UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
September 27, 2001
EPA-SAB-C AS AC-LTR-0 1-001 OFFICE OF THE ADMINISTRATOR
SCIENCE ADVISORY BOARD
Honorable Christine Todd Whitman
Administrator
U.S. Environmental Protection Agency
1200 Pennsylvania Avenue, NW
Washington, DC 20460
Subj ect: Review of the Air Quality Criteria Document for P articulate Matter:
Second External Review Draft (EPA 600/P-99/002bB): A CAS AC Review
Dear Governor Whitman:
The Clean Air Scientific Advisory Committee (CAS AC) of EPA' s Science Advisory
Board, supplemented by expert consultants (together referred to as the "Panel"), met on July 23-
24, 2001 to review the March 2001 draft document, Air Quality Criteria for Paniculate Matter -
Second External Review Draft (EPA 600/P-99/002bB) (EPA, 2001), in a public meeting in
Research Triangle Park, NC. This was the second CAS AC review of the draft Criteria
Document (CD) for particulate matter (PM) in the current cycle for reviewing the National
Ambient Air Quality Standards (NAAQS) for PM. CAS AC review of this document is required
by section 109 of the Clean Air Act.
As noted below, the CASAC could not come to closure on this draft document and has
requested that the Agency revise the draft for another review.
1. BACKGROUND
The Panel reviewed the First External Review Draft of the PM Criteria Document (EPA,
1999) in December 1999, focusing primarily on the organization, structure, and presentation of
material in the draft document. It was understood that this was an early incomplete draft and
that additional information would be incorporated in subsequent drafts. There was no intent nor
expectation that the Panel would close on the draft document at this stage of its development.
The Panel was generally complimentary about the content and quality of this draft, while noting
the need for considerable development both in structure and content (CASAC, 2000).
The EPA has assessed approximately 1800 new references published since the October
1999 First External Review Draft (EPA, 1999) was released for CASAC review. This
assessment, along with responses to CASAC's earlier comments (CASAC, 2000), are
incorporated into the Second External Review Draft.
-------
2. REVIEW OF THE APRIL 2001 SECOND EXTERNAL REVIEW DRAFT
CRITERIA DOCUMENT FOR PM
The Panel was impressed with the revised version of the Draft Criteria Document as
compared to the version that was reviewed in December 1999 (EPA, 1999). It is clear that the
comments provided by the Panel on the prior draft were seriously considered and efforts made to
address the issues and concerns that were raised. A large body of new literature has been
published in the intervening time and the staff has clearly made a substantial effort to incorporate
as much of it as appropriate. The Panel felt that this version represented a significant step
toward achieving an acceptable summary of the available science. The Panel was unanimous in
its view that the document was not yet ready for closure, but it was our opinion that appropriate
modifications to the present document should permit closure.
In this report, we summarize the major questions, concerns, and issues that were
expressed regarding the Draft Criteria Document. There were extensive comments from the
members of the CAS AC PM Panel and they are provided in appendices to this report. These
comments are considered an integral part of this report, and the Agency is encouraged to take
them into consideration in making further revisions to the document.
2.1 Major Issues
A major global concern of the Panel was the lack of adequate linkage among the chapters
so that the Document presents a consistent and coherent summary of the science. We suggest
that the chapters could be realigned using the paradigm presented by the National Research
Council Committee on Research Priorities for Airborne Particulate Matter (NAS, 1999). The
paradigm begins with emissions from sources and follows the path to concentrations, exposure,
dose, and effects. We suggest moving the dosimetry chapter (Chapter 7) to precede the
epidemiology chapter (Chapter 6) and it should be linked to exposure as described in Chapter 5.
This framework could also serve as a basis for organizing the synthesis chapter (Chapter 9).
There are other possible unifying themes that could be used, and we suggest that using such an
approach would provide more coherent flow to the document. The approach should be described
in Chapter 1 and would provide a clearer path through the wealth of new information that needs
to be reviewed and synthesized into a picture of our current state of knowledge.
A second major concern was the limited information on coarse thoracic particles (PM10_
25) in many of the chapters. The Court of Appeals (1999) decision in American Trucking
Associations vs. EPA precludes having both PMj0 and PM25 standards. Thus, it is critical that
the document clearly defines the information that is and is not available on PM,10.2 5). There is a
clear distinction made between the roles of PM25 and PM(10_2 5) in Chapters 4 and 6. However,
Chapters 2, 3 and 5 do not adequately distinguish the measurement methods, ambient
concentrations and distributions, and exposure issues associated with the two size ranges. There
is also no discussion of the toxicology of thoracic coarse particles in Chapter 8. Although
extensive information may not be available, it is critical that the document clearly describes what
scientific information is and is not known about PM,10_2 5).
-------
A third major issue is the need for more emphasis on the new information concerning
better time resolution in measurements. Our understanding of the atmospheric processes,
exposure, and epidemiology is currently limited by the fact that most data represent integrated
24-hour samples. One of the major differences between PM and the gaseous criteria pollutants
is the greater difficulty in making hourly or even shorter interval measurements. Such data
allow the evaluation of effects at various integrating times. Such evaluations are currently not
possible because of the limited amount of shorter sampling interval data.
However, major improvements in semi-continuous measurement methods for PM are
now beginning to provide one-hour or better time resolution. The evaluation of these
instruments is a major focus of several of the current Supersite activities. There have been
publications showing extremely high short-term excursions in the PM concentrations. For
example, Michaels (1996) does discuss some of the developments in measurement methods, but
there is not an explicit discussion of such data in the exposure or epidemiology chapters (current
Chapters 5 and 6). There have been a limited number of epidemiological studies based on short
term data which were not explicitly evaluated (e.g., Morgan et al, 1998; Simpson et al., 1997).
Thus, there is a developing information base that could be used in the future to reexamine the
averaging time interval for the short-term PM NAAQS. It would be useful for this version of the
PM Criteria Document to more explicitly recognize this direction. For example, it would be
useful to have a subsection in Chapter 6 discussing the limited epidemiological studies that have
been made with short interval data with a parallel section in Chapter 5 evaluating the
complementary exposure information.
The document provides a detailed description of our current knowledge, but does not
fully discuss its limitations. For example, the focus of epidemiological studies has been 24-hour
integrated, PM10 mass since there is a large available data base. Additional studies have made
use of the data that have been collected, but there cannot be extensive studies of specific
chemical constituents, different integrating times, etc., because the data do not exist. A broader
discussion of the limitations of our current state of knowledge particularly in the context of the
extensive ongoing research and monitoring programs that EPA is supporting would be helpful in
providing a better perspective on the current status of our knowledge of PM and its effects.
2.2 Chapter Issues
With respect to individual chapters, the specific Panel member comments provide
considerable guidance for the needed changes (see Appendix A). There were no major structural
issues with Chapters 2 and 3 other than the general framework question raised above. However,
Chapter 3 must include a stronger discussion on the emission of precursors and formation of
Secondary Organic Aerosol (SOA). At present, the emissions section almost exclusively
discusses primary particles, which do not represent most of the fine particle mass in many
locations. These chapters help to provide the background to the exposure and effects chapters
and need to reflect that direction.
Chapter 4 (Environmental Effects of Particulate Matter) was a substantial improvement
from the chapter on welfare effects in the earlier version of the CD. One important issue that is
not currently included in the chapter was the potential for effects of PM on urban vegetation.
