 2.21)
37-72 mos: 1.19(0.77, 1.82)
73-108 mos: 1.19(0.63,2.22)
Studnicka et al.
(1997)

Austria

(8 nonurban
communities)  1991-
1993
Longitudinal study of 843 children 7 yrs old from
8 nonurban Austrian communities. A logistic
regression was used to examine the
association between SO2 concentrations and
asthma and respiratory symptoms by
comparing low, regular, and high SO2
communities with very low SO2 communities.
Range:

Jan. 1991-Dec. 1993
(ppb): 6.0 (Krems), 12.0
(Mistel. and Cans)
SO2 was significantly associated with bronchial
asthma in the last 12 mos and positively
associated with parent-reported "ever asthma"
when comparing low SO2 concentration
communities with very low SO2 communities.

Adjusted Prevalence Odds Ratio
Wheeze last 12 mos
Low: 0.68. Regular: 0.88. High: 0.42

Cough apart from colds last 12 mos
Low: 0.75. Regular: 0.85. High: 0.72

Bronchitis last 12 mos
Low: 0.21. Regular: 0.45. High: 0.56

Bronchial asthma last 12  mos
Low: 2.35. Regular: 0.22. High: 0.33

Parent-reported "ever asthma"
Low: 1.70. Regular: 0.23. High: 0.67
May 2008
                                       F-96
                      DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
von Mutiuset al.
(1995)
l_6 ipz io E 8st

Oct 1991-Jul 1992










METHODS
The effects of high to moderate levels of air
pollution (SO2, NOX, and PM) on the incidence
of upper respiratory were investigated in 1 ,500
schoolchildren (9-11 yrs) in Leipzig, East
Germany. Logistic regression models controlled
for paternal education, passive smoke
exposure, number of siblings, temperature, and
humidity.








POLLUTANT
During winter mos, SO2
daily max concentrations
ranged from 40-1283
pg/m3.
During high pollution
period, mean
concentration of SO? was
188 pg/m3and during low
pollution mean was
57 pg/m3.








RESULTS
The daily mean values of SO2 and NOX were
significantly associated with increased risk of
developing upper respiratory illnesses during the
high concentration period. In the low
concentration period, only NOX daily mean values
were associated with increased risks. In a two-
pollutant model with PM, similar estimates to the
single-pollutant model were obtained, thus
collinearity of data may not account for the effects
of high mean concentrations of SO2.
Odds ratio and 95% Cl: (did not indicate per what
level of SO2 increase)
Daily mean SO2
High period: 1.72(1.19,2.49)
Low period: 1.40 (0.95, 2.07)
Daily max SO2
High period: 1.26(0.80, 1.96)
Low period: 0.99 (0.66, 1.47)
May 2008
F-97
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
METHODS
POLLUTANT
RESULTS
LATIN AMERICA
Sole et al. (2007)
Sao Paulo, Brazil
(Sao Paulo West
(SPW), Sao Paulo
South (SPS), Santo
Andre (SA),
Curitiba (CR), Porto
Alegre (PoA))














































Cohort of 16,209 adolescents (13-14 yrs old)
from the 21 centers involved in the International
Study of Asthma and Allergies in Childhood
(ISAAC). Each participant was given a
questionnaire to identify various allergy-related
symptoms that occurred in the last 12 mos. The
relationship between affirmative answer to a
question, socioeconomic status, and air
pollutants was analyzed by the Spearman
correlation coefficient. The location with the
lowest level of a specific air pollutant was
defined as the reference and the risk of an
affirmative answer to a question was presented
as an odds ratio for each location.











































NR















































In the analysis of the risk of allergy-related
symptoms due to SO2 levels in relation to the
center with the lowest annual mean SO2
concentrations SPW was significantly associated
with every symptom. Other significant
associations were observed in SA for current
wheezing; in CR for rhinitis and
rhinoconjunctivitis; and in PoA for current
wheezing, nighttime cough, rhinitis, and eczema.
Odds Ratio (95% CI)-Reference Center: Sao
Paulo South (SPS)
Current Wheezing
SPW: 1.21 (1.08, 1.38)
SA: 1.31 (1.16, 1.48)
CR: 1.02 (0.90, 1.15)
PoA: 1.68(0.85, 1.10)
Severe Asthma
SPW: 2.01 (1.56,2.60)
SA: 1.04(0.78, 1.40)
CR: 1.08(0.81, 1.42)
PoA: 1.01 (1.29,2.20)
Nighttime Cough
SPW: 1.14(1.03, 1.26)
SA: 0.94 (0.85, 1.04)
CR: 0.93 (0.84, 1 .02)
PoA: 1.25(0.91, 1.12)
Rhinitis Last Yr
SPW: 1.14(1.02, 1.27)
SA: 1.05(0.94, 1.18)
CR: 1.71 (1.54, 1.90)
PoA: 1.36(1.12, 1.40)
Rhinoconjunctivitis
SPW: 1.78(1.55,2.04)
SA: 1.15(0.99, 1.33)
CR: 1.50(1.31, 1.72)
PoA: 1.48(1.18, 1.57)
Severe Rhinitis
SPW: 1.50(1.32, 1.71)
SA: 1.08(0.94, 1.24)
CR: 1.52 (1.34, 1.73)
PoA: 0.97 (1.30, 1.69)
Eczema
SPW: 1.40(1.17, 1.68)
SA: 1.00(0.83, 1.21)
CR: 0.88 (0.72, 1.06)
PoA: 1.40(0.80, 1.18)
Flexural Eczema
SPW: 2.00 (1.58, 2.52)
SA: 0.95 (0.73, 1.24)
CR: 1.02 (0.79, 1.31)
PoA: 2.41 (1.09, 1.80)
Severe Eczema
SPW: 2. 58 (1.94, 3.44)
SA: 0.92 (0.65, 1.30)
CR: 0.71 (0.50, 1 .02)
PoA: NR (1.80, 3.22)
ASIA
Ho et al. (2007)
Tsiwsn

1995-1996


Survey of 69,367 children ages 12-15 by
questionnaire. The max likelihood estimation
was carried out with Fisher's scoring algorithm
and GEE.


NR





SO2 not significant in both genders. However,
SO2 showed a reversal effect on monthly asthma
attack rate.
(Authors state that this reversal effect could be
caused by the interaction of sulfur dioxide with the
lowest 5% monthly temperature avg)
May 2008
F-98
DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Hwang et al. (2005)
Tsiwsn

2001






Peters etal. (1996)
Hong Kong (Kwai
Tsing; Southern)
1989-1991





Wang etal. (1999)
Taiwan (Kaohsiung;
Pintonq)

1995-1996











METHODS
A cross-sectional study consisting of 32,672
Taiwanese school children aged 6-15 yrs old.
Using a modified Chinese version of the
International Study of Asthma and Allergies in
Childhood (ISAAC) questionnaire collected
information on each participant's health,
environmental exposures, and other variables.
A two-stage hierarchical model consisting of
logistic and linear regression analyses was
used to account for, in the first stage, variation
among subjects, and, in the second stage,
variation among municipalities.

Cohort of 3,521 children from two districts in
Hong Kong with good and poor air quality prior
to the 1990 legislation to reduce fuel sulfur
levels. Analyses consisted of multivariate
methods using logistic regression along with
generalized estimating equations (GEE) to
examine the effect of legislation implemented to
reduce fuel sulfur levels on respiratory health.


A cross-sectional study consisting of 165,173
high school students aged 11-16 yrs old
residing in the communities of Kaohsiung and
Pintong in Taiwan from Oct 1995 to Jun 1996.
Used a video and questionnaire developed by
the International Study of Asthma and Allergies
in Childhood (ISAAC). The association between
air pollution and asthma was examined using
logistic regression. In addition, the study
performed a multiple logistic regression to
examine the independent effects of risk factors
of asthma after adjusting for age, sex, parents'
education, and area of residence. The multiple
logistic regression included pollutant
concentrations to examine the combined effect.


POLLUTANT
2000 (ppb): 3.53 (2.00)









Annual avg (pg/m3)
Southern
1989: 11
1990' 8
1991: 7

Kwai Tsing
1989: 111
1990: 67
1991:23
Median:
1996 (ppm):0.013













RESULTS
Increased annual levels of NOX, CO, and O3were
associated with an increased risk of childhood
asthma levels. In both single- and co-pollutant
models SO2 was not found to be associated with
the risk of asthma.
Odds Ratio (95% Cl) (per 10 ppb SO2)
Single-pollutant model
0.874(0.729, 1.054)
Two-pollutant model
NOX + SO2: 0.724 (0.545, 0.963)
CO + SO2: 0.689 (0.542, 0.875)
SO2 + O3: 0.826 (0.674, 1.014)
SO2 emissions were reduced by 80% after
institution of the legislation. The study does not
provide effect estimates for individual pollutants.






In the univariate analysis, increasing
concentrations of TSP, SO2, NO2, CO, O3, and
airborne dust were all found to be significantly
associated with asthma. These univariate
estimates are associated with concentrations
above a cutoff (i.e., the median concentrations of
each pollutant). In the multivariate analysis
increasing concentrations of TSP, NO2, CO, O3,
and airborne dust were significantly associated
with asthma.
Odds Ratio (95% Cl) (per 0.013 ppm SO2)
Univariate analysis
> 0.01 3 ppm: 1.05(1.02, 1.09)
Adjusted Odds Ratio (95% Cl)
Multivariate analysis
0.98(0.95, 1.02)
MIDDLE EAST
Dubnov et al.
(2007)


(Hadera, Pardes-
Hanna)
1996 and 1999













Cohort of 1,492 schoolchildren (7-14 yrs old)
living near a major coal-fired power station.
Subjects underwent pulmonary function tests
(PFT) for forced vital capacity (FVC) and forced
expiratory volume during the first second
(FEV1) to examine the association between
pulmonary function and long-term exposure to
air pollution. Using stepwise multiple regression
(SMR) and ordinary leas squares regression
(OLS) examined the multiplicative effect of NOX
and SO2 on pulmonary function.









1996 and 1999 avg (SD)
(ppm):
12.9(11.3)

















Using an integrated concentration value (ICV),
which equals the product of NOX concentration
and SO2 concentration when both concentrations
individually exceed the half-hour reference level,
found significant associations between exposure
to air pollution and decrements in pulmonary
function.
All Children
NOX D SO2
AFVC (%)
3 = -0.004, p < 0.001
AFEV, (%)
3 = -0.004, p < 0.001
Children in zone of highest concentration of air
pollution
NOX D SO2
AFVC (%)
3 = -0.005, p < 0.001
AFEV,(%)
3 = -0.005, p < 0.001
AFRICA
Houssaini et al.
(2007)
Morocco


Cross-sectional study of 1,318 children with a
mean age of 12 yrs. Used a questionnaire and
medical diagnosis/reporting for asthma, and
evaluated using Student's t-test, Chi-square,
odds ratios, and Cochran-Armitage tests.
Annual Avg:
2000-2001: 60.2 pg/m3
2001-2002: 50.2 pg/m3
2002-2003: 49.6 pg/m3
2003-2004: 36.8 pg/m3
Significant prevalence for respiratory diseases,
asthma, and infectious disease, when combined
with TSP.


May 2008
F-99
DRAFT—DO NOT QUOTE OR CITE

-------
Table F-7.   Associations of long-term exposure to SO2 with lung cancer incidence and mortality.
STUDY
METHODS
POLLUTANTS
CONCLUSIONS
UNITED STATES
Abbey et al.
(1999)
Three California

Coast (Los
Angeles and
eastward), San
Diego
1977-1992




Krewski et al.
(2000)




Prospective cohort study of 6,338 nonsmoking
non- Hispanic white adult members of the
Adventist Health Study followed for all cause,
cardiopulmonary, nonmalignant respiratory, and
lung cancer mortality. Participants were aged 27-
95 yrs at enrollment in 1 977. 1 ,628 (989 females,
639 males) mortality events followed through
1992. All results were stratified by gender. Used
Cox proportional hazards analysis, adjusting for
age at enrollment, past smoking, environmental
tobacco smoke exposure, alcohol use, education,
occupation, and body mass index. Analyzed
mortality from all natural causes,
cardiopulmonary, nonmalignant respiratory, and
lung cancer.
Re-analysis and sensitivity analysis of Dockery
et al. (1993) Harvard Six Cities study.




Mean SO2 Levels:
24-h avg SO2: 5.6 ppb
Copollutants:
PMnn
r IVI1Q
5O4
O,
^3
NO2





Mean SO2 Levels: 24-h
avg SO2 ranged from 1 .6
(Topeka) to 24.0
(Steubenville) ppb
Copollutants: Fine
Particles, Sulfates
Lung cancer mortality showed large risk
estimates for most of the pollutants in either or
both sexes, but the number of lung cancer
deaths in this cohort was very small (12 for
female and 18 for male) Generally wide confi-
dence intervals (relative to other U.S. cohort
studies).
Adjusted Mortality Relative Risk
(95% Cl) (per 3.72 ppb SO2)
Lung Cancer
Males: 1.99(1.24,3.20)
Females: 3.01 (1.88, 4.84)


SO2 showed positive associations with lung
cancer deaths (1.03 (95% Cl: 0.91, 1.16)), but in
this dataset, SO2 was highly correlated with
PM25 (r = 0.85), sulfate (r = 0.85), and NO2 (r =
0.84)

EUROPE
Beelen et al.
(2008)
Th N th I
e e eran s

1987-1996.



Filleul et al.
(2005)
Seven French


1975-2001













Cohort study on diet and cancer with 120,852
subjects who were followed from 1987 to 1996.
BS, NO2, SO2, and PM2.5 and traffic-exposure
estimates were analyzed. Cox regression model
adjusted for age, sex, smoking, and area-level
socioeconomic status.


Cohort study of 14,284 adults who resided in 24
areas from seven French cities when enrolled in
the PAARC survey (air pollution and chronic
respiratory diseases) in 1974. Daily meas-
urements of SO2, TSP, black smoke, NO2, and
NO were made in 24 areas forthree yrs (1974-
1976). Cox proportional hazards models adjusted
for smoking, educational level, BMI, and
occupational exposure. Models were run before
and after exclusion of six area monitors
influenced by local traffic as determined by the
NO/NO2 ratio >3.







Mean SO2 Levels:
Mean: 4.8 ppb, with a
range of 1.5 to 11.8 ppb.

Copollutants:
PM-jc
rlv|2.5
BS
NO2
Mean SO2 Levels:
24-h avg SO2 ranged from
1 7 mg/m3 ("Area 3" in
Lille) to 85 mg/m3 ("Area
3" in Marseille) in the 24
areas in seven cities
during 1974-1976. Median
levels during 1990-1997
ranged from 8.5 mg/m3
(Bordeaux) to 23.4 mg/m3
(Rouen) in the five cities
where data were
available.

Copollutants:
TSP
Black Smoke
NO2
NO
Traffic intensity on the nearest road was not
associated with exposure SO2. Background SO2
levels were not associated with lung cancer
mortality.
Adjusted RR

(per 20 pg/m3 SO2)
1.00(0.79, 1.26)
The authors noted that inclusion of air monitoring
data from stations directly influenced by local
traffic could overestimate the mean population
exposure and bias the results. It should be noted
that the table describing air pollution levels in
Filleul et al.'s report indicates that the SO2 levels
in these French cities declined markedly from
1974-1976 and 1990-1997 period, by a factor of
2 to 3, depending on the city, whereas NO2
levels between the two periods were variable, in-
creased in some cities, and decreased in others.
These changes in air pollution levels over the
study period complicate interpretation of reported
risk estimates.

Relative Risk (95% Cl) for lung cancer mortality
(per 10 mg/m3 multi-year average)
All 24 areas: 0.99 (0.92, 1.07)
18areas:1.00(0.91, 1.11)
May 2008
F-100
DRAFT—DO NOT QUOTE OR CITE

-------
    STUDY
                                  METHODS
                                                                   POLLUTANTS
                                                                                                      CONCLUSIONS
Nafstad et al.
(2003)

Oslo, Norway

1972-1998
Retrospective study associating cardiovascular
risk factors to a national cancer register among
16,209 men ages 10-49 yrs. Survival analyses
and Cox proportional hazards regression were
used to estimate associations.
Estimated for each person
each yr from 1974 to 1998

Five-yr median  average
levels SO2 participants
home address,  1974-
1978: 9.4 pg/m3
(range 0.2 to 55.8)

Median levels within the
quartiles:
2.5 pg/m3
6.2 pg/m3
14.7pg/m3
31.3pg/m3

Copollutants:
NOX
Adjusted risk ratios (95% Cl) of developing lung
cancer:

Model!:
0-9.99 pg/m3: Ref
10-19.99 pg/m3: 1.05(0.81, 1.35)
20-29.99 ug/m3: 0.95 (0.72, 1.27)
30+pg/m3: 1.06(0.79, 1.43)

Model 2: PeMOpg/m3: 1.01 (0.94, 1.08)
Adjusted risk ratios (95% Cl) of developing non-
lung cancer

Model!:
0-9.99 pg/m3: Ref.
10-19.99 pg/m3: 1.07(0.96, 1.19)
20-29.99 ug/m3: 0.90 (0.80, 1.02)
30+pg/m3: 0.98 (0.86, 1.10)

Model 2:
Per 10 pg/m3: 0.99 (0.96, 1.02)
Nafstad et al.
(2004)

Oslo, Norway

1972-1998
Cohort study of 16,209 Norwegian men 40-49 yrs
of age living in Oslo, Norway, in 1972-1973. Data
from the Norwegian Death Register were linked
with estimates of avg yearly air pollution levels at
the participants' home addresses from 1974 to
1998. NOX, rather than NO2was used. Exposure
estimates for NOX and SO2 were constructed
using models based on the subject's address,
emission data for industry, heating, and traffic,
and measured concentrations. Addresses linked
to 50 of the busiest streets were given an addi-
tional exposure based on estimates of annual avg
daily traffic. Cox  proportional-hazards regression
was used to estimate associations between
exposure and total and cause-specific mortality,
adjusting for age strata, education, occupation,
smoking, physical activity level, and risk groups
for cardiovascular diseases
Mean SO2 Levels:
The yearly avg of 24-h
avg SO2 were reduced
with a factor of 7 during
the study period from 5.6
ppb in 1974 to 0.8 ppb in
1995.

Copollutants:
NOX
SO2 did not show any associations with lung
cancer, e.g., 1.00 (0.93, 1.08) per 10 pg/m3 in-
crease mortality in SO2. No association was also
observed when including SO2 in the model as a
categorical variable. Note the very low levels of
SO2.
Nyberg et al.
(2000)

Stockholm
County, Sweden

Jan 1, 1985-Dec
31, 1990
Case-control study of men 40-70 yrs, with 1,042
cases of lung cancer and 1,274 controls, to
evaluate the suitability of an indicator of air
pollution from heating.
Annual levels computed
for each yr between 1950
and 1990, but not
provided herein

NOX/NO2
Little effect of SOX in any time window, but
highest correlations in early yrs.

SOX RR (Cl 95%) from heating (per 10 pg/m3) for
30-yr avg
< 41.30 pg/m3: 1
> 41.30 to < 52.75: 1.06 (0.83, 1.35)
> 52.75 to < 67.14: 0.98 (0.77, 1.24)
> 67.14 to < 78.20:0.90(0.68, 1.19)
> 78.20: 1.00(0.73, 1.37)

SOX RR (Cl 95%) from heating (per 10 pg/m3) for
10-yr avg
< 66.20 pg/m3: 1
> 66.20 to < 87.60: 1.16(0.91, 1.47)
> 87.60 to < 110.30: 1.00(0.79, 1.27)
> 110.30to < 129.10:0.92 (0.70, 1.21)
2129.10: 1.21 (0.89, 1.66)
May 2008
                                       F-101
                      DRAFT—DO NOT QUOTE OR CITE

-------
Table F-8.   Associations of long-term exposure to SO2 with prenatal and neonatal outcomes.
STUDY
METHODS
POLLUTANTS
FINDINGS
UNITED STATES
Bell et al. (2007)
Connecticut and
Massachusetts
Period of Study:
1999-2002







Gilboa et al.
(2005)
Seven Texas
Counties
Period of Study:
1997-2000





































Outcome(s): LBW
Study design: Case-control
N: 358,504 live singleton births
Statistical Analysis: Linear models and logistic
regression
Covariates: Gestational length, prenatal care,
type of delivery, child's sex, birth order, weather,
yr, and mother's race, education, marital status,
age, and tobacco use.


Outcome(s): Selected birth defects
Study design: Case-control
N: 4,570 cases and 3,667 controls
Statistical Analysis: Logistic regression
Covariates: Maternal education, maternal
race/ethnicity, season of conception, plurality,
maternal age, maternal illness
Statistical package: SAS vs. 8.2




































Gestational exposure (ppb)
Mean: 4.7
SD: 1.2
IQR: 1.6
Copollutants:
NO2
CO
PM*n
riving
PM
rlVl2.5



Levels NR
Copollutants
PM10
03
NO2
CO






































No relationship between gestational exposure to
SO2 and birth weight. First trimester exposure to
SO2 was associated with low birth weight. No
statistical difference n the effect estimates of
SO2 for infants of black and white mothers.
Increment: 1 .6 ppb (IQR)
Change in birth weight:
Entire pregnancy: -0.9 g (-4.4, 2.6)
Black mother: 1.2 (-6.5, 8.8)
White mother: -1.4 (-5. 1,2.3)
1st trimester: -3.7 to -3.3 grams
LBW: OR 1 .003 (0.961 , 1 .046)
When the fourth quartile of exposure was
compared with the first, SO2 was associated
with increased risk of isolated ventricular septal
defects. Inverse associations were noted for
SO2 and risk of isolated atrial septal defects and
multiple endocardial cushion defects.
Aortic artery and valve defects
< 1.3 ppb: 1.00
1.3to< 1.9: NA
1. 9 to < 2.7: 1.06(0.34,3.29)
> 2.7: 0.83 (0.26, 2.68)
Atrial septal defects
< 1.3 ppb: 1.00
1.3to< 1.9: 1.22(0.79, 1.88)
1. 9 to < 2.7: 0.76 (0.47, 1.23)
> 2.7: 0.42 (0.22, 0.78)
Pulmonary artery and valve defects
< 1.3 ppb: 1.00
.3to< 1.9:0.63(0.23, 1.74)
1. 9 to < 2.7: 0.93 (0.36, 2.38)
22.7: 1.07(0.43,2.69)
Ventricular septal defects
< 1.3 ppb: 1.00
1.3to< 1.9: 1.02(0.68, 1.53)
1. 9 to < 2.7: 1.13(0.76, 1.68)
22.7:2.16(1.51,3.09)
Conotruncal defects
< 1.3 ppb: 1.00
1.3to< 1.9:0.71 (0.46, 1.09)
1. 9 to < 2.7: 0.71 (0.46, 1.09)
> 2.7: 0.58 (0.37, 0.91)
Endocardial cushion and mitral valve defects
< 1.3 ppb: 1.00
1.3to< 1.9:0.89(0.50, 1.61)
1. 9 to < 2.7: 0.89 (0.49, 1.62)
22.7: 1.18(0.68,2.06)
Cleft lip with or without cleft palate
< 1.3 ppb: 1.00
1.3to< 1.9:0.79(0.52, 1.20)
1. 9 to < 2.7: 0.95 (0.64, 1.43)
22.7:0.75(0.49, 1.15)
Cleft palate
< 1.3 ppb: 1.00
1.3to< 1.9:0.89(0.40, 1.97)
1. 9 to < 2.7: 1.49(0.72,3.06)
22.7: 1.22(0.56,2.66)
May 2008
F-102
DRAFT—DO NOT QUOTE OR CITE

-------
    STUDY
                                   METHODS
                                                                     POLLUTANTS
                                                                                                           FINDINGS
Lipfert et al.
(2000b)
United States
1990
Mortality
Outcome(s): SIDS Study design: Cohort
Statistical Analysis: Three logistic regression
analyses to examine the relation between annual
avg air pollutant values and various infant
mortality endpoints (i.e., all causes, SIDS,
respiratory, and other causes
Statistical package: NR
Age groups analyzed: 0-1
Covariates: Altitude, degree days (°F), median
income (U.S. $), population density
Lag(s): N/A
                           In a model that included states that lacked data
                           on maternal education and smoking along with
                           various personal and ecological variables, SO2
                           was not found to be a significant predictor of
                           SIDS  mortality in infants with birth weights
                           >2,500g.

