Clean Air Scientific Advisory Committee’s (CASAC) Peer Review of EPA’s Risk and Exposure Assessment to Support the Review of the NO2 Primary National Ambient Air Quality Standard: Second Draft


             UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
         \                       WASHINGTON D.C. 20460
                                                         OFFICE OF THE ADMINISTRATOR
                                                          SCIENCE ADVISORY BOARD

                              September 24, 2008

EPA-CASAC-08-021

The Honorable Stephen L. Johnson
Administrator
U.S. Environmental Protection Agency
1200 Pennsylvania Avenue, N.W.
Washington, D.C. 20460

Subject:      Clean Air Scientific Advisory Committee's (CASAC) Peer Review of
             EPA's Risk and Exposure Assessment to Support the Review of the NO 2
             Primary National Ambient Air Quality Standard: Second Draft

Dear Administrator Johnson:

       The Clean Air Scientific Advisory Committee (CASAC), augmented by subject-
matter-experts to form the CASAC Oxides of Nitrogen Primary National Ambient Air
Quality Standards (NAAQS) Review Panel (hereafter referred to as the panel, roster
provided in Enclosure A) held a public meeting on September 9-10, 2008 to review
EPA's Risk and Exposure Assessment to Support the Review of the NO2 Primary National
Ambient Air Quality Standard: Second Draft. EPA requested that CASAC address
charge questions listed below that fell into four categories (characterizations of air
quality, health effects evidence and selection of potential alternative standards for
analysis, exposure, and health risks).  Panel consensus comments on how the REA might
be further strengthened appear below in the form of responses to the Agency's charge
questions within those categories. Individual comments from CASAC panel members
are enclosed in Enclosure B.

       The CASAC Panel was generally pleased with the second draft of the Risk and
Exposure (REA) to support the review of the NO2 primary NAAQS and found that the
Agency had been responsive to the CASAC review of the first draft of the document.
However, the REA is incomplete and the CASAC should conduct further review prior to
the document becoming final. The next draft should include both a completed Chapter 8
and an integration of the results of all the analyses based on clinical and epidemiological
studies. CASAC will also review the Advance Notice of Proposed Rulemaking when it
is published. The following describes the CASAC comments on the four categories of
charge questions.

-------
Characterization of Air Quality (Chapters 2, 6, and 7)

1. To what extent are the air quality characterizations and analyses technically sound,
clearly communicated, appropriately characterized, and relevant to the review of the
primary NO2 NAAQS?

3. Because of the impact of mobile sources on ambient NO 2, we have estimated on-road
NO2 concentrations. To what extent is the approach taken technically sound, clearly
communicated, and appropriately characterized? Do Panel members have comments
on the relevance of this procedure for reviewing the primary NO2 NAAQS?

4. What are the views of the Panel regarding the adequacy of the assessment of
uncertainty and variability?

The air quality characterizations, analyses, and uncertainty and variability
discussions are generally improved, but in some cases additional clarification is needed.
There are inconsistencies in the air quality metrics used in the analyses and those
considered as alternative standards and benchmark levels. First, the concepts of
alternative standards and benchmark levels, and their differences, should be clarified.
Currently, the approach proposes using 98th and/or 99th percentile levels, but in different
analyses switches between using the overall 98th/99th percentiles of the hourly values, the
daily, 1-hr maximums and the annual means among the various monitors in a city. These
multiple metrics are confusing, and make some of the analyses less informative to setting
a standard.

The derivation and use of the on-road modification factor (m\ needs to be
strengthened, with improved documentation as to its basis and more explicit comparison
with observations. Staff should consider using different weightings over the range of m
values employed, based on a strengthened uncertainty characterization. The discussion
of the measurements upon which m are based needs to address how those measurements
represent on- and near-roadway exposures. This expansion should be part of additional
discussion about how well the monitoring network represents actual population exposures
and provides meaningful information for exposure analysis and model evaluation. This
should include a better characterization of vertical concentration gradients and how
monitoring height might affect the relationship between observed levels and exposure.
There is some concern that the importance of the biases associated with monitoring
height and monitor interferences might be misinterpreted.

-------
Characterization of Health Effects Evidence and Selection of Potential Alternative
Standards for Analysis (Chapters 3, 4, 5)

1. The presentation of the NO 2 health effects evidence is based on the information
contained in the NO2 Integrated Science Assessment. What are the views of the Panel
on the overall characterization of the health evidence for NO2? To what extent is the
presentation clear and appropriately balanced?

Chapter 3 covers susceptibility, describing the range of populations found to be
susceptible, both to air pollution generally and to NO2 specifically. The document would
be improved by sharpening its conclusions. Clearly, one important overall finding is that
a large number of people could be susceptible, when considering the full range of groups
identified. On the other hand, the experimental and epidemiological evidence would
appear to converge in finding that asthmatics are highly susceptible. The concept of
vulnerability, as distinct from susceptibility, is introduced, and appropriately followed
through.

This draft REA appropriately reflects the NOX Integrated Science Assessment
(ISA) in summarizing conclusions regarding the currently available health evidence
related to NO2 exposures. The choice to express the overall evaluation of the data on the
major findings in terms of five levels of "confidence" is applauded, since a consistent
application of this approach can bring a new level of rigor and consistency to this type of
evaluation. On page 32, lines 1-3, the staff makes the judgment that it will focus on
endpoints for which the ISA "concludes that the available evidence is sufficient to infer
either a causal or a likely casual relationship". This represents a decision that sets a
precedent with regard to the level of evidence in support of outcomes that will be
considered in the REA. Given the precedent-setting nature of the decision, clearer
justification is needed.

The REA concludes that a "likely causal relationship" can be inferred from the data
for short-term NO2 exposure and adverse effects on the respiratory system at near
ambient levels of exposure - and that the susceptible populations include people with
asthma or airways hyperresponsiveness (AHR) and the young and elderly. The ISA and
the REA conclude that there is suggestive, but not sufficient, data to infer a causal
relationship between short term concentrations near those associated with ambient NO2
exposure and cardiopulmonary mortality and between long-term NO2 exposure and
respiratory morbidity.

The bases for the above conclusions should be more clearly defined in the REA with
clear linkages to the ISA. Both the ISA and the REA build on primary conclusions
related to weight of evidence for causality. Ideally, an ISA needs to have a full
discussion of the application of the Hill criteria, as adapted by the Agency for its review
process: strength of association, experimental evidence, consistency, biological
plausibility, coherence, temporal relationship and the presence of an exposure-response
relationship. The ISA should refer to each of these criteria and assess the data with
respect to each for each of the major health outcomes considered. If done in the ISA, the
causal conclusions could then be summarized in the REA with explicit reference to the
ISA. It is not clear in the ISA that the seven criteria were consistently considered in
coming to the final conclusions for the various health outcomes for NO2. Absent such in-

-------
depth analyses in the ISA, the conclusions of the ISA and consequently the basis for the
RE A are weakened.

A remaining task for this document is to compare and synthesize the results of the
assessments based on the epidemiologic studies and the human clinical studies. For
example, one challenge in accomplishing this is addressing differences in doses received
in these two different contexts. Human clinical studies involve controlled exposures to
NO2 concentrations at the breathing zone of the subject while the epidemiology studies
rely on a small number of fixed monitors which may not necessarily represent the actual
human exposure concentrations. The next draft REA should describe the different dose-
responses from clinical and epidemiological studies and discuss how to interpret them.

A stronger justification is needed to set aside the studies of indoor NO2. The stated
rationale acknowledges that these studies focused on NO2 alone to the extent possible and
that the exposure situation indoors differs from that outdoors. On the other hand, the
experimental literature is based on exposure to NO2 alone as well. Given the emphasis
placed on the human clinical studies, there does not appear to be a solid rationale for
setting aside the studies directed at exposure to NO2 from indoor sources.

2. The specific potential alternative standards that have been selected for analysis are
based on both controlled human exposure studies and on epidemiological studies
conducted in the United States. What are the Panel's views on the appropriateness of
these potential alternative standards (in terms of indicator, averaging time, form, and
level) for the purpose of conducting air quality, exposure, and risk assessments and
on the rationale used to select them for that purpose?

In general, the bases for selecting the indicator, averaging time, form, and level
for the NO2 NAAQS are clearly stated. The CASAC agrees that NO2 is the best indicator
for gaseous forms of oxides of nitrogen because the majority of our information on health
effects and exposure involves this oxide form. The Integrated Science Assessment for
NOX provides a detailed description of the relationship between NO2 and other oxides of
nitrogen and can be used as a reference for other forms of gaseous nitrogen oxides that
may be present.

The averaging time of 1-hour is reflective of the duration of the experimental
studies and the finding that there are adverse health effects. CASAC would recommend
that consideration be given to the need to explore scenarios for the 24-hour averaging
time as well.

The proposed alternative form of the standard is considered appropriate. The
REA should better define the strengths and weaknesses of using the 98th or 99th percentile
form for the standard - including defining how well the exposure distribution influences
the magnitude and extent of high level exposures. The epidemiological studies that form
the basis for the proposed alternative standards are well described in the REA. However,
the REA should more clearly describe how controlled human exposures were used to
establish or validate the proposed range for NO2 analyses.

-------
With regard to level, the document provides a clear rationale for assessing a lower
range extending to 0.05 ppm, with which CAS AC agrees. The upper end of the range is
quite reasonable, based on the experimental findings.

The REA should develop a scientific foundation for any decision regarding
retaining or revising the long term NAAQS for NC>2. The REA does not establish that a
short-term standard alone would be sufficient to meet the public health protection
mandate of the Clean Air Act. Are there areas of the United States that would be in
compliance with a short-term standard but not with a long-term standard? The REA
needs a discussion of the utility of the current long-term standard for NO2.

Characterization of Exposure (Chapters 6 and 8):

1. To what extent is the assessment, interpretation, and presentation of the results of the
exposure analysis technically sound, clearly communicated, and appropriately
characterized?

2. The second draft assessment document evaluates exposures in Atlanta. What are the
views of the Panel on the approach taken and on the interpretation of the results of
this analysis?

3. What are the views of the Panel regarding the adequacy of the assessment of
uncertainty and variability?

Staff provided an update on progress, because Chapter 8 is still under revision.
The Atlanta case study location is a reasonable one. The panel commends the
responsiveness of staff and their ongoing consideration of adequate prediction of air
quality. The strategies Staff have outlined to improve the modeling are likely to bring the
model results closer to observed concentrations. There is some concern that the modeling
approach may underestimate high exposures to residents who live near roads. We
encourage Staff to include a clear characterization of biases and additional assessment of
the predicted versus observed concentrations. Though not discussed at this meeting, the
rest of the exposure modeling is expected to be similar to the first draft REA, which we
previously commented on. The personal exposure data from Atlanta should also be
compared with the model results. The CASAC looks forward to reviewing the completed
exposure chapter in the future.

Characterization of Health Risks (Chapters 7, 8, 9):

1. Based on conclusions in the final ISA regarding airway responsiveness, we have
expanded the range of potential health effect benchmark values to include 0.1 ppm.
Do Panel members have comments on the range of potential health effects benchmark
values chosen to characterize risks associated with 1-hour NO2 exposures?

2. To what extent are the assessment, interpretation, and presentation of health risk
results technically sound, clearly communicated, and appropriately characterized?

3. A focused risk assessment has been conducted for emergency department visits in
Atlanta, GA. To what extent are the assessment, interpretation, and presentation of

-------
health risk results technically sound, clearly communicated, and appropriately
characterized? What are the views of the Panel on the approach taken and on the
interpretation of the results of this analysis?

4. What are the views of the Panel regarding the clarity and adequacy of the discussion
of uncertainty and variability with respect to the characterization of health risks?

The health risk assessment methodology described in Chapters 7 and 9 is well-
developed and generally of high quality. The basis for expanding the range of exposure
levels considered in the REA to include 0.1 ppm NC>2 is well-developed in the document.
It is less clear, however, why a value as low as 0.05 ppm is not proposed, given results in
the ISA. This decision should be more clearly justified, or the range expanded downward
accordingly. At a minimum, 0.05 ppm and 0 .1 ppm should be included in the Chapter 7
exceedances tables (e.g., 7-5 thru 7-16) to allow comparisons across cities at relevant
ambient conditions. On a related note, it would be more informative for the tables and
discussion to include the rate of exceedances as well as the absolute number.

An improved description of the rationale leading to the selection of Atlanta as the
representative site for detailed exposure and risk calculations would improve
understanding of the selection's implications. Justification for Atlanta's results being
generalizable is needed, given the ultimate objective of assessing national health risks
and the potential for possible recommendation of an alternate national air quality
standard.

The topics of uncertainty and variability are central to interpretation of the
analyses in the REA. The presentation of these concepts throughout the document is
uneven, repetitive, lacks sufficient specificity, and should be more quantitative. The
discussion should highlight the most important and relevant sources of uncertainty and
variability for the main analyses. One uncertainty that needs to be mentioned is the
possible effect of lowering the level of one pollutant on the level of co-pollutants. The
document should address that multi-pollutant modeling in the risk assessment assumes
co-pollutants are unchanged across alternative standards and should discuss the
implications for such an assumption. Key points and issues should be addressed in the
document, with supporting additional details located in appropriate appendices.

In closing, the CAS AC was pleased to review the second draft of the REA for the
primary NOX review. We look forward to reviewing the completed draft in the near
future and to continue to advise you as the Agency completes the ANPR.
Sincerely,

/Signed/

Dr. Rogene Henderson, Chair
Clean Air Scientific Advisory Committee
Enclosures

-------
                               NOTICE

       This report has been written as part of the activities of the EPA's Clean Air
Scientific Advisory Committee (CASAC), a Federal advisory committee
independently chartered to provide extramural scientific information and advice to the
Administrator and other officials of the EPA. The CASAC provides balanced, expert
assessment of scientific matters related to issues and problems facing the Agency.
This report has not been reviewed for approval by the Agency and, hence, the
contents of this report do not necessarily represent the views and policies of the EPA,
nor of other agencies within the Executive Branch of the Federal government. In
addition, any mention of trade names or commercial products does not constitute a
recommendation for use. CASAC reports are posted on the EPA Web site at:
http://www.epa.gov/casac.

-------
Enclosure A
U.S. Environmental Protection Agency
Clean Air Scientific Advisory Committee
Oxides of Nitrogen Primary NAAQS Review Panel

CHAIR
Dr. Rogene Henderson, Scientist Emeritus, Lovelace Respiratory Research Institute,
Albuquerque, NM

CASAC MEMBERS
Dr. Ellis B. Cowling, University Distinguished Professor At-Large, Emeritus, Colleges of
Natural Resources and Agriculture and Life Sciences, North Carolina State University, Raleigh,
NC

Dr. James Crapo, Professor of Medicine, Department of Medicine, National Jewish Medical
and Research Center, Denver, CO

Dr. Douglas Crawford-Brown, Professor and Director, Department of Environmental Sciences
and Engineering, Carolina Environmental Program, University of North Carolina at Chapel Hill,
Chapel Hill, NC

Dr. Donna Kenski, Data Analyst, Lake Michigan Air Directors Consortium, Des Plaines, IL

Dr. Armistead (Ted) Russell, Professor, Department of Civil and Environmental Engineering,
Georgia Institute of Technology, Atlanta, GA

Dr. Jonathan M. Samet, Professor and Chair of the Department of Epidemiology, Bloomberg
School of Public Health, Johns Hopkins University, Baltimore, MD

CONSULTANTS

Professor Ed Avol, Professor, Preventive Medicine, Keck School of Medicine, University of
Southern California, Los Angeles, CA

Dr. John R. Balmes, Professor, Department of Medicine, Division of Occupational and
Environmental Medicine, University of California, San Francisco, CA

Dr. Terry Gordon, Professor, Environmental Medicine, NYU School of Medicine, Tuxedo, NY

Dr. Dale Hattis, Research Professor, Center for Technology, Environment, and Development,
George Perkins Marsh Institute, Clark University, Worcester, MA

Dr. Patrick Kinney, Associate Professor, Department of Environmental Health Sciences,
Mailman School of Public Health , Columbia University, New York, NY

-------
Dr. Steven Kleeberger, Professor, Lab Chief, Laboratory of Respiratory Biology,
National Institute of Environmental Health Sciences, National Institutes of Health,
Research Triangle Park, NC

Dr. Timothy V. Larson, Professor, Department of Civil and Environmental Engineering,
University of Washington, Seattle, WA, USA

Dr. Kent Pinkerton, Professor, Regents of the University of California, Center for
Health and the Environment, University of California, Davis, CA

Dr. Edward Postlethwait, Professor and Chair, Department of Environmental Health
Sciences, School of Public Health, University of Alabama at Birmingham, Birmingham,
AL

Dr. Richard Schlesinger, Associate Dean, Department of Biology, Dyson College, Pace
University, New York, NY

Dr. Christian Seigneur,  Vice President, Atmospheric & Environmental Research, Inc.,
San Ramon, CA

Dr. Elizabeth A. (Lianne) Sheppard, Research Professor, Biostatistics and
Environmental & Occupational Health Sciences, Public Health and Community
Medicine, University of Washington, Seattle, WA

Dr. Frank Speizer, Edward Kass Professor of Medicine, Channing Laboratory, Harvard
Medical School, Boston, MA

Dr. George Thurston, Professor, Environmental Medicine, NYU School of Medicine,
New York University, Tuxedo, NY

Dr. James Ultman, Professor, Chemical Engineering, Bioengineering Program,
Pennsylvania State University, University Park, PA

Dr. Ronald Wyzga, Technical Executive, Air Quality Health and Risk, Electric Power
Research Institute, Palo Alto, CA

SCIENCE ADVISORY BOARD STAFF
Dr. Angela Nugent, Designated Federal Officer, 1200 Pennsylvania Avenue, NW
1400F, Washington, DC,  Phone: 202-343-9981, Fax: 202-233-0643,
(nugent. angel a@epa. gov)

-------
Enclosure B: Compilation of Individual Panel Member Comments on EPA's Risk and
Exposure Assessment to Support the Review of the NOi Primary National Ambient Air
Quality Standard: First Draft
This enclosure contains the preliminary and/or final written comments of individual members of
the Clean Air Scientific Advisory Committee (CASAC) Oxides of Nitrogen Primary National
Ambient Air Quality Standards (NAAQS) Review Panel. The comments are included here to
provide both a full perspective and a range of individual views expressed by panel members
during the review process. These comments do not represent the views of the CASAC or the
CASAC Panel.