-------
The chapter would be improved if the ecosystem and other welfare effects presentations were
more focused on information that addresses key questions in the setting of standards, linking
exposure to effects. Currently, the document addresses putative effects without an adequate
discussion of what PM exposure and deposition are in the environment. Such a setting would
make it clearer as to purpose of the chapter and would also link the chapter on the environment
to the chapter on exposure. Finally, the purpose of the current economics discussion is unclear.
It needs to be either better focused or potentially eliminated.
In Chapter 5 (Human Exposure to Particulate Matter and its Constituents! a clearer
discussion of the chapter goals within the overall framework of the document would be helpful
in evaluating the information being presented. The processes that lead to exposure are non-
linear in nature, but the chapter appears to suggest that they are linear. There needs to be more
emphasis on distributions of exposure as well as point estimates of the average exposure. It also
needs to provide a better description of exposure error so that it provides the background for the
discussions of exposure error problems in the epidemiology chapter. It needs to better reflect the
complexity of the relationships between ambient concentrations and personal exposure. The
chapter also needs to provide the background on the data needed to estimate dose, and
particularly dose to target organs like the heart.
One of the major areas of the discussion was the epidemiology presented in Chapter 6
(Epidemiology of Human Health Effects from Ambient Particulate Matter! We recognize the
problem of summarizing such a large body of work and then presenting and evaluating a limited
number of specific studies that are most relevant to the understanding of the relationships
between exposure to PM and health effects. It would be useful to establish a well defined set of
criteria for choosing studies selected for detailed discussion. A clear description of the selection
criteria needs to be added to the chapter. There also needs to be a uniform strategy for
evaluating the studies that are discussed and for making comparisons among studies. We
recognize that it is important for there to be careful evaluation of the major new studies.
However, without well defined criteria for evaluating the studies, the presentation and evaluation
of the studies and their results may not be uniform. It is also important to be very careful in the
definition and consistent use of terms. Concepts such as "confounding" and "modifying"
especially need to be defined and used consistently.
We suggest that additional summary tables be provided that present the key features of
the studies and their findings. It may be useful to move the current detailed tables to appendices
to the chapter and utilize the summary tables in the text.
One of key findings of recent large studies (Samet et al., 2000a; Samet et al, 2000b) is
that there is heterogeneity in the relationships between PM10 and health outcomes among a
number of locations. There are several possible explanations for this observation and these
possibilities need to be presented. Is the apparent heterogeneity a random effect or is there real
systematic variation in PM10 toxicity from location to location? Do multiple populations need to
be considered separately, such as those living in air conditioned vs. non-air conditioned
buildings? Are there differences among locations because of differences in the relative sizes of
these multiple populations? Although the underlying cause of the heterogeneity is not yet
-------
known, the potential causes deserve more discussion with regard to their implications for
estimating exposure/response relationships.
The chapter currently contains little discussion of what is known regarding the
relationship between PM exposure and cancer. Given the extensive discussion of this topic with
respect to diesel exhaust particles and the ubiquitous presence of diesel as a component of the
ambient aerosol, this issue should not be ignored in the Criteria Document.
Finally, discussion of epidemiology studies on morbidity/mortality effects on the fetus,
neonates and infants should be expanded to better reflect the current state of knowledge.
Chapter 7 (Posimetry of Particulate Matter) provides an extensive discussion of
dosimetric models, but was no effort to use this knowledge to connect information on exposure,
dose, and health effects suggested by toxicology or epidemiology. The connections could be
greatly improved by moving this chapter to follow the exposure chapter, by including illustrative
examples of relationships between particle size and regional deposition, and by providing
examples of the magnitude of deposited and retained doses resulting from environmental
exposures. This information is critical to setting the stage for evaluating how lexicological
information might apply to the epidemiological observations in subsequent chapters.
Chapter 8 (Toxicology of Particulate Matter) is very selective in the choice of
lexicological studies presented. Again a clearly defined set of criteria for choosing studies is
needed as well as a discussion of how the toxicology helps to provide understanding of the
relationships observed in the epidemiological studies. The relationship of this chapter with the
rest of the document needs to be better defined.
The chapter should not only review key recent findings and advances since the last
Criteria Document, but it also needs to discuss how the toxicology results help our understanding
of the exposure-dose-response relationships observed in the epidemiological studies. An
important facet of the discussion should be the relationship between doses used in the different
lexicological approaches and doses received by people from environmental exposures. With
these modifications, the value of the chapter within the context of the entire document will be
greatly enhanced.
As indicated previously, the chapter needs to point out data which help our understanding
of the toxicology of thoracic PM,^ 5). It is not clear how much of such information exists, but
to the extent it is available, it needs to be presented and the limitations of our knowledge of
PM(10_2.5)
It is especially important that the lexicological studies using concenlraled ambienl
particles (CAP) be discussed Ihoroughly. Such sludies are potentially very valuable; however,
the CAP sludies differ from sludies of laboratory-generated atinospheres in thai Ihe cells and
animals are nol exposed to material lhal can be predicted in advance and reproduced as desired.
Because the composition of CAP varies in bolh time and location, a thorough physical-chemical
characterization is necessary to compare resulls among sludies or among exposures within
-------
studies, or to link particle composition to effect. The chapter should portray this issue and note
that while studies incorporating sufficient characterization have high value, those lacking
characterization have minimal value. Care also needs to be taken in discussing studies of
ultrafme particles. If, for example, the concentrations used in the experiments were sufficiently
high that particle coagulation would have been very rapid, it can be assumed that the animals
were not actually exposed to ultrafme particles unless data are presented showing otherwise.
The chapter provides an extensive introduction to molecular dosimetry, but there is no
effort to relate this information to possible mechanisms of action leading to the observed adverse
health effects or effects at the tissue and whole organism level. If this information has any
relevance to the standard setting process, there needs to be a clear description of how such
information informs the exposure/effects relationships or the extrapolation of the high dose
effects to likely effects at ambient concentrations.
Chapter 9 (Integrative Synthesis of Key Points: Particulate Matter Atmospheric Science.
Air Quality. Human Exposure. Dosimetry. and Health Risks^ requires major revisions. As EPA
Staff noted during the meeting, there was insufficient time to really provide a truly integrated
synthesis of the information in this chapter. The extensive discussion at the meeting along with
our individual comments should provide assistance in revising this chapter from a summary to a
synthesis.
We commend the EPA staff for the effort and attention to detail in preparing the current
draft Criteria Document. We think it has brought together a wealth of new information and
serves as a sound basis for a revised version that we look forward to reviewing in the near future.
We look forward to your response to our advice.
Sincerely;
/ Signed /
Dr. Philip K. Hopke, Chair
Clean Air Scientific Advisory Committee
-------
REFERENCES CITED
CASAC. 2000. Review of the Draft Air Quality Criteria for Particulate Matter. US EPA
Science Advisory Board, Clean Air Scientific Advisory Committee (CASAC),
Washington, DC February 15, 2000.
Court of Appeals. 1999. American Trucking Associations et al. vs. US EPA, 97-1440, US
Court of Appeals for the District if Columbia Circuit.
EPA. 1999. Air Quality Criteria for Particulate Matter. US EPA Office of Research and
Development (ORD), EPA 600/P-99/002a, Washington, DC, October 1999.
EPA. 2001. Air Quality Criteria for Particulate Matter TSecond External Review Draft! US
EPA Office of Research and Development (ORD), EPA 600/P-99/002bB, Washington,
DC, March 2001.