                           3 = -0.0118(0.0094)

                           Mean Risk (95% Cl)

                           0.95(0.87,  1.03)
Maisonet et al.
(2001)

6 Northeastern
cities of U.S.

Period of Study:
1994-1996
Outcome(s): Term LBW
Study design: Case-control
N: 89,557 live singleton births
Statistical Analysis: Logistic regression models
linear regression models
Covariates: Maternal age, race, season of the yr,
smoking and alcohol use  during pregnancy,
firstborn,  gender, marital status, and previous
terminations, prenatal care (ordinal variable),
weight gain, and gestational age
Stratified  by race/ethnicity
Statistical package: STATA
Exposure distribution
(< 25th, 25th to < 50th, 50th
to < 75th, 75th to < 95th, >
95th)

First trimester: < 7.09, 7.090
to 8.906, 8.907to 11.969,
11.970to 18.447,218.448

Second trimester: < 6.596,
6.596 to 8.896, 8.897 to
11.959, 11.960to 18.275,2
18.276

Third trimester: < 5.810,
5.810 to 8.453,
8.454 to 11.777, 11.778 to
18.134,218.135

Copollutants:
CO
PM10
This study provides evidence of an increased
risk for term LBW in relation to increased
ambient air levels of SO2 at concentrations well
below the established standards. Higher risk
estimates among whites when stratified by
race/ethnicity

First trimester:
< 25th: Referent
25th-50th: 1.04(0.88, 1.23)
50th-75th: 1.04(0.94, 1.15)
75th-95th: 0.98 (0.81, 1.17)
> 95th: 0.88 (0.73, 1.07)
Increment (10 ppm): 0.98 (0.93, 1.03)

Second trimester:
25th-50th: 1.18(1.12, 1.25)
50th-75th: 1.12(1.07, 1.17)
75th-95th: 1.13(1.05, 1.22)
> 95th: 0.87 (0.80, 0.95)
Increment (10 ppm): 1.01 (0.93, 1.10)

Third trimester:
25th-50th: 1.04(0.92, 1.18)
50th-75th: 1.02(0.87, 1.18)
75th-95th: 1.04(0.84, 1.28)
>95th: 1.06(0.76, 1.47)
Increment (10 ppm): 1.01 (0.86, 1.20)
Sagiv et al.
(2005)

4 Pennsylvania
counties

Period of Study:
1997-2001
Outcome(s): Pre-term birth
Study design: Time-series
N: 187,997 births
Study design: Poisson-regression models
Covariates: Long-term trends, copollutants,
temperature, dew point temperature, and day of
wk.
Lag: Daily lags ranging from 1-7 days
Mean SO2 Levels:
6-wk Mean: 7.9 ± 3.5 ppb
(Range: 0.8, 17),MediaN:
8.1
Daily Mean: 7.9 ±6.2
(Range: 0, 54.1), MediaN:
6.4
Copollutants:
PM10; r= 0.46
CO
NO2
This study found an increased risk for preterm
delivery during the last 6 wks of pregnancy with
exposure to SO2.

Increment: 15 ppb
Mean: 6-wk SO2:
RR = 1.15(1.00, 1.32)
< 4.9 ppb: Referent
4.9 to 8.1 ppb: 1.02 (0.97, 1.06)
8.1 to 10.6 ppb: 1.04(0.98, 1.10)
10.6 to 17.0 ppb: 1.06(0.99, 1.14)

Mean: Daily SO2:
RR = 1.07(0.99, 1.15) lag 3
                                                              CANADA
Dales et al.
(2004)

12 Canadian
cities

Period of Study:
1984-1999
Outcome(s): SIDS
Study design: Time-series
N: 1556 SIDS deaths
Statistical Analysis: Random effects regression
model
Covariates: Temperature, humidity, barometric
pressure, season
Lag: 0-5 days
Mean SO2 Levels:
24-h avg: 5.51 ppb
IQR: 4.92

Copollutants:
CO
NO2
03
PM10
PM2.5
PMio.2.5
SIDS was associated with air pollution, with the
effects of SO2 seeming to be independent of
sociodemographic factors, temporal trends, and
weather.
Increment: 4.92 ppb (IQR)
Increase in SIDS incidence: 8.49%;
p = 0.0079 lag  1
May 2008
                                        F-103
                       DRAFT—DO NOT QUOTE OR CITE

-------
    STUDY
                                   METHODS
                                                                      POLLUTANTS
                                                                                                            FINDINGS
Dales et al.
(2006)

11 Canadian
cities

Period of Study:
1986-2000
Outcome(s): Hospitalization for respiratory
disease in the neonatal period

Study design: Time-series

N: 9,542

Statistical Analysis: Random effects regression
model; Poisson using fixed- or random-effects
model

Covariates: Fay of wk, temperature, humidity,
pressure

Lag: 0-5 days

Statistical package:
S-PLUS vs. 6.2
Mean SO2 Levels:

24-h avg: 4.3 ppb
IQR: 3.8

Copollutants:
NO2;r = 0.20, 0.67
CO; r = 0.19, 0.66
O3;r =-0.41, 0.13
PM10;r=-0.09, 0.61
SO4
This study detected a significant association for
respiratory disease among neonates and
gaseous air pollutants.
Increment: 3.8 ppb (IQR)
Increase in neonatal respiratory
hospital admissions:
SO2 alone: 2.06% (1.04, 3.08)
Multipollutant model: 1.66% (0.63, 2.69)
Multipollutant model restricted to days with PMio
measures: 1.41%
(0.35, 2.47)
Dugandzic et al.
(2006)

Nova Scotia,
Canada

Period of Study:
1988-2000
Outcome(s): Term LBW
Study design: Retrospective cohort study
N: 74,284 term, singleton births
Statistical Analysis: Logistic regression models
Covariates: Maternal age, parity, prior fetal death,
prior neonatal death, and prior low birth weight
infant, smoking during pregnancy, neighborhood
family income, infant gender, gestational age,
weight change, and yr of birth.
Statistical package:
SAS vs. 8.0
Mean: SO210 ppb
MediaN: 10
25th%: 7
75th%: 14
Max: 38
Copollutants:
03
PM10
In the analyses unadjusted for birth yr, first
trimester exposures in the highest quartile for
SO2 associated with increased risk of LBW.
After adjusting for birth yr, RR attenuated and
not statistically significant. There was a linear
concentration-response effect with increasing
levels of SO2 during the first trimester.

First Trimester
25th-50th: 0.96 (0.73, 1.28)
51st-75th: 1.18(0.88, 1.58)
>75th: 1.36(1.04, 1.78)
Increment (7 ppb): 1.20 (1.05, 1.38)

Second Trimester
25th-50th: 1.12(0.86, 1.46)
51st-75th: 1.13(0.85, 1.50)
>75th: 1.04(0.79, 1.37)
Increment (7 ppb): 0.99 (0.87, 1.13)

Third Trimester
25th-50th: 1.04(0.80, 1.34)
51st-75th: 0.85 (0.63, 1.15)
>75th: 0.88 (0.67, 1.15)
Increment (7 ppb): 0.93 (0.81, 1.06)
Liu et al. (2003)

Vancouver,
Canada

Period of Study:
1986-1998
Outcomes: Preterm birth, LBW, IUGR
Study design: Case-control
N: 229,085 singleton live births
Statistical Analysis: Multiple logistic regressions
Covariates: Maternal age, parity,  infant sex,
gestational age or birth weight and season of
birth
24-h avg: 4.9 ppb,
5th: 1.5
25th: 2.8
50th: 4.3
75th: 6.3
95th: 10.5
100th: 30.5

1-h max: 13.4 ppb,
5th: 4.3
25th: 7.8
50th: 11.7
75th: 16.8
95th: 28.3
100th: 128.5

Copollutants:
NO2(r = 0.61)
CO (r = 0.64)
O3 (r = -0.35)
LBW and IUGR were associated with maternal
exposure to SO2 during the first mo of
pregnancy and preterm birth was associated
with SO2 during the last mo. These results were
robust to adjustment for Copollutants.

Increment: 5 ppb

Low birth weight
First mo: OR 1.11  (1.01, 1.22)
Last mo: OR 0.98  (0.89, 1.08)

Preterm birth
First mo: OR 0.95  (0.88, 1.03)
Last mo: OR 1.09  (1.01, 1.19)

IUGR
First mo: OR 1.07  (1.01, 1.13)
Last mo: OR 1.00  (0.94, 1.06)

First trimester: OR 1.07 (1.00, 1.14)
Second trimester:  0.98 (0.91, 1.04)
Third trimester: 1.03 (0.96,  1.10)
May 2008
                                        F-104
                       DRAFT—DO NOT QUOTE OR CITE

-------
    STUDY
                                  METHODS
                                                                     POLLUTANTS
                                                                                                           FINDINGS
Liu et al. (2006)
Calgary,
Edmonton and
Montreal, Canada

Period of Study:
1986-2000
Outcome(s): IUGR
Study design: Case-control
N: 386,202 singleton live births
Statistical Analysis: Multiple logistic regression
Covariates: Maternal age,  parity, infant sex,
season of birth, city of residence
24-h avg: 3.9 ppb, 25% 2.0
ppb

50% 3.0 ppb
75% 5.0 ppb
95% 10.0 ppb

1-h max: 10.8 ppb, 25% 5.0
ppb
50% 8.6 ppb
75% 14.0 ppb
95% 28.0 ppb

Copolllutants:
NO2 (r = 0.34)
CO (r = 0.21)
03 (r = -0.30)
PM25(r = 0.44)
IUGR did not increase with maternal exposure
to SO2. Risk decreased during first 3 mos.

Increment: 3.0 ppb

ORs estimated from graph:
1st mo: OR -0.966  (0.94, 0.99)
2nd mo: OR -0.97 (0.95, 0.995)
3rd mo: OR -0.97 (0.95, 0.995)
1st trimester: OR -0 .96 (0.93, 0.99)
Bobak et al.
(2000)
Czech Republic

Period of Study:
1990-1991
Outcomes: LBW, preterm birth
Study design: Case-control
N: 108,173 live singleton births
Statistical Analysis: Logistic regression
Covariates: Temperature, humidity, day of wk,
season, residential area, maternal age, gender
Statistical package: STATA
Mean trimester exposures

25th: 17.5pg/m3
50th: 32.0 pg/m3
75th: 55.5 pg/m3

Copollutants:
TSP; r = 0.68 0.73
NOX; r = 0.53, 0.63
LBW and preterm birth were associated with
maternal exposure to SO2, though the
association between SO2 and LBW was
explained to a large extent by low gestational
age.
Increment: 50 pg/m3
LBW (adjusted for sex, parity, maternal age
group, education, marital status, and nationality,
and mo of birth)
1st trimester: 1.20 (1.11, 1.30)
2nd trimester: 1.14(1.06, 1.22)
3rd trimester: 1.14(1.06, 1.23)
LBW (also adjusted for gestational age)
1st trimester: 1.01 (0.88, 1.17)
2nd trimester: 0.95 (0.82, 1.10)
3rd trimester: 0.97 (0.85, 1.10)
Preterm birth (AOR)
1st trimester: 1.27(1.16, 1.39)
2nd trimester: 1.25(1.14, 1.38)
3rd trimester: 1.24 (1.13, 1.36)
Reduction in mean birth weight:
1 st trimester: 1.4 g (5.9, 16.9)
                                                          LATIN AMERICA
Gouveia et al.
(2004)
Sao Paulo, Brazil
Period of Study:
1997
Outcome(s): LBW
Study design: Case-control
N: 179,460 live singleton births
Statistical Analysis: Logistic regression with GAM
Covariates: Gender, gestational age, maternal
age, maternal education, antenatal care, parity,
delivery method
Statistical package: S-Plus 2000
Annual Mean: SO2 (pg/m3)
Mean: 19.6
SD: 10.3
Range: 3.4, 56.9
Jan-Mar: 22.3 (7.7)
Apr-June: 28.1 (10.1)
Jul-Aug: 17.9(8.7)
Oct-Dec: 10.3 (3.9)

Copollutants:
PM10
CO
NO2
03
 First and second trimester exposures to SO2
 had a significant association with birth weight,
 though in different directions. When air
 pollutants were divided into quartiles and the
 lowest quartile was used as the referent
 exposure category, SO2 during the second
 trimester was marginally associated with low
 birth weight.

 Increment: 10 pg/m3

 Reduction in birth weight
 First trimester: -24.2 g (-55.5, 7.1)
 Second trimester: 33.7 g (1.6, 65.8)
 Third trimester: 9.7 g (-25.6, 44.9)

 First trimester:
 2nd: 0.902 (0.843, 0.966)
 3rd: 0.911 (0.819,  1.013)
 4th: 0.906 (0.793, 1.036)

 Second trimester:
 2nd: 0.986 (0.922, 1.053)
 3rd: 1.005(0.904,  1.117)
 4th: 1.017(0.883, 1.173)

 Third trimester:
 2nd: 1.203(0.861, 1.68)
 3rd: 1.225(0.872,  1.722)
 4th: 1.145(0.749, 1.752)
May 2008
                                        F-105
                       DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
METHODS
POLLUTANTS
FINDINGS
EUROPE
Mohorovic (2004)
Labin, Istra,

Period of Study:
1987-1989















Pereira et al.
(1998)
Sao Paulo, Brazil
Period of Study:
1991-1992



Outcomes: LBW and preterm delivery
Study design: Cross-sectional
N: 704 births
Statistical Analysis: Multiple correlation analyses,
factor analyses, chi-square
Statistical package: DBASE IV, SPSS













Outcome(s): Intrauterine mortality
Study design: Time-series
Statistical Analysis: Poisson regression models
Covariates: Mo, day of wk, min daily temperature,
relative humidity
Lag: 2 to 14 days

Monthly ground levels of SO2:
Range: 34.1, 252.9 pg/m3

















24-h avg SO2:
18.90(8.53) mg/m3
Range: 3.80, 59.70
Copollutants:
PM10; r= 0.45
NO2; r = 0.41
O3;r = 0.17
CO; r=0.24
The results show an association between SO2
exposure at the end of the first and second mo
of pregnancy and a negative correlation
between length of gestations and lower birth
weight of newborns.
Correlation coefficients:
1st mo: Gestation length: -0.09, p = 0.008
Birthweight: -0.08, p = 0.016
2nd mo: Gestation length: -0.08, p = 0.016
Birthweight: -0.07, p = 0.026
3rd mo: Gestation length: -0.04, p = 0.147
Birthweight: -0.04, p = 0.135
6th mo: Gestation length: -0.02, p = 0.266
Birthweight: -0.04, p = 0.151
Whole pregnancy: Gestation length: -0.09, p =
0.007
Birthweight: -0.04, p = 0.153
Weekly avg during whole pregnancy:
Gestation length: -0.05, p = 0.086
Birthweight: -0.06, p = 0.069
SO2 exhibited a marginal association with
intrauterine mortality, but only when Poisson
regression was employed. A concentration-
response relationship was found.
Estimated regression coefficients and standard
errors:
SO2 alone: 0.0038 (0.0020)
SO2 + NO2 + CO + PM10 + O3: 0.0029 (0.0031)
ASIA
Ha et al. (2001)
Seoul, Korea
Period of Study:
1996-1997





Lee et al. (2003)
Seoul, Korea
Period of Study:
1996-1998





Outcome(s): LBW
Study design: Case-control
N: 276,763
Statistical Analysis: Logistic regression, GAM
Covariates: Gestational age, maternal age,
parental education level, infant's birth order,
dender


Outcome(s): Term LBW
Study design:
N: 388,105 full-term singleton births
Statistical Analysis: GAM
Covariates: Infant sex, birth order, maternal age,
parental education level, time trend, and
gestational age.


24-h avg:
1st trimester: 25th: 10.0 ppb
50th: 13.2 ppbJSth: 16.2
ppb
3rd trimester:25th: 8.4 ppb
50th: 12.2 ppbJSth: 16.3
ppb
Copollutants:
CO; r=0.83. NO2;r = 0.70
TSP; r = 0.67. O3;r = -0.29
Avg concentration (ppb)
Mean: 12.1
SD: 7.4
Range: 3, 46
25th: 6.8
50th' 9 8
75th' 156

Copollutants:
PM10; r= 0.78, 0.85
CO; r = 0.79, 0.86
NO2' r = 0 75 0 76

Ambient SO2 concentrations during the first
trimester of pregnancy were associated with
LBW
Increment: 1st trimester: 6.2 ppb;
3rd trimester: 7.9 ppb
1 st trimester: RR 1.06 (1.02, 1.10)
3rd trimester: RR 0.93 (0.88, 0.98)
Reduction in birth weight: 8.06 g (5.59, 10.53)

Second trimester exposures to SO2 as well as
during the entire pregnancy were associated
with LBW. Reduction in birth weight was 14.6 g
for IQR increase in SO2 in the second trimester.
When the exposure for each mo of pregnancy
was evaluated separately, SO2 exposure during
3 to 5 mos of pregnancy associated with LBW.
Increment: 8.8 ppb (IQR)
First trimester: 1 .02 (0.99, 1 .06)
Second trimester: 1.06(1.02, 1.11)
Third trimester: 0.96 (0.91 , 1 .00)
All trimesters: 1.14(1.04, 1.24)
May 2008
F-106
DRAFT—DO NOT QUOTE OR CITE

-------
    STUDY
                                   METHODS
                                                                      POLLUTANTS
                                                                                                            FINDINGS
Leem et al.
(2006)

Incheon, Korea

Period of Study:
2001-2002
Outcome(s): Preterm delivery
Study design:
N: 52,113 singleton births
Statistical Analysis: Log-binomial regression
Covariates: Maternal age, parity, sex, season,
maternal education, paternal education
Mean:SO2 Concentrations
by trimester:

1st trimester: MiN: 7.86
pg/m3
25th: 17.61. 50th: 22.74
75th: 45.85. Max: 103.96

3rd trimester: MiN:
6.55 pg/m3
25th: 17.03. 50th: 25.62
75th: 46.53. Max: 103.15

Copollutants:
NO2; r = 0.54. CO; r = 0.31
PM10; r=0.13
This study found the highest SO2 concentrations
during the first trimester to be significantly
associated with  elevated risks of preterm
delivery.

1st trimester: 7.86 to 17.61 pg/m3: referent
17.62 to 22.74: 1.13(0.99, 1.28)
22.75 to 45.85: 1.13(0.98, 1.30)
45.86 to 103.96: 1.21 (1.04, 1.42)

3rd trimester: 6.55 to 17.03 pg/m3: referent
17.04 to 25.62: 0.87 (0.76, 1.01)
25.63 to 46.53: 0.97 (0.83, 1.13)
46.54 to 103.15: 1.11 (0.94, 1.31)
Lin et al. (2004)

Kaohsiung and
Taipei, Taiwan
(1995-1997)

Kaohsiung and
Taipei, Taiwan
(1995-1997)
Outcome(s): LBW
Study design: Case-control
N: 92,288 live births
Statistical Analysis: Multiple logisistic regression
Covariates: Gestational period, gender, birth
order, maternal age, maternal education, season
of birth
24-h avg: Kaohsiung
Range: 10.07, 25.36 ppb
Taipei: Range: 5.65, 9.33
ppb

Copollutants:
CO
NO2
03
PM10
Few women living in Taipei were exposed to
high levels of SO2. In Kaohsiung, almost all
women were exposed to high levels of SO2.
Women living in Kaohsiung had significantly
higher risk of term LBW compared with women
living in Taipei.
OR for Kaoshiung births (compared to Taipei
births) All births: OR: 1.13(1.03, 1.24)
Female births only: OR: 1.14(1.01, 1.28)
Lin et al. (2004a)

Kaohsiung and
Taipei, Taiwan

Period of Study:
1995-1997
Outcome(s): Term LBW
Study design: Cohort
N: 92,288 live births
Statistical Analysis: Multiple logisistic regression
Covariates: Gestational period, gender, birth
order, maternal age, maternal education, season
of birth
24-h avg: Kaohsiung
Range: 10.07, 25.36 ppb

Taipei:
Range: 5.65, 9.33 ppb

Copollutants:
CO
NO2
03
PH10
This study found a 26% higher risk of term LBW
delivery for mothers exposed to mean SO2
concentrations exceeding 11.4 ppb during the
entire pregnancy, as compared with mothers
exposed to mean concentrations less than 7.1
ppb. Trimester specific analysis showed a
significant association only for the third
trimester. Lowest quartile of exposure = referent

Entire pregnancy: 25th-75th: 1.16 (1.02, 1.33)
>75th: 1.26(1.04, 1.53)
1st trimester: 25th-75th: 1.02 (0.90, 1.16)
>75th: 1.11 (0.94, 1.33)
2nd trimester: 25th-75th: 1.09 (0.96, 1.24)
>75th: 1.17(0.99, 1.37)
3rd trimester:25th-75th: 1.13 (0.99, 1.28)
>75th: 1.20(1.01, 1.41)
Wang et al.
(1997)

Four residential
areas:
Dongcheng,
Xicheng, Cong-
wen, Xuanwu
Beijing, China

Period of Study:
1988-1991
Outcome(s): Term LBW
Study design: Cohort study
N: 74,671 first parity live births
Statistical Analysis: Multiple linear regression and
logistic regression with GAM
Covariates: Gestational age, residence, yr of
birth, maternal age, and infant gender.
Mean pollution
concentrations provided in
graph
TSP; r = 0.92
Exposure-response relationship between SO2
during the third trimester of pregnancy and low
birth weight.

3rd trimester: 9 to 18 pg/m3 (reference)
18 to 55: 1.09(0.94, 1.26)
55 to 146: 1.12 (0.97, 1.29)
146 to 239:  1.16(1.01, 1.34)
239 to 308:  1.39(1.22, 1.60)

SO2 as continuous variable: Odds ratio per 100
pg/m3: 1.11  (1.06, 1.16)
Xuetal. (1995)
Four residential
areas:
Dongchen,
Xichen,
Congwen,
Xuanwu
Beijing, China
Period of Study:
1988
Outcome(s): Preterm delivery
Study design: Prospective cohort study
N: 25,370 singleton first live births
Statistical Analysis: Multiple linear and logistic
regression
Covariates: Temperature, humidity, day of wk,
season, residential area, maternal age, and
gender of child.
2 monitors for SO2:
Dongcheng and Xicheng
Dongcheng Annual Mean:
108pg/m3SD: 141 pg/m3)
Xicheng annual Mean:
93pg/m3(SD: 122 pg/m3)
Copollutants:
TSP
Exposure response relationship between quar-
tiles of SO2 and crude incidence rates of pre-
term birth. Dose dependent relationship
between SO2 and gestational age. The
estimated reduced length of gestation was
0.075 wks or 12.6 h per 100/m3 increase in SO2.
When TSP and SO2 included in a multipollutant
model, the effect of SO2 was reduced by 32%.
Effect on gestational age (wk) per 100 pg/m3
regression coef and SE for lagged moving avg
of SO2.
lag 0: -0.016 (0.021). lag 1: -0.022 (0.021)
lag 6: -0.067 (0.024), p < 0.01
lag 7: -0.075 (0.024) , p < 0.01
lag 8: -0.075 (0.025), p < 0.01

OR for each quartile of SO2
1st: 1.00.2nd: 1.70(1.15,2.52)
3rd: 1.74 (1.03, 2.92). 4th: 1.58 (0.87, 2.86)

Adjusted OR for preterm delivery:
1.21  (1.01, 1.46) per In pg/m3 increase in SO2
May 2008
                                        F-107
                       DRAFT—DO NOT QUOTE OR CITE

-------
    STUDY
                                  METHODS
                                                                    POLLUTANTS
                                                                                                         FINDINGS
Yang et al.
(2003a)

Kaohsiung,
Taiwan

Period of Study:
1995-1997
Outcome(s): Term LBW

Study design: Case-control

N: 13,396 first parity singleton live births

Statistical Analysis: Multiple linear regression

Covariates: Maternal age, season, marital status,
maternal education, gender

Statistical package: SAS
                       Mean: trimester exposure
                       (Mg/m3)

                       1st trimester
                       33rd: 26.02. 67th: 36.07

                       2nd trimester
                       33rd: 25.76. 67th: 35.63

                       3rd trimester
                       33rd: 25.39. 67th: 36.96

                       Copollutants:
                       r= 0.45, 0.46
   A significant exposure-response relationship
   between maternal exposures to SO2 and birth
   weight was found first trimester of pregnancy.