Comments Received:

Comments from Prof. Ed Avol 11
Comments from Dr. JohnBalmes 16
Comments from Dr. James Crapo 19
Comments from Dr. Douglas Crawford-Brown 21
Comments from Dr. Terry Gordon 27
Comments from Dr. Dale Hattis 31
Comments from Dr. Donna Kenski 33
Comments from Dr. Patrick Kinney 37
Comments from Dr. Steven Kleeberger 39
Comments from Dr. Timothy Larson 41
Comments from Dr. Kent Pinkerton 44
Comments from Dr. Edward Postlethwait 48
Comments from Dr. Armistead Russell 50
Comments from Dr. Jonathan Samet 54
Comments from Dr. Richard Schlesinger 57
Comments from Dr. Christian Seigneur 59
Comments from Dr. Elizabeth "Lianne" Sheppard 60
Comments from Dr. Frank Speizer 64
Comments from Dr. George Thurston 69
Comments from Dr. James Ultman 72
Comments from Dr. Ronald Wyzga 73
10

-------
Comments from Prof. Ed Avol

Comments on the Second Draft N02 Risk Exposure Assessment Document
Ed Avol, CASAC N02 Primary Review Panel
Characterization of Air Quality (Chaps 2, 6, and 7):
This second draft is a marked improvement over the previous version. There is
additional detail, explanation, characterization, and continuity of presentation in the
revised document. Several alternative approaches to the air characterization analyses
have been developed and presented, and provide a useful perspective on consideration
of the current standard and the basis for re-consideration.

The discussion of on-road and roadway-related exposures is timely, important, useful,
and insightful. To be complete, however, it is important to note that not ALL exposure is
determined by roadside vehicle exhaust at a height of a few meters. Even on-road, the
hot exhaust from on-road heavy-duty trucks (most of which have elevated pipe exhaust
systems) will tend to rise. Similarly, exhaust from rail locomotives, from cruise and
cargo vessels in port communities, and from stationary boilers and power plants will
contribute to higher exposure concentrations at elevations of several meters.
Accordingly, blanket assumptions about several-fold increases in ambient air
concentrations at ground level compared to several meters above the ground may be
mistaken in several important exposure scenarios.

The discussion of uncertainty and variability is an important one and should be in the
document, but in my opinion, is somewhat out of place and better located in an
appendix, with a shortened summary in the main text. Moreover, essentially the same
discussion is presented in more than one section of the document (again in Chapter 9),
so some editing is suggested.

Characterization of Health Effects Evidence and Selection of Potential Alternative
Standards (Chaps 3,4,5):
The health evidence has been presented in a useful and detailed manner, although in
some sections (see detailed comments below), it seemed a presentation of the data
was being made rather than a summary of findings from the ISA document. The
challenging topics of susceptibility and vulnerability were again presented and
described, and in this presentation, a clearer differentiation between the two has
partially emerged. The challenge and objective for staff should now be to maintain
consistent and discrete separation between these working definitions.

Characterization of Exposure (Chaps 6, 8):
It was somewhat surprising that the occasionally presented lower benchmark value of
SOppb N02 was not consistently carried through the various presentations of health,
alternative standards, and exposure characterizations. If this was an intentional
decision, then a justification for this omission and path of action should be presented for
evaluation.
11

-------
The issues of on-road and in-vehicle exposures are important ones thankfully discussed
in the revised document.  While there are still some portions of the discussion that
would benefit from additional comment and data (including consideration of increased
on-road speeds leading to higher air exchange rates, and potentially important
differences in decay of pollutants with distance from roadside as a function of time of
day and meteorological conditions (interwoven with temporal and spatial activities and
opportunities for exposure),  the presentation was generally focused, well-done, and
appropriate.

(Review of Chapter 8 is delayed until the revised draft chapter is provided by staff)

Characterization of Health Risks (Chaps 7,8,9):
The range of potential health effects benchmarks does seem appropriately discussed,
but the appearance/disappearance of SOppb as a presented level for consideration in
the presentation seems odd. The ED visits discussion related to Atlanta appeared to be
well-done (but out of my immediate are of expertise).  Given that Atlanta is the single
location evaluated, some discussion would be appropriate as to why Atlanta is
representative of the country (aside from the fact that data was available for it).

(I await receipt of Chapter 8 to provide additional comments  for consideration)
Specific Editing Comments and Questions
Page ii: Section 5. font changes in listing; make consistent

Page vi: several abbreviations missing from listing: BAL, CAMP study, ECP,
EDAA, EDRA, MS, PMN

P4, Figure 1-1: (top right box, "qualitative characterization of US epidemiology studies"
- Throughout the document, there are references to relevant
European/Australian/Mexican/Norwegian research results, so this comment is
inaccurate.
(Note - if the notation "N02" were changed to "pollutant", this figure would be
generically applicable to virtually all subsequent REA documents)

P9, lines 25-26 ("...model-base estimates indicate that N02 levels in many non-urban
areas of the  United States are less than 1ppb..."): What level of confidence is there for
assigning modeled concentrations to areas with little or no actual measurements? Have
spot assessments confirmed that levels in non-urban areas are in fact less than 1ppb,
or is this a modeling artifice imposed by boundary conditions of the application?

P11, line 21  (..."because most sources of N02 are near ground level...") - I assume
you're actually talking about on-road traffic tailpipe emissions, so why not be explicitly
clear? Also,  this is not always true (see discussion above regarding emissions from
HDD trucks,  locomotives, ships, boilers, etc.
                                       12

-------
P11. line 24 (..."levels are likely even higher at elevations below 4 meters...")- I would
certainly agree that this is broadly true, but there are notable exceptions with elevated
exhaust, such as from trucks, trains, ships, boilers, power plants, etc, so this is not a
universal truth...

P14, Iine4: "evaluates" should be "evaluated"

P14, lines 24-25 (...there is only limited supporting evidence from clinical or
toxicological studies on potential susceptibility to N02 in persons with cardiovascular
disease..."): but aren't there also just limited studies of these people, period?

P15, lines 11-17 (paragraph on criteria that must be met for establishing useful links
between polymorphisms and adverse respiratory effects): I agree with these points, but
don't these three criteria (involvement in the pathogenesis pathway,  observable
functional change, and careful consideration of possible confounding) equally apply to
every potential adverse health effect?...so it seems to me, this paragraph is either
broadly generic and applicable or unnecessary in this specific section.

P17, lines 1 -4 (discussion of violence and elevated risk of exposure): what about
confounders, such as being more likely to be "out on the streets", or in closer proximity
to traffic, or other environmental justice angles (more likely to be closer to higher
exposure areas such as train yards, truck depots, shipping docks - less affluent areas)?
In other words, there a number of potentially confounding variables,  so making any
sweeping general conclusions is simplistic and probably incorrect.

P19, line 12 (and in multiple locations throughout the document)" repeated reference is
made to the "last review of the N02 NAAQS", but it would be more useful to readers
and more accurate to refer to the date of the last review (1995?) rather than "the last
review", to underscore the time period until now and the opportunity  for new and more
refined research to have been performed but not yet have been considered...

P19, lines 14 and 15 ("...children and older adults...."): reference in a consistent manner
to ages would be helpful to anchor the discussion,  e.g. "...children (<18yrs) and older
adults (>65yrs)..."
P22, line 2 - spell out what CAMP stands for (and place it in the abbreviation table)

P22, line 5 ("...each subject having an approximate average of two months of data..."):
This is poorly worded and inaccurate (an "approximate" average was not presented, the
mathematical average was; I assume the "approximate" part refers to the two months'
time period?), and should be re-worded.
                                       13

-------
P22, Iine16 ("...intervention study in Australia..."): this is scientifically fine (as far as I am
concerned), but it is inconsistent with your earlier declaration about emphasizing only
US studies.

P24, lines - "sites" should be "cites"

P25, lines 20-31: This section seems pretty detailed and more suited to the ISA; the
findings should be summarized here, not presented in detail. Additionally, BAL, ECP,
and PMN should also be placed on the Abbreviations listing and defined in the text
when they first appear.

P26, line 11  - replace "and" with "or"

P29, Iine12-"thrombosis", notthombosis

P32, line 4 - the presentation would be markedly improved with a two-column summary
table for short-term N02 exposure effects, listing the endpoint (respiratory, mortality,...)
and the ISA determination (causal, insufficient,...)

P41, line 3-7 - In a previous discussion (P19, section 4.2.2, lines 23+), the Linn study
was down-weighted (if not dismissed) for using having employed one-pollutant
modeling;  here it is identified as appropriate for use and included; isn't this inconsistent?

P42, Figure 5.1:  include EDRA and EDAA to Abbreviations listing

P43, Figure 5.2:  include MS in Abbreviations listing

P45, line 9 - change "thee" to "three"

P51, lines 8-10: Delete the sentence beginning with "while an individual ambient
monitor..." since it adds nothing to the discussion and is a duplication of part of the next
sentence.

P58, lines 14 and 20 seem to be contradictory; it seems to say near-road
measurements were used to calculate on-road N02 in Line 14, then say near-road
measurements were not used in Line 20...???

P72, lines 16-18: if N02 levels are generally declining and vehicle exhaust is the
primary source, how to explain the increase in exceedances between 2004-2006 and
2001 -2003 data sets?

P72, Iine18: replace "that" with "than"

P91, line 1 ("...if the monitors are not evenly distributed (causing a bias)..."): one
doesn't necessarily always follow the other. If the monitoring surface  is both smooth
                                       14

-------
and rough terrain, evenly distributed samplers can miss the variability in the rough
terrain (in valleys, up hills, etc) and over-report the smooth terrain.

P92, line 15: change "being" to "of this bias would be"

P95, line 16: change "froma" to "from"

P95, line 18: change "selecting monitors these monitors" to "selecting these monitors"
                                        15

-------
Comments from Dr. John B alines

John Balmes

GENERAL COMMENTS

Characterization of Health Effects Evidence and Selection of Potential Alternative
Standards for Analysis (Chapters 3, 4, and 5)

1. The presentation of the NO2 health effects evidence is based on the information
contained in the NO2 Integrated Science Assessment. What are the views of the
Panel on the overall characterization of the health evidence for NO2? To what extent
is the presentation clear and appropriately balanced?

Overall, I feel that the presentation of the NO2 health effects evidence in Chapters 3 and
4 is clear and appropriately balanced.

I feel quite strongly that the epidemiological evidence supports a short-term averaging
time, i.e., an annual standard is not the averaging time appropriate to protecting persons
with asthma from developing exacerbations that result in health care utilization. I think
the epidemiological data best support a 24-hour averaging time, but controlled human
exposure study data better support a 1-hour averaging time. The alternative standard
forms and levels that Chapter 5 states will be considered in the subsequent risk analysis
are appropriate and reasonably well justified.

Characterization of Health Risks (Chapters 7, 8, 9)

The range of potential health effects benchmark values chosen as described in Chapter 5
is appropriate, but there is inconsistency in how these values are applied in Chapters 7
and 9. In Chapter 5 (p. 44, lines 16-17), it is stated that 0.05 ppm will be considered in
the risk assessment. I support doing so, but in Chapter 7, exceedance data for 0.05 ppm
are not presented in the tables. In fact, data for 0.1 ppm are not consistently presented.
To improve clarity and avoid confusion, I suggest being consistent in the application of
potential health effect benchmark values.
16

-------
2. To what extent are the assessment, interpretation, and presentation of health risk
results technically sound, clearly communicated, and appropriately characterized?

The presentation of health risk results is technically sound, clearly communicated, and
appropriately characterized.

In addition to the promulgation of a short-term standard, I also think that consideration
should be given to retaining an annual standard to protect children from potential adverse
effects of NO2 on growth of lung function.

3. A focused risk assessment has been conducted for emergency department visits in
Atlanta, GA. To what extent are the assessment, interpretation, and presentation of
health risk results technically sound, clearly communicated, and appropriately
characterized? What are the views of the Panel on the approach taken and on the
interpretation of the results of this analysis?

The Atlanta-based risk assessment of the impact of varying NO2 air quality standards is
an appropriate approach to provide useful information to policy makers about the public
health impact of alternative standards.

4. What are the views of the Panel regarding the clarity and adequacy of the discussion
of uncertainty and variability with respect to the characterization of health risks?

The discussion of uncertainty and variability is reasonably clear, but there is some
needless repetition. Chapter 7 contains an in-depth discussion of uncertainty and
variability. It seems unnecessary for Chapter 9 (Section 9.6) to repeat this discussion in
so much detail (to the point of using some identical text. This section of Chapter 9 would
be improved if it placed the key issues regarding uncertainty and variability in the risk
assessment in context. Specifically, how important are the various sources of uncertainty
and variability in a relative sense, and how should they impact the interpretation of the
results? As currently written, these questions are lost in the systematic listing of all
sources of uncertainty and variability.

SPECIFIC COMMENTS

p. 13, lines 18-21 Diet is another factor that may affect vulnerability to NO2. A diet
low in antioxidant micronutrients likely increases vulnerability to NO2.

p. 18, lines 5-11 The REA uses the same five-level hierarchy to assess the level of
evidence for a causal relationship that was discussed by CASAC during the SOx meeting
on 7/30-31/08. The panel wrote about the use of this hierarchy as follows: "We concur
with using the five levels but recommend that the descriptions be changed to better reflect
the level of certainty or confidence in the classification of the level of evidence. The
phrasing of the second level is particularly problematic in its addition of the wording
17

-------
'likely causal relationship.'  The approach to evidence interpretation should avoid using
statistical significance as a criterion for evidence interpretation. CASAC recommends
that EPA reconsider the language used to describe the weight of evidence, particularly for
the first three categories which cover a range of certainty or confidence in causal
inference that extends from full certainty to lesser degrees. The language used should be
consistent with other such schemes used by EPA."  The panel's previous comments about
this hierarchy are relevant here.

p. 23, lines 21-24     The statement, "In the laboratory, airway responses can be
measured by assessing changes in pulmonary function (e.g., decline in FEV1) or changes
in the inflammatory response (e.g., using markers in bronchoalveolar lavage (BAL) fluid
or induced sputum) (ISA, section 3.1.3)," is somewhat misleading in a sub-section
entitled "Airway Responsiveness."   Bronchoalveolar lavage and induced sputum cannot
be used to directly assess airway responsiveness.

p. 25, line 21-p.26, line 11    Unlike enhanced lung function responses to inhaled
allergen after NO2 exposure, the discussion of airway inflammatory responses to inhaled
allergen do not properly fit under the sub-section title of "Airway Responsiveness."  I
would move this discussion to the next sub-section (4.2.6) on "Airway Inflammation."

p. 95, line 3    Ozone would be a better example of a reactive pollutant with a low
indoor/outdoor ratio than PM2.5.

p. 95, line 13   Should be "from a major road..."

p. 116, lines 1-4       Should be "For example,  changing from a 98th percentile 1-hour
daily maximum standard based on 0.1 ppm to one based on 0.05 ppm reduces the
estimated incidence of respiratory-related ED visits in Atlanta by about 49 percent
in 2007 (from 4700 to 2400);...
                                        18

-------
Comments from Dr. James Crapo

Review Comments
Prepared by: James D. Crapo MD
Risk and Exposure Assessment to Support the Review of the NO2 Primary National
Ambient Air Quality Standard: Second Draft
Characterization of Health Effects Evidence and Selection of Potential Alternative Standards for
Analysis (Chapters 3, 4 and 5).