Michaels, R. A. 1996. Airborne Particle Excursions Contributing to Daily Average Particle
Levels may be Managed via a 1 hr Standard, with Possible Public Health Benefits.
Aerosol Sci. Technol. 25:437.
Morgan, G.; Corbett, S.; Wlodarczyk, J.; Lewis, P. 1998. Air pollution and daily mortality in
Sydney, Australia, 1989 through 1993. Am. J. Public Health 88:759-764.
NAS. 1999. Research Priorities for Airborne Particulate Matter -1 - Immediate Priorities and a
Long Range Research Portfolio. National Research Council. National Academy press,
Washington, DC 195p.
Samet, J. M.; Dominici, F.; Zeger, S. L.; Schwartz, J.; Dockery, D. W. 2000a. National
morbidity, mortality, and air pollution study. Part I: methods and methodologic issues.
Cambridge, MA: Health Effects Institute; Research Report no. 94.
Samet, J. M.; Zeger, S. L.; Dominici, F.; Curriero, F.; Coursac, I; Dockery, D. W.; Schwartz, J.;
Zanobetti, A. 2000b. The national morbidity, mortality, and air pollution study. Part II:
morbidity, mortality, and air pollution in the United States. Cambridge, MA: Health
Effects Institute; Research Report no. 94.
Simpson, R. W.; Williams, G.; Petroeschevsky, A.; Morgan, G.; Rutherford, S. 1997.
Associations between outdoor air pollution and daily mortality in Brisbane, Australia.
Arch. Environ. Health 52:442-454
R- 1
-------
NOTICE
This report has been written as part of the activities of the EPA Science Advisory Board,
a public advisory group providing extramural scientific information and advice to the
Administrator and other officials of the Environmental Protection Agency. The Board is
structured to provide balanced, expert assessment of scientific matters related to problems facing
the Agency. This report has not been reviewed for approval by the Agency and, hence, the
contents of this report do not necessarily represent the views and policies of the Environmental
Protection Agency, nor of other agencies in the Executive Branch of the Federal government, nor
does mention of trade names or commercial products constitute a recommendation for use.
Distribution and Availability: This EPA Science Advisory Board report is provided to the EPA
Administrator, senior Agency management, appropriate program staff, interested members of the
public, and is posted on the SAB website (www.epa.gov/sab). Information on its availability is
also provided in the SAB's monthly newsletter (Happenings at the Science Advisory Board).
Additional copies and further information are available from the SAB Staff [US EPA Science
Advisory Board (1400A), 1200 Pennsylvania Avenue, NW, Washington, DC 20460-0001; 202-
564-4533].
-------
U.S. Environmental Protection Agency
EPA Science Advisory Board
Clean Air Scientific Advisory Committee
CASAC Particulate Matter Review Panel*
CHAIR
Dr. Philip Hopke, Clarkson University, Potsdam, NY
Also Member: Executive Committee
Research Strategies Advisory Committee
CASAC MEMBERS
Dr. John Elston, State of New Jersey Department of Environmental Protection and Energy,
Trenton, NJ (Did not attend meeting)
Dr. Joe Mauderly, Lovelace Respiratory Research Institute, Albuquerque, NM (outgoing Chair)
Dr. Frederick J. Miller, CUT Centers for Health Research, Research Triangle Park, NC
Dr. Arthur C. Upton, UMDNJ-Robert Wood Johnson Medical School, New Brunswick, NJ
Dr. Sverre Vedal, National Jewish Medical and Research Center, Denver, CO
Dr. Barbara Zielinska, Desert Research Institute, Reno, NV (Did not attend meeting)
OTHER SAB MEMBERS
Dr. Paul J. Lioy, UMDNJ - Robert Wood Johnson Medical School, Piscataway, NJ
Member: Advisory Council on Clean Air Compliance Analysis
Dr. Morton Lippmann, New York University Medical Center, Tuxedo, NY
Member: Executive Committee
CONSULTANTS
Dr. Jane Q. Koenig, University of Washington, Seattle, WA
Dr. Petros Koutrakis, Harvard University, Boston, MA
Dr. Allan Legge, Biosphere Solutions, Calgary, Alberta
Dr. Roger O. McClellan, Albuquerque, NM
Dr. Gunter Oberdorster, University of Rochester, Rochester, NY (Did not attend meeting)
Dr. Robert D. Rowe, Stratus Consulting, Inc., Boulder, CO
Dr. Jonathan M. Samet, Johns Hopkins University, Baltimore, MD
Dr. George E. Taylor, George Mason University, Fairfax, VA
Mr. Ronald White, National Osteoporosis Foundation, Washington, DC
Dr. Warren H. White, Washington University, St. Louis, MO
Dr. George T. Wolff, General Motors Corporation, Detroit, MI (Did not attend meeting)
-------
SCIENCE ADVISORY BOARD STAFF
Mr. A. Robert Flaak, Designated Federal Officer, USEPA Science Advisory Board,
Washington, DC
Ms. Rhonda Fortson, Management Assistant, USEPA Science Advisory Board, Washington,
DC
* Members of this SAB Panel consist of
a. SAB Members: Experts appointed by the Administrator to serve on one of the SAB Standing
Committees.
b. SAB Consultants: Experts appointed by the SAB Staff Director to a one-year term to serve on ad hoc
Panels formed to address a particular issue.
c. Liaisons: Members of other Federal Advisory Committees who are not Members or Consultants of the
Board.
d. Federal Experts: The SAB charter precludes Federal employees from being Members of the Board.
"Federal Experts" are federal employees who have technical knowledge and expertise relevant to the subject matter
under review or study by a particular panel.
Ill
-------
APPENDIX A - INDIVIDUAL PANELIST WRITTEN COMMENTS
Note: These are the final written comments provided by individual Panelists following
the July 23-24, 2001 meeting. They are included here to present the full range of opinion and to
document all edits suggested by Panelists. These are individual comments and do not
necessarily represent the views of the Clean Air Scientific Advisory Board (CASAC) nor the
EPA Science Advisory Board (SAB).
Table of Contents
Philip Hopke, PhD A - 1
Frederick Miller, PhD A - 4
Arthur C. Upton, MD A - 10
Sverre Vedal, MD A - 10
Barbara Zielinska, PhD A - 17
Joe L. Mauderly, DVM A - 20
Allan Legge, PhD A - 33
Paul J. Lioy, PhD A - 35
Mort Lippmann, PhD A - 40
Jane Q. Koenig, PhD A - 50
Petros Koutrakis, PhD A - 52
Roger McClellan, DVM A - 57
Giinter Oberdorster, PhD A - 64
Robert Rowe, PhD A - 70
Jonathan Samet, MD A - 73
George Taylor, PhD A - 78
Ronald H. White, M.S.T A - 80
Warren White, PhD A - 81
George T. Wolff, PhD A - 85
* Did not attend the July 23-24, 2001 meeting.
-------
Philip Hopke, PhD
Guiding Principles
There are two major questions facing EPA with respect to PM standards. First, they cannot have
both a PM10 and a PM25 standard. There can be one or the other. Thus, what science is
available to help make this choice. One suspects that the answer is that there is more scientific
support for a PM2 5 standard.
This leads to the second issue: Should there be a standard for larger particle sizes? If so, what
should be the indicator, concentration, averaging times, and statistical form of the standard?