   Reduction in birth weight:
   1st trimester: 33rd-67th: 3.68 g (-12.45, 19.21)
   >67th: 18.11 g (1.88, 34.34)
   Continuous: 0.52 g (0.09, 2.63)

   2nd trimester: 33rd-67th: 1.78g (-17.91, 14.35)
   >67th: 13.53 g (-2.62, 29.68)
   Continuous: 0.19g (-0.78, 1.8)

   3rd trimester: 33rd-67th: 0.43 g (-16.56, 15.70)
   >67th: 1.97g (-18.24, 14.30)
   Continuous: 0.03 g (-1.21, 1.37)
Table F-9.    Associations of long-term exposure to SO2 with mortality.
    STUDY
                        CONC.
                                                         METHOD
                                                                                                    CONCLUSIONS
                                                        UNITED STATES
Abbey et al.
(1999)

Three California
air basins: San
Francisco, South
Coast (Los
Angeles and
eastward), San
Diego

1977-1992
24-h avg SO2: 5.6 ppb
Prospective cohort study of 6,338 nonsmoking non-
Hispanic white adult members of theAdventist
Health Study followed for all cause,
cardiopulmonary, nonmalignant respiratory, and
lung cancer mortality. Participants were aged 27-95
yrs at enrollment in 1977.  1,628 (989 females,
639 males) mortality events followed through 1992.
All results were stratified by gender. Used Cox
proportional hazards analysis, adjusting for age at
enrollment, past smoking, environmental tobacco
smoke exposure, alcohol use, education,
occupation, and body mass index. Analyzed
mortality from all natural causes, cardiopulmonary,
nonmalignant respiratory, and lung cancer.
SO2 was not associated with total (RR = 1.07
(95% Cl: 0.92, 1.24) for male and 1.00 (95% Cl:
0.88, 1.14) for female per 5 ppb increase in
multiyear average SO2), cardiopulmonary, or
respiratory mortality for either sex. Lung cancer
mortality showed large risk estimates for most of
the pollutants in either or both sexes, but the
number of lung cancer deaths in this cohort was
very small (12 for female and 18 for male)
Generally wide confidence intervals (relative to
other U.S. cohort studies).
Beeson et al.
(1998)

Three California
air basins: San
Francisco, South
Coast (Los
Angeles and
eastward), San
Diego

1977-1992
24-h avg SO2: 5.6 ppb
Prospective cohort study of 6,338 nonsmoking non-
Hispanic white adult members of theAdventist
Health Study aged 27-95 yrs at time of enrollment.
36 (20 females, 16 males) histologically confirmed
lung cancers were diagnosed through 1992.
Extensive exposure assessment, with assignment
of individual long-term exposures to O3, PM10,
SO42~, and SO2, was a unique strength of this
study. All results were stratified by gender. Used
Cox proportional hazards analysis, adjusting for
age at enrollment, past smoking, education, and
alcohol use.
Lung cancer incidence relative risk:
Male: RR = 3.72 (95%CI: 1.91, 7.28);
Female: RR = 2.78 (95%CI: 1.51, 5.12) per 5 ppb
increase in SO2

Case number very small (16 for male, 20 for
female).
Dockery et al.
(1993)

Portage, Wl;
Topeka, KS;
Watertown,  MA;
Harriman, TN;
St. Louis, MO;
Steubenville, OH

1974-1991.
24-h avg NO2 ranged
from 1.6 (Topeka) to
24.0 (Steubenville)
ppb.
A prospective cohort study to study the effects of air
pollution with main focus on PM components in six
U.S. cities, which were chosen based on the levels
of air pollution (Portage, Wl, the least polluted to
Steubenville, OH, the most polluted). Cox propor-
tional hazards regression was conducted with data
from a 14-to-16-yr follow-up of 8,111 adults in the
six cities, adjusting for smoking, sex, BMI,
occupational exposures, etc. PM25 and sulfate were
associated with these causes of deaths.
SO2 result presented only graphically. Fine
particles and sulfate showed better fit than SO2.
Krewski et al.
(2000)

Re-analysis and
sensitivity analysis
of Dockery et al.
(1993) study.
24-h avg NO2 ranged
from 1.6 (Topeka) to
24.0 (Steubenville)
ppb
Gaseous pollutants risk estimates were presented.
SO2 showed positive associations with total (RR =
1.05 (95% Cl: 1.02, 1.09) per 5 ppb increase in
the average SO2 over the study period),
cardiopulmonary (1.05 (95% Cl: 1.00, 1.10)), and
lung cancer deaths (1.03 (95% Cl: 0.91, 1.16)),
but in this dataset, SO2 was highly correlated with
PM25 (r = 0.85), sulfate (r = 0.85), and NO2 (r =
0.84)
May 2008
                                       F-108
                                              DRAFT—DO NOT QUOTE OR CITE

-------
    STUDY
                         CONC.
                                                           METHOD
                                                                                                       CONCLUSIONS
Krewski et al.
(2000); Jerrett
et al. (2003);
Krewski et al.
(2003); Re-
analysis/sensitivity
analysis of Pope
etal. (1995) study.
Multiyear avg of 24-h
avg 9.3 ppb.
Re-analysis of Pope et al. (1995) study. Extensive
sensitivity analysis with ecological covariates and
spatial models to account of spatial pattern in the
ACS data.
In the Jerret et al. reanalysis the relative risk
estimates for total mortality was 1.06 (95% Cl:
1.05, 1.07) per 5 ppb increase in the annual avg
SO2. In the spatial filtering model (this was the
model that resulted in the largest reduction of SO2
risk estimate when sulfate was included), the SO2
total mortality risk estimate was 1.07 (95% Cl:
1.03, 1.11) in the single-pollutant model and 1.04
(95% Cl: 1.02, 1.06) with sulfate in the model.
The risk estimates for PM2.5 and sulfate were
diminished when SO2 was included in the models.

In the sensitivity analysis conducted by Krewski
et al. SO2 was significantly associated  with all-
cause mortality in a single-pollutant model: 1.30
(1.23, 1.38). In the spatial analysis, SO2 was
found to be the only gaseous pollutant strongly
associated with all-cause and cardiopulmonary
disease mortality:

Relative Risk  (95% Cl) (per 19.9 pg/m3 Sulfates)
Spatial Analysis, All-Cause

Independent Observations
Sulfates + SO2: 1.05 (0.98, 1.12)
Fine particles + SO2: 1.03 (0.95, 1.13)

Independent Cities
Sulfates + SO2: 1.13 (1.02, 1.25)
Fine particles + SO2: 1.14(0.98, 1.32)

Regional Adjustment
Sulfates + SO2: 1.10(0.97, 1.24)
Fine particles + SO2: 1.11 (0.93, 1.33)

Spatial Filtering
Sulfates + SO2: 1.05 (0.97, 1.14)
Lipfert et al.
(2000a)

32 Veterans
Administration
hospitals
nationwide in the
U.S.

1976-2001
Mean of the 95th
percentile of the
24-h avg SO2 for
1997-2001 period:
15.8 ppb.
Update of the Lipfert et al. (2000a) study, with
follow-up period extended to 2001. Study focused
on the traffic density data. The county-level traffic
density was derived by dividing vehicle-km traveled
by the county land area. Because of the wide range
of the traffic density variable, log-transformed traffic
density was used in their analysis. They reported
that traffic density was a better predictor of mortality
than ambient air pollution variables, with the
possible exception of O3. The log-transformed
traffic density variable was weakly correlated with
SO2 (r = 0.32) in this data set.
RR using the 1997-2001  air quality data period:
0.99 (95% Cl: 0.97, 1.01) per 5 ppb increase; in a
single-pollutant model.

The 2-pollutant model with the traffic density
variable: 0.99 (95% Cl: 0.96, 1.01) per 5 ppb.
Lipfert et al.
(2000b)

32 Veterans
Administration
hospitals
nationwide in the
U.S.

1997-2001
Mean of the 95th
percentile of the
24-h avg SO2 for
1999-2001 period:
16.3 ppb.
Update of the Lipfert et al. (2000a) study, examined
PM25 chemical constituents data. The analysis
used county-level air pollution data for the period
1999-2001 and cohort mortality data for 1997-2001.
Traffic density was the most important predictor of
mortality, but associations were also seen for EC,
V, nitrate, and Ni. NO2, ozone, and PMio also
showed positive but weaker associations. The
risk estimate for SO2 was essentially the same as
that reported in the 2006a Lipfert et al. analysis
(0.99 (95% Cl: 0.96, 1.01) per 5 ppb) in a  single-
pollutant model. Multipollutant model results were
not presented for SO2.
Pope etal. (1995)

U.S. nationwide

1982-1989
Not analyzed/
reported.
Investigated associations between long-term
exposure to PM and the mortality outcomes in the
American Cancer Society cohort. Ambient air
pollution data from 151 U.S. metropolitan areas in
1981 were linked with individual risk factors in
552,138 adults who resided in these areas when
enrolled in the prospective study in 1982. Death
outcomes were ascertained through 1989.  Cox
proportional hazards model adjusted for smoking,
education, BMI, and occupational  exposures. PM25
and sulfate were associated with total,
cardiopulmonary, and lung cancer mortality, but not
with mortality for all other causes.
Gaseous pollutants not analyzed.
May 2008
                                        F-109
                                               DRAFT—DO  NOT QUOTE OR CITE

-------
STUDY
Pope et al. (2002)
U.S. nationwide
1982-1998
Willis et al. (2003)
Re-analysis/
sensitivity analysis
of Pope et al.
(1995) study.
Lipfert et al.
(2000b; 2003)
32 Veterans
Administration
hospitals
nationwide in the
U.S.
1976-1996
CONC.
24-h avg mean of
118 MSA's in1980: 9.7
ppb; mean of
126 MSA's during
1982-1998: 6.7 ppb.
Multiyear average of
24-h avg using MSA
scales: 9.3 ppb; using
county scales:
10.7 ppb.
SO2 mean levels NR.
METHOD
Prospective cohort study of approximately 500,000
members of American Cancer Society cohort en-
rolled in 1982 and followed through 1998 for all
cause, cardiopulmonary, lung cancer, and all other
cause mortality. Age at enrollment was 30+ yrs. Air
pollution concentrations in urban area of residence
at time of enrollment assessed from 1982 through
1998. Other pollutants considered include TSP,
PM15, PM10, PM25, PM1M5, SO42", SO2, NO2, and
CO.
Investigation of the effects of geographic scale over
which the air pollution exposures are averaged.
Exposure estimates were averaged over the county
scale, and compared the original ACS results in
which MSA scale average exposures were used.
Less than half of the cohort used in the MSA-based
study were used in the county scale based analysis
because of the limited availability of sulfate
monitors and because of the loss of subjects from
the use of five-digit zip codes
Cohort study of approximately 50,000 U.S. veterans
(all males) diagnosed with hypertension. Mean age
at recruitment was 51 yrs. Exposure to O3 during
four periods (1960-1 974, 1975-1981, 1982-1988,
1989-1996) associated with mortality over three
periods (1976-1981, 1982-1988, 1989-1996). Long-
term exposures to TSP, PM15, PM10, PM2.5, PM15_2.5,
SO42~, NO2, and CO also analyzed. Used Cox
proportional hazards regression, adjusting for race,
smoking, age, systolic and diastolic blood pressure,
body mass index, and socioeconomic factors.
CONCLUSIONS
PM25 was associated with total, cardiopulmonary,
lung cancer mortality, but not with deaths for all
other causes. SO2 was associated with all the
mortality outcomes, including all other causes of
deaths. SO2's risk estimate for total mortality was
1 .03 (95% Cl: 1 .02, 1 .05) per 5 ppb increase
(1982-1998 average). Residential location was
known only at enrollment to study in 1982. Thus,
exposure misclassification possible.
In the analysis comparing the 2-pollutant model
with sulfate and SO2, they found that, in the MSA-
scale model, the inclusion of SO2 reduced sulfate
risk estimates substantially (>25%), but not
substantially (< 25%) in the county-scale model.
In the MSA-level anlaysis (with 113 MSA's), SO2
relative risk estimate was 1 .04 (95% Cl: 1 .02,
1 .06) per 5 ppb increase, with sulfate in the
model. In the county-level anaysis (91 counties)
with sulfate in the model, the corresponding
estimate was smaller (RR = 1 .02 [95% Cl : 1 .00,
1 .05]). The correlation between covariates are
different between the MSA-level data and county-
level data.
"SO2 and Pb were considered less thoroughly.
"The authors presented only qualitative results for
SO2 from the "Screening regressions" which
indicated negatively significant risk estimate in
the univariate model and non-significant positive
estimate in the multivariate model.
CANADA
Finkelstein et al.
(2003)
Ontario, Canada
1992-1999
24-h avg (ppb): 4.9
(1.0)
Cohort consisting of 5,228 people >40 yrs old that
were referred for pulmonary function testing
between 1985 and 1999. Within the cohort
identified nonaccidental deaths that occurred from
1992 through 1999. Used air quality data for TSP
from 1992-1 994 and SO2 for 1993-1995. Analyzed
the association between TSP or SO2 and
socioeconomic status and mortality using a Cox
proportional hazards model stratified by sex and 5-
yr age groups.
Using the high income-low pollutant level as the
reference the following results were reported for
each of the mortality endpoints:
Relative Risk (95% Cl) All causes
High income-high pollutant level: 1.35 (1.05, 1.73)
Low income-low pollutant level: 1 .64 (1 .21 , 2.24)
Low income-high pollutant level: 2.40 (1.61, 3.58)
Interaction with age group: 0.97 (0.95, 0.99)
Cardiopulmonary causes
High income-high pollutant level: 1.54 (1.13, 2.10)
Low income-low pollutant level: 2.05 (1.45, 2.91)
Low income-high pollutant level: 3.36 (2.12, 5.32)
Interaction with age group: 0.95 (0.92, 0.97)
EUROPE
Beelen et al.
(2008)
The Netherlands
1987-1996.
Cohort study on diet
and cancer with
120,852 subjects
followed from 1987 to
1996. BS, NO2, SO2,
and PM2.s and traffic-
exposure estimates
were analyzed. Cox
regression model
adjusted for age, sex,
smoking, and area-
level socioeconomic
status.
Mean SO2 Levels:
Mean: 4.8 ppb, with a range of 1.5 to 11.8 ppb.
Copollutants:
PM2.5
BS
NO2
Traffic intensity on the nearest road was not
associated with exposure SO2. Background SO2
levels were not associated with mortality.
Adjusted RR
(per 20 pg/m3 SO2)
All cause: 0.97 (0.90,1. 05)
Cardiovascular: 0.94 (0.82, 1.06)
Respiratory: 0.88 (0.64, 1 .22)
May 2008
F-110
DRAFT—DO NOT QUOTE OR CITE

-------
    STUDY
                         CONC.
                                                           METHOD
                                                                                                      CONCLUSIONS
Elliott et al. (2007)

Great Britain;
1966-1994 air
pollution; 1982-
1998 mortality in
four periods.
24-h avg SO2 levels
declined from 41.4
ppbin 1966-197010
12.2 ppb in 1990-
1994
A small area analysis of mortality rates in electoral
ward, with the mean area of 7.4 km2 and the mean
population of 5,301 per electoral ward. Deaths rates
were computed for four successive 4-yr periods
from 1982 to 1994. The number of wards in these
four periods ranged from 118 in the 1994-1998
period to 393 in the 1982-1986 period. Poisson
model was fit to model observed deaths for each
ward with a linear function for pollutant and random
intercept, with and without adjustment for social
deprivation.
They observed associations for both BS and SO2
and mortality outcomes. The estimated effects
were stronger for respiratory illness than other
causes of mortality for the most recent exposure
periods and most recent mortality period
(pollution levels were lower). The adjustment for
social deprivation reduced the risk estimates for
both pollutants. The adjusted  risk estimates for
SO2 for the pooled mortality periods using the
most recent exposure windows were: 1.021 (95%
Cl: 1.018, 1.024) for all-cause; 1.015 (95% Cl:
1.011, 1.019) for cardiovascular; and 1.064%
(95% Cl:  1.056, 1.072) for respiratory causes per
5 ppb increase in SO2. The risk estimates for the
most recent mortality period using the most
recent exposure windows were larger.
Filleul et al. (2005)

Seven French
cities

1975-2001
24-h avg SO2 ranged
from 5.9 ppb ("Area 3"
in Lille) to 29.7 ppb
("Area 3" in Marseille)
in the 24 areas in
seven cities during
1974-1976. Median
levels during 1990-
1997 ranged from 3.0
ppb (Bordeaux) to 8.2
ppb (Rouen) in the
five cities where data
were available.
Cohort study of 14,284 adults who resided in 24
areas from seven French cities when enrolled in the
PAARC survey (air pollution and chronic respiratory
diseases) in 1974. Daily measurements of SO2,
TSP, black smoke, NO2, and NO were made in 24
areas for three yrs (1974-76). Cox proportional
hazards models adjusted for smoking, educational
level, BMI, and occupational exposure. Models
were run before and after exclusion of six area
monitors influenced by local traffic as determined
by the NO/NO2 ratio >3.
Before exclusion of the six areas, none of the air
pollutants were associated with mortality
outcomes. After exclusion of these areas,
analyses showed associations between total
mortality and TSP, BS, NO2, and NO, but not SO2
(1.01 (95% Cl: 0.97,  1.06) per 5 ppb multi-yr
average). From these results, the authors noted
that inclusion of air monitoring data from stations
directly influenced by local traffic could
overestimate the mean population exposure and
bias the results. It should be noted that the table
describing air pollution levels in Filleul et al.'s
report indicates that the SO2 levels in these
French cities declined markedly between 1974-76
and 1990-1997 period, by a factor of 2 to 3,
depending on the city, whereas NO2 levels
between the two periods were variable,  increased
in some cities, and decreased in others. These
changes in air pollution levels over the study
period complicates interpretation of reported risk
estimates.
Lepeule et al.
(2006)

Bordeaux, France

1988-1997
24-h avg (pg/m3):
10.3(6.6)
Identified 439 non-accidental deaths and 158
cardiorespiratory deaths from the PersonnesAgees
QUID (PAQUID) cohort. Used a Cox proportional
hazards model with time dependent covariates to
examine the association between black smoke (BS)
and sulfur dioxide-strong acidity (SO2-AF) and non-
accidental and cardiorespiratory mortality.
Relative Risk (per 10 pg/m SO2-AF)

Non-accidental Mortality
1.03 (0.86, 1.24) lag 0. 0.96 (0.88, 1.06) lag 1
0.96 (0.85, 1.09) lag 2. 1.03 (0.94, 1.12) lag 3
1.17(0.99, 1.39) lag 4> 1.16(0.86, 1.55)
cumulative

Cardiorespiratory mortality
0.84(0.65, 1.10) lag 0. 1.06(0.94, 1.19) lag 1
1.19(1.03, 1.37) lag 2. 1.19(1.03, 1.37) lag 3
1.07(0.95, 1.19) lag 4. 0.85(0.66, 1.10) lag 5
1.15(0.75, 1.77) cumulative
Nafstad et al.
(2004)

Oslo, Norway

1972-1998.
The yearly averages
of 24-h avg SO2 were
reduced with a factor
of 7 during the study
period from 5.6 ppb in
1974 to 0.8 ppb in
1995.
Cohort study of 16,209 Norwegian men 40-49 yrs of
age living in Oslo, Norway, in 1972-1973. Data from
the Norwegian Death Register were linked with
estimates of average yrly air pollution levels at the
participants' home addresses from 1974 to  1998.
NOX, rather than NO2 was used. Exposure
estimates for NOX and SO2 were constructed using
models based on the subject's address, emission
data for industry, heating, and traffic, and measured
concentrations. Addresses linked to 50 of the busi-
est streets were given an additional exposure
based on estimates of annual average daily traffic.
Cox proportional-hazards regression was used to
estimate associations between exposure and total
and cause-specific mortality, adjusting for age
strata, education, occupation, smoking, physical
activity level, and risk groups for cardiovascular
diseases.
NOX was associated with total, respiratory, lung
cancer, and ischemic heart disease deaths. SO2
did not show any associations with mortality (e.g.,
0.97 (95% Cl: 0.94, 1.01) per 5 ppb multi-yr
average). The risk estimates presented for
categorical levels of these pollutants showed
mostly monotonic exposure-response
relationships for NOX, but not for SO2. Note the
very low levels of SO2.
May 2008
                                        F-111
                                               DRAFT—DO NOT QUOTE OR CITE

-------
STUDY
Nafstad et al.
(2003)
Oslo, Norway
1972-1998


CONC.
Yearly averages of
24-h avg. SO2
reduced with a factor
of 7 during study
period from 5.6 ppb in
1974 to 0.8 ppb in
1995.
METHOD
Lang cancer incidence was examined in the above
cohort. During the follow-up period, 418 men
developed lung cancer.



CONCLUSIONS
NOX was associated with lung cancer incidence.
SO2 showed no association (1 .01 ; [95% Cl: 0.92,
1.12] per
5 ppb multi-yr average).



Table F-10.   Associations of long-term exposure to SO2 with lung cancer.
    STUDY
                               METHODS
                                                                 POLLUTANTS
                                                                                                     CONCLUSIONS
                                                        UNITED STATES
Abbey et al.
(1999)

Three California
air basins: San
Francisco, South
Coast (Los
Angeles and
eastward), San
Diego

1977-1992
Prospective cohort study of 6,338 nonsmoking
non- Hispanic white adult members of the
Adventist Health Study followed for all cause,
cardiopulmonary, nonmalignant respiratory,
and lung cancer mortality. Participants were
aged 27-95 yrs at enrollment in 1977. 1,628
(989 females, 639 males) mortality events
followed through 1992. All results were
stratified by gender. Used Cox proportional
hazards analysis, adjusting for age at
enrollment, past smoking, environmental
tobacco smoke exposure, alcohol use,
education, occupation, and body mass index.
Analyzed mortality from all natural causes,
cardiopulmonary, nonmalignant respiratory,
and lung cancer.
Mean SO2 Levels:
24-h avg SO2: 5.6 ppb

Co pollutants:
PM10
SO4
03
NO2
Lung cancer mortality showed large risk
estimates for most of the pollutants in either or
both sexes, but the number of lung cancer
deaths in this cohort was very small (12 for
female and 18 for male) Generally wide confi-
dence intervals (relative to other U.S. cohort
studies).

Adjusted Mortality Relative Risk
(95% Cl) (per 3.72 ppb SO2)

Lung Cancer
Males: 1.99(1.24,3.20)
Females: 3.01 (1.88, 4.84)
Krewski et al.
(2000)
Re-analysis and sensitivity analysis of
Dockery et al. (1993) Harvard Six Cities study.
Mean SO2 Levels: 24-h avg SO2
ranged from 1.6 (Topeka) to
24.0 (Steubenville) ppb

Copollutants: Fine Particles,
Sulfates
SO2 showed positive associations with lung
cancer deaths (1.03 [95% Cl: 0.91, 1.16]), but in
this dataset, SO2 was highly correlated with
PM25 (r = 0.85), sulfate (r = 0.85), and NO2 (r =
0.84)
                                                            EUROPE
Beelen et al.
(2008)

The Netherlands

1987-1996.
Cohort study on diet and cancer with 120,852
subjects who were followed from 1987 to
1996. BS, NO2, SO2, and PM25 and traffic-
exposure estimates were analyzed. Cox
regression model adjusted for age, sex,
smoking, and area-level socioeconomic
status.
Mean SO2 Levels:
Mean: 4.8 ppb, with a range of
1.5 to 11.8 ppb.

Copollutants:
PM2.5
BS
NO2
Traffic intensity on the nearest road was not
associated with exposure SO2. Background SO2
levels were not associated with lung cancer
mortality.

Adjusted RR
(per 20 pg/m3 SO2)
1.00(0.79, 1.26)
Filleul et al.
(2005)

Seven French
cities

1975-2001
Cohort study of 14,284 adults who resided in
24 areas from seven French cities when en-
rolled in  the PAARC survey (air pollution and
chronic respiratory diseases) in 1974. Daily
measurements of SO2, TSP, black smoke,
NO2, and NO were made in 24 areas for three
yrs (1974-1976). Cox proportional hazards
models adjusted for smoking, educational
level, BMI, and occupational exposure.
Models were run before and after exclusion of
six area  monitors influenced by local traffic as
determined by the NO/NO2 ratio >3.
Mean SO2 Levels:
24-h avg SO2 ranged from 17
mg/m3 ("Area 3" in Lille) to 85
mg/m3 ("Area 3" in Marseille) in
the 24 areas in seven cities
during 1974-1976. Median
levels during 1990-1997 ranged
from 8.5 mg/m3 (Bordeaux) to
23.4 mg/m  (Rouen) in the five
cities where data were
available.