Charge question 1: What are the views of the Panel on the overall characterization of
the health evidence for NO2?

This REA document appropriately reflects the NOx ISA in summarizing conclusions regarding the
currently available health evidence related to N02 exposures. This second draft of the REA is
substantially improved, and I would concur with the primary conclusions therein. The choice to o
express the overall evaluation of the data on the major findings in terms of five levels of
"confidence" is applauded, since a consistent application of this approach can bring a new level
of rigor and consistency to this type of evaluation. The REA concludes that a "likely causal
relationship" can be inferred from the data for short-term N02 exposure and adverse effects on
the respiratory system at near ambient levels of exposure - and that the susceptible populations
include subjects with asthma or AHR and the young and elderly. The ISA and the REA conclude
that there is suggestive, but not sufficient data to infer a causal relationship between short term
near ambient N02 exposure and cardiopulmonary mortality and between long-term N02
exposure and respiratory morbidity. The existing data is considered inadequate to infer the
presence or absence of a relationship between long-term near ambient N02 exposure and overall
mortality.

The basis for the above conclusions should be more clearly defined in the REA. Both the ISA and
the REA allege that the primary conclusions on Health Evidence are based on an application of 7
of the Hill criteria: strength of association, experimental evidence, consistency, biological
plausibility, coherence, temporal relationship and the presence of an exposure-response
relationship. The REA should refer to each of these criteria and assess the data with respect to
them for each of the major conclusions. This would greatly strengthen the presentation of the
REA and make the basis for the final conclusions clear. As currently written one gets the
impression that the 7 criteria were inconsistently considered in coming to the final conclusions.
In the judgment of this reviewer, the primary conclusion regarding adverse respiratory health
effects on sensitive populations exposed to near ambient N02 would meet all 7 of the Hill criteria
that were put forward as standards for data assessment.

There is difficulty in comparing the results of epidemiologic studies and those of human clinical
studies in terms of exposure levels for short-term dose. Human clinical studies involve
measurements of N02 exposure at the breathing zone of the subject while the epidemiology
studies rely on distant, fixed monitors that are commonly 4-5 meters high. There needs to be a
discussion of the projected relationship between the effective dose in these two very different
circumstances. If breathing zone N02 levels are commonly substantially higher than that
measured at a fixed monitor at 4-5 meters height, then the REA should make adjustments in its
discussion of how epidemiology and clinical studies are used to project and assess levels that are
used in risk assessment.
19

-------
Charge question 2: What are the Panel's views on the appropriateness of the
potential alternative standards for the purpose of conducting air quality, exposure,
and risk assessments and the rational used to select them?

Review Comments-Crapo
Page 2
The proposed alternative standards for analysis are appropriate. I would concur with the choice
of N02 as the indicator, 1 hour as the averaging time, the 98th or 99th percentile as the form and
levels from 0.05 to 0.20 ppm as the levels to be considered when modeling the effect of a
revised standard. Having agreed with the ultimate recommendations, I would like to see a
stronger and more detailed defense of these decisions in the text of the REA. The choice for
averaging time also included a consideration of a 24 hour standard. The strengths and
limitations of focusing on 1 hr instead of 24 hrs need to be more fully defended. The rationale
for choosing the 98th or 99th percentile as the form needs a substantially greater discussion rather
than just noting that this form was chosen for analysis of PM. The strengths and weaknesses of
using this type of form depend on the shape of the exposure curve and the degree to which the
98th or 99th percentile reflects both the magnitude and extent of high level exposures than can be
expected for these populations. The choice of exposure levels to be considered was said to be
based on both epidemiology studies and controlled human exposures - however the subsequent
discussion and tables 5-1 and 5-2 discuss only the epidemiology. This section should more
clearly describe how the controlled human exposures were used to establish or validate the
proposed range for N02 analysis. Since the subsequent exposure analysis will be based on
ambient air measurements of N02 it is essential to discuss the comparability of levels measured
in the breathing zone for controlled human exposures with those obtained from ambient monitors
located at heights of 4-5 meters.

The REA also needs a discussion of the utility of the current long-term standard for N02. A
revised standard focused on short-term N02 levels is being considered. Would this replace the
long-term standard?—or would it be an additional standard. If it is contemplated to retain the
current long-term standard, then this needs a detailed analysis.
20

-------
Comments from Dr. Douglas Crawford-Brown
Comments on Risk and Exposure Assessment (REA) to Support the Review of the
NO2 Primary National Ambient Air Quality Standard: Second Draft

Douglas Crawford-Brown (3-9-08)

This review is built around the charge questions provided. I focus on Chapters 3, 4 and 5,
and then on chapters 7, 8 and 9.

First, a more general comment is appropriate. The overall REA is both a good step
forward from the first draft, and contains a wealth of useful material. I will need to
withhold complete judgment until Chapter 8 is complete, but the existing chapters are
generally in good shape and make for a compelling argument. There is strong consistency
between this document and the ISA, with caveats noted later.

There is, however, a need for a global edit of the document. The chapters differ in writing
style and even, to some extent, in the way they have used the ISA material. There also
seems to me more detail in the Appendices than are reflected in the document. It is
appropriate for their to be more technical detail, but I felt that the Appendices suggested
that more detailed studies had been done yet not reflected in the main document. I may be
wrong about this, but I did get this sense, and it again suggests the need for a global edit
to be sure all of the chapters, and the main document and Appendices, are consistent.
Finally, the levels of exposure considered don't seem to me consistent throughout. Each
chapter and section should use identical ranges of values.

For Chapters 3, 4 and 5:

1. The presentation of the NO2 health effects evidence is based on the information
contained in the NO2 Integrated Science Assessment. What are the views of the
Panel on the overall characterization of the health evidence for NO2? To what extent is
the presentation clear and appropriately balanced?
21

-------
The authors have done a good job of tracking through the ISA and culling the more
important results. These results have then been summarized quite well in the REA an
from the basis for the assessment.! found it quite easy to follow the discussion although,
as noted in my general comments, I had a sense that there was more detail in the
Appendix than is evident in the body of these chapters.

I notice that only the human data seem to play any sort of role in the assessment. I had
thought that the animal data were at least going to play some modifying role, although I
admit we had decided on the CASAC that the human data would form the primary basis
for setting the health benchmarks. At the least, the document should explain the lack of
use of the animal data.

I also feel the ISA provided a better understanding of the issue of sensitive, susceptible,
vulnerable populations. Why is there no mention of infants throughout? There is
significantly more nuance and subdivision of populations in the ISA, and it was just odd
to find so little included in the REA. I suppose the decision may have been made to
abstract the ISA discussion and use only the most important results for the REA, but if
this is the case, it should be explained. I also suppose the results might be the same if
they were culled out as a special subpopulation, but can't tell from the analysis.

I agree with the decision to use the studies for which the conclusion is "... the available
evidence is sufficient to infer either a causal or a likely casual relationship". However,
this judgment is introduced with little supporting justification. As it may set a precedent
for further REAs, it deserves a bit of discussion both in the document and in CASAC
deliberations. Fortunately, the issue is not so pressing since, even if one uses this set of
criteria, the evidence is sufficient to support the EPA assessment and to make a good case
for regulatory consideration.

Finally, am I correct that the indoor NO2 exposure studies are playing no role? This is a
bit odd, given that they seem to me to be studies focused largely on NO2 exposures by
22

-------
themselves, which is precisely what we are trying to get at here. Or is the assumption
made that they, too, are confounded by other exposures? At the least, there needs to be a
better explanation as to why these studies are being rejected for use.

2. The specific potential alternative standards that have been selected for analysis are
based on both controlled human exposure studies and on epidemiological studies
conducted in the United States. What are the Panel's views on the appropriateness of
these potential alternative standards (in terms of indicator, averaging time, form, and
level) for the purpose of conducting air quality, exposure, and risk assessments and on
the rationale used to select them for that purpose?

I first note that I was pleased to see an assessment based on both the clinical and
epidemiological studies. This was a good step forward in letting the latter results play a
role.

I support the use of NO2 as the Indicator, if for no other reason than that I don't know
how any other indicator would be implemented.

I support the Averaging Time suggested, although would like to see some discussion of
the implications of a daily averaging time, as it links more directly to the epidemiological
studies. But I doubt this would change the level of exposure selected. And in nay event, I
don't think the epidemiological studies can fully support the choice of a daily average
value, since there is a lot of intra-daily variability at a location, and this means the
appearance of health effects may not be related to the daily average (but rather to shorter-
term periods of elevation throughout the day).

On the Form, I wasn't clear how this 98th or 99th percentile issue is to be dealt with
methodologically. The document would profit from having an example stream of
measurements, then showing how they are to be abstracted down to some sort of
cumulative distribution function and the 98th or 99th percentile estimated and then
averaged. I support the use of the 98th or 99th percentile as at least a good policy choice (I
23

-------
am not sure it has any real scientific basis - which is not to say it goes against the
science, only that it is not really a scientific issue), but just am not clear what it means
methodologically.

As to the Level, I am again supportive, both of the lower and upper bounds selected.
However, there is a reasonable argument to be made that the incidence of hospital
admission for asthmatics may be high even at these levels. And the levels don't reflect, in
my kind, any sort of margin of safety. So there is merit in at least considering shifting the
range down by factor of 2 or so. Still, I suspect the current range contains a value likely
to emerge as the Level for the final NAAQS.

For chapters 7, 8 and 9:

I support this decision, especially since the fraction of asthmatics showing response at
between 0.1 and 0.2 ppm (or 0.15) is so high. As I noted in the answer to Question 2
above, there is at least an argument to be made for a further reduction below 0.1 ppm
based on a margin of safety, but I am comfortable that the expansion of the range down to
0.1 ppm now encompasses the range of values likely to be considered seriously in
regulation.

2. To what extent are the assessment, interpretation, and presentation of health risk
results technically sound, clearly communicated, and appropriately characterized?

I found this a section that was easy to follow, and support the underlying methodology. I
will withhold full judgment until Chapter 8 is presented and integrated, as that is where
the wheels may fall off. But the methodology for estimating health impacts contingent on
24

-------
exposure is well developed and sophisticated. The one point that will need clarity is how
the health effects can be related properly to the spatial gradient of exposures in the
epidemiological studies and in actual population exposures. This is an area that will
require more substantive discussion of uncertainties, as I suspect it is an important driver
of the results. It is also important to note that exposures close to roadways are likely to be
controlled not just by the concentration field (which is considered in the assessment) but
by mode choice in travel, which in turn is known to be affected by the quality of the built
environment (i.e. how attractive and safe it is to walk or bike near roads). This again is an
issue of uncertainty and not assessment methodology, since there is no good way to
account for this at present.

I am generally supportive of this analysis. It was well written and simple to follow, and
makes a good addition to the other analyses in supporting the case being made overall in
the REA. Atlanta seems to have been a good choice, leading to (somewhat) clear
conclusions). My sole concern is whether the baseline can really be established well. One
year is used, but I worry about such a short baseline given the climate in Atlanta, which
can swing substantially from year to year. This can change both exposures and the
background incidence of related respiratory diseases. I would have preferred to see an
average taken over at least several years as a baseline, and suspect the results are driven
significantly by the choice of this baseline year.

4. What are the views of the Panel regarding the clarity and adequacy of the discussion
of uncertainty and variability with respect to the characterization of health risks.
25

-------
As with the other REAs, I remain concerned that the analysis of uncertainty is so
qualitative. There has been so much work done within the EPA on development of tools
for uncertainty analysis, I don't understand why there is the continued reliance on
qualitative studies. The reader is left with the impression that all we can say is that the
results are uncertain to some unspecified degree. An uncertainty analysis need not be a
full Monte Carlo approach, with PDFs for each parameter and nesting of model
uncertainty, but there should at least be more quantitative statements about the key
parameters and model forms that affect uncertainty, the sensitivity of the estimates to
these components, and some idea as to how large of an error might be introduced (and in
what direction). This would also greatly improve the utility of the uncertainty results in
the Appendix (around page 86) concerning exceedences. These seem to me particularly
important results that deserve a bit more reflection and analysis for uncertainty.

Variability is treated a bit better in existing REA, although there is still not a good
discussion of how variability and uncertainty are being separated; the purpose behind
each kind of analysis; or how variability can at times affect uncertainty in the analyses
preformed.

To be helpful, the uncertainty analysis should also provide better understanding as to why
the uncertainty exists. A summary code could be given for each source indicating
whether the issue is one of representativeness (as in monitor locations), measurement
methodology, number of samples, incomplete science to back a model, etc. It would then
be possible to target future resources on reduction of uncertainly. At some point many
years ago, the EPA had a significant program aimed at reducing residual uncertainty, and
that work or kind of approach doesn't seem to be showing up in these REAs.
26

-------
Comments from Dr. Terry Gordon

Terry Gordon

Charge Question Responses

Characterization of Air Quality (Chapters 2, 6, and 7)

1. To what extent are the air quality characterizations and analyses technically sound,
clearly communicated, appropriately characterized, and relevant to the review of the
primary NO2 NAAQS?

The characterizations and analyses appear to be sound and clearly communicated.

To the extent of my somewhat low level of expertise in this area, I understood a great
deal more of the approach in this draft of the RE A. Therefore, I'd say the approach
appears to be clearly communicated and characterized.

This is quite relevant to the exposure of individuals to ambient NOx. Because the time
spent 'on-road' may differ significantly amongst MSA's (i.e., traffic-dominated Los
Angeles vs. other U.S. cities), it may be appropriate to extend this analysis with different
factors for low or high traffic/commuting.

4. What are the views of the Panel regarding the adequacy of the assessment of
uncertainty and variability?

The uncertainty and variability assessment were thoroughly covered.

Characterization of Health Effects Evidence and Selection of Potential Alternative
Standards for Analysis (Chapters 3, 4, 5)

1. The presentation of the NO2 health effects evidence is based on the information
contained in the NO2 Integrated Science Assessment. What are the views of the
Panel on the overall characterization of the health evidence for NO 2? To what extent
is the presentation clear and appropriately balanced?
27

-------
The characterization of the health evidence was quite clear and the decision pathway to
choosing the appropriate benchmarks was excellent. The one exception might be the
explanation of the different choices of NOx exposure concentrations in the risk
analyses. It took careful and repeated reading to understand why some analyses would
range from 50 ppb and up and why others started at higher values. Perhaps, a summary
paragraph at the beginning or end of the section (5.5) would provide a clearer explanation
of the pathway forward for the choices.

The potential alternative standards are appropriately chosen. It is unclear, however, what
to make of the current annual standard given the language in the REA that states the
"evidence is suggestive but not sufficient to infer a causal relationship between long-term
NO2 exposure and respiratory morbidity".

Characterization of Exposure (Chapters 6 and 8):

1. To what extent is the assessment, interpretation, and presentation of the results of the
exposure analysis technically sound, clearly communicated, and appropriately
characterized?
Waiting for Chapter 8.
2. The second draft assessment document evaluates exposures in Atlanta. What are the
views of the Panel on the approach taken and on the interpretation of the results of
this analysis?

Waiting for Chapter 8.
3. What are the views of the Panel regarding the adequacy of the assessment of
uncertainty and variability?

The assessment of uncertainty was very thorough although presenting it twice in the text
and once in a table may be overkill.

Characterization of Health Risks (Chapters 7, 8, 9):

I concur with the selected benchmark and value.
28

-------
2. To what extent are the assessment, interpretation, and presentation of health risk
results technically sound, clearly communicated, and appropriately characterized?

Waiting for Chapter 8.