Thus, there needs to be a clear summary of the science that is or is not available to answer these
questions
The current document is much too diffuse in its focus and fails to provide enough real evaluation
to provide clear statements of what is and is not know about the health and welfare effects of
various components of the ambient aerosol.
Chapter 2
Page 2-2, Lines 9-15: Complements on the good definitions.
Page 2-13, LinelS, Prior to 1987 (not 1997)
Page 2-17, lines 2 to 15, There needs to be discussions of the nucleation events that are being
observed by several groups (O'Dowd; Kulmala) of nucleation events. Most have been reported
for remote areas like Macehead and Hyytila, but Kulmala has indicated to me that they have seen
such events in Helsinki and it may be that McMurry has also seen them in Atlanta. Thus,
homogeneous nucleation can be an important process for new particle formation although the
details need to be investigated.
Page 2-18, lines 1 to 7: Kulmala suggests that ternary nucleation with NHg along with sulfuric
acid and water is necessary to observe nucleation events. This is mentioned in their Nature
article from last year. I can provide other references if needed.
Page 2-72, lines 7-12: The CASAC Subcommittee on Particle Monitoring has raised questions
about the process of demonstrating equivalency for continuous monitors and the need to greater
flexibility in bringing new technology into the compliance monitoring program. There needs to
be some recognition of these problems here.
Page 2-72, lines 14-23: The 1996 CASAC PM Monitoring Subcommittee recommended a
performance standard rather than a design standard. The fact that EPA made the PM10 criteria
too loose does not mean that a performance standard would not work. It would not be as stifling
on technology development.
The discussion fails to really highlight the fact that the FRM provides high precision
measurements of totally unknown accuracy with respect to airborne particle mass concentration.
It is time to face the need to really define what you want to measure and it should not simply be
duplicating measurements made with dichotomous samplers that have now been discredited as
being adequate for measurement of PM2 5. The whole FRM program is full of contradictions and
false assertions.
The discussion fails to discuss the need to greater time resolution in mass measurements. Right
now we have no idea what the proper time interval is for setting a standard. We use 24 hours
because that is what has been measured, but this interval has no basis in health effects.
A- 1
-------
Page 2-77, line 19-22: Should mention new developments in Synchotron XRF. It appears to
have much higher sensitivity that will permit much higher time resolution.
Page 2-81, lines 5 and 6: The uncertainties in the reported slope and intercepts should be
reported. There should always be uncertainty estimates with any reported data or data analyses.
Page 2-82, line 6: There are a number of references to papers in the PM2000 issue of Aerosol
Sci. Technol. This issue was published in January 2001 and thus, all of these references like
Chow et al. need to be updated.
Page 2-88, line 20: Same date problem. Also need to point out why we want better time resolved
data. We have been adjusting the time interval of the O3 standard because we have the data to
aggregated the data at various time intervals. We do not have that advantage with PM unless
"continuous" monitors are more widely used.
Page 2-94, lines 1-10: Fergenson et al. (In press, 2001) have shown that quantitative estimates
can be made of the aerosol composition from the single particle MS data.
Page 2-96, lines 1-12: Need to add the Continuous Sulfate Monitors that are now available.
There is no discussion of measurement methods for Coarse particles and how you would get
samples that could be used to characterize the particle compositions, do source apportionment,
etc.
Chapter 3.
There needs to be a clearer focus on what is know about the concentrations, sources,
distributions, compositions of coarse particles. There is too much emphasis on PM10 and not
enough on coarse. Chapters 4 and 6 single out the effects of coarse. This chapter needs to
coordinate with the background atmospheric behavior so that the context for the subsequent
chapters has been created. There is only a short section on Page 3-11. If this is all the
information that is available, then there has to be a clearer statement that information on PM10_2 5
is limited and that this is all there is that is known regarding its distribution nationwide.
Page 3-19, Because of the very large variability in the relative amounts of fine and coarse
particles, it is misleading to provide only a mean PM2 5/PM10 ratio. The only sensible way to
provide such information is with distributions.
Page 3-26, lines 4 -10: Need to have the discussion of active homogeneous nucleation in Chapter
2 and need a pointer back to that discussion here.
Page 3-36, lines 10-21: A major part of secondary particles in the east and maybe elsewhere is
secondary organic aerosol. Thus, there needs to be a discussion of the sources of the precursor
gases that include both anthropogenic and biogenic sources.
Page 3-37, lines 23-31: Should add a discussion of UNMIX. EPA has invested in getting this
model ready for use and should be included in this discussion.
Page 3-42, lines 11-28: This discussion is out of date. More recent data suggests a much more
important role for SO A in the eastern US ambient aerosol.
Page 3-51, lines 8-11: Agriculture is a major source of ammonia. Take for example the Chino
feedlot. With dairy farming widespread in the NE and North Central US, it could be important
in the eastern US as well.
A-2
-------
Page 3-54, lines 10-15: Point out that CEMs are available to look at SO2. Other pollutants could
be done if it were important (like needing verification for trading rights).
Page 3-55, lines 18-21: Later in the chapter, it is pointed out that the El gives average emissions
rather that the specific emission rates needed for modeling. It would be better to start that
discussion here.
Page 3-56: Biogenic sources are important for precursors for SO A. What is known about those
emission inventories. How good is BIES?
Chapter 4
This chapter should be complemented for making clear distinctions between the effects of fine
and coarse particles.
Page 4-8, lines 19-28: The outflow of Asian dust to the mid-Pacific is now thought to be a
critical source of micronutrients to the phytoplankton there. There should be a note of the work
of Bob Duce and coworkers in this area.
Chapter 5
This chapter is very weak on describing what is known regarding the exposure to PM10_2 5. Since
Chapter 6 highlights the epidemiology of PM10_2 5 separate from PM25 and PM10, it is important
that there be parallel coverage of the exposure in this chapter to set the scene for that discussion.
Page 5-54- 5-55, What can be said more definitively about the penetration efficiency of PM10_2 5?
How does this affect our ability to relate PM10_2 5 to total exposure and thus to adverse health
effects? There needs to be a focused section on this issue comparing and contrasting the
difference between PM10_2 5 and PM2 5 in this respect.
Pages 5-66, lines 23 - 26: UNMIX is rather different from PMF. It uses an eigenvector analysis
which has been shown to provide non-optimal data point weighting because it fundamentally
assumes homoscedastic data that is generally a poor assumption.
Page 5-92, lines 14 to 20: Is this discussion only about PM2 5 or about both PM2 5 and PM10_2 5
Page 5-92, lines 25-31: Because of the poorer penetration of coarse particles and the much larger
spatial inhomogeneities, there will be much larger errors in the exposure estimates for coarse.
There needs to be a discussion of this problem here.
Chapter 6
Given the variability of PM25/PM10, there needs to be a discussion of whether or not the study of
PM10 will really provide any insights into the quantitative risk assessment of PM10_2 5.
Chapter 7
No comments
Chapter 8
There is no real discussion of organic aerosol mechanisms. The work that has been done is very
important and useful, but there are other types of particles particularly primary and secondary
organics. Since we can only separate and identify the specific compounds in about 20% of the
organic mass under the best of conditions, it makes it very difficult to fully characterize the
lexicological properties of these particles. It is very difficult to generate secondary organic
aerosol in the lab and thus, providing reproducible conditions for exposing animal models is
difficult.