Copollutants:
TSP
Black Smoke
NO2
NO
The authors noted that inclusion of air
monitoring data from stations directly influenced
by local traffic could overestimate the mean
population exposure and bias the results. It
should be noted that the table describing air
pollution levels in  Filleul et al.'s report indicates
that the SO2 levels in these French cities
declined markedly from 1974-1976 and 1990-
1997 period, by a factor of 2 to 3, depending on
the city, whereas NO2 levels between the two
periods were variable, increased in some cities,
and decreased in  others. These changes in air
pollution levels over the study period complicate
interpretation of reported risk estimates.
Relative Risk (95% Cl) for lung cancer mortality
(per 10 mg/m3 multi-year average). All 24 areas:
0.99(0.92, 1.07).  18areas:1.00 (0.91, 1.11)
May 2008
                                        F-112
                          DRAFT—DO NOT QUOTE OR CITE

-------
    STUDY
                               METHODS
                                                                POLLUTANTS
                                                                                                   CONCLUSIONS
Nafstad et al.
(2004)

Oslo, Norway

1972-1998.
Cohort study of 16,209 Norwegian men 40-49
yrs of age living in Oslo, Norway, in 1972-
1973. Data from Norwegian Death Register
linked with estimates of avg yearly air
pollution levels at the participants' home
addresses from 1974 to 1998. NOX, rather
than  NO2 was used. Exposure estimates for
NOX  and SO2 were constructed using models
based on the subject's address, emission data
for industry, heating, and traffic, and
measured concentrations. Addresses linked to
50 of the busiest streets were given an addi-
tional exposure based on estimates of annual
avg daily traffic. Cox proportional-hazards
regression was used to estimate associations
between exposure and total and cause-
specific mortality, adjusting for age strata,
education, occupation, smoking, physical
activity level, and risk groups for cardiovascu-
lar diseases
Mean SO2 Levels:
The yearly avg of 24-h avg SO2
were reduced with a factor of 7
during the study period from 5.6
ppb in 1974 to 0.8 ppb in 1995.

Co pollutants:
NOX
SO2 did not show any associations with lung
cancer, e.g., 1.00 (0.93, 1.08) per 10 pg/m3 in-
crease mortality in SO2. No association was
also observed when including SO2 in the model
as a categorical variable. Note the very low
levels of SO2.
May 2008
                                       F-113
                         DRAFT—DO NOT QUOTE OR CITE

-------
                                   References
IARC. (1989) Diesel and gasoline engine exhausts and some nitroarenes. Lyon, France: International
       Agency for Research on Cancer. (IARC monographs on the evaluation of carcinogenic risks to
       humans: v. 46).

World Health Organization (WHO). (2004) IPCS risk assessment terminology. Part 2: IPCS glossary of
       key exposure assessment terminology. Geneva, Switzerland: IPCS Harmonization Project
       document no. 1. Available:
       http://www.who.int/ipcs/methods/harmonization/areas/ipcsterminologyparts1 and2.pdf [19 April,
       2007].

World Health Organization (WHO). (2005) Principles of characterizing and applying human exposure
       models. Geneva, Switzerland: World Health Organization. (IPCS harmonization project
       document; no. 3). Available: http://whqlibdoc.who.int/publications/2005/9241563117_eng.pdf [7
       March, 2007].

Abbey DE; Nishino N; McDonnell WF; Burchette RJ; Knutsen SF; Beeson WL; Yang JX. (1999) Long-
       term inhalable particles and other air pollutants related to mortality in nonsmokers. Am J Respir
       Grit Care  Med 159:373-382.

Abraham WM; Oliver W, Jr.; Welker MJ; King MM; Wanner A; Sackner MA. (1981) Differences in airway
       reactivity  in normal and allergic sheep after exposure to sulfur dioxide. J Appl Physiol. 51:1651-6.

Abraham WM; Sielczak MW; Delehunt JC; Marchette B; Wanner A. (1986) Impairment of tracheal
       mucociliary clearance but not ciliary beat frequency by a combination of low level ozone and
       sulfur dioxide in sheep. European Journal Respiratory Dis. 68:114-120.

Ackermann-Liebrich U; Leuenberger P; Schwartz J; SchindlerC; Monn C; Bolognini B; Bongard JP;
       Brandli O; Domenighetti G; ElsasserS; Grize L; KarrerW; Keller R; Keller-Wossidlo H; Kunzli N;
       Martin BW; Medici TC; Perruchoud AP; Schb'ni MH; Tschopp JM; Villiger B; Wuthrich B; Zellweger
       JP; Zemp E.  (1997) Lung function and long term exposure to air pollutants in Switzerland. Am J
       Respir Grit Care Med. 155:122-129.

Agarwal R; Jayaraman G; Anand S; Marimuthu P.  (2006) Assessing Respiratory Morbidity Through
       Pollution Status and Meteorological Conditions for Delhi. Environmental Monitoring and
       Assessment. 114:489-504.

Ainsworth N; Fumagalli I; Giorcelli A; Mignanego L; Schenone G; Vietto L. (1996) Assessment of EDU
       stem injections as a technique to investigate the response of trees to ambient ozone in field
       conditions. Agriculture, Ecosystems and Environment, 59 (1-2)  pp. 33-42, 1996.

Amdur MO; McCarthy JF; Gill MW. (1983) Effect of mixing conditions on irritant potency of zinc oxide and
       sulfur dioxide. Am Ind Hyg Assoc J. 44:7-13.

Amdur MO; Chen LC; Guty J; Lam HF; Miller PD. (1988) Speciation and pulmonary effects of acidic SOX
       formed on the surface of ultrafine zinc oxide aerosols. Atmospheric Environment. 22:557-560.

Anderson HR; Ponce de Leon A; Bland JM; Bower JS; Strachan DP. (1996) Air pollution and daily
       mortality in London: 1987-92. British Medical Journal. 312:665-669.
May 2008                                  1                 DRAFT—DO NOT QUOTE OR CITE

-------
Anderson HR; SpixC; Medina S; Schouten JP; Castellsague J; Rossi G; Zmirou D; Touloumi G;
       Wojtyniak B; Ponka A; Bacharova L; Schwartz J; Katsouyanni K. (1997) Air pollution and daily
       admissions for chronic obstructive pulmonary disease in 6 European cities: results from the
       APHEA project. European Respiratory Journal. 10:1064-1071.

Anderson HR; Ponce de Leon A; Bland JM; Bower JS; Emberlin J; Strachan DP. (1998) Air pollution,
       pollens, and daily admissions for asthma in London 1987-92. Thorax. 53:842-8.

Anderson HR; Bremner SA; Atkinson RW; Harrison RM; Walters S. (2001) Particulate matter and daily
       mortality and hospital admissions in the west midlands conurbation of the United Kingdom:
       associations with fine and  coarse particles, black smoke and sulphate. Occup Environ Med.
       58:504-510.

Andreae, M. O. (1991) Biomass burning: its history, use, and distribution and its impact on environmental
       quality and  global climate. In:  Levine, J. S., ed. Global biomass burning: atmospheric, climatic,
       and biospheric implications. Cambridge, MA: MIT Press; pp. 1-21.

Appel, K. W; Gilliland, A.; Eder,  B. (2005) An operational evaluation of the 2005 release  of models-3
       CMAQ version 4.5. Washington, DC: National Oceanic and Atmospheric Administration (NOAA)
       Air Resources Laboratory, Atmospheric Sciences Modeling Division.

Aranyi C; Vana SC; Thomas PT; Bradof JN; Renters JD; Graham JA; Miller FJ. (1983) Effects of
       subchronic exposure to  a mixture of O3, SO2,  and (NH4)2SO4 on host defenses of mice. Journal of
       toxicology and environmental health. 12:55-71.

Atkinson RW; Anderson HR; Strachan DP; Bland JM; Bremner SA; Ponce  de Leon A. (1999a) Short-term
       associations between outdoor air pollution and visits to accident and emergency departments in
       London  for  respiratory complaints. EurRespir J. 13:257-65.

Atkinson RW; Bremner SA; Anderson HR; Strachan DP; Bland JM; de Leon AP. (1999b)  Short-term
       associations between emergency hospital admissions for respiratory and cardiovascular disease
       and outdoor air pollution in London. Arch Environ Health. 54:398-411.

Atkinson RW; Anderson HR; Sunyer J; Ayres J; Baccini M; Vonk JM; BoumgharA; Forastiere F; Forsberg
       B; Touloumi G; Schwartz J; Katsouyanni K. (2001) Acute effects of particulate air pollution on
       respiratory  admissions:  results from APHEA 2 project. American journal of respiratory and critical
       care medicine. 164:1860-1866.

Azoulay-Dupuis E; Bouley G; Blayo MC. (1982) Effects of sulfur dioxide on resistance to bacterial
       infection in  mice. Environ Res. 29:312-9.

Bai J; Meng Z. (2005a) Effects of sulfur dioxide on apoptosis-related gene expressions in lungs from rats.
       Regulatory  Toxicology and Pharmacology. 43:272-279.

Bai J; Meng Z. (2005b) Expression of apoptosis-related genes in livers from rats exposed to sulfur
       dioxide. Toxicology. 216:253-260.

Ballester F; Tenias JM; Perez-Hoyos S.  (2001) Air pollution and emergency hospital admissions for
       cardiovascular diseases in Valencia, Spain. J  Epidemiol Community Health. 55:57-65.

Ballester F; Rodriguez P; Iniguez C; Saez M; Daponte A; Galan I; Taracido M; Arribas F; Bellido J; Cirarda
       FB; Canada A; Guillen JJ;  Guillen-Grima F; Lopez E; Perez-Hoyos S; Lertxundi A; Toro S. (2006)
       Air pollution and cardiovascular admisisons association in  Spain: results within the EMECAS
       project. Journal of epidemiology and community health. 60:328-336.
May 2008                                 2                  DRAFT—DO NOT QUOTE OR CITE

-------
Barnes I; Bastian V; Becker KH; Overath RD. (1991) Kinetic studies of the reactions of IO, BrO and CIO
       with Dimethylsulfide. International Journal of Chemical Kinetics. 23:579-591.

BarnettAG; Williams GM; Schwartz J; NellerAH; BestTL; Petroeschevsky AL; Simpson RW. (2005) Air
       pollution and child respiratory health: a case-crossover study in Australia and New Zealand. Am J
       Respir Grit Care Med. 171:1272-8.

Barthelemy P; BadierM; Jammes Y. (1988) Interaction between SO2 and cold-induced bronchospasm in
       anesthetized rabbits. Respir Physiol. 71:1-10.

Baskurt OK. (1988) Acute hematologic and hemorheologic effects of sulfur dioxide inhalation. Arch
       Environ Health. 43:344-8.

Bates DV; Baker-Anderson M; Sizto R. (1990) Asthma attack periodicity: a study of hospital emergency
       visits in Vancouver. Environ Res. 51:51-70.

Beelen R; Hoek G; van den Brandt PA; Goldbohm RA; Fischer P; Schouten LJ; Jerrett M; Hughes E;
       Armstrong B; Brunekreef B. (2008) Long-term effects of traffic-related air pollution on mortality in
       a Dutch cohort (NLCS-AIR study). Environ Health Perspect.  116:196-202.

Beeson WL; Abbey DE; Knutsen SF. (1998) Long-term concentrations of ambient air pollutants and
       incident lung cancer in California adults: results from the AHSMOG study. Environmental health
       perspectives. 106:813-823.

Bell ML; Ebisu K; Belanger K. (2007) Ambient air pollution and  low birth weight in Connecticut and
       Massachusetts. Environ Health Perspect. 115:1118-24.

Benner RL; Stedman DH. (1989) Universal sulfur detection by chemiluminescence. Analytical chemistry.
       61:1268-1271.

Benner RL; Stedman DH. (1990) Field evaluation of the sulfur chemiluminescence detector.
       Environmental  Science and Technology. 24:1592-1596.

Berresheim, H.; Wine, P. H.; Davis, D. D. (1995) Sulfur in the atmosphere. In: Singh, H. B., ed.
       Composition, chemistry, and climate of the atmosphere. New York, NY: Van Nostrand Reinhold;
       pp. 251-307.

Bigby B; Boushey H. (1993) Effects of nedocromil sodium on the bronchomotor response to sulfur dioxide
       in asthmatic patients. J Allergy Clin Immunol.  92:195-7.

Biggeri A; Baccini M; Bellini P; Terracini B. (2005) Meta-analysis of the  Italian studies of short-term effects
       of air pollution (MISA), 1990-1999. International Journal of Occupational and Environmental
       Health. 11:107-122.

Biller, W F.; Feagans, T. B.; Johnson, T. R.; Duggan, G. M.; Paul, R. A.; McCurdy, T.; Thomas, 4 H. C.
       (1981) A general model  for estimating exposure associated with alternative 5 NAAQS. Presented
       at: 74th annual meeting  of the Air Pollution Control Association; 6 June; Philadelphia, PA.
       Pittsburgh, PA: Air Pollution Control Association; paper no. 81-7 18.4.

Binkowski FS; Arunachalam S; Adelman Z; Pinto JP. (2007) Examining photolysis rates with a prototype
       online photolysis module in CMAQ. Journal of Applied Metereology and Climatology Climatol
       46:1252-1256.

Blanchard P; Brook JR; Brazil P. (2002) Chemical characterization of the organic fraction of atmospheric
       aerosol at two sites in Ontario, Canada. Journal of Geophysical Research.  107.
May 2008                                  3                  DRAFT—DO NOT QUOTE OR CITE

-------
Blanquart C; Giuliani I; Houcine O; Jeulin C; Guennou C; Marano F. (1995) In vitro exposure of rabbit
       tracheal epithelium to SO2: effects on morphology and ciliary beating. Toxicology in Vitro. 9:123-
       132.

Bobak M. (2000) Outdoor air pollution, low birth weight, and prematurity. Environ Health Perspect.
       108:173-6.

Boezen HM; van derZee SC; Postma DS; Vonk JM; Gerritsen J; Hoek G; Brunekreef B; Rijcken B;
       Schouten JP. (1999) Effects of ambient air pollution on upper and lower respiratory symptoms
       and peak expiratory flow in children. Lancet. 353:859-60.

Borja-Aburto VH; Loomis DP; Bangdiwala SI; Shy CM; Rascon-Pacheco RA. (1997) Ozone, suspended
       particulates, and daily mortality in Mexico City. American Journal of Epidemiology. 145:258-268.

Borja-Aburto VH; Castillejos M; Gold DR; Bierzwinski S; Loomis D. (1998) Mortality and ambient fine
       particles in southwest Mexico City, 1993-1995. Environ Health Perspect. 106:849-55.

Braga AL; Zanobetti A; Schwartz J. (2001) The lag structure between particulate air pollution and
       respiratory and cardiovascular deaths in 10 US cities. J Occup Environ Med. 43:927-33.

Braga ALF; Conceicao CMS; Pereira LAA; Kishi HS; Pereira JCR; Andrade MF; Goncalves FLT; Saldiva
       PHN; Latorre MRDO. (1999) Air pollution and  pediatric respiratory hospital admissions in Sao
       Paulo, Brazil. Journal of Environmental Med. 1:95-102.

Braun-Fahrlander C; Vuille JC; Sennhauser FH; Neu  U;  Kunzle T; Grize L; Gassner M; Minder C;
       SchindlerC; VaronierHS; Wuthrich B; team. S. (1997) Respiratory health and long-term
       exposure to air pollutants in Swiss schoolchildren. American journal of respiratory and critical care
       medicine. 155:1042-1049.

Buchdahl R; Parker A; Stebbings T; BabikerA. (1996) Association between air pollution and acute
       childhood wheezy episodes: prospective observational study. BMJ. 312:661-5.

Burke JM; Zufall MJ; Ozkaynak H. (2001) A population exposure model for particulate matter: case study
       results for PM2.5 in Philadelphia, PA. Journal  of Exposure Science and Environmental
       Epidemiology. 11:470-489.

Burnett RT; Brook JR; Yung WT; Dales RE; Krewski D. (1997a) Association between ozone and
       hospitalization for respiratory diseases in 16 Canadian cities. Environmental Research. 72:24-31.

Burnett RT; Cakmak S; Brook JR; Krewski D. (1997b)  The role of particulate size and chemistry in the
       association between summertime ambient air pollution and  hospitalization for cardiorespiratory
       diseases. Environ  Health  Perspect. 105:614-20.

Burnett RT; Cakmak S; Raizenne  ME; Stieb D; Vincent R; Krewski D; Brook JR; Philips O; Ozkaynak H.
       (1998) The  association between ambient carbon monoxide  levels and daily mortality in Toronto,
       Canada. J Air Waste Manag Assoc. 48:689-700.

Burnett RT; Smith-Doiron M; Stieb D; Cakmak S; Brook JR. (1999) Effects of particulate and gaseous air
       pollution on cardiorespiratory hospitalizations. Archives of Environmental Health. 54:130-139.

Burnett RT; Smith-Doiron M; Stieb D; Raizenne ME; Brook JR; Dales RE; Leech JA; Cakmak S;  Krewski
       D. (2001) Association between ozone and  hospitalization for acute respiratory diseases in
       children less than 2 years of age. Am J Epidemiol.  153:444-52.
May 2008                                  4                 DRAFT—DO NOT QUOTE OR CITE

-------
Burnett RT; Stieb D; Brook JR; Cakmak S; Dales R; Raizenne M; Vincent R; Dann T. (2004) Associations
       between Short-Term Changes in  Nitrogen Dioxide and Mortality in Canadian Cities. Archives of
       Environmental Health: An International Journal. 59:228-236.

Byun D; Schere KL. (2006) Review of the Governing Equations, Computational Algorithms, and Other
       Components of the Models-3 Community Multiscale Air Quality (CMAQ) Modeling System.
       Applied Mechanics Reviews. 59:51.

Byun DW; Ching JKS. (1999) Science Algorithms of the EPA Models-3 Community Multiscale Air Quality
       (CMAQ) Modeling System. Atmospheric modeling division, National Exposure Research
       Laboratory, US Environmental Protection Agency, Research Triangle Park, NC. 27711.

Cakmak S; Dales RE; Judek S. (2006) Do gender, education, and income modify the effect of air pollution
       gases on cardiac disease? J Occup Environ Med. 48:89-94.

Camp DC; Stevens RK; Cobourn WG; HusarRB; Collins JF; Huntzicker JJ; Husar JD; JaklevicJM;
       McKenzie RL; Tanner RL; Tesch  JW. (1982) Intercomparison of concentration results from fine
       particle sulfur monitors. Atmospheric Environment. 16:911-916.

IARC. (2006) Report of the advisory group to review the amended preamble to the IARC monographs.
       Lyon, France: International Agency for Research on Cancer. (IARC monographs on the
       evaluation of carcinogenic risks to humans: Report 06/001.

Castellsague J; Sunyer J; Saez M; Anto JM. (1995)  Short-term association between  air pollution and
       emergency room visits for asthma in Barcelona. Thorax. 50:1051-1056.

Chameides WL. (1984) The photochemistry of a remote marine stratiform cloud. Journal of Geophysical
       Research. 89:4739-4755.

Chan CC; Chuang KJ; Chien  LC; Chen WJ; Chang  WT. (2006) Urban air pollution and emergency
       admissions for cerebrovascular diseases in Taipei, Taiwan. Eur Heart J. 27:1238-44.

Chang CC; Tsai SS; Ho SC; Yang CY. (2005) Air pollution and hospital admissions for cardiovascular
       disease in Taipei, Taiwan. Environ Res. 98:114-9.

Charpin D; Pascal L; Birnbaum J; Armengaud A; Sambuc R;  LanteaumeA; Vervloet  D. (1999) Gaseous
       air pollution and atopy. Clin Exp Allergy. 29:1474-80.

Chatfield RB; Crutzen PJ. (1984) Sulfur dioxide in remote oceanic air: cloud transport of reactive
       precursors. Journal of Geophysical Research. 89:7111-7132.

Chen LC; Schlesinger RB. (1983) Response of the bronchial mucociliary clearance system in rabbits to
       inhaled sulfite and sulfuricacid aerosols. ToxicolAppI Pharmacol. 71:123-31.

Chen LC; Lam HF; Ainsworth D; Guty J; Amdur MO. (1987) Functional changes in the lungs of guinea
       pigs exposed to sodium sulfite aerosols. ToxicolAppI Pharmacol. 89:1-8.

Chen LC; Peoples SM; Amdur MO. (1991) Pulmonary effects of sulfur oxides on the  surface of copper
       oxide aerosol. Journal of occupational and environmental hygiene. 52:187-191.

Chen LC; Miller PD; Amdur MO; Gordon T. (1992) Airway hyperresponsiveness in guinea pigs exposed to
       acid-coated ultrafine particles. Journal of toxicology and environmental health. 35:165-174.

Chen PC; Lai YM; Chan CC; Hwang JS; Yang CY; Wang JD. (1999) Short-term effect of ozone on the
       pulmonary function of children in  primary school. Environmental health perspectives. 107:921.
May 2008                                  5                 DRAFT—DO NOT QUOTE OR CITE

-------
Chew FT; Goh DYT; Ooi BC; Saharom R; Hui JKS; Lee BW. (1999) Association of ambient air-pollution
       levels with acute asthma exacerbation among children in Singapore. Allergy. 54:320-329.

Ching J; Herwehe J; Swall J. (2006) On joint deterministic grid modeling and sub-grid variability
       conceptual framework for model evaluation. Atmospheric Environment. 40:4935-4945.

Chock DP; WinklerSL. (1994) A comparison of advection algorithms coupled with chemistry. Atmospheric
       Environment. 28:2659-2675.

Chock DP; WinklerSL; Chen C. (2000) A study of the association between daily mortality and ambient air
       pollutant concentrations in Pittsburgh, Pennsylvania. J Air Waste Manag Assoc. 50:1481-500.

Christakos, G. (2000) Modern spatiotemporal geostatistics. New York, NY: Oxford University 19 Press.
       (International Association for Mathematical Geology; studies in mathematical 20 geology: 6).

Christakos, G.; Hristopulos, D. T. (1998) Spatiotemporal environmental health modelling. A 22 tractatus
       stochasticus. Boston, MA: Kluwer Academic Publishers.

Christakos G; Kolovos A. (1999) A study of the spatiotemporal health impacts of ozone exposure. Journal
       of Exposure Science and Environmental Epidemiology. 9:322-335.

Clancy L; Goodman P;  Sinclair H; Dockery DW. (2002) Effect of air pollution  control on death rates in
       Dublin, Ireland: an intervention study. Lancet. 360:1210-1214.

Clegg NA; Toumi R. (1998) Non-sea-salt-sulphate formation in sea-salt aerosol. Journal of Geophysical
       Research. 103:31,095-31,31,102.

Clench-Aas J; BartonovaA; BohlerT; Gronskei KE; Sivertsen B; Larssen S. (1999) Air pollution exposure
       monitoring and estimating. Part I. Integrated  air quality monitoring system. Journal of
       Environmental Monit. 1:313-319.

Health Effects Institute. (2004) Health effects of outdoor air pollution in developing countries of Asia: a
       literature review. Special Report 15.

Conner MW; Lam HF; Rogers AE; Fitzgerald S; AmdurMO. (1985) Lung injury in guinea pigs caused  by
       multiple exposures to submicron zinc oxide mixed with sulfur dioxide in a humidified furnace. J
       Toxicol Environ Health. 16:101-14.

Conner MW; Flood WH; Rogers AE; AmdurMO. (1989) Changes in pulmonary lavage fluid  of guinea pigs
       exposed to  ultrafine zinc oxide with adsorbed sulfuric acid. J Toxicol  Environ Health. 26:223-34.

ICF Consulting.  (2005)  Decision points for consulting APEX for air toxics exposure assessments.
       Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of Air Quality Planning
       and Standards. Available:
       http://www.epa.gov/ttnmain1/fera/data/apex/APEXDecisionPoints_121205.pdf [19 September,
       2007].

Crutzen PJ. (1976) Upper limits on atmospheric ozone reductions following increased application of fixed
       nitrogen to the soil. Geophysical Research Letters. 3:169-172.

Crutzen PJ; Gidel LT. (1983) A two-dimensional photochemical model of the atmosphere. 2. The
       tropospheric budgets of the anthropogenic chlorocarbons, CO, CH4,  CH3CI and the effect of
       various NOX sources on tropospheric ozone. Journal of Geophysical  Research. 88:6641-6661.
May 2008                                 6                  DRAFT—DO NOT QUOTE OR CITE

-------
Cuijpers CEJ; Swaen GMH; Wesseling G; Wouters EFM. (1994) Acute respiratory effects of summer
       smog in primary school children. Toxicology letters. 72:227-235.