The approach is excellent and clearly communicated. If this focus on one city will be
used to project risk across the nation, it might be appropriate to include other cities for
validation/cross comparison (Philadelphia was done previously?).

4. What are the views of the Panel regarding the clarity and adequacy of the discussion
of uncertainty and variability with respect to the characterization of health risks.
The assessment of uncertainty was very thorough although presenting it twice in the text
and once in a table/conclusion may be overkill.
Minor Comments:

Page 13, line 27 - conduction disorders - heart or nerve?
Page 13, line 28 - hypertension - vascular or pulmonary?
Page 17, line 1 - Is the 4.3 ppb increase 24 hr or annual or 1 hr?
Page 20, line 14 - Delete space after ' 12'
Page 24, line 6 - Change 'sites' to 'cites'
Page 24, Table 4-1 and line 17 - The 76% listed on line 17 does not match the 0.75
fraction in the Table. Also, there appears to be a typo on the 2nd row of data - the
fraction for All Exposures (0.68) is higher than both of its components (Exercise is 0.59
and Rest is 0.67)
Page 30, line 12 - chronic is misspelled
Page 31, line 15 - fetal 'growth retardation'?
Page 42, Figure 5-1 - The legend says 95% CI but such data are not presented in the
Figure.
Page 46, line 11 - delete the 2nd 'to'
Page 52, line 27 - a space is needed between scenario and 4.
Page 53, line 16 - It is unclear what is meant by "A screening..." because the previous
sentence already said the ambient air quality data would be screened.
Page 73, line 15 - 'other time not' is unclear
Page 73, line 26 - insert 'are' before 'provided'
Page 91, lines 7-17 - It is unclear why the vertical height discussion is included here
rather than in the Spatial Representation section 7.4.4. I consider spatial/space to have 3
dimensions.
29

-------
Page 109, lines 1-4 - The wording is such that someone could grab this
reasoning/sentence, even after the earlier careful elucidation of the potential for
confounding by co-pollutants, to dismiss the conclusions of the entire risk assessment.
                                        30

-------
Comments from Dr. Dale Hattis

Dale Hattis—Premeeting Comment on Charge Question 4 for the Analysis of Air Quality

3. Estimation of On-Road NO2 Concentrations—Technically Sound? Adequately
Characterized

In view of the prominence of on-road exposures in the exceedance analysis, I also think
that it is important for the authors to document the underlying data they use and do
quantitative uncertainty analysis for key parameters that determine the results.  In
particular they should document their conclusions about the distribution of values for "m"
in equation 4 on p. A-107. The authors should document the m values they derived from
each study and evaluate the uncertainty in this value they infer for this key parameter
from each study. Documenting the individual m values for each study would allow an
independent assessment of the authors' conclusion that the data cannot be well described
by a parametric distriibution (e.g., lognormal, normal, etc.) whose values are not limited
to those directly inferred from the existing studies. The current empirical distribution
treatment implies that each observed "m" should be treated as having equal
uncertainty/weight and that the differences in "m" values observed by the various authors
under various circumstances should be considered to adequately represent the real
variability that would be seen in a representative sample of roads in the locations
modeled.  In fact it is likely that the different measurements carry different uncertainties
that could be assessed in a comparative analysis of the likely errors in different
measurements.  Such an analysis could potentially lead to different weighting of (1)
measurements that were done with greater vs lesser accuracy (taking account relative
confidence/measurement uncertainties) and (2) measurements that were done under
conditions of traffic volume and meteorology that are relatively common vs relatively
rare (taking into account the representativeness of the measured conditions relative to the
real variability among sites, etc., that is being modeled).

4. What are the views of the panel regarding the adequacy of the assessment of
uncertainty and variability?

The assessment at present is entirely qualitative. In my opinion there should be at least
some quantitative assessment of some of the important sources of uncertainty.  For
example in previous comments on an  earlier draft  of the REAI reported that model-based
analyses of the likely effect of distribution of elevations of monitoring heights indicated
that NO2 concentrations observed at those heights were likely to understate NO2 levels at
breathing elevations of about 2 meters by 17-35% depending on the mathematical form
of the function used for the decline of concentrations with height. Instead of creating its
own more sophisticated analysis of this issue, the  current document simply repeats a
dismissal of the issue as not likely to be substantial in the light of the fact that most
monitors are not as elevated as the monitor studied in the paper that documented the
likely bias with monitoring height. Where an approximate quantitative analysis of this
bias is possible, I think it  should be done.

It is also important to quantitatively analyze  and discuss the model biases and
uncertainties indicated in  Figures A-97 through A-99 on pages A-86 and A-87 of

                                       31

-------
Appendix A. To my eye, these figures suggest quite a bit of low bias in model
predictions of observed exceedances, although the figures are not especially clear on this
point. It would be better to put observed exceedances on the y axis and predicted
exceedances on the x axis, and also to include an identity line (y = x) to better show the
reader the magnitude and direction of the biases.

Also please see my comment in response to question 3 on the uncertainties in the present
estimation of on-road concentrations.
                                        32

-------
Comments from Dr. Donna Kenski

Pre-meeting comments on 2nd Draft RE A for NOx Primary NAAQS
Donna Kenski

Characterization of Air Quality

1. To what extent are the air quality characterizations and analyses technically
sound, clearly communicated, appropriately characterized, and relevant to the
review of the primary NO2 NAAQS?

Chapter 2 was very brief but accurately summarized the air quality data from the ISA.
The summary of sources of NO2 in Sec. 2.1 was succinct but adequately quantitative.

This discussion was fine.

3. Because of the impact of mobile sources on ambient NO2 we have estimated on-
roadNO2 concentrations. To what extent is the approach taken technically
sound, clearly communicated, and appropriately characterized. Do Panel
members have comments on the relevance of this procedure for reviewing the
primary NO2 NAAQS?

I think the approach is probably fine, and Appendix 8 did a nice job describing the
derivation of the on-road factors. Nevertheless, the REA is still lacking information
validating this particular model. I was expecting a plot or two showing how well the
model fit the data it was derived from at the very least, and hopefully also an assessment
of how it fit an independent dataset or a subset of the original data. Both demonstrations
are necessary to support this analysis. It is difficult to have confidence in the estimated
on-road concentrations if we don't have a sense of how well the model fits the data it was
derived from. I do appreciate that Appendix B included additional comparisons of
estimated on-road concentrations from the ambient data and AERMOD, but that's still
not demonstrating the original model validity (and the Appendix B comparison wasn't all
that encouraging in matching estimated numbers of benchmark exceedances).

4. What are the views of the Panel regarding the adequacy of the assessment of
uncertainty and variability?

Again, the discussion in Sees. 2.3.2 and 2.3.3 were very brief, but captured the essence of
the discussion in the ISA so I think they were adequate for this document.

Characterization of Health Effects Evidence and Selection of Potential Alternative
Standards for Analysis (Chapters 3,4,5)

-------
1. What are the views of the Panel on the overall characterization of the health
evidence for NO2 ? To what extent is the presentation clear and appropriately
balanced?

Chapters 3 and 4 were essentially distillations of the ISA. I found them faithful to the
more comprehensive presentation in that document and as such, a clear, balanced, and
adequate summary for this analysis.

I'm not quite sure what to make of Section 4.4.1 and its conclusion that evidence is
suggestive but not sufficient to infer a causal relationship between long-term NO2
exposure and respiratory morbidity. Does this mean the current annual standard should
be dropped entirely, to be replaced with a short-term standard? The evidence for a short-
term standard is convincing and the REA appropriately focuses on that, but it would be
helpful to have some discussion, before the ANPR arrives, about the ramifications of
replacing the annual standard, perhaps in Section 5. Even without such discussion,
however, Section 5 was a very helpful addition in weighing the various combinations of
averaging time, form, and level.

2. What are the panel's views on the appropriateness of the potential alternative
standards (in terms of indicator, averaging time, form, and level) for the purpose
of conducting air quality, exposure, and risk assessments and on the rationale
used to select them for that purpose?

As noted above, Chapter 5 was a great addition.

Characterization of Exposure

1. To what extent is the assessment, interpretation, and presentation of results of the
exposure analysis technically sound, clearly communicated, and appropriately
characterized?
2. The second draft assessment devaluates exposures in Atlanta. What are the views
of the panel on the approach taken and on the interpretation of the results of this
analysis?
3. What are the views of the panel regarding the adequacy of the assessment of
uncertainty and variability?

(1,2,3 to be answered after Chapter 8 is delivered)

Characterization of Health Risks

1. Based on conclusions in the final ISA regarding airway responsiveness, we have
expanded the range of potential benchmark values to include 0.1 ppm. Do panel
members have comments on the range of potential health effect benchmark values
chosen to characterize risks associated with 1-hour NO2 exposures?

The expanded range of benchmarks is an improvement. However, the document was
quite confusing in the presentation of results with respect to those benchmarks. Some

-------
additional clarification should be made somewhere up front, explaining why some
analyses used lower benchmarks of 100 ppb and some used 150. The use of a potential
standard of 50 ppb added to the confusion as well. Although it's possible that a standard
of 50 ppb might be proposed in order to limit exposures to 100 ppb, it's not necessarily
intuitive, so warrants further discussion.

2. To what extent are the assessment, interpretation, and presentation of health risk
results technically sound, clearly communicated, and appropriately
characterized? (to be answered after Chapter 8 is delivered)

3. A focused risk assessment has been conducted for emergency department visits in
Atlanta. To what extent are the assessment, interpretation, and presentation of
health risk results technically sound, clearly communicated, and appropriately
characterized? What are the views of the panel on the approach taken and on the
interpretation of the results of this analysis?

This analysis was a nice addition to the REA. It would be helpful to have added an
explanation of why Atlanta was selected (this was documented in Appendix A but not
explained so concisely in Chapter 9) and also to make some attempt to put these results in
the context of other cities. How similar is Atlanta to other US cities with respect to NO2
concentrations and population characteristics?

4. What are the views of the panel regarding the clarity and adequacy of the discussion
of uncertainty and variability with respect to the characterization of health risks.

I'm not sure we can really answer this without Chapter 8. Section 9.6 seems quite
thorough, but again in a qualitative way. Can the bulleted items be ranked or prioritized?

Minor editorial comments:
p.9, several lines: anthropogenic is misspelled.
p. 10, line 9: though -> through
p.30, line 23: chonic -> chronic
p.45, line 9: thee -> three
p. 51, line 26: The discussion of the empirical model is in App. A Sec. 7, not Sec.6.
p. 72, line 18: that-> than
p. 73, lines 23-26; fix grammar
p. 82, line 28: percentile is misspelled
p. 83, line 21: range of estimates provided
p. 108, line 15: though -> through
p. 108, line 18: thee -> three
p. 108, line 26: Refers to Sec 3.8 in App. C, but App. C doesn't have Section numbers.
p. Ill, lines 2 and 6: thee -> three
Appendix C: Table numbers are incorrect.
A-107, line 42 and p. A-108, line 16: References to equation 1 should be to equation 3 as
numbered in this draft.
35

-------
Please review rules for hyphenation. Especially in section 4, hyphens are used erratically
and incorrectly, frequently in the phrase "statistically significant."
The phrase "as is" is sometimes italicized, sometimes not. Should be consistent.
                                         36

-------
Comments from Dr. Patrick Kinney

Kinney Comments on NO2 REA
September 8, 2008

I commend EPA staff on a much-improved document for review. My comments are
largely either editorial or focus on improving the precision of the writing.

p. 11, lines 4-7: Edit sentence for clarity. This should be understandable to an educated
lay person.

p. 11, line 15: at end of sentence, insert examples of oxidation products of concern in
parentheses.

p. 12, lines 11-12: change "change the principal conclusions" to "bias conclusions in a
positive direction"

p. 17, line 4: It would be helpful to mention here though that there is some evidence that
low SES people tend to live nearer to busy roadways than more wealthy people. I think
Margo Schwab's research demonstrated this in Washington, DC.

p. 29, line 13: mention residences too in this context.

p. 18, line 28: the noted range of increased risks (2-20%) means little without
information on the exposure range and time scale. Better would be something like,
"Effect estimates from epidemiologic studies conducted in the US and Canada indicate
that ED visits may increase by 2-20% on days when NO2 concentrations are X ppb
higher than on other days." This same problem recurs on lines 29-30 of p. 19, and
elsewhere when presenting epi findings. See also page 27, line 14.

p. 20, line 1:1 don't think the word "detected" is appropriate here. Better would be
"inferred" or "estimated." Any calculation that is based on an arbitrary concentration
range (say 30 ppb) is an inferred value, not a finding or observation from a specific study,
especially where observed concentrations in the study were typically much lower than 30
ppb, and day to day changes as large as 30 ppb may never have been, or only rarely,
observed. This problem occurs again at lines 14-15 of page 20, where it is stated that
"The authors found a 12% increase in risk per 20 ppb increase in 24-hr ambient NO2."
They may have calculated that, but I doubt they found that.

p. 20, line 26: this statement that "NO2 associations in multicity studies are generally
robust" is contradicted by one of the three studies noted. Seems like a more nuanced
summary is needed, e.g., in the few multi city studies that have been conducted, more
often than not, NO2 was robust.

p. 28, line 5: should the word "generally" be changed to "more often than not"?

p. 42, figure e5-l: I believe the NYC DOH results are for all ages, not just children.
Also, caption should include 24 hour.

-------
p. 43, figure 5-2:  caption should also mention 4 and 24 hour results.

p. 45, line 9: "three" is misspelled.

p. 59, line 19: change "two year-groups" to "two six-year groups" This occurs elsewhere
also.

p. 70, Table 7-10  and tables which follow:  In spite of the footnote, I have a hard time
understanding what these numbers mean in terms of spatial and temporal averaging.
Need a better explanation to aid interpretation.

p. 72, line 18: change "that" to "than"

p. 93, lines 4-6: please expand further on why you think this is likely to a be a problem
and why it would go in a particular direction. Same goes for line 16.  Not obvious to this
reader why this would be the case.

p. 94, lines 19-23: this sentence needs  editing for clarity.

p. 107, line 7:  Since some of the observed baseline incidence is due to as-is NO2
concentrations,  do these ED visits need to be subtracted out to get the true "baseline"?

p. Ill, line 6: "three" is misspelled.

p. 115, line 5: however,  as noted above, NO2 was only moderately correlated with PM10
and O3 in this study.  The text as-written is misleading in this context.

p. 116, line 2:1 think the concentrations should be swapped in this sentence.
                                        38

-------
Comments from Dr. Steven Kleeberger

I thought the second draft of the REA for NO2 was very well written, and overall have
very few substantive comments. The sections were largely very clearly presented and
approaches adequately explained. Minor issues are raised below in response to specific
questions.

Characterization of Air Quality (Chapters 2, 6, and 7)

1. To what extent are the air quality characterizations and analyses technically sound,
clearly communicated, appropriately characterized, and relevant to the review of the
primary NO2NAAQS?

I thought the air quality characterizations and analyses were appropriately presented.

This approach was well communicated and characterized.

3. Because of the impact of mobile sources on ambient NO2, we have estimated on-road
NO2 concentrations. To what extent is the approach taken technically sound, clearly
communicated, and appropriately characterized? Do Panel members have comments on
the relevance of this procedure for reviewing the primary NO2NAAQS?

No comments. As above, I the approach was technically sound and well communicated.

4. What are the views of the Panel regarding the adequacy of the assessment of
uncertainty and variability?

This is not my particular area of expertise.

Characterization of Health Effects Evidence and Selection of Potential Alternative
Standards for Analysis (Chapters 3, 4, 5)

1. The presentation of the NO2 health effects evidence is based on the information
contained in the NO2 Integrated Science Assessment. What are the views of the
Panel on the overall characterization of the health evidence for NO 2? To what extent
is the presentation clear and appropriately balanced?

This area was largely well-written, and the presentation was appropriately balanced.
However, the general lack of inclusion of animal modeling data was somewhat
surprising. Most of the discussion focused on epidemiology and chamber studies.

I also thought that a table was going to be created to illustrate/list susceptibility factors
known to be, or potentially could be, important in responsivity to NO2. In the "Age"
section, perhaps a statement that indicates, in addition to children and the elderly, infants

-------
could also be a particularly susceptible subgroup. Some evidence, though not sufficient
to indicate absolute causality, would suggest that this is the case (e.g. increased incidence
of SIDS associated with NO2).

In general, I believe it is important to emphasize that strong biological plausibility exists
for other subgroups to be susceptible to NO2-induced effects. The evidence that
children, elderly, and individuals with asthma are susceptible is clear, and correctly
indicated in the REA. However, other factors could also be important but simply haven't
been investigated sufficiently. A strong statement to this effect is warranted.