A-
-------
Page 8-1, lines 26-28: Combustion aerosol are rarely "dominant" particle types. In most of the
country, the dominant fine particle types are secondary aerosols. They do often derive from
combustion generated SO2, NOX, and some organic compounds, but there are also important
biogenic precursor compounds. Thus, there needs to be some rewording of this sentence.
Page 8-87: There is no separate discussion of the toxicology of coarse particles even to say that
nothing is know regarding the toxicology of such particles.
Chapter 9
Page 9-10, linesl? to 29: This data is very outdated. The needs to be examination of the
extensive data showing the frequent presence of an Aitken mode in rural aerosols as well as the
occasional nuclei mode as well.
Section 9.3 has no discussion of coarse mode sampling.
Section 9.3.2.3: why is there a discussion of spatial variability of PM25 without parallel
Frederick Miller, PhD
discussions of PM10 and PM10_25?
Chapter 6 - General Comments:
The chapter in its current form represents an extensive review of the available literature from
epidemiological studies on the effects of parti culate matter. The organization of the chapter into
the major subheadings is appropriate. As one reads the chapter, there is a tendency for the PM^ 5
effects to be discussed in great detail and for the conclusion to be drawn that PM2 5 is of more concern
than PMjQ_9 5. However, as the chapter develops, studies are presented showing the potential for coarse
particles to nave an effect. The balance of this discussion should be examined in particular as it is brought
forth to the synthesis chapter.
Throughout the document values of PM2 5 and PM^Q_2 5 are presented. However, the document fails to
make clear when PM2 5 is a derived measurement vs. a direct measurement. This is critically important
for standard setting purposes as correlation analyses provide different weight of evidence on average
values compared to direct measurement. To help the reader in evaluating the strengths of the different
studies, under Study Description it would be of value to simply indicate if measurements on exposure
levels are direct or derived measurements.
Specific Comments: 3
p. 6-20 The Schwartz (2000c) study in the table reports a PM2 5 mean of 15.6 mg/m .
The study was conducted using data from 1979 1986. Were PIV^ 5
measurements available in the late 1970s? How was the mean for PJV^ 5
arrived at?
p. 6-23 The entry for the Smith study under Results and Comments brings up the topic
of threshold. No discussion of this study follows until page 6-247. It is not clear
why the emphasis in the discussion of Table 6-1 should be restricted to multi-
city studies, particularly when individual studies bring up topics that are
important for standard setting such as the concept of threshold, the statistical
averaging time, or additional potential sensitive subpopulations.
p. 6-53 The figure presented here shows that 10 of 13 PlVt^ 5 studies and 4 of 13 coarse
mode studies show statistical significance. While this gives greater emphasis to
the importance of both the fine and the coarse mode for standard setting, the
discussion in the text does not bring this point out as strongly as it should be.
For example, in the section on crustal particle effects on page 6-56, the studies
are discussed with a tendency for not showing an effect and little discussion is
involved for the four studies that did demonstrate effects of coarse mode
particles.
p. 6-77, 1. 826 The slant towards interpretation of PM2 5 and relative dismissal of the
importance of the coarse mode is continued in this section here on fine and
A-4
-------
coarse particle effects. The paragraph clearly comes across as there may be
some PM coarse mode effects but they probably are specific in location and
they may even be due to biogenetically-derived particles. In addition, the
statements throughout the chapter reflect strong statements of PM2 5 causing
effects and then the statements around the coarse mode, i.e., PM^Q.oj use
phraseology such as may also be important. This comes across to the reader as
a bias of the authors relative to fine vs. coarse mode effects. This tone is
continued on page 6-78,1. 24 where the statement is made that crustal particles
do not appear overall to support associations with mortality in the source
oriented evaluations. While clear recognition must be given that there are more
studies demonstrating the importance of PM2 5, the dismissal of coarse
particles in the presence of positive studies is disconcerting, particularly given
that much of the western part of the United States has PMjQ dominated by the
coarse mode fraction.
p. 6-84, Table 6-6 In the cardiopulmonary mortality column for the six cities original vs. the HEI
reanalysis, a consistency of the point estimate is what one would expect.
However, the much larger difference in the confidence limit bounds is
surprising. It would be worth checking the entry in this table to ensure that a
typographical error has not occurred.
p. 6-105,1. 11 26 The actuarial and statistical calculations presented based upon Brunekreef are
hard to believe. The implication that the life span of persons exposed to and
dying from air pollution is a reduction of more than 10 years, if true, would
surely have been detected without the kind of sophisticated statistical analyses
that are currently being required. In addition, what exactly is meant by implying
that up through age 25 a loss of 1.31 years occurs for the entire population? Is
this life span reduction? If so, actuarial numbers likely contradict this
conclusion.
p. 6-107,1. 17 The conclusion from the Krewski et al. study that mortality may be associated
with more than one component of the complex of ambient pollutants in urban
areas bears emphasis in the synthesis chapter and is appropriately highlighted in
various sections of the epidemiological discussions, o
p. 6-107,1. 30 The mortality log hazard ratio increasing to 15 mg/m and then being flat
before continuing to increase again, while being a statistical model that appears
to fit the data, has little biological motivation to support it (i.e., such a model
makes little biological sense).
p. 6-108,1.8 13 The Krewski et al. study looking at the relative risk and incorporating time-
dependent estimates is particularly important for the standard setting process.
EPA must factor the temporal decline in PM that has been occurring in its
assessment of the need for revisions of or new standards for particulate matter.
This is particularly important with the various implementation strategies that
have yet to take effect that are clearly leading to a reduction in overall pollution
levels in this country.
p. 6-205,1. 10 19 A number of studies on long term effects from PM are cited as having been
conducted in California but with inconsistent results. Yet the authors choose to
describe the McConnell study as the most notable because it showed an increase
that is similar to results reported by Dockery. Why is this study notable? It
appears the authors have considered it such because it found effects when others
didn t. This does not appear to be a balanced representation and discussion of
the newly available studies.
p. 6-230,1.17 20 The nonlinear model for fine PM effects in the study by->Smith et al. is of
potential interest since a threshold between 20 25 mg/m for PM2 5 was seen in
this study. Has the type of model presented by Smith et al. been applied in other
data sets?
p. 6-247,1. 25 The summation of the Smith study relative to threshold selection and
importance of fine vs. coarse is phrased as these results, if they in fact reflect
reality, make it difficult to evaluate the relative role of different PM
components One might interpret the authors use of the phrase if they in fact
reflect reality as a bias for wanting to attribute one of the two modes as being
more important. Alternatively, the sentence is an excellent summary of why the
PM issue is so entangled and difficult to separate on a causative basis for one
A-5
-------
mode vs. the other. In fact, such kinds of difficulties are precisely why the
Agency must look carefully at standards for PM that encompass the full
spectrum of potential effects in different locations.
Chapter 7: Dosimetry of Particulate Matter - General Comments:
Chapter 7 on the dosimetry of particulate matter primarily focuses on an update of new studies since the
1996 Criteria Document (CD). While the chapter provides a reasonable review of the available literature,
the review is lacking in details in a number of areas. Given the importance of and reference to dosimetry
considerations elsewhere in the document, the chapter should be strengthened with more specific
presentations of some of the latest results.
The chapter fails to take advantage of a graphical representation of the more recent data. Such graphical
representations covering susceptible subgroups in comparison to normal subjects would be of great value.