D'lppoliti D; Forastiere F; Ancona C; Agabiti N; Fusco D; Michelozzi P; Perucci CA. (2003) Air pollution
       and myocardial infarction in Rome: a case-crossover analysis.  Epidemiology. 14:528-35.

Dab W; Medina S; Quenel P; Le Moullec Y; Le Tertre A; Thelot B; Monteil C; Lameloise P; Pirard P;
       Momas I; Ferry R; Festy B. (1996) Short term respiratory health effects of ambient air pollution:
       results of the APHEA project in Paris.  J Epidemiol Community Health. 50 Suppl 1:s42-6.

Dagnall RM; Thompson KG; WestTS. (1967)  Molecular-emission in cool flames. Part I. the behaviour of
       sulphur species  in a hydrogen-nitrogen diffusion flame and in a shielded air-hydrogen flame.
       Analyst. 92:506-512.

Dales R; Burnett RT; Smith-Doiron M; Stieb DM; Brook JR. (2004) Air pollution and sudden infant death
       syndrome. Pediatrics. 113:e628-31.

Dales RE; Cakmak S; Doiron MS. (2006) Gaseous air pollutants and hospitalization for respiratory
       disease in the neonatal period. Environ Health Perspect. 114:1751-4.

de Diego Damia A; Leon Fabregas M; Perpina Tordera M; Compte Torrero L. (1999) Effects of air pollution
       and weather conditions on asthma exacerbation. Respiration. 66:52-8.

De Leon SF; Thurston GD; Ito K.  (2003) Contribution of respiratory disease to nonrespiratory mortality
       associations with air pollution. Am J Respir Grit Care Med. 167:1117-23.

Delfino RJ; Gong Jr H; Linn WS; Pellizzari ED; Hu Y. (2003) Asthma Symptoms in Hispanic Children and
       Daily Ambient Exposures to Toxic and Criteria Air Pollutants. Environmental health perspectives.
       111:647-656.

Delmas R. (1982) On the emission of carbon,  nitrogen and sulfur in the atmosphere during bushfires in
       intertropical savannah zones. Geophysical Research Letters. 9:761-764.

Dentener F; Stevenson D; Cofala J; Mechler R; Amann M; Bergamaschi P; Raes F; Derwent R.  (2005)
       The impact of air pollutant and methane emission controls on tropospheric ozone and radiative
       forcing: CTM calculations for the period 1990-2030. Atmospheric Chemistry and Physics. 5:1731-
       1755.

Dentener F; Drevet J; Lamarque JF; Bey  I; Eickhout B; Fiore AM; Hauglustaine D; Horowitz LW; Krol M;
       Kulshrestha  DC; Lawrence M; Galy-Lacaux C; Rast S; Shindell D; Stevenson D; Van Noije T;
       Atherton C; Bell N; Bergman D; Butler T; Cofala J; Collins B; Doherty R; Ellingsen K; Galloway J;
       Gauss M; Montanaro V; Miiller JF; Pitari G; Rodriguez J; Sanderson M; Solman F; Strahan  S;
       Schultz M; Sudo K; Szopa S; Wild O.  (2006) Nitrogen and sulfur deposition on regional and
       global scales: a multi-model evaluation.  Global Biogeochemical Cycles. 20:GB4003,
       doi:10.1029/2005GB002672.

Desqueyroux H; Pujet JC; Prosper M; Squinazi F;  Momas I. (2002) Short-term effects of low-level air
       pollution on respiratory health of adults suffering from moderate to severe asthma. Environ Res.
       89:29-37.

Devalia JL; Rusznak C; Herdman MJ; Trigg CJ; Tarraf H;  Davies RJ. (1994) Effect of nitrogen dioxide  and
       sulphur dioxide on airway response of mild asthmatic patients to allergen inhalation. Lancet.
       344:1668-71.
May 2008                                 7                  DRAFT—DO NOT QUOTE OR CITE

-------
Diaz J; Garcia R; Ribera P; Alberdi JC; Hernandez E; Pajares MS; Otero A. (1999) Modeling of air
       pollution and its relationship with mortality and morbidity in Madrid, Spain. IntArch Occup Environ
       Health. 72:366-76.

Dickerson RR; Huffman GJ; Luke WT; Nunneracker LJ; Pickering KE; Leslie ACD; Lindsey CG; Slinn
       WGN; Kelly TJ; et al. (1987) Thunderstorms: An important mechanism in the transport of air
       pollutants. Science. 235:460-464.

Dockery DW; Speizer FE;  Stram DO; Ware JH; Spengler JD; Ferris BG, Jr. (1989) Effects of inhalable
       particles on respiratory health of children. Am Rev Respir Dis.  139:587-94.

Dockery DW; Schwartz J;  Spengler JD.  (1992) Air pollution and daily mortality: associations with
       particulates and acid aerosols. Environ Res. 59:362-73.

Dockery DW; Pope CA, 3rd; Xu X; Spengler JD; Ware JH; Fay ME; Ferris BG, Jr.; Speizer FE. (1993) An
       association between air pollution and mortality in six U.S. cities. N Engl J Med. 329:1753-9.

Dockery DW; Cunningham J; Damokosh Al; Neas LM; Spengler JD; Koutrakis P; Ware JH; Raizenne M;
       Speizer FE. (1996) Health effects of acid aerosols on North American children: respiratory
       symptoms. Environ Health Perspect. 104:500-5.

Dominici F; McDermott A;  Daniels M; Zeger SL; Samet JM. (2003) Mortality among residents of 90 cities.
       In: Revised analyses of time-series studies of air pollution and health. Special report. Boston, MA:
       Health Effects Institute; pp. 9-24. Available: http://www.healtheffects.org/Pubs/TimeSeries.pdf [12
       May, 2004].

Douglas GJ; Price JF; Page CP. (1994) A method for the long-term exposure of rabbits to environmental
       pollutant gases. Eur Respir J. 7:1516-1526.

Du Z;  Meng Z. (2004) Effects of derivatives of sulfur dioxide on transient outward potassium currents in
       acutely isolated hippocampal neurons. Food ChemToxicol. 42:1211-6.

Du Z;  Meng Z. (2006) Sulfur dioxide derivatives modulation of high-threshold calcium currents in rat
       dorsal root ganglion neurons. Neurosci Lett. 405:147-152.

Dugandzic R; Dodds L; Stieb D; Smith-Doiron M. (2006) The association between low level exposures to
       ambient air pollution and term low birth weight: a retrospective cohort study. Environ Health. 5:3.

Elliott  P; Shaddick G;  Wakefield JC; de Hoogh C; Briggs DJ. (2007)  Long-term associations of outdoor air
       pollution with  mortality in Great  Britain. Thorax. 62:1088-94.

U.S. Environmental Protection Agency. (1991) Guideline for regulatory application of the urban  airshed
       model. Research Triangle Park, NC: Office of Air Quality Planning and Standards; report no. EPA-
       450/4-91-013. Available from: NTIS, Springfield, VA; PB92-108760.

U.S. Environmental Protection Agency. (1993) Ambient air quality surveillance. Final rule. F. R.  (February
       12): 8452-8475.

U.S. Environmental Protection Agency. (1996) Air quality criteria for ozone and related photochemical
       oxidants. Research Triangle Park, NC: Office of Research and Development; report nos.
       EPA/600/AP-93/004aF-cF. 3v. Available from: NTIS, Springfield, VA; PB96-185582, PB96-
       185590, and PB96-185608. Available: http://cfpub2.epa.gov/ncea/.
May 2008                                  8                 DRAFT—DO NOT QUOTE OR CITE

-------
U.S. Environmental Protection Agency. (1997) National air pollutant emission trends 1990-1996.
       Research Triangle Park, NC: Office of Air Quality Planning and Standards; report no. EPA/454/R-
       97/011.

U.S. Environmental Protection Agency. (1999) Getting started: emissions inventory methods for PM-2.5.
       Research Triangle Park, NC: Office of Air Quality Planning and Standards, Emission Factor and
       Inventory Group; EIIP document series - volume IX. Available:
       http://www.epa.gov/ttn/chief/eiip/techreport/volume09/index.html [7 May 2003].

U.S. Environmental Protection Agency. (2000) Air quality criteria for carbon monoxide. Research Triangle
       Park, NC: National Center for Environmental Assessment; report no. EPA/600/P-99/001F.
       Available: http://www.epa.gov/ncea/pdfs/coaqcd.pdf [19 April 2007].

U.S. Environmental Protection Agency. (2004) Air quality criteria for particulate matter. Research Triangle
       Park, NC: National Center for Environmental Assessment; report no. EPA/600/P-99/002aF-bF. 2v.
       Available: http://cfpub.epa.gov/ncea/ [9 November, 2004].

U.S. Environmental Protection Agency. (2005) Technical assistance document (TAD) for precursor gas
       measurements in the NCore multi-pollutant monitoring network. Version 4. Research Triangle
       Park, NC: Office of Air Quality Planning and Standards; report no. EPA-454/R-05-003.

U.S. Environmental Protection Agency. (2006a) Air quality criteria for ozone and related photochemical
       oxidants. Research Triangle Park,  NC: National Center for Environmental Assessment; report no.
       EPA/600/R-05/004aF-cF. 3v. Available:  http://cfpub.epa.gov/ncea/ [24 March, 2006].

U.S. Environmental Protection Agency. (2006b) Technology Transfer Network clearinghouse for
       inventories and emission factors. Available: http://www.epa.gov/ttn/chief/net/2002inventory.html
       [19 September, 2007].

EtlikO; TomurA; Kutman MN; Yorukan S; Duman O. (1995) The effects of sulfur dioxide inhalation and
       antioxidant vitamins on red blood cell lipoperoxidation. Environmental Research. 71:25-28.

EtlikO; TomurA; Tuncer M; Ridvanagaoglu AY; Andac O. (1997) Protective effect of antioxidant vitamins
       on red blood cell lipoperoxidation induced by SO2 inhalation. Journal  of Basic Clinical Physiol
       Pharmacol. 8:31-43.

Euler GL; Abbey DE; Magie AR; Hodgkin JE. (1987) Chronic obstructive pulmonary disease symptom
       effects of long-term cumulative exposure to ambient levels of total suspended particulates and
       sulfur dioxide in California Seventh-Day Adventist residents. Arch Environ Health. 42:213-22.

Evans MV; Simmons JE. (1996) Physiologically Based Pharmacokinetic Estimated Metabolic Constants
       and Hepatotoxicity of Carbon Tetrachloride after Methanol Pretreatment in Rats. Toxicology and
       Applied  Pharmacology. 140:245-253.

Faraji M; et al. (2005) Comparison of the carbon bond and SAPRC photochemical mechanisms event to
       southeast Texas. Houston, TX: Texas Environmental Research Consortium (TERC); project report
       H12.8HRB.

Farhat SCL; Paulo RLP; Shimoda TM; Conceicao CMS; Lin CA; Braga ALF; Warth MPN; Saldiva PHN.
       (2005) Effect of air pollution on pediatric respiratory emergency room  visits and hospital
       admissions. Brazilian Journal  of Medical and Biological Research. 38:227-235.

Farwell SO; Rasmussen RA. (1976) Limitations of the FPD and ECD in atmospheric analysis: a review.
       Journal  of Chromatogr Sci. 14:224-234.
May 2008                                 9                  DRAFT—DO NOT QUOTE OR CITE

-------
Field PI; Simmul R; Bell SC; Allen DH; Berend N. (1996) Evidence foropioid modulation and generation
       of prostaglandins in sulphur dioxide (SO)2-induced bronchoconstriction. Thorax. 51:159-63.

Finkelstein MM; Jerrett M; DeLuca P; Finkelstein N; Verma DK; Chapman K; Sears MR. (2003) Relation
       between income, air pollution and mortality: a cohort study. CMAJ. 169:397-402.

Fiore M; Petruzzi S; Dell'Omo G; Alleva  E. (1998) Prenatal sulfur dioxide exposure induces changes in
       the behavior of adult male mice during agonistic encounters. Neurotoxicol Teratol. 20:543-8.

Fischer P; Hoek G; Brunekreef B; Verhoeff A; van Wijnen J. (2003) Air pollution and mortality in the
       Netherlands: are the elderly more at risk? European Respiratory Journal. 21:34S-38S.

Forsberg B; Stjernberg N; Falk M; Lundback B; Wall S. (1993) Air pollution levels,  meteorological
       conditions and asthma symptoms. European Respiratory Journal. 6:1109-1115.

Fox-Rabinovitz MS; Takacs LL; Govindaraju RC. (2002) A variable-resolution stretched-grid  general
       circulation model and data assimilation system with multiple areas of interest: studying the
       anomalous regional climate events of 1998. Journal of Geophysical Research.  107:024.

Freijer Jl; Bloemen HJT; de Loos S; Marra M; Rombout PJA; Steentjes GM; Van Veen MP. (1998)
       Modelling  exposure of the Dutch population to air pollution. Journal of Hazard Mater. 61:107-114.

FrischerT; Studnicka M; Gartner C; TauberE; Horak F; VeiterA; Spengler J; Kuhr J; Urbanek R. (1999)
       Lung function growth and ambient ozone: a three-year population study in school children. Am J
       Respir Grit Care Med. 160:390-6.

FrischerT; Studnicka M; HalmerbauerG; Horak F; Gartner C; TauberE; KollerDY. (2001) Ambient ozone
       exposure is associated with eosinophil activation in healthy children. Clinical and experimental
       allergy. 31:1213-1219.

Fuentes M; RafteryAE. (2005) Model evaluation and spatial interpolation by Bayesian combination of
       observations with outputs from numerical models. Biometrics. 61:36-45.

Fung K; Krewski D; Burnett R; Dominici  F. (2005) Testing the harvesting hypothesis by time-domain
       regression analysis. I: baseline analysis. J Toxicol Environ Health A. 68:1137-54.

Fung KY; Khan S;  Krewski D; Chen Y. (2006) Association between Air Pollution and Multiple Respiratory
       Hospitalizations among the Elderly in Vancouver, Canada. Inhalation Toxicology. 18:1005-1011.

Fusco D; Forastiere F; Michelozzi P; Spadea T; Ostro B; Area M; Perucci CA. (2001) Air pollution and
       hospital admissions for respiratory conditions in Rome,  Italy. Eur Respir J. 17:1143-50.

Galan I; Tobias A;  Banegas JR; Aranguez E. (2003) Short-term effects of air pollution on daily asthma
       emergency room admissions. Eur Respir J. 22:802-8.

Gamble JL. (1998) Effects of ambient air pollution on daily mortality: a time series analysis of Dallas,
       Texas, 1990-1994.

Garcia-Aymerich J; Tobias A; Anto JM; Sunyer J. (2000) Air pollution and mortality in a cohort of patients
       with chronic obstructive pulmonary disease: a time series analysis. J Epidemiol Community
       Health. 54:73-4.

Garcia-Marcos L; Guillen JJ; Dinwiddie R; Guillen A; Barbero P. (1999) The relative importance of socio-
       economic  status, parental smoking and air pollution (SO2) on asthma symptoms, spirometry and
       bronchodilator response  in  11-year-old children. Pediatr Allergy Immunol.  10:96-100.
May 2008                                 10                 DRAFT—DO NOT QUOTE OR CITE

-------
U.S. Surgeon General. (2004) The Health Consequences of Smoking: A Report of the Surgeon General.
       Atlanta, GA: U.S. Department of Health and Human Services, Centers for Disease Control and
       Prevention. Washington, DC.

U.S. Surgeon General. (1964) Smoking and Health: Report of the Advisory Committee of the Surgeon
       General of the Public Health Service. Atlanta, GA: U.S. Department of Health and Human
       Services, Centers for Disease Control and Prevention. Washington, DC.

Georgopoulos PG; Lioy PJ. (1994) Conceptual and theoretical aspects of human  exposure and dose
       assessment. Journal of Exposure Science and Environmental Epidemiology. 4:253-285.

Georgopoulos PG; Purushothaman V; Chiou R. (1997) Comparative evaluation of methods for estimating
       potential human exposure to ozone: photochemical modeling and ambient monitoring. Journal of
       Exposure Science and Environmental Epidemiology. 7:191-215.

Georgopoulos PG; Wang S-W; Vyas VM; Sun Q; Burke J; Vedantham R; McCurdy T; Ozkaynak H. (2005)
       A source-to-dose assessment of population exposures to fine PM and ozone in Philadelphia, PA,
       during a summer 1999 episode. Journal of Exposure Science and Environmental Epidemiology.
       15:439-457.

Georgopoulos PG; Lioy PJ. (2006) From theoretical aspects of human exposure and dose assessment to
       computational model implementation: the Modeling ENvironment for Total Risk studies
       (MENTOR). Journal of Toxicology and Environmental Health, Part B: Critical Reviews. 9:457-483.

Gery MW; Whitten GZ; Killus JP; Dodge  MC. (1989) A photochemical kinetics mechanism for urban and
       regional scale computer modeling. Journal of Geophysical Research.  94:12,925-12,956.

Gidel LT. (1983) Cumulus cloud transport of transient tracers. Journal of Geophysical Research. 88:6587-
       6599.

Gokirmak M; Yildirim Z; Hasanoglu HC; Koksal N; Mehmet N. (2003) The role of oxidative stress in
       bronchoconstriction due to occupational sulfur dioxide exposure. Clinica Chimica Acta. 331:119-
       126.

Gold DR; Litonjua A; Schwartz J; Lovett E; Larson A; Hearing B; Allen G; Verrier M;  Cherry R; Verrier R.
       (2000) Ambient pollution and  heart rate  variability. Circulation. 101:1267-73.

Goldberg, M. S.; Burnett, R. T. (2003) Revised analysis of the Montreal time-series study. In: Revised
       analyses of time-series studies of air pollution and health. Special report.  Boston, MA: Health
       Effects Institute; pp. 113-132. Available: http://www.healtheffects.org/Pubs/TimeSeries.pdf [18
       October, 2004].

Gondal MA. (1997) Laser photoacoustic spectrometer for remote monitoring of atmospheric pollutants.
       Laser photoacoustic spectrometer for remote  monitoring of atmospheric pollutantsAppI Opt.
       36:3195-3201.

Gondal MA; Mastromarino J.  (2001) Pulsed laser photoacoustic detection of SO2  near 225.7 nm. Applied
       optics 40:2010-2016.

Gong H; Linn WS; Shamoo DA; Anderson KR; Nugent CA; Clark KW; Lin AE.  (1996) Effect of inhaled
       salmeterol on sulfur dioxide-induced bronchoconstriction in asthmatic subjects. Chest. 110:1229-
       1235.
May 2008                                 11                 DRAFT—DO NOT QUOTE OR CITE

-------
Gong H, Jr.; Lachenbruch PA; Harber P; Linn WS. (1995) Comparative short-term health responses to
       sulfur dioxide exposure and other common stresses in a panel of asthmatics. Toxicol Ind Health.
       11:467-87.

Gong H, Jr.; Linn WS; Terrell SL; Anderson KR; Clark KW. (2001) Anti-inflammatory and lung function
       effects of montelukast in asthmatic volunteers exposed to sulfur dioxide. Chest. 119:402-8.

Goss CH; Newsom SA; Schildcrout JS; Sheppard L;  Kaufman JD. (2004) Effect of ambient air pollution
       on pulmonary exacerbations and lung function  in cystic fibrosis. Am J Respir Grit Care Med.
       169:816-21.

Gouveia N; Fletcher!. (2000) Respiratory diseases in children and outdoor air pollution in  Sao Paulo,
       Brazil: a time series analysis. Occup Environ Med. 57:477-83.

Greenhut GK; Ching JKS; Pearson R, Jr.; RepoffTP. (1984) Transport of ozone by turbulence and clouds
       in an urban boundary layer. Journal of Geophysical Research. 89:4757-4766.

Greenhut GK. (1986) Transport of ozone between boundary layer and cloud  layer by cumulus clouds.
       Journal of Geophysical Research. 91:8613-8622.

Griese M; Winzinger G; Schams A; Josten M; Ziesenis  A; Maier K. (1999) Health  effects of sulfur-related
       environmental air pollution: the pulmonary surfactant system is not disturbed by exposure to
       acidic sulfate and neutral sulfite aerosols. J Aerosol Med. 12:37-44.

Gross A; Stockwell WR. (2003) Comparison of the EMEP, RADM2 and RACM mechanisms. Journal of
       Atmospheric Chemistry. 44:151-170.

GuentherA; Geron C; Pierce T; Lamb B; Harley P; Fall R. (2000)  Natural emissions of non-methane
       volatile organic compounds,  carbon monoxide, and oxides of nitrogen from North America.
       Atmospheric Environment. 34:2205-2230.

Gumuslu S; Akbas H; Aliciguzel Y; AgarA; Kucukatay V; Yargicoglu P. (1998) Effects of sulfur dioxide
       inhalation on antioxidant enzyme activities in rat erythrocytes. Industrial Health. 36:70-73.

Gumuslu S; Bilmen S; Korgun DK; Yargicoglu P; AgarA. (2001) Age-related  changes in antioxidant
       enzyme activities and lipid peroxidation in lungs of control and sulfur dioxide exposed rats. Free
       Radical Res. 34:621-627.

Gunnison AF; SellakumarA; Currie D; Snyder EA. (1987) Distribution, metabolism and toxicity of inhaled
       sulfur dioxide and endogenously generated sulfite in the respiratory tract  of normal and sulfite
       oxidase-deficient rats. J Toxicol Environ Health. 21:141-62.

Gunnison AF; SellakumarA; Snyder  EA; Currie D. (1988) The effect of inhaled sulfur dioxide and
       systemic sulfite on the induction of lung carcinoma in rats by benzo[a]pyrene. Environ Res. 46:59-
       73.

Gwynn RC; Burnett RT; Thurston GD. (2000) A time-series analysis of acidic particulate matter and daily
       mortality and morbidity in the Buffalo, New York, region. Environ Health Perspect.  108:125-33.

Ha EH; Hong YC; Lee BE; Woo BH;  Schwartz J; Christian! DC. (2001) Is air  pollution a risk factor for low
       birth weight in Seoul? Epidemiology. 12:643-8.

Hagen JA; Nafstad P; SkrondalA; Bjorkly S; Magnus P. (2000) Associations  between outdoor air
       pollutants and hospitalization for respiratory diseases. Epidemiology. 11:136-40.
May 2008                                 12                 DRAFT—DO NOT QUOTE OR CITE

-------
Haider SS; Hasan M; Hasan SN; Khan SR; AN SF. (1981) Regional effects of sulfur dioxide exposure on
       the guinea pig brain lipids, lipid peroxidation and lipase activity. Neurotoxicology. 2:443-450.

Haider SS; Hasan M; Khan NH. (1982) Air pollutant sulfur dioxide-induced alterations on the levels of
       lipids, lipid peroxidation and lipase activity in various regions of the rat brain. Pharmacology and
       Toxicology. 51:45-50.

Haider SS; Hasan M. (1984) Neurochemical changes by inhalation of environmental pollutants sulfur
       dioxide and hydrogen sulfide: degradation of total lipids, elevation of lipid peroxidation and
       enzyme activity in discrete regions of the guinea pig brain and spinal cord. Ind Health. 22:23-31.

Hajat S; Haines A; Goubet SA; Atkinson RW; Anderson HR. (1999) Association of air pollution with daily
       GP consultations for asthma and other lower respiratory conditions in London. Thorax. 54:597-
       605.

Hajat S; Anderson HR; Atkinson RW; Haines A. (2002) Effects of air pollution on general practitioner
       consultations for upper respiratory diseases in London. Occup Environ Med. 59:294-9.

Hansen K; Draaijers  GPJ; Ivens WPMF; Gundersen P; van Leeuwen NFM. (1994) Concentration
       variations in  rain and canopy throughfall collected sequentially during individual rain events.
       Atmospheric Environment. 28:3195-3205.

Heinrich J; Hoelscher B; Frye C; Meyer I; Pitz M; Cyrys J; Wjst M; Neas L; Wichmann H-E. (2002)
       Improved air quality in reunified Germany and decreases in respiratory symptoms. Epidemiology.
       13:394-401.

Heinrich U; Mohr U; Fuhst R; Brockmeyer C. (1989) Investigation of a potential cotumorigenic effect of the
       dioxides of nitrogen and sulfur, and of diesel-engine exhaust, on the respiratory tract of Syrian
       golden hamsters.  Res Rep Health Eff lnst.1-27.