An important point to be emphasized in Section 3.2 (Susceptibility: Pre-existing Disease)
is that controlled NO2 exposures of asthmatics have been restricted to mild asthmatics for
ethical and safety concerns. It is likely that severe asthmatics are even more susceptible
to the effects of NO2 exposures than mild asthmatics, and further illustrates the health
concerns for this large subpopulation.

A minor point: it is not clear what is meant by "physiological susceptibility" (page 17,
line 12).

I thought the introduction and explanation of form was appropriate and nicely tied to
other criteria pollutants.

Characterization of Exposure (Chapters 6 and 8):

1. To what extent is the assessment, interpretation, and presentation of the results of the
exposure analysis technically sound, clearly communicated, and appropriately
characterized?

The assessment, interpretation and presentation were clearly presented in Chapter 6.

2. The second draft assessment document evaluates exposures in Atlanta. What are the
views of the Panel on the approach taken and on the interpretation of the results of this
analysis?

N/A

3. What are the views of the Panel regarding the adequacy of the assessment of
uncertainty and variability?

N/A
40

-------
Comments from Dr. Timothy Larson

Review of the Risk and Exposure Assessment to Support the Review of the
NO2 Primary National Ambient Air Quality Standard: Second Draft

Comments by Dr. Timothy Larson

This is a much improved draft, including the useful appendices. The staff has been very
responsive to the major comments from CAS AC. I look forward to hearing more about
the exposure assessment approaches to be presented in Chapter 8. In the meantime, I
have a few preliminary comments regarding exposure assessment in addition to
addressing several of the charge questions.

Air Quality Analyses

1. To what extent are the air quality characterizations and analyses technically sound,
clearly communicated, appropriately characterized, and relevant to the review of the
primary NO2 NAAQS?

These characterizations, including the accompanying appendices, are clearly
communicated. The different approaches to estimating the spatial distribution of NC>2 are
a good mix of direct measurements and statistical methods.

The choice of a 98th or 99th percentile 1-hour value is presumably done to provide a stable
target over time. It would be good to show how stable these targets are over time.
Specifically, it would be good to show the relative rankings of this metric between cities
over time. If these relative rankings are reasonable stable (non-parametric statistical tests
are available), then it would appear justified.

The downwind conversion of NO to NC>2 is mentioned briefly in the REA long after
introduction of equation 7-1. This conversion can produce a maximum value of Cx that is
not necessarily at x = 0. In the section introducing equation 7-2, it might be instructive to
emphasize that the distribution of m values reflects both decay (by dispersion) and
formation (via ozone reaction) of NO2.

In general, this approach is clearly communicated and the appropriate caveats are
provided. The distinctions between the benchmark levels versus the alternative standards
needs further clarification, possibly showing a graph of frequency distributions of air
concentrations under "as is" and "CS" levels and a representation of the accompanying
number of exceedances of a given benchmark level under different alternative standards.
The fact that considering an alternative 1-hr standard of 100 ppb at the 98th percentile will
allow some 1-hr 150 ppb concentration exceedances in a given year will be clarified in
such a figure and this might help to make the distinction between the benchmark values
41

-------
and the alternative standards and help to interpret the numbers in the relevant tables (e.g.
Table 7-15).

3. Because of the impact of mobile sources on ambient NO 2, we have estimated on-road
NO2 concentrations. To what extent is the approach taken technically sound, clearly
communicated, and appropriately characterized? Do Panel members have comments on
the relevance of this procedure for reviewing the primary NO 2 NAAQS?

The on-road and near-road exposures are an important aspect of the overall exposure
assessment. It would be interesting to know whether the monitors showing the extreme
values of the distribution in figure A-101 are different from the others, specifically are
these very near the road or confined in street canyons.

While I agree that the range of average m values discussed in the REA is reasonable for
both flat on-road conditions and street canyons, there can be larger values observed on an
hourly basis in street canyons under certain wind direction and traffic level combinations.
In a given hour, m can exceed 4 assuming rooftop levels are equivalent to urban
background levels [c.f. Xie et al, Atmos. Environ. V37 (2003) 3214-3224 and Ghenu et
al, Environmental Modeling and Software v23 (2008) 314-321].

4. What are the views of the Panel regarding the adequacy of the assessment of
uncertainty and variability?

It would be good to clearly describe which monitors were being used to compare with the
Aermod predictions. If these monitors are further than 100 meters from major roads, this
limitation should be included.
Exposure Assessment

This is a very useful Appendix that clarifies some of the issues described in the main
report. I think it should be referred to earlier in the main report to help put the major
ideas in context.

The other city that was seriously being considered for this analysis was New York. In
this major urban area, people live next to street canyons and spend more time there than
those that commute on roads. Zhou and Levy (Atmos. Environ. v42 (2008), 3087-3098)
looked at the issues of population exposures to NOx in street canyons in dense urban
areas from the perspective of NOx intake fraction. Due to the high population density in
the New York City urban canyons and to the fact that the canyons can trap the pollutants
relative to unobstructed roads, Zhou and Levy predicted intake fractions for NC>2 in New
York that approach those for second hand smoke (on the order of 10"3). These exposures

-------
may not occur in Atlanta and that should be clearly communicated. Can the intake
fractions be estimated in the upcoming exposure analyses in chapter 8?  This would help
put such analyses in a broader context.

Health Assessment

1. Based on conclusions in the final ISA regarding airway responsiveness, we have
expanded the range of potential health effect benchmark values to include 0.1 ppm. Do
Panel members have comments on the range of potential health effects benchmark values
chosen to characterize risks associated with 1-hour NO2 exposures?

I think the new range of benchmark values that include 0.1 ppm is reasonable.
                                       43

-------
Comments from Dr. Kent Pinkerton

Comments on Risk and Exposure Assessment to Support the Review of the NO2 Primary
Ambient Air Quality Standard: Second Draft

Kent E. Pinkerton, University of California, Davis
Characterization of Air Quality (Chapters 2, 6, and 7)

1. To what extent are the air quality characterizations and analyses technically sound,
clearly communicated, appropriately characterized, and relevant to the review of the
primary NO2NAAQS?

REPLY: The characterization of the sources of NO2, ambient levels of NO2 and
uncertainties associated with ambient NO2 monitoring are clearly presented in a brief, but
adequate fashion in Chapter 2. Approaches to assess exposures and risks to NO2 are
clearly outlined in Chapter 6 with reasonable explanations of formulas and equations
implemented to simulate current and alternative standards for NO2. Air quality
characterization and analyses appear to be scientifically sound, appropriately
characterized and communicated in Chapter 7. Each are highly relevant to the review of
the primary NO2 NAAQS, but need to clearly identify how these approaches may lead to
a completely different basis for a standard that is currently not the approach used for
assessment - i.e., a 1-hour standard.

REPLY: The approach to simulate just meeting the current annual standard is clearly
communicated and well characterized.

3. Because of the impact of mobile sources on ambient NO2, we have estimated on-road
NO2 concentrations. To what extent is the approach taken technically sound, clearly
communicated, and appropriately characterized? Do Panel members have comments on
the relevance of this procedure for reviewing the primary NO 2 NAAQS?

REPLY: Since mobile sources represent a primary origin for the genesis of NO2 and
personal exposure to NO2, to evaluate the impact of roadway levels of NO2 is extremely
logical. The formulas applied to estimate on-road NO2 concentrations are clearly stated
and explained with a highly reasonable degree of characterization. However, it was not
clear why some locations (Denver, Los Angeles, Phoenix and St. Louis - line 2, page 60)
found higher NO2 concentrations at greater distances from roadways that those within
100 m of roadways. The authors should clarify this point and/or provide further rationale
for estimating on-road NO2 concentrations.

4. What are the views of the Panel regarding the adequacy of the assessment of

-------
uncertainty and variability?

REPLY: Although the assessment of uncertainty and variability is not an area I have
much experience, this section is logically organized into specific sources of uncertainty
(air quality data, ambient measurement techniques [height of monitor, etc], temporal and
spatial representation, air quality adjustments, on-road estimations and the choice of
potential health benchmarks) that are all reasonably explained and presented. The
definitions for uncertainty and variability are also highly beneficial. The presentation of
all these factors provides great value to this portion of the document.
Characterization of Health Effects Evidence and Selection of Potential Alternative
Standards for Analysis (Chapters 3, 4, 5)

1. The presentation of the NO2 health effects evidence is based on the information
contained in the NO2 Integrated Science Assessment. What are the views of the
Panel on the overall characterization of the health evidence for NO 2? To what extent
is the presentation clear and appropriately balanced?

REPLY: This section is logical and well presented. The authors of these chapters have
been careful to provide balance in the studies reported. Although it is unfortunate that the
discussion of the biological plausibility and potential mechanisms for adverse health
effects due to NO2 is limited due to a lack of the presentation of animal toxicology, the
chapters are excellent in the presentation of health effects evidence. The discussion is in
large measure based on human clinical studies and epidemiology. At risk populations
are well described to include those with pre-existing disease, children and the elderly, as
well as those that those with genetic conditions that might involve the glutathione S-
tranferase (GST) gene. The evidence is clear that health implications occur well below
the current NAAQS standard for NC>2.

REPLY: This section is highly appropriate in the approach taken to define NC>2 risk and
exposure assessment based on current levels of ambient NCh and NC>2 levels associated
with just meeting the current standard. The section is well written and appropriate to
identify potential alternative standards in terms of indicator, averaging time, form and
level and well as providing the rationale used to select these alternative standards. The
selection of a one-hour averaging time to evaluate standards is highly appropriate. A
concentration-based form for the NCh standard would also much better reflect health
risks posed by elevated NC>2 concentrations. The level to include both the 98th and 99th
percentile NC>2 concentrations averaged over 3 years as recommended by the staff also
seems to be highly logical and appropriate.

-------
Characterization of Exposure (Chapters 6 and 8):

1. To what extent is the assessment, interpretation, and presentation of the results of the
exposure analysis technically sound, clearly communicated, and appropriately
characterized?

REPLY: The assessment, interpretation and presentation of the results of the exposure
analysis are technically sound, clearly communicated and appropriately characterized.
The approach to 1) compare NO2 air quality levels to potential health effect benchmark
values derived from the literature for controlled human exposures, 2) the evaluation of an
inhalation exposure model to generate more realistic estimates of personal exposures and
3) the estimation of emergency department (ED) visits are all highly appropriate and well
communicated in Chapter 6 of the document.

2. The second draft assessment document evaluates exposures in Atlanta. What are the
views of the Panel on the approach taken and on the interpretation of the results of this
analysis?

REPLY: The selection of Atlanta to evaluate exposures would appear to be highly
appropriate, based on the information available to perform this analysis. Appendix C
provides an excellent summary for the NO2 health risk assessment for Atlanta for the
consideration of the relationship between NO2 and adverse health effects. The methods
used for the selection of health endpoints based on epidemiologic studies as well as the
selection of concentration-response functions and air quality considerations are well-
described and appropriate. This type of analysis for Atlanta NO2 exposures is extremely
well done.

3. What are the views of the Panel regarding the adequacy of the assessment of
uncertainty and variability?

REPLY: Again the presentation of uncertainty and variability appear to be very
reasonable, however, this presentation also appears again in Chapters 7 and 9 as well as
to a limited (and appropriate degree) in Appendix C.
Characterization of Health Risks (Chapters 7, 8, 9):

REPLY: The range of potential health effects benchmark values chosen to characterize
risks associated with 1-hour NO2 exposures is highly appropriate. Based on the health
effects information available through human clinical and epidemiological studies using a
1-hour exposure air quality characterization, the most relevant NO2 concentration range
appears to be 100 to 300 ppb. However, in the body of the text 50, 100, 150 and 200 ppb

-------
are stated, but in the tables (7-5 to 7-8) the four potential health effect benchmark levels
used are 150, 200, 250 and 300 ppb to estimate the number of exceedances for NC>2. The
authors should clarify this point. Never-the-less, the selected potential health effect
benchmark levels are highly appropriate.

2. To what extent are the assessment, interpretation, and presentation of health risk
results technically sound, clearly communicated, and appropriately characterized?

REPLY: From the data provided to date, the assessment, interpretation, and presentation
of health risk results in the document have been technically sound, clearly communicated,
and appropriately characterized.

REPLY: The approach for conducting a risk assessment for emergency department visits
in Atlanta, GA is thorough with excellent presentation of the findings that have been
clearly communicated. The focus on Atlanta seems highly appropriate; however, further
comparisons using other cities may also be appropriate in making a decision for national
risk and possible consideration in the current air quality standard for NC>2.

4. What are the views of the Panel regarding the clarity and adequacy of the discussion
of uncertainty and variability with respect to the characterization of health risks?

REPLY: The uncertainty analysis throughout the document has been clearly stated with a
reasonable degree of discussion to provide a logical explanation of factors impacting on
uncertainty. The use of data such as the temporal representation of NC>2 levels into
distinct time periods (page 92) may also prove useful in better dealing with issues of
uncertainty and variability. Table 7-19 is useful as a summary of those factors that
impact on uncertainty and how it may affect (positive or negative) the analysis for the
characterization of air quality and health risk.
47

-------
Comments from Dr. Edward Postlethwait

Edward M. Postlethwait, Ph.D.

Comments on the NOX second draft REA for the Sept 2008 CASAC meeting (in no
particular order)

1. The document is clearly improved and the extensive efforts by EPA staff in
crafting the REA are evident.

2. Section 3.4 (Susceptibility: Genetics) needs some minor editing to correct
inaccuracies (e.g., "absent peroxide activity" which I assume is "peroxidase"; not all
ROS are electrophilic).

2. The statements in section 6 regarding the utility of airway hyperresponsiveness
versus ED visits are not always clear and thus some editing would strengthen the
document. "For example, the public health significance of the effect in question (i.e., ED
visits) is less ambiguous in terms of its impact on an individual than in the case of airway
hyperresponsiveness." While the rationale for this is likely evident to those in the field, it
does raise questions as to how ED visits are characterized with regard to NO2-specific
related pathophysiological events. Thus, in the absence of more characterizations,
counting all ED visits rather than attempting to relate AHR to ambient NO2 levels does
not necessarily appear to strengthen risk estimates.

3. The benchmark NO2 concentrations are not consistent throughout the document.
For example, on pg 49, 50 ppb 1 hr levels are included but to not appear elsewhere.

4. A graphic and/or pictorial representation of the best estimates of the spatial
distribution of NO2 levels around roadways would be useful. Although there appears to
be uncertainties in the precise spatial distribution, a consensus best estimate that would
give the reader an improved feel for exposures relative to highway proximity and that
would help identify where the 100 m boundaries exist (middle of the median outward,
edge of road, etc) would seem germane.

5. The estimated risks/exposures appear to be based on Atlanta due to a single
publication. However, Atlanta represents neither a worst or best case scenario and has
missing data in many of the presented tables. Thus, staff should consider either including
additional cities or at least providing a more compelling rationale for the focus on
Atlanta. Furthermore, to this reader all the figures/tables in chapter 7 do not lend clarity
but rather make this section more difficult to readily decipher. Perhaps final compilations
should be presented with the numerous iterations moved to the appendix.

6. Tables 9-1,2,3 suggest that the inclusion of PMio and Os reduce the NOX related
ED visit incidence by factors of > 4 fold bringing into question the strong statements
regarding causality and the "robust" NOX related outcomes in multi-pollutant models.
These discrepancies should be resolved to improve internal consistency.
48

-------
7.      While the numerous potential limitations and confounding factors are noted and
discussed, it is unclear how they factor into the risk analyses. Based on the presented
information, there appears to be a large number of uncertainties that could have
substantial impacts on estimating health impacts.  Thus, it seems critical that sensitivity
analyses are incorporated to evaluate the extent of confounding and how any and all of
the uncertainties affect the health/exposure outcomes boundaries.
                                        49

-------
Comments from Dr. Armi stead Russell

Pre-meeting Review of EPA's "Risk and Exposure Assessment to Support the Review of
the NO2 Primary National Ambient Air Quality Standard"
Armi stead (Ted) Russell

General Review:

My overall view of the REA (sans Chapter 8) is positive. I think it begins to lay out the
information needed to inform setting of a NAAQS. One aspect I felt missing was that as
it focuses on the need, and the respective level, etc., of a short term standard, I was a bit
disappointed at the lack of discussion and analysis of "what about the old one"? Will it
still be needed and why? What happens if it is removed? The latter question has to go
beyond just the health issues, but welfare as well. Certainly, a one hour standard for NO2
has little relevance to addressing environmental acidification or nutrient enrichment.
Given the recent decision by the administrator to not promulgate a separate welfare
NAAQS (i.e., for ozone), not having a long term NC>2 standard has major implications.
What might be needed in this case is to look specifically at the sources likely to be
affected by a short term standard and assess how controls targeted at a one hour standard
might impact longer term concentrations and emissions overall.