The authors failed to report whether increases compared from one group to another are actually
statistically significant or just represent general trends. Without showing the data and the standard
deviations or error bars, the reader is left with a general uncertainty about the significance of any
differences that are reported.
Section 7.5 on the comparisons of deposition and clearance patterns of particles administered by
inhalation intratracheal instillation adds little to the chapter. This section, while accurate, is of little value
for the risk assessment of particulate matter. There is no mention of the role that intratracheal
administration can play in hazard identification and in mechanism of action studies. If this section is
retained, clearer identification of the value of the animal toxicological studies using this method should be
discussed. This is particularly important since many of the studies presented in Chapter 8 on animal
toxicological results arise from intratracheal administration experiments. Section 7.5 should be reduced in
size if it is retained.
Detailed tables or graphs contrasting deposition in children compared to adults should be presented in the
chapter. Since arguments are made elsewhere in the CD about children being a potential susceptible
population, dosimetric differences between children and adults need to be presented in greater detail than
they currently are. The logic of having the only figure in the dosimetry chapter be one of total deposition
is not apparent. While such data are of general interest, the types of effects and standard setting concerns
focus on the major regions of the respiratory tract. Regional deposition should be presented and should
incorporate recent research on different subpopulations and disease groups.
Specific Comments:
p. 7-2,1. 16 The reference to information related to the phenomenon of particle overload is
stretching the case for inclusion of this material. Clearly, there are no ambient
exposures of particulate matter that approximate anything close to the exposure
levels needed to induce overload of alveolar macrophage-mediated clearance
that is the basis for this phenomenon in animals.
p. 7-5,1. 7 11 The authors should clarify that the importance being described for various
deposition mechanisms in respiratory tract regions applies to humans. The
importance of some of these mechanisms differs on a relative sense for some
and on an absolute sense for others when referring to particle deposition in
animals.
p. 7-6,1. 21 27 The cast studies with charged particles are not very relevant to real world
ambient aerosols. If this material is retained, a better explanation of where these
results might be applicable for potential real world exposures should be
provided.
p. 7-7,1. 6 16 It is important in this paragraph to contrast inhalability in humans compared to
inhalability of particles in animals. Otherwise the reader has no insight as to
why this is an important concept to introduce and further has no reference for
determining some of the relevance of concentrations used in animals when
judging the potential for effects in humans.
p. 7-8,1.17 Of value would have been to compare the recent results of Kim to those
previously published by the GSF group for various combinations of tidal
volumes and respiratory frequencies.
p. 7-10,1. 14 19 Since the study by Lenin used a fairly narrow size range (0.3 2.5 mm), the
A-6
-------
statements concerning particle size and flow rate and various breathing modes,
while accurate, should be stated in such a way that the reader understands that
these conclusions do not hold for a wider range of particle sizes.
p. 7-13,1. 24 The study by Kim and Fisher using sequential double bifurcation tube models,
while yielding interesting results, should be put into perspective given that
downstream flow affects deposition in the whole lung and is not necessarily
approximated by sequential series of double bifurcation models.
p. 7-14. 1. 26 The study by Venkataram and Kao 1999 used totally unrealistic breathing
conditions in that they assumed breathing for 24 hours at conditions that are not
physiologically sustainable. Only general trends can be inferred from their
calculations as the quantitative values are not useful.
p. 7-15,1. 25 The paragraph beginning with this line should be reworked. The statements
made in this paragraph are inconsistent with earlier statements of a decrease in
deposition for particles with an initial diameter less than 0.5 mm and an
increase in deposition with an initial diameter greater than 0.5 mm.
p. 7-17,1. 21 A gender difference of about 15% at rest for particle deposition is stated for the
studies of Kim et al. Was the 15% change statistically significant? Without this
information the reader cant really interpret the significance of the findings.
p. 7-19,1. 18 30 The way the Bennett et al. study is presented the reader cannot really judge the
importance of the reported data, on ET deposition. ET deposition as a
percentage of total respiratory tract deposition is the basis for making
statements about differences in percentages. While these differences are
statistically different, they are restricted to 4.5 mm particles since this was the
only particle size Bennett et al. studied. However, the statement in the CD about
the trend for ET deposition tending to increase as age decreased is not a
statistically significant observation. The contention that the deposition seen in
the cystic fibrosis children studied by Bennett et al. likely reflects what one
would expect in normal children is suspect. The argument presented by Bennett
et al is not convincing in that just because lung deposition is expected to be
increased in cystic fibrosis children does not infer that ET deposition would
tend to be decreased in these kids. Since ET deposition is upstream relative to
lung deposition, one can not infer the negative (i.e., increased lung deposition
does not confer that ET deposition should be decreased in cystic fibrotic
children compared to normal children).
p. 7-20,1. 12 15 Again, are the differences reported statistically significant?
p. 7-24,1. 3 10 Recent results published by Asgharian et al. (Aerosol Sci. 32, 817 832, 2001)
also support the influence of lung size on the retention of particles in the
tracheobronchial region for periods longer than 24 hours after deposition.
p. 7-27,1.10 22 The way the paragraph comes across in describing the results in Musante and
Martonen to infer that the rat may not be a good model for the resting human
masks the fact that one has to account for differences in doing interspecies
extrapolations. To make the argument that a greater activity level yields a more
similar distribution of dose on a regional basis does not necessarily imply that
this mode would be better since, for example, the distribution of types of cells
within the respiratory tract differ by airway generation between the rat and the
human. This paragraph could be expanded upon to point out some of the
differences that must be taken into account when extrapolating between species.
p. 7-29,1. 20 Rather than starting the sentence with the phrase for the most part, the author
should indicate that for hygroscopic particles and liquid droplets, clearance
mechanisms are different compared to poorly soluble particles.
p. 7-34,1. 25 Asgharian et al. (Aerosol Sci. 32, 817 832, 2001) recently showed that it is not
necessary to invoke a slow- and a fast-phase for tracheobronchial clearance to
have particles retained in the TB region longer than 24 hours. Intersubject
variability in retained mass arising from the periphery of the TB based upon
lungs with variable number of airways can explain the experimental
observations while still fitting a single compartment clearance model.
p. 7-40,1. 4 7 References should be supplied to support the statement made in this paragraph.
p. 7-40,1. 14 Physical activity is not really a biological factor in comparison to the other
subsections covering age, gender, and the like. Why not simply entitle Section
7.3.4 Factors Modulating Clearance?
A-7
-------
p. 7-49,1. 20 In an effort to make the chapter brief, the authors have indicated that additional
work on modeling deposition in animals has been published but that it merely
expands on work and approaches already noted in the 1996 PM Criteria
Document. The text would leave most readers with the idea that the additional
work is not of value. Since the work of Hoffman et al. (2000) is described on
the next page, surely the inference is not that this is the only work that has made
additional contributions. Some of the features and some of the additional
references should be included here to provide a perspective on what the thrust
of the additional work has been. To merely say that it has expanded upon
previous work is not sufficient. For example, recent experimental and modeling
work on particle deposition with pulsating flow in a rat nasal mold by
Asgharian et al. (Inhal. Toxicol. 13: 577-588, 2001) demonstrates that
deposition efficiencies for pulsating flows are markedly higher than for steady
flows.
p. 7-50,1. 7 12 The statement that models have not been adapted to examine low level
exposures to particles of low toxicity and poor solubility is incorrect. Koch and
Stober (Inhal. Toxicol. 13: 129-148, 2001) published a pulmonary retention
model that accounts for dissolution and macrophage-mediated removal of
deposited polydisperse particles. Their model and the results arising therefrom
should be discussed.
p. 7-50,1. 13 The Asgharian et al. reference has the incorrect year. 2000 is cited in the text,
but the correct year is 1995.
p. 7-51, Section There does not appear to be a compelling reason that a separate section should
7.6.2 be devoted to models that estimate retained dose. Estimation of retained dose is
a natural extension of models that handle both deposition and clearance
processes. The material discussed in this section should be integrated into the
clearance discussion because the various topics that are presented form the basis
of clearance models of varying degrees of sophistication depending upon how
much is known about the biological process.
p. 7-52,1. 25 Strike recently, from the sentence describing the work of Nikula et al. (1997).