Herbarth O; Fritz G; Krumbiegel P; Diez U;  Franck U; RichterM. (2001) Effect of sulfur dioxide and
       particulate pollutants on bronchitis in children-a risk analysis. Environ Toxicol. 16:269-76.

Heyder J; Beck-Speier I; Ferron GA; Heilmann P; Karg E; Kreyling WG; LenzAG; Maier K; Schulz H;
       Takenaka S; Tuch T. (1992) Early response of the canine respiratory tract following long-term
       exposure to a sulfur(IV) aerosol at low concentration. I. Rationale, design, methodology, and
       summary. . Inhalation Toxicology. 4:159-174.

Higgins BG; Francis  HC; Yates CJ; Warburton CJ; Fletcher AM; Reid JA; Pickering CA; Woodcock AA.
       (1995) Effects of air pollution on symptoms and peak expiratory flow measurements in subjects
       with obstructive airways disease. Thorax. 50:149-155.

HillAB. (1965) The Environment and Disease: Association or Causation? ProcRSocMed. 58:295-300.

Hiltermann TJN; Stolk J; Van der Zee SC; Brunekreef B; De Bruijne CR; Fischer PH; Ameling CB; Sterk
       PJ; Hiemstra PS; Van  Bree L. (1998) Asthma severity and susceptibility to air pollution. European
       Respiratory Journal. 11:686-693.

Hoek G; Brunekreef B. (1993) Acute effects of a winter air pollution episode on  pulmonary function and
       respiratory symptoms of children. Arch Environ Health. 48:328-35.

Hoek G; Brunekreef B. (1995)  Effect of photochemical air pollution on acute respiratory symptoms in
       children. Am J Respir Grit Care Med. 151:27-32.
May 2008                                 13                 DRAFT—DO NOT QUOTE OR CITE

-------
Hoek G; Brunekreef B; Goldbohm S; Fischer P; van den Brandt PA. (2002) Association between mortality
       and indicators of traffic-related air pollution in the Netherlands: a cohort study. Lancet. 360:1203-
       9.

Holguin F; Tellez-Rojo MM; Hernandez M; Cortez M; Chow JC; Watson JG; Mannino D; Romieu I. (2003)
       Air pollution and heart rate variability among the elderly in Mexico  City. Epidemiology. 14:521-7.

Holland HD. (1978) The chemistry of the atmosphere and oceans. New York, NY: Wiley. Wiley.

HorakF, Jr.; Studnicka M; Gartner C; Spengler JD; TauberE; Urbanek R; VeiterA; FrischerT. (2002)
       Particulate matter and lung function growth in children: a 3-yr follow-up study in Austrian
       schoolchildren [author response].

Hosseinpoor AR; Forouzanfar MH; Yunesian M; Asghari F; Naieni KH; Farhood D. (2005) Air pollution and
       hospitalization due to angina pectoris in Tehran, Iran: a time-series study. Environ Res. 99:126-
       31.

Huebert BJ; Zhuang L; Howell S; Noone K; Noone B. (1996) Sulfate, nitrate, methanesulfonate, chloride,
       ammonium, and sodium measurements from ship, island, and aircraft during the Atlantic
       Stratocumulus Transition Experiment/Marine Aerosol Gas Exchange. Journal of Geophysical
       Research. 101:4413-4423.

Hwang BF; Lee YL; Lin YC; Jaakkola JJ; Guo YL. (2005) Traffic related air pollution as a determinant of
       asthma among Taiwanese school children. Thorax. 60:467-73.

Hwang JS; Chan CC. (2002) Effects of air pollution on daily clinic visits for lower respiratory tract illness.
       Am J Epidemiol. 155:1-10.

IARC. (2006) Report of the advisory group to review the amended preamble to the IARC monographs.
       lARCInt Rep No 06/001.

Ibald-Mulli A; Stieber J; Wichmann HE; Koenig W; Peters A. (2001) Effects of air pollution on blood
       pressure: a population-based approach. Am J Public Health. 91:571-7.

llabaca M; Olaeta I; Campos E; Villaire J; Tellez-Rojo MM; Romieu I. (1999) Association between levels of
       fine particulate and emergency visits for pneumonia and other respiratory illnesses among
       children in Santiago, Chile. J Air Waste Manag Assoc. 49:154-63.

IOM. (2007) Improving the Presumptive Disability Decision-Making Process for Veterans. In: Samet
       James M, Bodurow CC, editors: Institute of Medicine of the National Academies.

Ito, K. (2003) Associations of particulate matter components with daily mortality and morbidity in Detroit,
       Michigan. In: Revised analyses of time-series studies of air pollution and health. Special report.
       Boston, MA: Health Effects Institute; pp. 143-156. Available:
       http://www.healtheffects.org/Pubs/TimeSeries.pdf [12 May, 2004].

Ito T; Ohyama K; Kusano T; Usuda Y; Nozawa A; Hayashi H; Ohji H; Kitamura H; Kanisawa M. (1997)
       Pulmonary endocrine cell hyperplasia and papilloma in rats induced by intratracheal injections of
       extract from particulate air pollutants. Exp Toxicol Pathol. 49:65-70.

Jaffe DH; Singer ME;  Rimm AA. (2003) Air pollution and emergency department visits for asthma among
       Ohio Medicaid recipients, 1991-1996. Environ Res. 91:21-8.
May 2008                                 14                 DRAFT—DO NOT QUOTE OR CITE

-------
Jakab GJ; Clarke RW; Hemenway DR; Longphre MV; KleebergerSR; Frank R. (1996) Inhalation of acid
       coated carbon black particles impairs alveolar macrophage phagocytosis. Toxicology letters.
       88:243-248.

Jaklevic JM; Loo BW; Fujita TY. (1981) Automatic participate sulfur measurements with a dichotomous
       sampler and on-line x-ray fluorescence analysis. Environmental Science and Technology. 15:687-
       690.

Jalaludin  B; Morgan G; Lincoln D; Sheppeard V; Simpson R; Corbett S. (2006) Associations between
       ambient air pollution and daily emergency department attendances for cardiovascular disease in
       the  elderly (65+ years), Sydney, Australia. J Expo Sci Environ Epidemiol. 16:225-37.

Jedrychowski W. (1999) Ambient air pollution and respiratory health  in the east Baltic region. Scand J
       Work Environ Health. 25 Suppl 3:5-16.

Jerrett M; Burnett RT; Goldberg MS; Sears M; Krewski D; Catalan R; Kanaroglou P; Giovis C; Finkelstein
       N. (2003) Spatial analysis for environmental health research: concepts, methods, and examples.
       (Special issue on Health and air quality: interpreting science for decision makers.). Journal of
       Toxicology and Environmental Health Part A Taylor & Francis, London, UK: 2003 66: 16/19, 1783-
       1810.

Jimenez P;  Baldasano JM; Dabdub D. (2003) Comparison of photochemical mechanisms for air quality
       modeling. Atmospheric Environment. 37:4179-4194.

Johnson CW; Williams WC; Copeland CB; DeVito MJ; Smialowicz RJ. (2000) Sensitivity of the SRBC
       PFC assay versus ELISA for detection of immunosuppression by TCDD and TCDD-like
       congeners. Toxicology. 156:1-11.

Johnson T. (2002) A guide to selected algorithms, distributions, and databases used in exposure models
       developed by the Office of Air Quality Planning and  Standards. U.S Environmental Protection
       Agency.

Junge CE; Chagnon  CW; Manson  JE. (1961) Stratospheric aerosols. Journal of Meteorol. 18:81-108.

Kasibhatla P; Chameides WL. (2000) Seasonal modeling of regional ozone pollution in the eastern United
       States. Geophysical Research Letters. 27:1415-1418.

Katsouyanni K; Touloumi G; Spix C; Schwartz J; Balducci F; Medina S; Rossi G; Wojtyniak B; Sunyer J;
       Bacharova L; Schouten JP; Ponka A; Anderson HR. (1997) Short term effects of ambient sulphur
       dioxide and particulate matter on mortality in 12 European cities: results from time series data
       from the APHEA project. British Medical Journal. 314:1658-1663.

Keatinge  WR; Donaldson GC. (2006) Heat acclimatization and sunshine cause false indications of
       mortality due to ozone. Environ Res. 100:387-93.

Keck L; Wittmaack K. (2005) Effect of filter type and temperature on volatilisation losses from ammonium
       salts in aerosol matter. Atmospheric Environment. 39:4093-4100.

Keene WC;  Jacob DJ; Fan S-M. (1996) Reactive chlorine: a potential sink for dimethylsulfide and
       hydrocarbons in the marine boundary layer. Atmospheric Environment. 306:i-iii.

Keene WC;  Pszenny AAP; Maben  JR; Sander R. (2002) Variation of marine aerosol acidity with particle
       size. Geophysical Research  Letters. 297.
May 2008                                 15                DRAFT—DO NOT QUOTE OR CITE

-------
Keene WC; PszennyAAP; Maben JR; Stevenson E; Wall A. (2004) Closure evaluation of size-resolved
       aerosol pH in the New England coastal atmosphere during summer. Journal of Geophysical
       Research. 109D23307.

Kesten S; Szalai J; Dzyngel B. (1995) Air quality and the frequency of emergency room visits for asthma.
       Ann Allergy Asthma Immunol. 74:269-73.

Kilic D. (2003) The effects of ageing and sulfur dioxide inhalation exposure on visual-evoked potentials,
       antioxidant enzyme systems, and lipid-peroxidation levels of the brain and eye. Neurotoxicol
       Teratol. 25:587-98.

Kinney PL; Ozkaynak H. (1991) Associations of daily mortality and air pollution in Los Angeles County.
       Environmental research. 54:99-120.

Kittelson DB; McKenzie R; Vermeersch M; Dorman F; Pui D; Linne M; Liu B; Whitby K. (1978) Total sulfur
       aerosol concentration with an electrostatically pulsed flame photometric detector system.
       Atmospheric Environment. 12:105-111.

Klein RG; Janowsky  I; Schmezer P; Hermann R; Spiegelhalder B; Zeller WJ;  Pool BL. (1989) Effect of
       long-term inhalation of N-nitroso-dimethylamine (NDMA) and SO2/NOX in rats. Exp Pathol.
       37:273-280.

Klemm RJ;  Mason RM, Jr. (2000) Aerosol Research and Inhalation Epidemiological Study (ARIES): air
       quality and daily mortality statistical modeling-interim results. J Air Waste Manag Assoc. 50:1433-
       9.

Klemm RJ;  Lipfert FW; Wyzga RE; Gust C. (2004) Daily mortality and air pollution in Atlanta: two years of
       data from ARIES. Inhal Toxicol. 16 Suppl 1:131-41.

Knorst MM; Kienast K; Riechelmann H; Muller-Quernheim  J; Ferlinz R. (1994) Effect of sulfur dioxide on
       mucociliary activity and ciliary beat frequency in guinea pig trachea. International Archives of
       Occupational and Environmental Health. 65:325-328.

KohlhammerY; Rzehak P; Behr J; Wichmann HE; Heinrich J. (2007) High prevalence of pneumonia in
       children of a  smelter town. International Journal of Occupational and  Environmental Health
       13(2)(pp 167-174), 2007 Date of Publication: Apr 2007.

Koken PJ; PiverWT;  Ye F; ElixhauserA; Olsen LM; PortierCJ. (2003) Temperature, air pollution, and
       hospitalization for cardiovascular diseases among  elderly people in Denver. Environ Health
       Perspect. 111:1312-7.

Koksal  N; Yildirim Z;  Gokirmak M; Hasanoglu HC; Mehmet N; Avci H. (2003) The role of nitric oxide and
       cytokines in asthma-like syndrome induced by sulfur dioxide exposure in agricultural
       environment. Clinica Chimica Acta. 336:115-122.

Kopp MV; Bohnet W; FrischerT;  UlmerC; Studnicka M; IhorstG; Gardner C;  Forster J; Urbanek R; Kuehr
       J. (2000) Effects of ambient ozone on lung function in children over a two-summer period.
       European Respiratory Journal.  16:893-900.

Korhonen P; Kulmala M; Laaksonen A;  Viisanen Y; McGraw R; Seinfeld JH. (1999) Ternary nucleation of
       H2SO4, NH3,  and H2O in the atmosphere. Journal of Geophysical Research. 104:26,349-26,353.

Kotesovec F; Skorkovsky J; Brynda J; Peters A; Heinrich J. (2000) Diurnal  mortality and air pollution in
       the  North Bohemia. Casopis Lekaru Ceskych 2000 139: 6, 166-173 28 ref.
May 2008                                  16                DRAFT—DO NOT QUOTE OR CITE

-------
Health Effects Institute. (2000) Reanalysis of the Harvard Six Cities study and the American Cancer
       Society study of particulate air pollution and mortality: a special report of the Institute's Particle
       Epidemiology Reanalysis Project. Special report.

Kreyling WG; Perron GA; Furst G; Heilmann P; Neuner M; Ruprecht L; Schumann G; Takenaka S; Heyder
       J. (1992) Early response of the canine respiratory tract following long-term exposure to a
       sulfur(IV) aerosol at low concentration. III. Macrophage-mediated long-term particle clearance.
       Inhalation toxicology. 4:197-233.

Kulmala M; Pirjola L; Makela JM. (2000) Stable sulphate clusters as a source of new atmospheric
       particles. Nature. 404:66-69.

Kumar N; Russell AG. (1996) Development of a computationally efficient, reactive subgrid-scale plume
       model and the impact in the northeastern United States using increasing levels of chemical detail.
       Journal of Geophysical Research. 101:16,737-16,744.

Kwon HJ; Cho SH; Nyberg  F; Pershagen G. (2001) Effects of ambient air pollution on daily mortality in a
       cohort of patients with  congestive heart failure. Epidemiology. 12:413-9.

Lam HF; Peisch R; Amdur MO. (1982) Changes in lung volumes and diffusing capacity in guinea pigs
       exposed to a combination of sulfur dioxide and submicron zinc oxide mixed in a humidified
       furnace. Toxicology and Applied Pharmacology. 66:427-433.

Langley-Evans S. (2007) [Letter to Annette lannucci on sulfur oxides  concentrations].

Langley-Evans SC; Phillips GJ; Jackson AA. (1996)  Sulphur dioxide:  a potent glutathione depleting agent.
       Comp Biochem Physiol C Pharmacol Toxicol Endocrinol. 114:89-98.

Langley-Evans SC; Phillips GJ; Jackson AA. (1997)  Fetal exposure to low protein maternal diet alters the
       susceptibility of young adult rats to sulfur dioxide-induced lung injury. Journal of Nutrition.
       127:202-209.

Lazarus SC; Wong HH; Watts MJ; Boushey HA; Lavins BJ; Minkwitz  MC. (1997) The leukotriene receptor
       antagonist zafirlukast inhibits  sulfur dioxide-induced bronchoconstriction in patients with asthma.
       AmJ Respir Grit Care Med. 156:1725-30.

Le Tertre A; Quenel P; Eilstein  D; Medina S; Prouvost H; Pascal L; Boumghar A; Saviuc P; Zeghnoun A;
       Filleul L; Declercq C; Cassadou S; Le Coaster C. (2002) Short-term effects of air pollution on
       mortality in nine French cities: a quantitative summary. Arch Environ Health. 57:311-9.

Lee IM; Tsai SS; Chang CC; Ho CK; Yang CY. (2007) Air pollution and hospital admissions for chronic
       obstructive pulmonary disease in a tropical city: Kaohsiung, Taiwan. Inhal Toxicol. 19:393-8.

Lee J-T; Shin D; Chung Y (1999) Air pollution and daily mortality in Seoul and Ulsan, Korea.
       Environmental health perspectives. 107:149-154.

Lee J; Schwartz J. (1999) Reanalysis of the effects of air pollution on daily mortality in Seoul, Korea:  a
       case-crossover design. Environmental Health Perspectives National Institute of Environmental
       Health  Sciences, Research Triangle Park, USA: 1999 107: 8, 633-636 17 ref.

Lee JT; Kim H;  Song H; Hong YC; Cho YS; Shin SY; Hyun YJ; Kim YS. (2002) Air pollution and asthma
       among children in Seoul, Korea. Epidemiology. 13:481-4.
May 2008                                 17                 DRAFT—DO NOT QUOTE OR CITE

-------
Lee JT; Kim H; Cho YS; Hong YC; Ha EH; Park H. (2003) Air pollution and hospital admissions for
       ischemic heart diseases among individuals 64+ years of age residing in Seoul, Korea. Arch
       Environ  Health. 58:617-23.

Lee SL; Wong WH; Lau YL. (2006) Association between air pollution and asthma admission among
       children in Hong Kong. Clin Exp Allergy.  36:1138-46.

Leem JH; Kaplan BM; Shim YK; Pohl HR; Gotway CA; Bullard SM; Rogers JF; Smith MM; Tylenda CA.
       (2006) Exposures to air pollutants during pregnancy and preterm delivery. Environ Health
       Perspect. 114:905-10.

Lemos M; Lichtenfels AJFC; Amaro E, Jr.; Macchione M; Martins MA; King M; Bohm GM; Saldiva PHN.
       (1994) Quantitative pathology of nasal passages in rats exposed to urban levels  of air pollution.
       Environmental research. 66:87-95.

Lepeule J; Rondeau V; Filleul L; Dartigues JF. (2006) Survival analysis to  estimate association between
       short-term mortality and air pollution. Environ Health  Perspect. 114:242-7.

Leung FT, A. J. Colussi, M. R. Hoffmann, and G.  C. Toon. (2002) Isotopic fractionation of carbonyl sulfide
       in the atmosphere: implications for the source of background stratospheric sulfate aerosol.
       Geophysical Research Letters. 29.

Levine JS; Bobbe T; Ray N; Witt RG; Singh A. (1999) Wildland fires and the environment: a global
       synthesis.

Lewis AJ; KirchnerT (1984) Modulation of sulfur dioxide-induced airways hyperresponsiveness in the
       conscious dog. IntArch Allergy Appl Immunol. 75:188-90.

Li R; Meng Z. (2007) Effects of SO2 derivatives on expressions of MUC5AC and IL-13 in  human bronchial
       epithelial cells. Arch Toxicol. 81:867-74.

Liao D; Duan Y;  Whitsel EA; Zheng  ZJ; Heiss G; Chinchilli VM; Lin  HM. (2004) Association of higher
       levels of ambient criteria pollutants with impaired cardiac autonomic control: a population-based
       study. Am J Epidemiol. 159:768-77.

Liebhart J; Malolepszy J; Wojtyniak B; Pisiewicz  K; Plusa T; Gladysz U; Chyrek-Borowska S; Siergiejko Z;
       Rogalewska A; Szymanski W; Dziedziczko A; Banach-Wawrzynczak E; Tlappa J; Kurek M;
       Malaczynska T; Grubska-Suchanska E; Lademan A;  Szczeklik A; Sladek K; Bochenek G;
       Duplaga M; Milanowski J; Trebas-Pietras E; Kuna P;  Elgalal A; Kuprys I; Alkiewicz J; Breborowicz
       A; Mlynarczyk W; Silny W; Kurzawa R; Sak I; Kruszewski J; Chmielewska D; Droszcz W; Zaras
       E; Boznanski A; Jankowska R; Nittner-Marszalska M; Machaj G; Wrzyszcz M; Balinska W;
       Rogala E;  Rogala B; Gawlik R; Dobek R; Obojski A. (2007) Prevalence and risk factors for
       asthma  in Poland: Results from the PMSEAD study. Journal of Investigational Allergology and
       Clinical  Immunology 17:367-374.

Lin CM; Li CY; Yang GY; Mao IF. (2004a) Association between maternal exposure to elevated ambient
       sulfur dioxide  during pregnancy and term low birth weight.  Environ Res. 96:41-50.

Lin M; Chen Y; Burnett RT; Villeneuve PJ; Krewski D. (2003)  Effect of short-term exposure to gaseous
       pollution on asthma hospitalisation in children: a bi-directional case-crossover analysis. J
       Epidemiol Community Health. 57:50-5.

Lin M; Chen Y; Villeneuve PJ; Burnett RT; Lemyre L; Hertzman C; McGrail KM; Krewski D. (2004b)
       Gaseous air pollutants and asthma hospitalization of children with low household income in
       Vancouver, British Columbia, Canada. Am J Epidemiol. 159:294-303.
May 2008                                  18                DRAFT—DO NOT QUOTE OR CITE

-------
Lin M; Stieb DM; Chen Y. (2005) Coarse Particulate Matter and Hospitalization for Respiratory Infections
       in Children Younger Than 15 Years in Toronto: A Case-Crossover Analysis. Pediatrics. 116:e235-
       240.

Lipfert FW; Morris SC; Wyzga RE. (2000a) Daily mortality in the Philadelphia metropolitan area and size-
       classified particulate matter. J Air Waste Manag Assoc. 50:1501-13.

Lipfert FW; Perry HM, Jr.; Miller JP; Baty JD; Wyzga RE; Carmody SE. (2000b) The Washington
       University-EPRI Veterans' Cohort  Mortality Study: preliminary results. Inhal Toxicol. 12 Suppl
       4:41-73.

Lippmann M. (2000) Sulfur oxides: acidic aerosols and SO2. Environmental toxicants: human exposures
       and their health  effects John Wiley and Sons, New York, USA: 2000 Ed2, 771-809 7 pp of ref.

Liu H; Jacob DJ; Bey  I; Yantosca RM. (2001) Constraints from 210Pb and 7Be on wet deposition and
       transport in a  global three-dimensional chemical tracer model driven by assimilated
       meteorological fields. Journal of Geophysical Research. 106:12,109-12,128.

Llorca J; Salas A; Prieto-Salceda D; Chinchon-Bengoechea V; Delgado-Rodriguez M. (2005) Nitrogen
       dioxide increases cardiorespiratory admissions in Torrelavega (Spain). J Environ Health.  68:30-5.

Lovati MR; Manzoni C; Daldossi M; Spolti  S; Sirtori CR. (1996) Effects of sub-chronic exposure to SO2 on
       lipid and carbohydrate metabolism in rats. Arch Toxicol. 70:164-73.

Low RB; Bielory L; QureshiAl; Dunn V; Stuhlmiller DF; Dickey DA. (2006) The relation of stroke
       admissions to recent weather, airborne allergens, air pollution, seasons, upper respiratory
       infections, and asthma incidence,  September 11, 2001, and day of the week. Stroke. 37:951-7.

Lu R; Turco RP; Jacobson MZ. (1997) An  integrated air pollution modeling system for urban and  regional
       scales: 1. Structure and performance. Journal of Geophysical Research. 102:6063-6079.

Luginaah IN; Fung KY; Gorey KM; Webster G; Wills C. (2005) Association of ambient air pollution with
       respiratory hospitalization in a government designated area of concern: the case of Windsor,
       Ontario. Environmental health perspectives. 113:290-296.

Mahrt L. (1998) Stratified atmospheric boundary layers and breakdown of models. Theor Comput Fluid
       Dyn. 11:263-279.

Maier K; Beck-Speier  I; Dayal N; Heilmann P; Hinze H; Lenz A-G; Leuschel L; Matejkova E;  Miaskowski
       U; J. H. (1992) Early response of the canine respiratory tract following long-term exposure to a
       sulfur (IV) aerosol at low concentration. II. Biochemistry and cell biology of lung lavage fluid.
       Inhal Toxicol.  4:175-195.

Maisonet M; Bush TJ; Correa A; Jaakkola  JJ. (2001) Relation between ambient air pollution  and low birth
       weight in the Northeastern United  States. Environ Health Perspect. 109 Suppl 3:351-6.

MannixRC; Phalen RF;  Kenoyer JL; Crocker TT. (1982) Effect of sulfur dioxide-sulfate exposure  on rat
       respiratory tract clearance. Journal of occupational and environmental hygiene.  43:679-685.

Martins CR; de Andrade JB. (2002) Atmospheric chemistry of sulfur (IV):  Emissions, aqueous phase
       reactions and environmental effects. Quimica Nova. 25:259-272.

Matsumi Y; Shigemori H; Takahashi K. (2005) Laser-induced fluorescence instrument for measuring
       atmospheric SO2. Atmospheric Environment. 39:3177-3185.
May 2008                                 19                 DRAFT—DO NOT QUOTE OR CITE

-------
Air Pollution Health Effects Laboratory. (1988) Respiratory effects of acid containing multicomponent
       pollutant atmospheres. California Air Resources Board. Final Report.

McCurdy T. (2000) Conceptual basis for multi-route intake dose modeling using an energy expenditure
       approach. J Expo Anal Environ Epidemiol. 10:86-97.