I might also suggest against the wording on page 5, line 4 "... and of any potential
alternative standards." This is not limiting in any way shape or form, and in particular,
suggests that standards could be introduced here that have not been subject to any prior
relevant analysis. Given the current approach to the NAAQS reviews, it would be
appropriate to have this altered to suggest standards for which informative analyses have
been conducted and reviewed.

One part that needs to be much more precise is section 5.4 as to the Form of the standard.
In particular, you note that it is the 98th percentile averaged over 3 years. By that, I
assume one means taking the 98%ile from the daily maximum one hour average for each
of three calendar years, and taking that average. An alternative is to take all 24-hr
measurements, and taking the 98th%ile, and then averaging. (This was less ambiguous
for PM since only 24-hr measurements are used.) This should be spelled out specifically
and precisely, along with the mathematical formula.

It seems as though the levels of concern examined vary between chapters, sometimes not
including the 0.05 ppb mark (e.g., figures in Chapter 7).
1. To what extent are the air quality characterizations and analyses technically sound,
clearly communicated, appropriately characterized, and relevant to the review of the
primary NO2 NAAQS?

There is a current weakness here, that raises its ugly little head later in the document. It
would help, at this point, to have a more comprehensive description of the range of
measured NO2 levels across the US. As noted below, we are led to believe that there are
areas where the monitored mean is 3 ppb (page 32, line 8). Is this true? What type of

-------
area might this be? Given the proposed form of the short term standard, it would be best
if, instead of showing just the annual mean, the 98th%ile (or other characterization of the
peak 1-hr levels) also be provided (as available).

It would be good if the air quality characterizations done in Chapter 7 were related to the
epidemiologic studies considered.

The order of cities in Table 7-1 is not apparent at this point, becoming apparent when
looking at Table 7-2. I prefer alphabetizing.

Tables 7-3 and 7-4 need to be more precisely described in the footnotes. In particular,
what annual means are being described? Are those the %iles of the annual means?
(Apparently, yes). Are these the annual means of the 98th%iles of the daily one hour
maximums? (Apparently not, but this is more pertinent.) Are they the %iles of the whole
distributions? The document should be made clear, and the analyses consistent,
throughout the document. Please make the statement of the standard in a precise
mathematical form.

The relationship between annual average and specific target standards should be
provided, i.e. in a table in the REA for the cities analyzed, what is the current annual
average, the 98th and 99th%iles (using the proposed averaging) should be given. I think
this might be provided in 7-4... but the footnotes have to be clearer that this is given.

Much of the more pertinent information is given in Appendix A, and it seems that some
of the less pertinent info is in the tables and discussion in the body of the report. There
should be a continuous link between the air quality characterization and the metrics given
there and the following exposure and risk assessments later on.
2. In order to simulate just meeting potential alternative 1-hour daily maximum
standards, we have adjusted NO2 air quality levels using the same approach that was
used in the first draft to simulate just meeting the current annual standard. To what
extent is this approach clearly communicated and appropriately characterized?

The approach used is reasonable, though the assumptions (and justifications) should be
stated, preferably with some foundation as to why the assumptions are appropriate. This
may involve considering specific sources and control approaches and saying "There is no
more appropriate approach given what is currently known." However, this should be
done after consideration of specific sources.

As a detail, the mathematical formula description should be tightened up some, i.e., :

Fy= Adjustment factor for location i and year j (unitless)

This tightening should be done for both Eq. 6-1 and 6-2 (though j stands for two different
things in the two equations). Further, they should specifically state, here, that the PRB is
taken as insignificant.
51

-------
In Table 7-12,1 would add a characterization of the F's for each area. This info is in the
Appendix, but some info here would be good (not as much detail needed).

3. Because of the impact of mobile sources on ambient NO2, we have estimated on-road
NO2 concentrations. To what extent is the approach taken technically sound, clearly
communicated, and appropriately characterized? Do Panel members have comments
on the relevance of this procedure for reviewing the primary NO 2 NAAQS?

My major concern about this analysis is how well founded is the calculation of "m".
While I guess this would come later, this is an area where some quantitative uncertainty
analysis should be done.

In equation 7-1, k is the "decay" constant, not the "rate" constant.

4. What are the views of the Panel regarding the adequacy of the assessment of
uncertainty and variability?

At present, the discussion is largely qualitative, with little information as to even the
ordering of major versus minor uncertainties. Section 7.4.2 has much of the discussion
on monitor location not technique. This should be moved to 7.4.4. The authors should
be specific on the extent of biases and uncertainties. In regards to a standard based on the
1-hr maximum of NO2,1 would argue that the 50% value for the technique bias is
irrelevant. Further, it would be good if you can be much more quantitative as to the
likely bias introduced from having a 4-to-5 m sampling height at a monitor 100 m away
from a major road. Bring in the AERMOD results here.

It is here that we get some discussion on the adjustment factor that should also be present
when the process is introduced. Introduce the issues earlier on, giving the more detailed
discussion later on. This uncertainty analysis could also be a bit more quantitative (or
semi-quantitative).

I appreciate Table 7-19., but would like to see a column characterizing level of
uncertainty.

P 93,1 26: Not sure what you mean by "concentration profile".

Specific Points:

P 20,1 26: Make sure that everyone would concur that the results are "robust". This
word is subjective, and can mean different things to different people. I would be more
specific, specifying how many studies still find statistically significant results when
multi-pollutant models are used. My take of 3.1 -7 does not suggest the word "robust" is
quite correct. Further, they only control for PM10, not PM2.5 or, say, EC. Looking at
3.1-10 and 3.3-2 leads me to further suggest the word "robust" should not be used here.

P24,1 6: "cites"

P27,114: 2-20% increase per "what"

-------
P32,1 8: This sentence, as written, indicates that adverse health effects can be identified
for a mean 24-hr concentration of 3 ppb. This seems a bit far fetched. Indeed, looking at
the Table 5.4-1 of the ISA, I think the 3 ppb comes from the Linn study, and if that is the
case, it was 3.4 pphm, not 3.4 ppb. I might re-check the units on the various studies
summarized in this table. Looking at Figure 2.4-13, it appears as though the minimum
mean NO2 is 5 ppb, and I would have to imagine that is at a pretty sparsely populated
location. I am curious, at what locations where epi studies have been conducted is the
mean NO2 under 10 ppb? I will also add, the fact that an adverse outcome associated
with NO2 exposure is found in areas where the mean is, say, 10 ppb, does not mean that
the adverse outcome results from an exposure of 10 ppb.

P36, 111: This line should be "NOx, for the purpose of this document, includes..."

P41,1 16: "Figures 1 and 2" (Capitalize). Check capitalization of "Figure", "Chapter"
and "Table", etc. throughout.

P45,1 9: three (not thee)

P 67: footnote: Table 7^0.
P72 118: "than" not "that"
P72,128: remove "that"
P73,126:") as provided..."

P107,1 17-18. One should note that in the Peel et al., and Tolbert et al., studies, that
having only 36 out of 42 hospitals providing usable data does not mean that a similar
fraction of potential ED visits are captured (or, conversely that 6/42 are missed). Often, it
is the smaller hospitals that are not captured, so a very high fraction of ED's are included.
Also, are you sure o the 36 number, particularly for 2004?

Section 9.6 seems out of place. Shouldn't it go later on?

Still check on the Philadelphia NOx emissions from aircraft.
53

-------
Comments from Dr. Jonathan Satnet
Review Comments
Prepared by: Jonathan M. Samet, MD, MS
Risk and Exposure Assessment to Support the Review of the NC>2 Primary National
Ambient Air Quality Standard: Second Draft
Characterization of Health Effects Evidence and Selection of Potential Alternative
Standards for Analysis (Chapters 3,4,5).

1. Charge Question 1:

In general, the document adequately represents the key evidence in the ISA with
regard to the overall findings. Chapter 3 covers susceptibility, describing the
range of populations found to be susceptible, both to air pollution generally and to
NC>2 specifically. The document might be improved by a sharpening of its
conclusions. Clearly, one important overall finding is that a large number of
people could be susceptible, when considering the full range of groups identified.
On the other hand, the experimental and epidemiological evidence would appear
to converge in finding that asthmatics are the most susceptible. This conclusion
may be stated obliquely on page 14, lines 17-19 in Chapter 3.

The concept of vulnerability, as distinct from susceptibility, is introduced, and
appropriately followed through. It will be important to continue to maintain this
same distinction between vulnerability and susceptibility in subsequent
documents related to the NAAQS for other pollutants.

Chapter 4 provides the relevant findings of the ISA from the literature on the
health effects of NC>2. As with Chapter 3, the ISA is satisfactorily distilled for the
main points relevant to the REA. There are several important issues to be
discussed. This set of evaluations for NO2 uses the five-level classification of
strength of evidence for causation. On page 32, lines 1-3, the staff makes the
judgment that it will focus on endpoints for which the ISA "concludes that the
available evidence is sufficient to infer either a causal or a likely casual
relationship". This represents a decision that sets a precedent with regard to the
level of evidence in support of outcomes that will be considered in the REA. I do
not dispute the choice, but I would suggest that it be better justified, given the
future significance of this decision.

I also suggest that a stronger justification is needed to set aside the studies of
indoor NO2. The stated rationale acknowledges that these studies focused on NO2
alone to the extent possible and that the exposure situation indoors differs from
that outdoors. On the other hand, the experimental literature is based on exposure
to NO2 alone. Given the emphasis placed on the experimental studies, there does
not appear to be a solid rationale for setting aside the studies directed at exposure
to NO2 from indoor sources.

-------
The results of selected epidemiological studies will be considered in the REA. A
rather weak argument is made for the appropriateness for this purpose in lines 12-
13 on page 33. This sentence, giving staff judgment, follows a reasonably
comprehensive discussion of the strengths and limitations of the epidemiological
data.
Charge Question 2:
In general, the basis for selecting the indicator, averaging time, form, and level for
the NC>2 NAAQS are clearly stated. The averaging time of 1-hour is reflective of
the duration of the experimental studies and the finding that there are adverse
health effects. Should consideration be given to exploring scenarios for the 24-
hour averaging time as well? It might be useful to learn of maintenance of this
averaging time would accord the same protection for deeper exposures as would
be reached by having the one-hour standard.

With regard to level, the document provides a clear rationale for assessing a lower
range extending to 0.05 ppm. The upper end of the range is quite reasonable, due
to the experimental findings.

The REA states that alternative long-term standards to the current annual value
will not be considered. On the other hand, the evidence does not provide certain
evidence that there are long-term consequences. Would a short-term standard
alone be sufficient? Are there areas that would be in compliance with a shirt-term
standard but not with a long-term standard?
55

-------
Specific Comments:
Page#
9
9
10
10
11
11
12
13
13
14
15
15
16
18
23
24
24
27
27
27
28
33
33
34
39
45
Line#
5
11
19
28
21
23
11
3
25
19
1
13
9
21
12
6

15
22
27
10
12
19
20
7
9
Comment
Correct spelling anthropogenic
cc cc cc
Can't the concentrations vary over the time of exposure?
"On average. (add in the United States) people spend. . ."
Delete "this produces" insert there is.
"2.5-fold (delete increase in) substitute higher
"However, (insert the possible consequences of this exposure error do
not) delete "is not expected to"
Delete subpopulations (insert groups within the general population)
are at increased risk for suffering (adverse effects from NO2
exposure.)
Delete are believed (insert have been found)
What is meant by "most sensitive"?
Delete agreed (insert had comparable findings)
"First the product " Necessarily? What about regulations?
"the vicinity" (Insert of roadways)
"recent studies" (or the entire body of evidence?)
Delete lead to the type of outcomes assessed (Insert be the basis for
the effects observed) (insert increased before respiratory illness
"In addition the ISA" delete sites (insert cities)
Table row 3 column 2 Should be 66%?
In association with? What exposure?
Redundant phrasing
Adverse (insert respiratory) health
Add (and the potential for confounding)
This sentence seems overly strong in view of the prior discussion of
limitations of the epidemiological findings.
But don't these studies address NO2 alone as in the exposure studies?
"levels is (increased) airway hyperresponsiveness"
Would it be reasonable to consider the 24-hr averaging time as well?
". . . we have employed thee approaches" Change to three.
                                   56

-------
Comments from Dr. Richard Schlesinger

RICHARD SCHLESINGER

Overall, this draft is an improvement over the first. Certain concepts that were not clear
in the first draft have been clarified in this one. Some comments on the specific questions
follow:

To what extent are the air quality characterizations and analyses technically sound,
clearly communicated, appropriately characterized, and relevant to the review of the
primary NO2 NAAQS?

From my perspective, these are clearly communicated.

In order to simulate just meeting potential alternative 1-hour daily maximum
standards, we have adjusted NO2 air quality levels using the same approach that was
used in the first draft to simulate just meeting the current annual standard. To what
extent is this approach clearly communicated and appropriately characterized?

This is explained much better in this than in the first version of the document.
Because of the impact of mobile sources on ambient NO2, we have estimated on-road
NO2 concentrations. To what extent is the approach taken technically sound, clearly
communicated, and appropriately characterized? Do Panel members have comments
on the relevance of this procedure for reviewing the primary NO2 NAAQS?

No specific comment.

. What are the views of the Panel regarding the adequacy of the assessment of
uncertainty and variability?

These are clearly addressed in this version of the document.

Chapters 3. 4. 5

The presentation of the NO2 health effects evidence is based on the information
contained in the NO2 Integrated Science Assessment. What are the views of the
Panel on the overall characterization of the health evidence for NO2? To what extent
is the presentation clear and appropriately balanced?

The health outcomes evidence was clearly presented. However more animal toxicology
studies should have been included as basis to support findings in epi studies.

The specific potential alternative standards that have been selected for analysis are
based on both controlled human exposure studies and on epidemiological studies
conducted in the United States. What are the Panel's views on the appropriateness of
these potential alternative standards (in terms of indicator, averaging time, form, and

-------
level) for the purpose of conducting air quality, exposure, and risk assessments and
on the rationale used to select them for that purpose?

I agree with the chosen alternatives

Chapters 7. 8. 9

 Based on conclusions in the final ISA regarding airway responsiveness, we have
expanded the range of potential health effect benchmark values to include 0.1 ppm.
Do Panel members have comments on  the range of potential health effects benchmark
values chosen to characterize risks associated with 1-hour NO2 exposures?

I agree with this approach. However, there does seem to be some inconsistency in the
specific benchmark concentrations in different sections of the document.

A focused risk assessment has been conducted for emergency department visits in
Atlanta, GA. To what extent are the assessment, interpretation, and presentation of
health risk results technically sound, clearly communicated, and appropriately
characterized?  What are the views of the Panel on the approach taken and on the
interpretation of the results of this analysis?

I agree with this approach

What are the views of the Panel regarding the clarity and adequacy of the discussion
of uncertainty and variability with respect to the characterization of health risks.

The discussion is quite sound.
                                        58

-------
Comments from Dr. Christian Seigneur

 Comments on the NOi Risk and Exposure Assessment- Second Draft - September
                                     2008.

                               Christian Seigneur
                    CEREA - Joint Laboratory ENPC/EDF R&D
                               Universite Paris-Est
                            Champs sur Marne, France
The second Draft shows some clear improvement over the first draft. For example, the
uncertainty in the measurement of NO2 ambient concentrations is now qualified as being
small near emission sources (p. 11, lines 17-19).  Furthermore, EPA has added
benchmark concentrations as low as 0.05 ppm and has added a new analysis based on the
epidemiological studies where the effect of NO2 concentrations on emergency department
visits are compared for different NO2 concentration benchmark values. These latter
revisions required a significant amount of new work by EPA staff and they clearly
demonstrate that EPA listened and responded in an effective manner to the CASAC
Panel's comments.