The year 1997 is no longer recent compared to 2001.
p. 7-52,1. 22 31 This paragraph lacks a punch line. While interspecies differences in interstitial
translocation and retention of particles is established, the statement is made that
these interspecies differences may not occur at low levels of exposure. What is
the justification for this statement? Are there any references to support this
conclusion?
Chapter 8: Toxicology of Particulate Matter - General Comments:
Since toxicological studies are presented for both animals and humans, the title of this chapter should
reflect such. In the past, toxicology has been usually restricted for description of animal results. This
chapter provides a reasonable summation of the findings of studies that have been conducted since the
1996 Criteria Document. Unfortunately, as reflected in the summary, the biological plausibility of various
constituents and mechanisms of action for effects are still not clearly established.
Section 8.5 of the chapter is labeled as Mechanisms of PM Toxicity from In Vitro Exposures. In actuality
much of the material presented is simply effects from in vitro studies and really not insightful on
mechanisms of actions of PM. The organization of the chapter in this way begs the question as to whether
any mechanistic insights have been or can be gained from in vivo studies. Since I do not think that is the
intent, cross referencing to in vivo and inhalation studies that correlate types of responses or effects seen
with those in in vitro studies should be made whenever possible.
Specific Comments:
p. 8-9,1. 19 22 The statement is made that it is not clear that the total dose of iron oxide
delivered acutely to the lungs of human subjects would be relevant to
deposition of iron given its concentration in ambient environment. A much
stronger statement can be made. Just consider a minute ventilation of 15 liters
per minute. Doing the calculations for 1 mg/m in the air, the amount instilled
bears no semblance to reality of what could be deposited in any reasonable
acute exposure to these levels (e.g., assuming no clearance of particles and 100
A-8
-------
p. 8-16
p. 8-17
p. 8-18, Table 8-5
p. 8-24
p. 8-28, Table 8-7
p. 8-29,1. 1120
p. 8-32,1. 619
p. 8-31,1. 18 19
p. 8-33,1. 69
p. 834,1. 414
p. 8-37,1.10
p. 8-41,1. 18
p. 8-46,1. 13
p. 8-61,1.3
p. 8-62,1. 3
p. 8-67,1. 4
% deposition, more than 7 months would be needed to deposit 5 mg of the iron
oxide particles in the lung since only about 20 mg would be deposited in a day).
The concentration stated in the table for the Madden et al. study should be 1000
mg in 0.5 ml.
For the Watkinson et al. study, what were the nose-only inhalation
concentrations? 3
Given the low exposure of 10 mg/m for 4 hours in the Ohtsuka et al. study,
this paper warrants expanded discussion in the text.
The symbol for the geometric standard deviation is not as it appears in the table
but rather should be the Greek symbol s. The same statement can be made for
Table 8-7.
This reviewer finds it of great interest that intertracheal instillation of ROFA in
the Watkinson et al. study showed effects but inhalation of 15 mg/m six hours
per day for three days of the same compound showed no effects.
In the Killingsworth et al. Studies using monocrotaline-MCT, mortality and
changes in MIP-2 were noted. What human condition does this model mimic?
This paragraph comes across as if the Godleski et al. HEI Report is considered
peer reviewed and the study by Muggenberg et al. appearing in an Inhalation
Toxicology Supplement from the PM 2000 Meeting is not peer reviewed. The
fact that these studies differed in their findings is what should be emphasized
because Godleski used concentrated ambient particles and Muggenberg used
high concentrations of ROFA. If EPA has criteria for what the agency considers
peer reviewed versus not peer reviewed, these criteria should be so stated and
applied uniformly throughout the Criteria Document.
The statement is made that the different findings between the dog studies
illustrate the difficulties in extrapolating animal toxicological data to human
health effects. The sentence falls short in that it fails to note that lack of
understanding of mechanism of action is the primary problem with
extrapolating animal results that are disparate in nature to humans.
The results from the Gordon et al. study are interpreted in this paragraph to
suggest that day-to-day changes in particle composition may play an important
role in the systemic effects of inhaled particles. This is an overinterpretation of
In addition to the potential mechanisms discussed in this paragraph, the role of
endothelins should be mentioned. Vincent et al. (Inhalation Toxicology of
Ambient Particulate Matter: Acute Cardiovascular Effects of Resuspended
EHC-93 Urban Particles in Wistar Rats. Final Report to the Health Effects
Institute for the Collaborative Study 98-32, In Press, 2001) have shown that
particles can affect endothelin 1 and 3 more than 30 hours post exposure.
Replace the word although with the word after.
Broad statements such as what Nell et al. made in their article on suggesting
that the rise in the U.S. prevalence rate for allergic rhinitis may be related to
increased diesel emissions in addition to other combustion sources is highly
speculative. Anyone can suggest a material is the culpritive agent for an effect
but the emphasis in a criteria document ought to be on the proof for such
relationships based upon experimental data.
The astronomically high carbon black exposure level used by Jakab produced
no effect on susceptibility to bacterial infection in contrast to high exposure
studies with titanium dioxide. Comparing such results implies that a particle is
not a particle and that composition or the nature of the particle is important for
the effects on the host. The Criteria Document does not put as much emphasis
on pointing out concepts such as this as what might be appropriate.
Round 11.9-fold to 12-fold. Such rounding is undoubtedly more in accord with
the accuracy of the data.
The concept discussed here that a combination of several components rather
than a single metal in PM is likely responsible for cellular effects is worth
bringing forward as one of the major conclusions that can be gained from
examining the toxicological data on PM.
The Lee et al. studies described here involved sulfuric acid aerosol
concentrations so high as to make the results of little value to the discussion of
A-9
-------
ambient PM effects. The paragraph describing this study should be deleted.
p. 8-70,1.3 Insert the word to after the word shown.
p. 8-71,1. 5 24 Perhaps the authors of this chapter would comment on the paradoxical outcome
of results found by Churg contrasting fine and ultrafine particles. Is the rat
tracheal explant model a reasonable one for making the kinds of comparisons
that were done by Churg et al.?
Arthur C. Upton, MD
Transmitted herewith, as requested, are my comments on chapters 6 and 9 of the draft Criteria
Document on PM. In general, I consider these chapters to be excellent, and I have no
substantive changes to suggest on either of them. Both chapters do, however, need careful
editing to deal with such problems as the following:
Pages 6-6, line 23 and 6-39, line 1: "most all" is ambiguous.
Page 6-267, lines 19-20: grammatically incorrect (words missing?).
Page 9-7, line 10: "this chapter and" should be deleted.
Page 9-8, last line: the reference to "Wilson and Suh" is missing from the bibliography,
as are many of the other references cited elsewhere in the chapter.