McCurdy T; Glen G; Smith L; Lakkadi Y. (2000) The national exposure research laboratory's consolidated
       human activity database. J Expo Anal Environ Epidemiol. 10:566-78.

McDonnell WF; Abbey DE; Nishino N; Lebowitz MD. (1999) Long-term ambient ozone concentration and
       the incidence of asthma in nonsmoking adults: the Ahsmog study. Environmental research.
       80:110-121.

McKenna DS; Konopka P; Grooss J-U; Gunther G; Muller R; Spang R; Offermann D; Orsolini Y. (2002) A
       new chemical Lagrangian model of the stratosphere (CLaMS) 1. formulation of advection and
       mixing. Journal of Geophysical Research. 107.

Mebust MR; Eder BK; Binkowski FS; Roselle SJ.  (2003) Models-3 community multiscale air quality
       (CMAQ) model aerosol component. 2. Model evaluation. Journal of Geophysical Research.
       108D6.

Meng Z; Zhang B; Ruan A; Sang N; Zhang J. (2002) Micronuclei induced by sulfur dioxide inhalation in
       mouse bone-marrow cells in vivo. Inhalation Toxicology. 14:303-309.

Meng Z; Geng H; Bai J; Yan G. (2003a) Blood pressure of rats lowered by sulfur dioxide and its
       derivatives. Inhal Toxicol. 15:951-9.

Meng Z; Qin G; Zhang B; Geng H; Bai Q; Bai W; Liu C. (2003b) Oxidative damage of sulfur dioxide
       inhalation on lungs and hearts of mice. Environ Res. 93:285-92.

Meng Z; Bai W. (2004) Oxidation damage of sulfur dioxide on testicles of mice.  Environ Res. 96:298-304.

Meng Z; Liu Y; Wu D. (2005a) Effect of Sulfur Dioxide Inhalation on Cytokine Levels in Lungs and Serum
       of Mice. Inhalation Toxicology. 17:303-307.

Meng Z; Qin G; Zhang B. (2005b) DMA damage in mice treated with sulfur dioxide by inhalation.
       Environmental and Molecular Mutagenesis. 46:150-155.

Menzel DB; Keller DA; Leung K-H. (1986) Covalent reactions in the toxicity of SO2 and sulfite. Adv Exp
       Med Biol 197:477-92.

Metzger KB; Tolbert PE; Klein M; Peel JL; Flanders WD; Todd K; Mulholland JA; Ryan PB; Frumkin  H.
       (2004) Ambient air pollution and cardiovascular emergency department visits. Epidemiology.
       15:46-56.

Metzger KB; Klein M; Flanders WD; Peel JL; Mulholland JA;  Langberg JJ; Tolbert PE. (2007) Ambient air
       pollution and cardiac arrhythmias in patients with implantable defibrillators. Epidemiology. 18:585-
       92.

Michaud J-P; Grove JS;  Krupitsky D. (2004) Emergency department visits and "vog"-related air quality in
       Hilo, Hawai'i. Environmental Research. 95:11-19.

Michelozzi P; Forastiere F; Fusco D; Perucci CA;  Ostro B; Ancona C; Pallotti G. (1998) Air pollution and
       daily mortality in Rome, Italy. Occup Environ Med. 55:605-10.
May 2008                                 20                DRAFT—DO NOT QUOTE OR CITE

-------
Milner, J. T.; Dimitroulopoulou, C.; ApSimon, H. (2005) Indoor concentrations in buildings from sources
       outdoors. In: Atmospheric Dispersion Modelling Liaison Committee Annual Report 2004-2005,
       Annex B. Available: http:www.admlc.org.uk/ar04-05.htm [6 March, 2007].

Min B; Paul WE. (2008) Basophils and type 2 immunity. CurrOpin Hematol.  15:59-63.

Mohorovic L. (2004) First two months of pregnancy - critical time for preterm delivery and low birthweight
       caused by adverse effects of coal combustion toxics. Early Human Development Elsevier
       Science Ltd, Oxford, UK: 2004 80: 2, 115-123.

Moolgavkar SH; Luebeck EG; Hall TA; Anderson EL. (1995) Air pollution and daily mortality in
       Philadelphia. Epidemiology. 6:476-84.

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

Moolgavkar SH. (2003a) Air pollution and daily deaths and hospital admissions in Los Angeles and Cook
       counties. Revised Analyses of Time-Series Studies of Air Pollution and Health Special Report
       Boston:  Health Effects Institute.183-198.

Moolgavkar SH. (2003b) Air pollution and daily mortality in two U.S. counties: season-specific analyses
       and exposure-response relationships. Inhal Toxicol. 15:877-907.

Morris RD; Naumova EN; Munasinghe RL. (1995) Ambient air pollution and hospitalization for congestive
       heart failure among elderly people in seven large US cities. Am J Public Health. 85:1361-1365.

Mortimer KM;  Neas LM; Dockery DW; Redline S; Tager IB. (2002) The effect of air pollution on inner-city
       children with asthma. EurRespir J. 19:699-705.

Myhre G; Berglen TF; Myhre CE; Isaksen ISA. (2004) The radiative effect of the anthropogenic influence
       on the stratospheric sulfate aerosol layer. Tellus Ser B. 56:294-299.

Nadziejko C; Fang K; Narciso S; Zhong M; Su WC; Gordon T; Nadas A; Chen LC. (2004) Effect of
       particulate and gaseous pollutants on spontaneous arrhythmias in aged rats. Inhal Toxicol.
       16:373-80.

Nafstad P; Haheim  LL; Oftedal B; Gram F; Holme I; Hjermann I; Leren P. (2003) Lung cancer and air
       pollution: a  27 year follow up of 16 209 Norwegian men. Thorax. 58:1071-6.

Nafstad P; Haheim  LL; Wisloff T; Gram F; Oftedal B; Holme I; Hjermann I; Leren P. (2004) Urban air
       pollution and mortality in a cohort of Norwegian men. Environ Health Perspect. 112:610-5.

Neukirch F; Segala C; Le Moullec Y; Korobaeff M; Aubier M.  (1998) Short-term effects of low-level winter
       pollution on respiratory health of asthmatic adults. Arch Environ Health. 53:320-8.

Newhouse CP; Levetin E. (2004) Correlation of environmental factors with asthma and rhinitis symptoms
       in Tulsa, OK. Ann Allergy Asthma Immunol. 92:356-66.

Nie A; Meng Z. (2005) Study of the interaction of sulfur dioxide derivative with cardiac sodium channel.
       Biochim Biophys Acta. 1718:67-73.

Nie A; Meng Z. (2006) Modulation of L-type calcium current in rat cardiac myocytes by sulfur dioxide
       derivatives. Food Chem Toxicol. 44:355-63.
May 2008                                 21                 DRAFT—DO NOT QUOTE OR CITE

-------
Morris G; YoungPong SN; Koenig JQ; Larson TV; Sheppard L; Stout JW. (1999) An association between
       fine particles and asthma emergency department visits for children in Seattle. Environ Health
       Perspect. 107:489-93.

NTP. (2005) National Toxicology Program's Report on Carcinogens. National Institute of Environmental
       Health Sciences, National Institutes of Health.

Odman, M. T; Ingram, C. I. (1996) Multiscale air quality simulation platform (MAQSIP):  source code
       documentation and validation. Research Triangle Park, NC: MCNC; technical report ENV-
       96TR002-v1.0. Available: http://www.ce.gatech.edu/~todman/maqsip.pdf (9 September 2003).

Odman MT; Russell AG. (2000) Mass conservative coupling of non-hydrostatic meteorological models
       with air quality models. Air Pollution Modeling and its Application XIII.651-660.

Oftedal B;  Nafstad P; Magnus P; Bj0rkly S; Skrondal A. (2003) Traffic related air pollution  and acute
       hospital admission for respiratory diseases in Drammen, Norway 1995-2000. European Journal
       of Epidemiology. 18:671-676.

Ostro B; Sanchez JM; Aranda C; Eskeland GS. (1996) Air pollution and mortality: results from a study of
       Santiago, Chile.

Ott WR. (1982) Concepts of human exposure to air pollution. Environment International. 7:179-196.

Park H; Lee B; Ha EH; Lee JT;  Kim H; Hong YC. (2002) Association of air pollution with school
       absenteeism due to illness. Arch PediatrAdolesc Med.  156:1235-9.

Park JK; Kim YK; Lee SR; Cho SH; Min KU; Kim YY. (2001 a) Repeated exposure to low levels of sulfur
       dioxide (SO2) enhances the development of ovalbumin-induced asthmatic reactions  in guinea
       pigs. Ann Allergy Asthma Immunol. 86:62-7.

Park RJ; Stenchikov GL; Pickering; Dickerson RR; Allen DJ; Kondragunta S. (2001 b) Regional air
       pollution and its radiative forcing: studies with a single column chemical and radiation transport
       model. Journal of Geophysical Research. 106:28,751-28,770.

Park SK; O'Neill MS; Vokonas PS; Sparrow D; Schwartz J. (2005)  Effects of air pollution on heart rate
       variability: the VA normative aging study. Environ Health Perspect. 113:304-9.

Peacock JL; Symonds P; Jackson P; BremnerSA; Scarlett JF;  Strachan DP; Anderson  HR. (2003) Acute
       effects of winter air pollution on respiratory function in schoolchildren in southern England. Occup
       Environ  Med. 60:82-89.

Peel JL; Tolbert PE; Klein M; Metzger KB; Flanders WD; Todd K; Mulholland JA; Ryan PB; Frumkin H.
       (2005) Ambient  air pollution and respiratory emergency department visits. Epidemiology. 16:164-
       74.

Peel JL; Metzger KB; Klein M; Flanders WD; Mulholland JA; Tolbert PE. (2007) Ambient air pollution and
       cardiovascular emergency department visits in potentially sensitive groups. Am J  Epidemiol.
       165:625-33.

Penard-Morand C; SchillingerC; Armengaud A; Debotte G; Chretien E; PellierS; Annesi-Maesano I.
       (2006) Assessment of schoolchildren's exposure to traffic-related air pollution in the French Six
       Cities Study using a dispersion model. Atmospheric Environment 40(13)(pp 2274-2287), 2006
       Date of Publication: Apr 2006.
May 2008                                 22                 DRAFT—DO NOT QUOTE OR CITE

-------
Pereira PM; Saldiva PHN; Sakae RS; Bohm GM; Martins MA. (1995) Urban levels of air pollution increase
       lung responsiveness in rats. Environmental research. 69:96-101.

Peters A; Perz S; Doring A; Stieber J; Koenig W; Wichmann HE. (1999) Increases in heart rate during an
       air pollution episode. Am J  Epidemiol. 150:1094-8.

Peters A; Skorkovsky J; Kotesovec F; Brynda J; Spix C; Wichmann HE; Heinrich J. (2000) Associations
       between mortality and air pollution in central Europe. Environ Health Perspect. 108:283-7.

Peters A; Dockery DW; Muller JE; Mittleman MA. (2001) Increased particulate air pollution and the
       triggering of myocardial infarction. Circulation. 103:2810-5.

PetroeschevskyA; Simpson RW; Thalib L; Rutherford S. (2001) Associations between outdoor air
       pollution and hospital admissions in  Brisbane, Australia. Arch Environ  Health. 56:37-52.

Petruzzi S; Dell'Omo G; Fiore M; Chiarotti F; Bignami G; Alleva E. (1996)  Behavioural disturbances in
       adult CD-1  mice and absence of effects on their offspring upon SO2 exposure. Archives of
       Toxicology. 70:757-766.

Phalen RF; Kleinman MT. (1987) 21-day exposure to mixed air pollutants: effects on lung airways and
       macrophages. California Air Resources Board.

Pikhart H; Bobak M; Gorynski P; Wojtyniak B; Danova J; Celko MA; Kriz B; Briggs D; Elliot P. (2001)
       Outdoor sulphur dioxide and respiratory symptoms in Czech and Polish school children: a small-
       area study (SAVIAH). International Archives of Occupational and Environmental Health. 74:574-
       578.

Pino P; Walter!; Oyarzun M; Villegas R; Romieu I. (2004) Fine particulate matter and wheezing illnesses
       in the first year of life. Epidemiology. 15:702-8.

Pinter A; Rudnai P; Sarkany E; Goczan  M; PaldyA. (1996) Air pollution and  childrens' respiratory
       morbidity in the  Tata area, Hungary.  Central  European Journal of Public Health. 4:17-20.

Pinto JP; Turco  RP; Toon OB. (1989) Self-limiting physical and chemical effects in volcanic eruption
       clouds. Journal  of Geophysical Research. 94:11,165-11,174.

Poloniecki JD; Atkinson  RW; de Leon AP; Anderson  HR. (1997) Daily time series for cardiovascular
       hospital admissions and previous day's air pollution in London, UK.  Occup Environ Med. 54:535-
       40.

Ponka A. (1990) Absenteeism and respiratory disease among children and adults in Helsinki in relation to
       low-level air pollution  and temperature. Environ Res. 52:34-46.

Ponka A. (1991) Asthma and low level air pollution in Helsinki. Archives of Environmental and
       Occupational Health.  46:262-270.

Ponka A; Virtanen M. (1994) Chronic bronchitis, emphysema, and low-level  air pollution in Helsinki, 1987-
       1989. Environmental research. 65:207-217.

Ponka A; Virtanen M. (1996) Low-level air pollution and hospital admissions for cardiac and
       cerebrovascular diseases in Helsinki. American journal of public health. 86:1273-1280.

Ponka A; Savela M; Virtanen M. (1998) Mortality and air pollution in  Helsinki. Arch Environ Health.
       53:281-6.
May 2008                                 23                 DRAFT—DO NOT QUOTE OR CITE

-------
Pool-Zobel BL; Schmezer P; Zeller WJ; Klein RG. (1990) In vitro and ex vivo effects of the air pollutants
       SO2 and NOX on benzo(a)pyrene activating enzymes of the rat liver. Exp Pathol. 39:207-212.

Pool BL; BrendlerS; Klein RG; Monarca S; Pasquini R; Schmezer P; Zeller WJ. (1988)  Effects of SO2 or
       NOX on toxic and genotoxic properties of chemical carcinogens. II. Short term in vivo studies.
       Carcinogenesis. 9:1247-1252.

Pope CA, 3rd; Thun MJ; Namboodiri MM; Dockery DW; Evans JS; Speizer FE; Heath CW, Jr. (1995)
       Particulate air pollution as a predictor of mortality in a prospective study of U.S. adults. Am J
       Respir Grit Care Med. 151:669-674.

Pope CA, 3rd; Burnett RT; Thun MJ; Calle EE; Krewski D; Ito K; Thurston GD. (2002) Lung cancer,
       cardiopulmonary mortality, and long-term exposure to fine particulate air pollution. JAMA.
       287:1132-41.

Possanzini M; De Santis F; Di Palo V. (1999) Measurements  of nitric acid and ammonium salts in lower
       Bavaria. Atmospheric Environment. 33:3597-3602.

Prescott GJ; Cohen GR; Elton RA; Fowkes FG; Agius RM. (1998) Urban air pollution  and
       cardiopulmonary ill health: a 14.5 year time series study. Occup Environ  Med. 55:697-704.

Price, P. S.; Chaisson, C. F.; Koontz, M.; Wilkes, C.; Ryan, B.; Macintosh, D.; Georgopoulos, P. G. (2003)
       Construction of a comprehensive chemical exposure framework using person oriented modeling.
       Prepared for: The Exposure Technical Implementation Panel, American Chemistry Council;
       contract #1338. Annandale, VA: The LifeLine Group. Available:
       http://www.thelifelinegroup.org/lifeline/docs.htm [6 March, 2007].

Pszenny AAP; Moldanova J; Keene WC; Sander R; Maben JR; Martinez M; Crutzen PJ; Perner D; Prinn
       RG. (2004) Halogen cycling and aerosol pH in the Hawaiian marine boundary layer. Atmospheric
       Chemistry and Physics. 4:147-168.

Qian Z; He Q; Lin HM; Kong L; Liao D; Yang N; Bentley CM; Xu S. (2007) Short-term effects of gaseous
       pollutants on cause-specific mortality in Wuhan, China. J Air Waste Manag Assoc. 57:785-93.

Qin G; Meng Z. (2005) Effect of sulfur dioxide inhalation on CYP1A1 and CYP1A2 in  rat liver and lung.
       Toxicology letters. 160:34-42.

RahlenbeckSI; Kahl H. (1996) Air pollution and mortality in East Berlin during the winters of 1981-1989.
       International journal of epidemiology. 25:1220-1226.

RamadourM; Burel C; Lanteaume A; Vervloet D; Charpin D;  Brisse F; Dutau H. (2000)  Prevalence of
       asthma and rhinitis in  relation to long-term exposure to  gaseous air pollutants. Allergy. 55:1163-9.

Raub JA; Miller FJ; Graham JA; Gardner DE; O'Neil JJ. (1983)  Pulmonary function in normal and
       elastase-treated hamsters exposed to a complex mixture of olefin-ozone-sulfur dioxide reaction
       products. Environmental research. 31:302-310.

Ravishankara AR. (1997) Heterogeneous and multiphase chemistry in the troposphere. Science.
       276:1058-1065.

ReithmeierC; Sausen R. (2002) ATTILA: atmospheric tracer transport in a Lagrangian model. Tellus.
       546:278-299.
May 2008                                 24                DRAFT—DO NOT QUOTE OR CITE

-------
Rich DQ; Schwartz J; Mittleman MA; Link M; Luttmann-Gibson H; Catalano PJ; Speizer FE; Dockery DW.
       (2005) Association of short-term ambient air pollution concentrations and ventricular arrhythmias.
       Am J Epidemiol. 161:1123-32.

Rich DQ; Kim MH; Turner JR; Mittleman MA; Schwartz J; Catalano PJ; Dockery DW. (2006a) Association
       of ventricular arrhythmias detected by implantable cardioverterdefibrillatorand ambient air
       pollutants in the St Louis, Missouri metropolitan area. Occup Environ Med. 63:591-6.

Rich DQ; Mittleman MA; Link MS; Schwartz J; Luttmann-Gibson H; Catalano PJ; Speizer FE; Gold DR;
       Dockery DW. (2006b) Increased risk of paroxysmal atrial fibrillation episodes associated with
       acute increases in ambient air pollution. Environ Health Perspect. 114:120-3.

Rich KE;  Petkau J; Vedal S; Brauer M. (2004) A case-crossover analysis of particulate air pollution and
       cardiac arrhythmia in patients with implantable cardioverterdefibrillators. Inhal Toxicol. 16:363-72.

Riechelmann H; Maurer J;  Kienast K; Hafner B;  Mann WJ. (1995) Respiratory epithelium exposure to
       sulfur dioxide-functional and ultrastructural alterations. Laryngoscope. 105:295-299.

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. 147:118-24.

Roemer W; HoekG;  Brunekreef B; Haluszka J; KalandidiA; Pekkanen J. (1998) Daily variations in air
       pollution and respiratory health in a multicentre study: the PEACE project. European Respiratory
       Journal. 12:1354-1361.

Roemer WH; van Wijnen JH. (2001)  Daily mortality and air pollution along busy streets in Amsterdam,
       1987-1998. Epidemiology. 12:649-53.

Romieu I; Meneses F; Ruiz S; Sienra JJ; Huerta J; White MC; Etzel RA. (1996) Effects of air pollution on
       the respiratory health of asthmatic children living in Mexico City. Am J Respir Grit Care Med.
       154:300-307.

Palma T.  (2005) The  HAPEM5 user's guide: hazardous air pollutant exposure model, version 5. U.S.
       Environmental Protection Agency: Office of Air Quality  Planning and Standards.

Ross MA; Persky VW; Scheff PA; Chung J; Curtis L; Ramakrishnan V; Wadden RA; Hryhorczuk DO.
       (2002) Effect of ozone and aeroallergens on the respiratory health of asthmatics. Archives of
       Environmental and Occupational Health. 57:568-578.

Rossi OV; Kinnula VL; Tienari J; Huhti E. (1993) Association of severe asthma attacks with weather,
       pollen, and air pollutants. Thorax. 48:244-8.

Ruidavets JB; Cassadou S; Cournot M; Bataille V; Meybeck M; Ferrieres J. (2005) Increased resting heart
       rate with pollutants in a population based study. J Epidemiol Community Health. 59:685-93.

Russell A; Dennis R. (2000) NARSTO critical review of photochemical models and modeling. Atmospheric
       Environment. 34:2283-2324.

Russell KM; Keene WC; Maben JR; Galloway JN; Moody JL. (2003) Phase-partitioning and dry
       deposition of atmospheric nitrogen at the mid-Atlantic U.S. coast. Journal of Geophysical
       Research. 108:021.

Rusznak  C; Devalia JL; Davies RJ. (1996) Airway response of asthmatic subjects to inhaled allergen after
       exposure to pollutants. Thorax. 51:1105-8.
May 2008                                 25                 DRAFT—DO NOT QUOTE OR CITE

-------
Saez M; Perez-Hoyos S; Tobias A; Saurina C; Barcelo MA; Ballester F. (1999) Time series methods in
       epidemiological studies on air pollution]. Rev Esp Salud Publica. 73:133-43.

SagivSK; Mendola P; Loomis D; Herring AH; Neas LM; Savitz DA; Poole C. (2005) A time-series analysis
       of air pollution and preterm birth in Pennsylvania, 1997-2001. Environ Health Perspect. 113:602-
       6.

Saiz-Lopez A; Plane JMC; Shillito JA. (2004) Bromine oxide in the mid-latitude marine boundary layer.
       Geophysical Research Letters. 31:103111.

Saldiva PH; Pope CA, 3rd; Schwartz J; Dockery DW; Lichtenfels AJ; Salge JM; Barone I; Bohm GM.
       (1995) Air pollution and mortality in elderly people: a time-series study in Sao Paulo, Brazil. Arch
       Environ Health. 50:159-63.

Saldiva PHN; King M;  Delmonte VLC; Macchione M; Parada MAC; Daliberto ML; Sakae RS; Criado PMP;
       Silveira PLP; Zin WA; Bohm  GM. (1992) Respiratory alterations due to urban air pollution: an
       experimental study in rats. Environmental research. 57:19-33.

Saldiva PHN; Lichtenfels AJFC; Paiva PSO; Barone IA; Martins  MA; Massad E; Pereira JCR; Xavier VP;
       Singer JM;  Bohm GM. (1994) Association between air pollution and mortality due to respiratory
       diseases in children in Sao Paulo, Brazil: a preliminary report. Environmental research. 65:218-
       225.

Samet, J. M.; Dominici, F.; Zeger, S.  L.; Schwartz, J.; Dockery, D. W. (2000a) National morbidity, mortality,
       and air pollution study. Part I: methods and methodologic issues. Cambridge, MA: Health Effects
       Institute; research report no.  94.

Samet, J. M.; Zeger, S. L.; Dominici,  F.; Curriero, F.; Coursac,  I.; Dockery, D. W; Schwartz, J.; Zanobetti,
       A. (2000b) The national morbidity, mortality, and air pollution study. Part II: morbidity, mortality,
       and air pollution in the United States. Cambridge, MA: Health Effects Institute; research report no.
       94, part II.

Sarnat SE; Coull BA; Schwartz J; Gold DR; Suh HH. (2006) Factors affecting the association between
       ambient concentrations and personal exposures to particles and gases.  Environmental health
       perspectives.  114:649-654.

Scala JM; Garstang M; Tao W-K; Pickering KE; Thompson AM; Simpson J; Kirchhoff VWJH; Browell EV;
       Sachse GW; L. TA; Gregory  GL; Rasmussen RW; Khalil MAK. (1990) Cloud draft structure and
       trace gas transport. Journal of Geophysical Research. 95:17,015-17,030.

Scanlon PD; Seltzer J; Ingram RH, Jr.; Reid L;  Drazen JM. (1987) Chronic exposure to sulfur dioxide.
       Physiologic and histologic evaluation of dogs exposed to 50 or 15 ppm. Am Rev Respir Dis.
       135:831-9.

Schildcrout JS; Sheppard L; Lumley T; Slaughter JC; Koenig JQ; Shapiro GG. (2006) Ambient Air
       Pollution and Asthma Exacerbations in Children: An Eight-City Analysis. Am J Epidemiol.
       164:505-517.