Most of my comments on the previous draft focused on the air quality modeling. It
appears that the modeling of NO2 with AERMOD is more challenging than expected,
possibly because NO2 is both a primary pollutant (emitted directly from vehicle exhaust
and power plants, for example) and a secondary pollutant (formed by oxidation of NO by
HO2 and Os for example). The  simplest approach to calculate secondary NO2
concentrations near an emission source is to use an "oxidant-limited" approach, which
assumes that the reaction of NO with Os is fast and stops when all Os has been depleted.
The reaction of NO with HO2 radicals and other peroxyl radicals can generally be
neglected near aNOx source because the concentrations of those radicals are typically
negligible when NO  concentrations are high. Nevertheless, such  a simple chemical
scheme must be implemented carefully to correctly represent the atmospheric chemistry
(for example, it is implemented incorrectly in the current regulatory  version of
CALPUFF).

Other air quality modeling related comments made on the previous draft pertained to the
model performance of AERMOD, taking into account the uncertainty in model
formulation in the uncertainty analysis, and the assumption made for the NO2/NOX
emission ratio. I will wait to see Chapter 8 and related material before providing
additional comments on the revised air quality modeling of this second REA draft.

Minor editorial comments:

p. 45, line 9 and p. 109, line 41: three instead of thee

p. 57, line 17; 0 meter instead of 0 meters (since it is less than 1)
                                       59

-------
Comments from Dr. Elizabeth "Lianne" Sheppard

General comments: This second draft REA is significant improvement over the first
draft. I commend EPA staff for their success. The document is appropriate in its length
and depth of coverage of the material, and it gives a concise yet clear discussion with
improved flow. Chapter 5 is a good addition. It uses the ISA results to justify the
standards to be considered in the risk and exposure assessment. Chapter 6 discusses the
risk and exposure assessment approaches used in the remainder of the document. It is
easier to follow this revised discussion. [Notably, there is no mention of the tiers
described in the Health Assessment Plan. This language should be revisited in future
rounds of the review process so there is more cross-document integration.] Chapter 9 has
been added in response to CASAC comments and appropriately presents an example
scenario to put the epidemiological results in context. There is no concluding chapter.
The document may further benefit from this addition in order to integrate the conclusions
from the three risk and exposure assessment approaches.

Now that it is becoming apparent that the REA is to more directly take on some of the
role of the previous Staff Paper, modifications are needed to bring forward all policy-
relevant aspects. As only one of many possible examples, it is important to explicitly
recognize in the REA that NO2 is the indicator for NOx.

There was good support and referencing to relevant ISA sections. The same should be
done with the appendices.

Air quality characterization: Chapter 7 is much improved. I have several major
comments as well as a host of specific comments noted below.
• I appreciate the additional stratification of monitors by distance from major roads
and the ability to see the differences in the analyses as a function of this
stratification. More refinement is needed prior to presenting this analysis to
describe the stratification and also the features of the two strata. Specifically,
Table 7-1 should be expanded to show the number of complete monitors in each
time period that fall into each stratum. In addition, it would be valuable to know
how much heterogeneity there is across monitors in each of the strata on
important characteristics such as the population density in a buffer surrounding
the monitor, the dominant land use within the surrounding buffer, the types of
major roads nearby, and the distribution of distances to major roads. An entire
subsection should be added to Section 7.2 to further describe these features.
Much of this information is described in Appendix A (as tabular lists); now all
that is needed is an analysis and concise description.
• This chapter still suffers from the assumption that the monitoring network
actually represents a sample of monitors that provides meaningful information
about population exposure. This assumption should be stated up front and
evaluated in the beginning of the chapter. I suggest adding a section that
discusses the monitoring objectives of the NOx regulatory monitoring network,
presents an analysis of the proportion of monitors sited to meet each objective,
and provides a qualitative discussion of how these objectives align with the
intended inference from data analysis presented in chapter.

-------
• The discussion of monitoring height suggests that in general further analysis of
monitoring data with respect to monitor features should be conducted for eventual
incorporation into the REAs. The most important features will vary by pollutant;
for NOx this will include vertical placement, proximity to nearby roads, and other
pertinent geographic features

Health effects evidence characterization and selection of potential alternative standards:
This appears appropriate.

Characterization of exposure: The plans for Chapter 8 suggest the content of the chapter
will be similar to the previous version with an improved air quality analysis and a new
location (Atlanta). EPA staff have been very responsive to CAS AC's comments and are
still struggling with providing predictions that are acceptable.

• For the air quality modeling, expand the prediction assessment with additional
figures and tables. In addition to the diurnal and cumulative distribution figures
we saw, ideas include observed vs. predicted scatterplots, time series plots over 3
years of predicted and observed series on the same plot (probably aggregated to
the daily average), scatterplots stratified by season, etc.
• Develop criteria for predictions that are "good enough". For instance, over-
predicting the tails of the distribution for NO2 particularly when these are still
below NOx may not be so bad from the benchmark assessment perspective.
Alternatively, and particularly since the effort is time-limited, clearly quantify
how good the predictions are.

Characterization of health risks: The addition of the health risk assessment for ED visits
in Atlanta is generally appropriate and appears to be sound. I appreciate this added
perspective.

Specific comments:
• p. 7 1.3: Expand this section to discuss the full scope of the document,
particularly given the September 8, 2008 memo from Marcus Peacock that
suggests an expanded role for the REA.
• p. 111 4-7: I think this statement is too general. The consequences of exposure
misclassification will depend upon the use of the data.
• p. 11: The distinction between exposure and concentration are blurred. The focus
is on exposure but the discussion is about concentration.
• p 12 111-13: Again I think this statement is too general here. In the context of
personal exposure-response vs. concentration-response estimates from
epidemiological studies, I think an important distinction is that the two types of
studies are estimating different parameters. This is more important than a focus
on measurement error bias (which I would agree shouldn't be neglected).
• P 22 1 7-8: This mistake came forward from the ISA and must be fixed. The
outcome for the CAMP study is asthma symptoms, not cough. Also the "multi-
pollutant" models reported are joint models, i.e. the effect of changing both
pollutants at the same time. The sentence implies the more common use of multi-
pollutant which is the effect of a single pollutant when the other pollutant is held

-------
constant in the model. I suggest replacing "Multi-pollutant" with "joint" and
"included" with "added".
P. 24 Table 4-1:  There is an error in the 0.1-0.15 row since the all exposure
average is outside the range of the separate strata.
P 39 4-7:  For completeness, inclusion (or citation) of an analysis of the
comparability of the  1-hr and 24-hr data would (hopefully) help support this
statement.
P 45 1 24-25 and rest of page: The results being  reasonably "considered a broad
characterization" is only appropriate if the monitors actually characterize air
quality and human exposure. The existing monitors don't capture all types of
locations where people spend their time and to the degree that they are a biased
representation of these locations, they may be far from a good "broad
characterization".
P 46 1 8: Justify why Atlanta here?
P 46 footnote: Say why Philadelphia is inadequate/only reserved for the
appendix?
P 51 1 10: insert "quality of the" after "The".
P 51 1 23-25: Unclear at this early place in the document.
P 52 1 28-29: Even though all monitors are included, it is quite possible that the
existing network doesn't broadly characterize national air quality.  There is no
discussion in the document of what the monitoring network represents.  The
stated assumption is too strong and not justified.  It should be evaluated
thoroughly.
P 53: Add a discussion of the regulatory monitoring objectives and how they
align with population exposure.
P 54 1 2-5: This validity criterion could be met but still allow most or all of an
entire season to be missing.
P 54 1 16: insert "long-term" before "temporal".
P 54 Table 7-1: Expand the description to include number of near vs. far from
road monitors in the "complete" columns, e.g. as "58 (40,  18)" for the early years
in Chicago.
P 55 section 7.2.2.: This assumes these criteria really capture areas with high
NO2 levels and that geographic location is the most important determinant.
P 58 1 17: What percent of locations are such monitors available?
P 58 1 20-1: How many such monitors?
P 58 123: What does the empirical distribution represent? Is the random
assignment appropriate?
P 58 1 29: Is there a new m each time the data are simulated?
P 57: The notation on this page appears to be unnecessarily complex. Equation
7-2 is essentially Cb+Cv.
P 59 7.3:  This discussion following the on-road  simulation needs some kind of
transition so the reader is clear that we've moved from a simulation approach to a
data analysis. This section has much better focus than the previous version.
Tables 7-5 - 7-8:  These tables need to better reflect the variable number of site-
years that go into each row of data. Otherwise cross-row comparisons within the
table have no meaning. Ditto for 7-10-7-11.
P 67 1 6-8: Is this assumption reasonable?
                                 62

-------
P 68 1 16: This is a strong and unsupported assumption.
P 72 1  14:  I had trouble digesting "subtle" particularly given the dramatic
differences illustrated in e.g Figure 7-2.
P 72 Figure 7-2 and others: These are nice summaries. Consider adding the 25th
to 75th percentiles (or the 5th to 95th) as vertical bands to suggest variation.
P 90 7.4.1:  Discuss here and/or earlier how regulatory siting rules affect this
application and the quality of inference.
P 91 1 1: "not evenly distributed" is too superficial.  Cover more carefully.
P 91 1 20-22: This potential bias can be assessed.  How often is there unequal
representation?
P 92 7.4: The assumption that the monitoring data are spatially representative is
strong  and unevaluated. In addition, there must be a discussion of network
monitoring design in this document and its implications for the inference in this
chapter.  Finally, it is unclear why it obviously follows that there would be more
bias in locations with fewer monitors.  While on average one expects more
coverage to better represent the population,  specifically the degree of bias will
depend more on where the monitors are sited in the various locations. If monitors
aren't placed to represent population exposure, more monitors won't fix the bias
problem.
P 93 top paragraph. I had difficulty with this discussion.
P 95 1 21: Insert "and higher mean" after "variable"
P 97 table 7-19:  Not sure I agree with the bias assessment of both air quality
adjustment types.
P 110 baseline incidence: Previous  years can be used to estimate missing baseline
data.
Table 9-1 and others: Reflect that these numbers are for one year in the title.
P 119 1 11-13:  I suggest replacing tables 9-1 - 9-3 with tables 4-7 - 4-9 from the
appendix.
                                 63

-------
Comments from Dr. Frank Speizer
Comments on the NO2 Risk and Exposure Assessment Document dated August 2008
Submitted by: Frank E. Speizer, MD
Date: Sept.  1, 2008

Chapter 3:
       Discussion of Susceptibility by dependent variables:  The REA document
appropriately presents the conclusions of the ISA as to the adequacy of the evidence (and
were appropriate) lack of adequacy of the evidence for a wide variety of risk groups. For
preexisting disease it appears that only asthma appears to be supported by both
epidemiological and clinical or toxicological studies.  For cardiovascular disease and
diabetes only epidemiological studies are cited. Further comment needs to be presented
that the reason there is not more supporting evidence is because the studies really have
not (and maybe cannot) been done, rather than there being negative clinical or
toxicological studies to report.  For age the only data appear to be epidemiologic and this
point although indicated by the studies cited could be more specifically stated.  The
discussion on the genetics of susceptibility is appropriately qualified. One genetic
marker is identified and though plausible the studies do not provide specific support.
Finally, a little more could have been said to rule out gender as a specific susceptibility
factor as it seems illogical not to indicate that it really must relate to degree of exposure
rather than anything inherent in being male or female.
       Issues of vulnerability:  The Staff has made the definition of vulnerability to
essentially mean increased risk of higher exposure. This is an acceptable definition but
not clear that all would agree. It is also not clear that it will define groups separate from
those that are at increased susceptibility.  There own example of selected ethnic groups at
higher risk of asthma indicates that susceptibility cannot be separated from vulnerability.
       It is not clear why in section 3.7 Staff has focused on the study of Clougherty et al
where violence is mentioned as a modifier.  Surely figure 5.3.1 of the ISA gives the
strong impression that most of the studies of asthma ED admissions are positive, without
suggestion that these are all associated with chronic violence exposure.
Chapter 4.
                                        64

-------
The table presented on page 24 (table 4.1), and text on page 25 provides an
important summary statistic that needs to be carried forward in analysis (I hope I find it).
Two-thirds of asthmatics appear to be responsive as defined as airway responsiveness
with exposure between 0.1 ppm- 0.15 ppm.
Top of page 32: Staff appears to have made a decision that only outcomes that
result in a consensus that the overall effect is either causal or likely causal would be
considered for formal quantitative risk assessment. This is reasonably well supported but
could be considered too conservative. Since the differences between likely causal and
suggestive but not sufficient to be considered causal are for the most part related to the
numbers of studies rather than the potential plausibility of results, it seems to me to
include this category at least initially would be warranted.
Choice of studies: Preferred US based studies. OK but would add or at least
consider some of the European studies, in which NO2 exposure in particular may very
well be better documented.
Ambient over indoor. Again, OK but the blanket rejection of studies that include
an indoor component seems inappropriate.
ED visits and hospital admission, OK there are more data but this works for me.
Single and multiple pollutant models—good.
Range of 0.1-0.3 ppm chosen for constructing health risk estimates appropriate.
However, some of the latter tables go down to 0.050 ppm.
Chapter 5
Indicator of NO2 still reasonable.
Averaging time: Issue here is whether suggestive is enough to make
consideration of a standard appropriate. This is one of the concerns brought about by the
formalization of causation status. It must be remembered that whatever criteria are use
they are a "suggested plan" for formalizing reasoning and should not be considered as
"written in stone".
The fact that previous work lead to the conclusion that a long term standard was
needed and new work in the ISA only indicates there is a suggestive relationship cannot
be the sole criteria for dropping the standard. This issue will need to be discussed further.
A better argument for dropping the standard will have to do with the degree to which the
65

-------
public is protected by an alternative short term standard and whether the levels of
exposure for a long term standard simply do not exist.
Form of standard: consideration of the concentration level rather than frequency
of exceedences seems reasonable. The issue will be how this concentration is determined
in the 1 hour averages and if a longer time period is required. Although the Staff appears
to want to be consistent with what was done with PM, my recollection is that the final
choice of 98%ile over 3 years was an administrative decision and not a scientific one.
The fact that both 98 and 99%ile is being looked at is good.
Level: Upper bound of 0.2 ppm is appropriate, however data presented in Figure
5.1 suggest a lower upper bound should be considered. Given that 5% excess risk for
asthmatics for admission is occurring with a 98 or 99%ile level of 0.09-0.1, respectively,
to allow the upper bound to go higher will most likely double the risk if one speculates
from Figure 5.2 that 30-40% excess symptoms of asthma occurs with a 98 or 99%ile at
0.05. This becomes too high. With regard to the lower bound Staff seems to be
accepting this relatively high risk of asthmatic symptoms as the norm. This is not
reasonable and some margin of safety or measure of uncertainty must be factored in. I
would suggest at least cutting the lower bound by a factor of 2.
Chapter 6
A brief description of apparently what is being done in Chapter 8. If I understand
it what is being done with the "role down' is to bring the calculated estimates of
benchmark levels to below currently found levels. Given the change from Annual
averages to 1-24 hour averages, I would like to be assured, and it will be interesting to
see, if we have sufficient short term measure at Background levels to do this.
(Unfortunately, Chapter 8 did not appear in time for the meeting).
Chapter 7
Approach: The selection criteria of sites seems reasonable Unfortunately, -20 of
sites selected did not meet criteria and in looking over Table 7-1 as a summary of the
sites it is apparent that the sites at altitude (Denver, Colorado Springs) as well as the hot
dry climate of Phoenix in the 2001-06 period did much worse than the sea level sites in
terms of completeness of data. This will need to be discussed and considered in terms
both of exposure representativeness as well as potential for health risk at altitude. This is
66

-------
particularly troublesome since two of these three sites had the most exceedences in the
study period summarized in table 7-2.
Although I may have missed it, it needs to be made clear in discussion of tables 7-
5 to 7-8, and 7-10-7-11, that the number of exceedences at 150 contains the total of those
at levels over 150 (if that is the case). Otherwise looking at the tables can be confusing
and misinterpreted.
Not clear why in tables 7-12, 7-13, 7-16 p98 is not included in the tables. Up to
this point this value is included and is likely to be part of the discussion on the form for
setting the standard.
Uncertainty analysis: Exposure discussion is good with consideration of a
number of issues. More could be said of the "missingness" and potential biased
representativeness of the available data (for example, the concern about the selective
nature of the missingness at altitude). With regard to the statement on page 95 of the
overestimated effects of in-vehicle levels, it seems a little too strong for the available
limited data. It is conceivable that under particular traffic conditions (heavy traffic
resulting in idealing) the in vehicle values could be considerably higher than roadside
measures.
Table 7-19 is particularly troubling. Taken out of context it suggests that there
are very defined bias directions that can be accounted for in each of the source questions.
The text provides much of the discussion of just how complex these issues are and seems
to stand alone. Even though the Table is titled "Summary of qualitative uncertainty..." it
doesn't add very much and might be poorly used to simply confuse the administrator in
assuming that the results for each category of source can be defined better than is really
the case.
Chapter 9
It would appear that the choice of Atlanta to model the risk assessment is quite
appropriate. (It might also be useful to consider another city with more sustained excesses
or more traffic like Los Angeles if time permitted). However, it seems to me that the
selection of one year (2004) as the baseline incidence may not be appropriate. Although
it is probably true that the general incidence of respiratory ED visits does not change over
a period of a few years, the potential for a unique year to be influenced by events such as
67

-------
a high or low flue season is great and therefore averaging an incidence figure over even 2
consecutive years would have been better.
       In tables 9-2 and 9-3 it is not clear what is used in the calculation of the figures
for the "current annual standard".  (I may have missed the explanation of how the daily
max hourly is derived). Is a footnote necessary to clarify this for the reader?
Although it is useful to summarize the uncertainties at the end of the chapter, there is
considerable confusion and variability of the nature of the kinds of uncertainty expressed.
For example uncertainty about the level of certainty of causality is quite different for the
uncertainty of the representativeness of Atlanta to other urban areas. Similarly the
statistical uncertainty due to sampling error is  quite different from the estimate of
baseline incidence.  The more thoughtful discussion of each of the topics as presented in
Section 9.6 is there is not helped by this effort to put it all in one paragraph.
                                         68

-------
Comments from Dr. George Thurston

Prof. George Thurston's Initial Pre-Meeting Comments on CASAC Oxides of Nitrogen
Primary NAAQS Review Panel on EPA's Second Draft of:
Risk and Exposure Assessment to Support the Review of the NO2 Primary National
Ambient Air Quality Standard

Overall, I find the document much improved, and feel that the EPA staffers are to
be congratulated on an excellent job responding to CASAC's concerns regarding the first
draft. I am especially pleased to see the epidemiology-based risk assessment provided in
Chapter 9, and am hoping to see something similar provided in the next draft of the SOx
REA.
My specific responses to the assigned questions for this 2nd draft NOX REA are
provided below.