Schouten JP; Vonk JM; de Graaf A. (1996) Short term effects  of air pollution on emergency hospital
       admissions for respiratory disease: results of the APHEA project in two major cities in The
       Netherlands, 1977-89. In:  The APHEA project. Short term effects of air pollution on health: a
       European approach using epidemiological time series data.  J Epidemiol Community Health.
       50:S22-S29.
May 2008                                 26                 DRAFT—DO NOT QUOTE OR CITE

-------
Schulz H; EderG; Heilmann P; Ruprecht L; Schumann G; Takenaka S; J. H. (1992) Early response of the
       canine respiratory tract following long-term exposure to a sulfur (IV) aerosol at low concentration.
       IV. Respiratory lung function. Inhal Toxicol. 4:235-246.

Schwartz J. (1989) Lung function and chronic exposure to air pollution: a cross-sectional analysis of
       NHANES II. Environ Res. 50:309-21.

Schwartz J. (1991) Particulate air pollution and daily mortality in Detroit. Environmental research. 56:204-
       213.

Schwartz J; Dockery DW; Neas  LM; Wypij D; Ware JH; Spengler JD; Koutrakis P; Speizer FE; Ferris BG,
       Jr. (1994) Acute effects of summer air pollution on respiratory symptom reporting in children. Am J
       Respir Grit Care Med. 150:1234-42.

Schwartz J. (1995) Short term fluctuations in air pollution and hospital  admissions of the elderly for
       respiratory disease. Thorax. 50:531.

Schwartz J; Morris R. (1995) Air pollution and hospital admissions for cardiovascular disease in Detroit,
       Michigan. Am J Epidemiol.  142:23-35.

Schwartz J; SpixC; Touloumi G; Bacharova L; Barumamdzadeh T; le Tertre A; Piekarksi T; Ponce de Leon
       A; Ponka A; Rossi G; Saez M; Schouten JP. (1996) Methodological issues in studies of air
       pollution and daily counts of deaths or hospital admissions. J Epidemiol Community Health. 50
       Suppl 1:83-11.

Schwartz J. (1997) Air pollution and hospital admissions for cardiovascular disease in Tucson.
       Epidemiology. 8:371-7.

Schwartz J. (2000) Daily deaths are associated with combustion particles rather than SO(2) in
       Philadelphia. Occup Environ Med. 57:692-7.

Schwartz J. (2004) Is the association of airborne particles with daily deaths confounded by gaseous air
       pollutants? An approach to  control by matching. Environ Health Perspect. 112:557-61.

Seaman NL. (2000) Meteorological modeling for air quality assessments. Atmospheric Environment.
       34:2231-2259.

Segala C; FaurouxB; Just J; Pascual L; Grimfeld A; Neukirch F. (1998) Short-term effect of winter air
       pollution on respiratory health of asthmatic children in Paris. Eur Respir J.  11:677-685.

Seinfeld JH; Pandis SN. (1998b) Atmospheric Chemistry and Physics: From Air Pollution to Climate
       Change,  1326 pp. John  Wiley, New York.

Shami SG; Wolff RK; Hahn FF; Brooks AL; Griffith WC. (1985) Early cytokinetic and morphological
       response of rat lungs to inhaled benzo(a)pyrene, gallium oxide, and SO2. Environmental
       research. 37:12-25.

Sheppard L; Levy D; Morris G; Larson TV; Koenig  JQ. (1999) Effects of ambient air pollution on
       nonelderly asthma  hospital admissions in  Seattle, Washington, 1987-1994. Epidemiology. 10:23-
       30.

Sheppard, L. (2003) Ambient air pollution and nonelderly asthma hospital admissions in Seattle,
       Washington, 1987-1994. In: Revised analyses of time-series studies of air pollution and health.
       Special report. Boston, MA: Health Effects Institute; pp. 227-230. Available:
       http://www.healtheffects.org/Pubs/TimeSeries.pdf [18 October, 2004].
May 2008                                 27                 DRAFT—DO NOT QUOTE OR CITE

-------
Shi X. (1994) Generation of SO3" and OH radicals in SO32" reactions with inorganic environmental
       pollutants and its implications to SO32" toxicity. Journal of Inorg Biochemistry. 56:155-165.

Shi X; Mao Y. (1994) 8-hydroxy-2'-deoxyguanosine formation and DMA damage induced by sulfur trioxide
       anion radicals. Biochemical and Biophysical Research Communications. 205:141-147.

Sillman S. (1995) The use of NOy, H2O2 and HNO3 as indicators for ozone-NOx-hydrocarbon sensitivity in
       urban locations. Journal of Geophysical Research. 100:14,175-14,188.

Simpson RW; Williams G; Petroeschevsky A; Morgan G; Rutherford S. (1997) Associations between
       outdoor air pollution and daily mortality in Brisbane, Australia. Archives of Environmental and
       Occupational Health. 52:442-454.

Singh J. (1989) Neonatal development altered by maternal sulfur dioxide exposure. Neurotoxicology.
       10:523-7.

Smith AM; Keene WC; Maben JR; Pszenny AAP; Fischer E; Stohl A. (2007) Ammonia sources, transport,
       transformation, and deposition in coastal New England during summer. Journal of Geophysical
       Research.  112.

Soyseth V; Kongerud J; Broen P; Lilleng P; Boe J. (1995) Bronchial responsiveness, eosinophilia, and
       short term exposure to air pollution. Arch Dis Child. 73:418-22.

Spix C; Wichmann HE. (1996) Daily mortality and air pollutants: findings from Koln, Germany. Journal of
       epidemiology and community health. 1:S52-S58.

Spix C; Anderson HR; Schwartz J; Vigotti MA; LeTertre A; Vonk JM; Touloumi G; Balducci F; Piekarski T;
       Bacharova L; Tobias A; Ponka A; Katsouyanni K. (1998) Short-term effects of air pollution on
       hospital admissions of respiratory diseases in Europe: a quantitative summary of APHEA study
       results. Archives of Environmental and Occupational Health. 53:54-64.

Stein AF;  Lamb D; Draxler RR. (2000) Incorporation of detailed chemistry into a three-dimensional
       Lagrangian-Eulerian hybrid  model: application to regional tropospheric ozone. Atmospheric
       Environment. 34:4361-4372.

StenchikovG; Dickerson R; Pickering K; Ellis WJ; Doddridge B; Kondragunta S; Poulida O; Scala J; Tao
       W-K. (1996)  Stratosphere-troposphere exchange in a midlatitude mesoscale convective complex.
       2. Numerical simulation. Journal of Geophysical Research. 101:6837-6851.

Stevens RK; O'Keeffe AE; Ortman GC.  (1969) Absolute calibration of a flame photometric detector to
       volatile sulfur compounds at sub-part-per-million levels. Environmental Science and Technology.
       3:652-655.

Stevens RK; Mulik JD; O'Keefe AE;  Krost KJ. (1971) Gas chromatography of reactive sulfur gases in air
       at the parts-per-billion level. Analytical chemistry. 43:827-831.

Stevenson DS;  Dentener FJ; Schultz MG; Ellingsen K; van Noije TPC; Wild O; Zeng G; Amann M;
       Atherton CS; Bell N; Bergmann DJ; Bey I; Butler T; Cofala J; Collins WJ; Derwent RG; Doherty
       RM; Drevet J; Eskes HJ; Fiore AM; Gauss M; Hauglustaine DA; Horowitz LW; Isaksen ISA; Krol
       MC; Lamarque JF; Lawrence MG; Montanaro V; Mueller JF; Pitari G; Prather MJ; Pyle JA; Rast
       S; Rodriguez JM; Savage NH; Shindell DT; Strahan SE; Sudo K; Szopa S. (2006) Multimodel
       ensemble simulations of present-day and near-future tropospheric ozone. 111 :D08301
May 2008                                  28                DRAFT—DO NOT QUOTE OR CITE

-------
Stieb DM; Burnett RT; Beveridge RC; Brook JR. (1996) Association between ozone and asthma
       emergency department visits in Saint John, New Brunswick, Canada. Environ Health Perspect.
       104:1354-60.

Stieb DM; Beveridge RC; Brook JR; Smith-Doiron M; Burnett RT; Dales RE; Beaulieu S; Judek S;
       MamedovA. (2000) Air pollution, aeroallergens and cardiorespiratory emergency department
       visits in Saint John, Canada. J Expo Anal Environ Epidemiol. 10:461-77.

Stieb DM; Judek S; Burnett RT. (2002) Meta-analysis of time-series studies of air pollution and  mortality:
       effects of gases and particles and the influence of cause of death, age,  and season. J Air Waste
       Manag Assoc. 52:470-84.

Stieb DM; Judek S; Burnett RT. (2003) Meta-analysis of time-series studies of air pollution and  mortality:
       update in relation to the use of generalized additive models. J Air Waste Manag Assoc. 53:258-
       61.

Stockwell W; Middleton P; Chang J; Tang X. (1990) The second generation Regional Acid Deposition
       Model chemical mechanism for regional air quality modeling. Journal of Geophysical Research.
       95:16343-16367.

Studnicka M; Hackl E; Pischinger J; FangmeyerC; Haschke N; Kuhr J; Urbanek R; Neumann M; Frischer
       T. (1997) Traffic-related NO2 and the prevalence of asthma and respiratory symptoms in seven
       year olds. EurRespirJ. 10:2275-8.

Sunyer J; Anto JM; Murillo C; Saez M. (1991) Effects of urban  air pollution on emergency room
       admissions for chronic obstructive pulmonary disease. American Journal of Epidemiology.
       134:277-286.

Sunyer J; Saez M; Murillo C; Castellsague J; Martinez F; Anto  JM. (1993) Air pollution and emergency
       room admissions for chronic obstructive pulmonary disease: a 5-year study. Am J Epidemiol.
       137:701-5.

Sunyer J; Castellsague J; Saez M; Tobias A; Anto JM. (1996) Air pollution and mortality in Barcelona.
       Journal of epidemiology and community health. 1 :S76-S80.

Sunyer J; SpixC; Quenel P; Ponce-de-Leon A; Ponka A; Barumandzadeh T; Touloumi G; Bacharova L;
       Wojtyniak B; Vonk J; Bisanti L; Schwartz J; Katsouyanni  K. (1997) Urban air pollution and
       emergency admissions for asthma in four European cities: the APHEA Project. Thorax. 52:760-5.

Sunyer J. (2002) [Atmospheric pollution and mortality]. Med Clin  (Bare). 119:453-4.

Sunyer J; Atkinson R; Ballester F; Le Tertre A; Ayres JG; Forastiere F; Forsberg B; Vonk JM; Bisanti L;
       Anderson RH; Schwartz J; Katsouyanni K. (2003) Respiratory effects of sulphur dioxide: a
       hierarchical multicity analysis in the APHEA 2 study. Occup Environ Med. 60:e2.

Sutton MA; Dragosits U; Tang  YS; Fowler D. (2000) Ammonia  emissions from non-agricultural sources in
       the UK. Atmospheric Environment. 34:855-869.

Taggart SCO; Custovic A; Francis HC; Faragher EB; Yates CJ; Higgins BG; Woodcock A. (1996)
       Asthmatic bronchial hyperresponsiveness varies with ambient levels of  summertime air pollution.
       European Respiratory Journal. 9:1146-1154.

Takenaka S; Furst G; Heilman P; Heini A; Heinzmann U; Kreyling WG; Murray AB; Schulz H; Heyder J.
       (1992) Early response of the canine respiratory tract following long-term exposure to a sulfur (IV)
       aerosol at low concentration. V Morphology and morphometry. Inhal Toxicol. 4:247-272.
May 2008                                29                DRAFT—DO NOT QUOTE OR CITE

-------
Takenaka S; Godleski JJ; Heini A; Karg E; Kreyling WG; Ritter B; Schulz H; Ziesenis A; Heyder J. (1999)
       Health effects of sulfur-related environmental air pollution. V. Lung structure. Inhal Toxicol.
       11:439-54.

Tanaka H; Honma S; Nishi M; Igarashi T; Teramoto S; Nishio F; Abe S. (1998) Acid fog and hospital visits
       for asthma: an epidemiological study. Eur Respir J. 11:1301-6.

Tao W-K; Simpson J. (1993) The Goddard Cumulus Ensemble  Model. Part I: model description. Terr
       Atmos Oceanic Sci (TAO). 4:35-71.

Tenias JM; Ballester F; Rivera ML. (1998) Association between hospital emergency visits for asthma and
       air pollution in Valencia, Spain. Occup Environ Med. 55:541-7.

Tenias JM; Ballester F; Perez-Hoyos S; Rivera ML. (2002) Air pollution and hospital emergency room
       admissions for chronic obstructive pulmonary disease in Valencia, Spain. Arch Environ Health.
       57:41-7.

Thompson AJ; Shields MD; Patterson CC. (2001) Acute asthma exacerbations and air pollutants in
       children living in Belfast, Northern  Ireland. Arch Environ Health. 56:234-41.

Thornton DC; Bandy AR. (1993) Sulfur dioxide and dimethyl sulfide in the central Pacific troposphere.
       Journal of Atmospheric Chemistry. 17:1-13.

Thornton DC; Bandy AR; Blomquist BW; Anderson BE. (1996) Impact of anthropogenic and biogenic
       sources and sinks on carbonyl sulfide in the North Pacific troposphere. Journal of Geophysical
       Research. 101:1873-1881.

Thornton DCB, A. R.; Tu, F. H.; Blomquist, B. W; Mitchell,  G. M.; Nadler, W; Lenschow, D. H.  (2002) Fast
       airborne sulfur dioxide measurements by atmospheric pressure ionization mass spectrometry
       (APIMS). J Geophys Res [Atmos]. 107.

Tobias A; Campbell MJ; Saez M.  (1999) Modelling asthma epidemics on the relationship between air
       pollution and asthma emergency visits in Barcelona, Spain. European journal  of epidemiology.
       15:799-803.

Tolbert PE; Klein M; Peel JL; Sarnat SE; Sarnat JA. (2007) Multipollutant modeling issues in a study of
       ambient air quality and emergency department visits in Atlanta. J Expo Sci Environ Epidemiol. 17
       Suppl 2:S29-35.

Toumi R. (1994) BrO as  a sink for dimethylsulfide in the marine atmosphere. Geophysical Research
       Letters. 21:117-120.

Trenga CA; Koenig JQ; Williams PV (1999) Sulphur dioxide sensitivity and plasma antioxidants in adult
       subjects with asthma. Occup Environ Med. 56:544-7.

Trenga CA; Koenig JQ; Williams PV. (2001) Dietary antioxidants and ozone-induced bronchial
       hyperresponsiveness in adults with asthma. Arch Environ Health. 56:242-9.

Tsai SS; Huang CH; Goggins WB; Wu TN; Yang CY. (2003) Relationship between air pollution and daily
       mortality in a tropical city: Kaohsiung, Taiwan. J Toxicol Environ Health A. 66:1341-9.

Tsai SS; Cheng MH; Chiu HF; Wu TN; Yang CY. (2006) Air pollution and hospital admissions for asthma
       in a tropical city: Kaohsiung, Taiwan. Inhal Toxicol. 18:549-54.
May 2008                                 30                 DRAFT—DO NOT QUOTE OR CITE

-------
Tunnicliffe WS; Hilton MF; Harrison RM; Ayres JG. (2001) The effect of sulphur dioxide exposure on
       indices of heart rate variability in normal and asthmatic adults. European Respiratory Journal.
       17:604-608.

Turco RP; Toon OB; Whitten RC; Hamill P; Keesee RG. (1983) The 1980 eruptions of Mount St. Helens:
       physical and chemical processes in the stratospheric clouds. Journal of Geophysical  Research.
       88:5299-5319.

Turekian VC; Macko SA;  Keene WC.  (2001) Application of stable sulfur isotopes to differentiate sources
       of size-resolved particulate sulfate in polluted marine air at Bermuda during spring. Geophysical
       Research Letters. 28:1491-1494.

Van der Werf GR; Randerson JT; Collatz J; Giglio L. (2003) Carbon emissions from fires in tropical and
       subtropical ecosystems. Glob Change Biol. 9:547-562.

Vedal SR, K.; Brauer, M.; White, R.; Petkau, J. (2004) Air pollution and cardiac arrhythmias in patients
       with implantable cardiovascular defibrillators.  Inhalation Toxicol.  16:353-362.

Verhoeff AP; Hoek G; Schwartz J; van Wijnen JH. (1996) Air pollution and daily mortality in Amsterdam.
       Epidemiology. 7:225-30.

Vigotti MA; Rossi G; Bisanti L; Zanobetti A; Schwartz J. (1996) Short term effects of urban air pollution on
       respiratory  health in Milan, Italy, 1980-89. J Epidemiol Community Health. 50 Suppl 1:s71-5.

Villeneuve PJ; Doiron MS; Stieb D; Dales R; Burnett RT; Dugandzic R. (2006)  Is outdoor air pollution
       associated  with physician visits  for allergic rhinitis among the elderly in Toronto, Canada? Allergy.
       61:750-8.

Von Glasow S, R.; Bott A; Crutzen PJ. (2002) Modeling halogen chemistry in the marine boundary  layer.
       2. Interactions with sulfur and cloud-covered MBL. Journal of Geophysical  Research. 107.

Von Mutius E; Sherrill DL; Fritzsch C;  Martinez FD;  Lebowitz MD. (1995) Air pollution and  upper
       respiratory  symptoms in children from East Germany. European Respiratory Journal. 8:723-728.

Walters S; Griffiths  RK; Ayres JG. (1994) Temporal association between hospital admissions for asthma in
       Birmingham and  ambient levels of sulphur dioxide and smoke. Thorax. 49:133-140.

Wang TN; Ko YC; Chao YY; Huang CC; Lin RS. (1999) Association between indoor and outdoor air
       pollution  and adolescent asthma from 1995 to  1996 in Taiwan. Environ Res. 81:239-47.

Ward DJ; Roberts KT; Jones N; Harrison RM; Ayres JG; Hussain S; Walters S. (2002) Effects of daily
       variation  in outdoor particulates and ambient acid species in normal and asthmatic children
       (correction). Thorax. 57:752.

Warneck P. (1999) The relative importance of various pathways for the oxidation of sulfur dioxide and
       nitrogen dioxide in sunlit continental fair weather clouds. Phys Chem Chem Phys. 1:5471-5483.

NAPAP. (1991) The experience and legacy of NAPAP. Report of the Oversight Review Board. National
       Acid Precipitation Assessment Program. U.S. Environmental Protection Agency.

Wellenius GA; Bateson TF; Mittleman MA; Schwartz J. (2005a)  Particulate air pollution and the rate of
       hospitalization for congestive  heart failure among medicare beneficiaries in Pittsburgh,
       Pennsylvania. Am J Epidemiol.  161:1030-6.
May 2008                                 31                 DRAFT—DO NOT QUOTE OR CITE

-------
Wellenius GA; Schwartz J; Mittleman MA. (2005b) Air pollution and hospital admissions for ischemic and
       hemorrhagic stroke among medicare beneficiaries. Stroke. 36:2549-53.

Wesely ML; Hicks BB. (2000) A review of the current status of knowledge on dry deposition. Atmospheric
       Environment. 34:2261-2282.

Wheeler A; Zanobetti A; Gold DR; Schwartz J; Stone P; Suh HH. (2006) The relationship between
       ambient air pollution and heart rate variability differs for individuals with heart and pulmonary
       disease. Environ  Health Perspect. 114:560-6.

Whitfield, R. G.; Richmond, H. M.; Johnson, T. R. (1997) Overview of ozone human exposure and health
       risk analyses used in the U.S. EPA's review of the ozone air quality standard. In: Proceedings of
       the U.S.-Dutch International Symposium on Air Pollution in the 21st Century; April; Noordwijk, The
       Netherlands. Argonne, IL: Argonne National Laboratory; ANL/DIS/CP-92660. Available:
       http://www.osti.gov/dublincore/gpo/servlets/purl/9748-Nh3avx/webviewable/9748.pdf (22 July
       2003).

Winterton DL; Kaufman J; Keener CV; Quigley S; Farin FM; Williams PV; Koenig JQ. (2001) Genetic
       polymorphisms as biomarkers of sensitivity to inhaled sulfur dioxide in subjects with asthma. Ann
       Allergy Asthma Immunol. 86:232-238.

Wong C-M; Atkinson RW; Anderson HR; HedleyAJ; Ma S; Chau PY-K; LamT-H. (2002a) A tale of two
       cities: effects of air pollution on hospital admissions in Hong Kong and  London  compared.
       Environmental  health perspectives. 110:67-77.

Wong TW; Lau TS; Yu TS; Neller A; Wong SL; Tarn W; Pang SW. (1999) Air pollution and hospital
       admissions for respiratory  and cardiovascular diseases in  Hong Kong.  Occup Environ Med.
       56:679-683.

Wong TW; Tarn WS; Yu TS; Wong  AH. (2002b) Associations between daily mortalities from respiratory
       and cardiovascular diseases and air pollution in  Hong Kong,  China. Occup Environ Med. 59:30-5.

Wu D; Meng Z. (2003) Effect of sulfur dioxide inhalation on the glutathione redox system in mice and
       protective role of sea buckthorn seed oil. Arch Environ Contam Toxicol. 45:423-8.

Xu X; Ding H; Wang X. (1995) Acute effects of total suspended particles and sulfur dioxides on preterm
       delivery: a community-based cohort study. Arch Environ Health. 50:407-15.

Xu XP; Dockery DW; Wang LH. (1991) Effects of air pollution on adult pulmonary function. Arch Environ
       Health. 46:198-206.

Yallop D; Duncan ER; Norris E; Fuller GW; Thomas N; Walters J; Dick MC; Height SE; Thein SL; Rees
       DC. (2007) The associations between air quality and the number of hospital admissions for acute
       pain and sickle-cell disease in an urban  environment.  Br J Haematol. 136:844-8.

Yang CY; Tseng YT; Chang CC. (2003a) Effects  of air pollution on birth weight among children born
       between 1995 and 1997 in Kaohsiung, Taiwan. J Toxicol Environ  Health A. 66:807-16.

Yang CY; Chang CC; Chuang HY;  Tsai SS; Wu TN; Ho CK. (2004) Relationship between air pollution and
       daily mortality in a subtropical city: Taipei, Taiwan.  Environ Int. 30:519-23.

Yang CY; Chen CJ. (2007) Air pollution and hospital admissions for chronic obstructive pulmonary
       disease in a subtropical city: Taipei, Taiwan. J Toxicol Environ Health A. 70:1214-9.
May 2008                                 32                 DRAFT—DO NOT QUOTE OR CITE

-------
Yang Q; Chen Y; Shi Y; Burnett RT; McGrail KM; Krewski D. (2003b) Association between ozone and
       respiratory admissions among children and the elderly in Vancouver, Canada. Inhal Toxicol.
       15:1297-308.

Yang Q; Chen Y; Krewski D; Burnett RT; Shi Y; McGrail KM. (2005) Effect of short-term exposure to low
       levels of gaseous pollutants on chronic obstructive pulmonary disease hospitalizations. Environ
       Res. 99:99-105.

Yargicoglu P; AgarA; Gumuslu S; Bilmen S; Oguz Y.  (1999) Age-related alterations in antioxidant
       enzymes, lipid peroxide levels, and somatosensory-evoked potentials: effect of sulfur dioxide.
       Arch Environ Contam Toxicol. 37:554-60.

Zanis P; Trickl T; Stohl A; Wernli H; Cooper O; Zerefos C; Gaeggeler H; Schnabel C; Tobler L; Kubik PW;
       PrillerA; Scheel HE; KanterHJ; Cristofanelli  P; ForsterC; James P; Gerasopoulos E; DelclooA;
       Papayannis A; Claude H. (2003) Forecast, observation and modelling of a deep stratospheric
       intrusion event over Europe. Atmos Chem Phys. 3:763-777.

Zmirou D; Barumandzadeh T; Balducci F; RitterP; Laham G; Ghilardi J-P. (1996) Short term effects of air
       pollution on mortality in the city of Lyon, France, 1985-90. Journal of epidemiology and
       community health. 1:S30-S35.

Zmirou D; Schwartz J; Saez M; Zanobetti A; Wojtyniak B; Touloumi G; Spix C; Ponce de Leon A; Le
       MoullecY; Bacharova L; Schouten J; Ponka A;  Katsouyanni K. (1998) Time-series analysis of air
       pollution and cause-specific mortality. Epidemiology. 9:495-503.

Code of Federal Regulations, Title 40, Part 53.23c. (2006) Ambient air monitoring references and
       equivalent methods: procedures for testing performance characteristics of automated methods:
       SO2, CO, O3, and NO2. Test procedures.
May 2008                                 33                DRAFT—DO NOT QUOTE OR CITE

-------