Characterization of Air Quality (Chapters 2, 6, and 7)
1. To what extent are the air quality characterizations and analyses technically sound,
clearly communicated, appropriately characterized, and relevant to the review of the
primary NO2NAAQS?

RESPONSE: This is done well, in general. However, the reference benchmarks
considered seem to vary in different parts of the document: Page 46 (line 18) says 100,
150, 200, 250, and 300 will be considered, but Table 7-8 considers only exceedances of
150, 200, 250, and 300 (no 100?). In addition, Chapter 5 says that a standard of 50 ppb
will be considered (on line 16), so shouldn't EPA also present the number of exceedances
of that level in Table 7-8 etc., too? Similarly, Table 7-5 and 7-6 only consider 150, 200,
250, and 300 ppb. Shouldn't these all consider the same benchmarks to allow maximum
comparability?
Also, with regard to term usage in the text, the word "historic" is used sometimes
applied to describe data collected in the past (e.g., footnote on pg. 52, line 15 of page 54,
line 14 on pg. 62, etc.), while sometimes it is call "historical" data : I believe the latter is
the more correct usage. Similarly, "less than" is occasionally used (e..g, line 25, pg. 67),
but I think "fewer than" is the more accepted usage when referring to counts.

I think that the explanation of, and justification for, the rolling upward and rolling
downward could be worded a bit more clearly on page 47, lines 21-30. I think it has to
be made clearer that each roll-up/back model is seeking to allow us to see how much
more acute NOX effects protection a potential new short-term standard might provide, vs.
the existing annual standard, and vs. other potential short-term standard options.

-------
on the relevance of this procedure for reviewing the primary NO 2 NAAQS?

I think this is a very important analysis, as it is said that people spend something
like 6% of their time commuting, but get something like 60% of their daily outdoor air
pollution exposures during that time (Source: CARB). I think the approach taken is a
reasonable one.

4. What are the views of the Panel regarding the adequacy of the assessment of
uncertainty and variability?

This is a very good qualitative assessment of the factors affecting uncertainty and
variability, but I would have liked a more quantitative assessment by which to inter-
compare the importance of each factor: are they of differing orders of magnitude? A
figure with a range of possible uncertainty effect (e.g., max-min range of % change)
would be more helpful, if it could be developed.

Characterization of Health Effects Evidence and Selection of Potential Alternative
Standards for Analysis (Chapters 3, 4, 5)

1. The presentation of the NO2 health effects evidence is based on the information
contained in the NO2 Integrated Science Assessment. What are the views of the
Panel on the overall characterization of the health evidence for NO 2? To what extent
is the presentation clear and appropriately balanced?

This section is concise and well done: sufficiently clear and balanced.

The choice of 50, 100, 200, 250, and 300 is appropriate, but not all levels are consistently
considered in all sections of the document. For example, Table 7-5 considers only 15,
200, 25, and 300, ignoring 50 and 100 ppb. Similarly, on page 55, line 7, the minimum
benchmark is stated to be 200, not 50. The lower benchmarks should also be addressed
throughout to allow comparison with other Tables and Figures in the document.
Characterization of Exposure (Chapters 6 and 8):
1. To what extent is the assessment, interpretation, and presentation of the results of the
exposure analysis technically sound, clearly communicated, and appropriately
characterized?

Chapter 6 does a good job of this. Chapter 8 remains to be seen.

2. The second draft assessment document evaluates exposures in Atlanta. What are the
views of the Panel on the approach taken and on the interpretation of the results of
this analysis?

-------
This remains to be seen, as Chapter 8 not provided yet.

3. What are the views of the Panel regarding the adequacy of the assessment of
uncertainty and variability?

I assume that this will be provided in Chapter 8.

Characterization of Health Risks (Chapters 7, 8, 9):
1. Based on conclusions in the final ISA regarding airway responsiveness, we have
expanded the range of potential health effect benchmark values to include 0.1 ppm.
Do Panel members have comments on the range of potential health effects benchmark
values chosen to characterize risks associated with 1-hour NO2 exposures?

I feel that the benchmarks should also include 50 ppb across all analyses, as it is
already used in the epidemiological rise assessment.

2. To what extent are the assessment, interpretation, and presentation of health risk
results technically sound, clearly communicated, and appropriately characterized?

The application of the epi-based risks are sound and clearly and appropriately
characterized.

I highly commend Staff for this effort: this work is excellent, and highly useful in
the standard-setting process. I strongly recommend that this approach be emulated in
future REA's (e.g., SOx).

4. What are the views of the Panel regarding the clarity and adequacy of the discussion
of uncertainty and variability with respect to the characterization of health risks.

This seems adequate, though the addition of quantitative estimates (e.g., order of
magnitude) would help in inter-comparing the potential importance of each source of
uncertainty.
71

-------
Comments from Dr. James Ulttnan
Comments on the Risk and Assessment Document - James Ulttnan (Sept. 8, 2008)

Much thoughtful work went into development of the analyses that are included in this
document. There is, however, an organizational shortcoming in the present document.
In particular, the various analyses of air quality, exposure modeling and health effects are
not evaluated and prioritized as a whole. An effective way of accomplishing this would
be to add a final integrative chapter in which the framing questions posed on pages 1 and
2 of the REA are revisited. I understand that it is not the role of this document to argue
for or against a particular NAAQS, but the results of the analyses need to put in a context
that expedites the policy-setting phase of the standard setting.

Specific comments on chapter 6: There is a lack of clarity in the explanation of the basis
of the equations on pages 38 and 40 (both labeled as eq. 2). This leads to confusion as to
the meaning of all the variables—particularly the difference between Ca and Q,. Given
the high frequency of on-road exceedances, it is important that these equations be better
explained and justified.

Answers to Specific Questions on Air Characterization

1. To what extent are the air quality characterizations and analyses technically sound, clearly
communicated, appropriately characterized, and relevant to the review of the primary NO2 NAAQS?
This was summarized very well in the main body and well-characterized in Appendix A.

2. In order to simulate just meeting potential alternative 1-hour daily maximum standards, we have adjusted
NO2 air quality levels using the same approach that was used in the first draft to simulate just meeting the
current annual standard. To what extent is this approach clearly communicated and appropriately
characterized?
Proportional roll-up and roll-down have appeared to become a common process for
adjusting ambient air quality data. The properties of the data that allow such an
extrapolation should be more explicitly stated. Does the NO2 data set meet these
criterion. Is there any way to validate the process?

3. Because of the impact of mobile sources on ambient NO2, we have estimated on-road
NO2 concentrations. To what extent is the approach taken technically sound, clearly
communicated, and appropriately characterized? Do Panel members have comments
on the relevance of this procedure for reviewing the primary NO2 NAAQS?
Better explanation of the equations in chapter 6 is needed (see my specific comments on
chapter 6.

4. What are the views of the Panel regarding the adequacy of the assessment of uncertainty and variability?
The use of ambient data monitored above ground level may underpredict the actual
concentration at ground level where the subjects reside. It is possible to conduct a
quantitative uncertainty analysis to see how this will affect on-road exceedances?
72

-------
Comments from Dr. Ronald Wyzga

Comments on CASAC Oxides of Nitrogen Primary NAAQS Review Panel on EPA's Second
Draft of:
Risk and Exposure Assessment to Support the Review of the NO2 Primary National Ambient Air
Quality Standard
Comments on Second Draft of:
Risk and Exposure Assessment to Support the Review of the NO2 Primary National Ambient Air
Quality Standard
Overall comment:

There is a very impressive amount of work that has gone into this document; I am clearly
impressed by the volume of materials and the thought behind much of the written material. My
major problem is that it appears as if the various chapters were written by different individuals
and it is unclear how much of the material fits together. Indeed at times the material in the
various chapters/appendices appears to contradict material in other sections. I personally
believe that the authors of this document would have benefited substantially if additional time
had been available to them before the release of this draft.

Characterization of Air Quality (Chapters 2, 6, and 7)
1. To what extent are the air quality characterizations and analyses technically sound, clearly
communicated, appropriately characterized, and relevant to the review of the primary NO2
NAAQS?

The air quality characterizations are well presented and summarized. I have two major
questions, however. The document uses a dichotomy of >/<100 m of a roadway in order to
characterize monitors and their resulting data. Yet it also states that under most circumstances
there is a 90% reduction of the NOx roadway emissions within 10m of a roadway. Given these
information, is the 100m breakpoint appropriate? Should an alternative breakpoint have been
considered? Are there any monitors within 10 m of a roadway? What do they show? Also I
would like to see data on the diurnal patterns of NOx levels, especially around roadways. APEX
says it includes such information, but it is not explicitly covered elsewhere; hence I am confused.
Intuitively it seems to me that the highest exposures could occur around roadways during rush
hour, but I don't see anything in this document that addresses this issue.

It would also help me to understand the data if Tables 7-3 and 4 explicitly indicated the number
of monitors considered for each city; this would give some indication of the extent there was
coverage in a given city.

Tables 7-5 through 7-8, Tables 7-10-7-18 also indicated the total number of opportunities for
there to be exceedances; in pother words, what is the denominator for the rate of exceedance. It
is not clear from the text.
73

-------
2. In order to simulate just meeting potential alternative 1-hour daily maximum standards, we
have adjusted NCh air quality levels using the same approach that was used in the first draft to
simulate just meeting the current annual standard. To what extent is this approach clearly
communicated and appropriately characterized?

The material is clearly presented; I remain uncomfortable with a concept that shows far greater
risks and exposures resulting from conformance to the current standard than with an "as is"
scenario. I am also uncomfortable with the great variability in multiplier constants across
different geographic areas; I have no solution for this problem, but it is disturbing and suggests
to me that there are considerable uncertainties introduced by this approach.

3. Because of the impact of mobile sources on ambient NCh, we have estimated on-road NCh
concentrations. To what extent is the approach taken technically sound, clearly communicated,
and appropriately characterized? Do Panel members have comments on the relevance of this
procedure for reviewing the primary NChNAAQS?

See my response to question 1. I would like to see more data on diurnal variability near
roadways and from monitors closer than 100m roadways.

I am also unclear about how values of the on-road factors (m) were randomly assigned? What
distributions were assumed for ml p. 58,11. 23-24.

4. What are the views of the Panel regarding the adequacy of the assessment of uncertainty and
variability?

See responses to the above questions. Some of my questions may need to be addressed in the
uncertainty section.

Characterization of Health Effects Evidence and Selection of Potential Alternative
Standards for Analysis (Chapters 3, 4, 5)

1. The presentation of the NCh health effects evidence is based on the information contained
in the NCh Integrated Science Assessment. What are the views of the Panel on the overall
characterization of the health evidence for NCh? To what extent is the presentation clear
and appropriately balanced?

I believe it is accurate and fair.
2. The specific potential alternative standards that have been selected for analysis are based on
both controlled human exposure studies and on epidemiological studies conducted in the United
States. What are the Panel's views on the appropriateness of these potential alternative standards
(in terms of indicator, averaging time, form, and level) for the purpose of conducting air quality,
74

-------
exposure, and risk assessments and on the rationale used to select them for that purpose?

There is inconsistency within the document about the level chosen for analysis. Personally I
believe that a shorter-term exposure (e.g., 1 hour) is more appropriate than a longer-term one
(24-hour); this also allows greater congruence between the human clinical and epidemiological
studies. If I recall correctly, didn't the Clean Air Act Amendments of 1977 suggest that a
standard of Shours or less be considered; is this still relevant?

Characterization of Exposure (Chapters 6 and 8):

1. To what extent is the assessment, interpretation, and presentation of the results of the
exposure analysis technically sound, clearly communicated, and appropriately
characterized?

I have mixed feelings here. I read the Chapter 6 and feel that I have a good understanding of the
analytical approach taken. Then I read the Appendix, which is also well-written, but seems to
take a more detailed approach than is conveyed in the Chapter. For example, I am unclear how
and the extent to which APEX is applied from reading the document per se.
2. The second draft assessment document evaluates exposures in Atlanta. What are the
views of the Panel on the approach taken and on the interpretation of the results of this
analysis?

I am basing my comments on Appendix C, which I basically like.

In the discussion of the Tolbert et al. and Peel et al. studies, it should be noted that the
authors considered an a priori lag structure for hypothesis testing purposes. Peel et al does
present lag-specific results for asthma and upper respiratory illness, which show that most
response is for lags 0-2 although there is also some response for NO2 for lag 3 as well.

I would ask whether the risk assessment should consider time of day. Near-road exposures
could be considerably higher during certain periods, and I wonder given the capabilities of
APEX can this be addressed?

It should be noted the Peel et al. and Tolbert et al. papers were part of a larger study (ARIES)
that considered other health endpoints as well. In particular, daily respiratory mortality and
unscheduled physician visits were considered in papers by Klemm et al and Sinclair and
Tolsma. Cardiovascular responses were also addressed in other papers. It should be noted
that NO2 was highly correlated with some components on PM, especially EC, in this dataset.

Extrapolation of the underlying incidence data for emergency department visits to other areas
could be impacted by socio-economic considerations. For those without health insurance or
the ability to pay for healthcare, ED visits are more frequent as those with other options seek
them out before visiting an ED.
75

-------
3. What are the views of the Panel regarding the adequacy of the assessment of uncertainty
and variability?

See responses to the above. It should be noted that the statistical significance of results is
impacted by the presence of other pollutants in the model. The Peel et al. results particularly
found ozone to be of concern for respiratory endpoints.
Characterization of Health Risks (Chapters 7, 8, 9):
1. Based on conclusions in the final ISA regarding airway responsiveness, we have expanded the
range of potential health effect benchmark values to include 0.1 ppm.
Do Panel members have comments on the range of potential health effects benchmark values
chosen to characterize risks associated with 1-hour NCh exposures?

There appears to be inconsistencies in the document about benchmarks or exposure levels
utilized.

2. To what extent are the assessment, interpretation, and presentation of health risk results
technically sound, clearly communicated, and appropriately characterized?

See above comments. I also await the results of risk analysis for human clinical study data.
3. A focused risk assessment has been conducted for emergency department visits in Atlanta,
GA. To what extent are the assessment, interpretation, and presentation of health risk results
technically sound, clearly communicated, and appropriately characterized? What are the views of
the Panel on the approach taken and on the interpretation of the results of this analysis?

See above comments.
Minor comment:
P. 45,1. 9: "three"
76

-------