Page 9-16, line 3: the second "is" should be changed to "are".
The effect of educational level on the relative risk of mortality (mentioned on page 9-65)
deserves discussion at an appropriate place in the chapter.
The data suggesting an increased relative risk for lung cancer (mentioned on page 9-65) deserve
to be included in the appendix and discussed at an appropriate place in the chapter.
In addition to editorial corrections such as those noted above, the document needs a glossary to
define the many technical terms and acronyms that are used in these and other chapters.
Sverre Vedal, MD
Chapter 6. Epidemiology.
This chapter has been extensively revised and updated since the last version, and is much
improved. The uneven treatment of the more recent studies in the previous version is now less
evident, although persists to some extent (see below), and the very important studies supported
by HEI (NMMAPS I & II, Reanalysis Project) have been thoroughly reviewed and incorporated
into the synthesis. The most important issues are dealt with in the sections that immediately
follow, with less important issues following.
1. Coarse fraction findings
As documented in the CD, much more observational data are available to address
whether the fine PM fraction is more toxic than the coarse fraction. While some new studies
found larger effects of the fine fraction, many new studies found at least comparable, and
sometimes larger, effects for the coarse fraction: these include studies from Detroit (Lippmann
2000), from Phoenix (Clyde 2000, Mar 2000 & Smith 2000), the Coachella Valley in California
(Ostro 1999 & 2000), and Seattle (Sheppard 1999), as well as the Latin American studies from
Mexico City (Castillejos) and Santiago (Cifuentes). The general impression one gets from the
CD is that new findings are generally supportive of the hypothesis that fine particle effects are
dominant. Instead, I find the recent study findings, as a group, support an effect of coarse mode
particles, and sometimes crustal particles regardless of size. At the time of the 1996 CD, there
A- 10
-------
was a dire need for studies assessing coarse particle effects directly. Now that findings from
several are available, at the very least they show little consistency in supporting a dominant role
for fine PM.
The CD attempts to undermine the validity of the observations in the studies in which
coarse PM effects were detected, although the synthesis (p.6-229 (line 22)-230 & 235) provides
a more balanced assessment. For example, the statement that "several [studies] do show
statistically] distinctly larger and significant mortality associations with PM25 than for non-
significant PM10_2 5 effects" (p.6-54, line5-6) ignores the fact that several do not. And, while it
may be true that no study has the power to adequately compare effect estimates sizes between
the fine and coarse range, this has previously not prevented comparisons of effect sizes of many
particle metrics that are highly correlated. In response to the Lippmann findings in Detroit it is
argued that the coarse fraction findings are present because the coarse fraction is correlated with
the fine fraction [6-55, line 10-11; 6-127, line 13-16]. In response to the findings in Phoenix it is
argued that the apparent coarse effects may be due to biogenic particles in that fraction (6-55,
line 27 & 6-77, line 22-26). This argument is speculative and should be framed as such. I also
find it unlikely. None of the above arguments supporting a more toxic role for fine PM is
compelling. Given the new data on coarse PM which were not available at the time of the last
CD, it is difficult to argue strongly that fine PM effects are dominant, regardless of setting.
It is also my opinion that the conclusions regarding crustal effects (p.6-78, line 2-4 and 6-
267, line 10-11) are too strong. Although the studies making use of factor analyses to attempt to
attribute effects to various sources generally do not find much to support adverse effects of
crustal sources (Laden 2000, for example), and some studies incorporating wind patterns in
attempting to identify periods of large crustal contribution to PM (Spokane and Salt Lake City
studies) argue against a crustal PM effect, it is difficult to ignore the findings from studies where
PM is almost entirely crustal in nature (Anchorage, Phoenix (for coarse mode PM), Coachella
Valley, etc.).
If the authors of the CD disagree with these assessments of coarse fraction effects and
effects of crustal particles, at the least a better attempt at making the case should be made,
preferably in the summarizing sections.
A small point: it is not appropriate to compare PM2 5 and PM10 on a meg per meg basis
(6-231, line 19-22).
2. Balance in review of relevant studies
There is still an unfortunate, and unnecessary, tendency in the body of this chapter to use
a different (more stringent) yardstick in evaluating studies that report findings at odds with the
favored hypotheses (PM effects are more consistent than gaseous effects; fine PM effects are
stronger than coarse fraction effects). Some examples follow:
6-45 Most of the cities included in NMMAPS n only had every 6-day PM measurements, yet
this is never brought up as a criticism, whereas this is identified as a weakness in the
Moolgavkar study (2000) that stressed the importance of gaseous pollutant effects over
PM effects.
6-101 Criticisms of the EPRI study are based on the argument that factors that are in the "causal
chain" cannot confound an association, and that the population sample is
unrepresentative. However, equally severe criticisms regarding lack of
representativeness could have been leveled at the ACS study, but were not. The
discussion regarding high blood pressure as a potential step in the development of PM-
induced mortality is very much speculative and has no place in the description of this
study. Why is it noted that the study has "no matched control or placebo" (6-100, line
14) when these are not relevant given the study design, and are not considerations for the
other cohort studies?
6-127 The paradoxical findings from the first 5 years of the Atlanta hospitalization study are
downplayed since the AIRS database is used for PM, whereas the more expected findings
for one year using Supersite data are emphasized. Recall that NMMAPS made use of the
AIRS database.
6-129 In reviewing the Burnett hospitalization studies in which effects of gaseous pollutants are
A-ll
-------
dominant, one criticism is that "best lags" were reported, yet this use of best lags is
justified later (6-238). Almost all studies explore "data driven" lag structures.
6-131 It is bizarre that the mortality data are brought up in this section dealing with
hospitalizations to "shore up" the argument for PM & cardiovascular effects.
6-134 Why does this summary only include the US studies, which incidentally are all "positive"
studies, when important international studies, many of which are "negative" studies, are
not included?
To summarize, rather than attempting to shore up favored hypotheses, and through doing so,
revealing a bias, it would be preferable to do less editorializing during presentation of studies,
and stand back for a more objective look at the studies as a whole. This is what we expect from
a CD.
3. Confounding
Confounding remains an issue of concern. In the time-series studies, concerns regarding
meteorology, in the absence of more innovative approaches to specifying the form of
meteorology in the time series regression models, can probably be put to rest given the many
attempts to incorporate alternative specifications without significant impacts on the PM
estimates of effect. The CD is probably correct in this regard.
Confounding by co-pollutants, a perennial concern, has also been addressed in the CD.
Several points should be noted. First, it is correctly noted that effects based on attempts to
control for confounding in two-pollutant or multi-pollutant models are often difficult to interpret
because of the typically strong between-pollutant correlations that are present in the time-series
studies. However, this does not imply that effects from single-pollutant models of PM are
unconfounded estimates. The findings regarding PM effects, as well as estimates of PM effect in
the CD, are largely reported only from single-pollutant models (as one example, p.6-142, line
17). Second, results from various alternatives to the use of multi-pollutant models in
estimating PM effects unconfounded by co-pollutants are presented. These approaches are
motivated by frustration at interpreting PM effects from multi-pollutant models. In NMMAPS n
gaseous pollutant effects were controlled in a second stage (multiple city) analysis after the
individual-city single-pollutant PM effects were estimated. This is probably justified in this
setting given the relatively large number of cities included, although it seems difficult to imagine
that adequate control for co-pollutants could be adequately accomplished without attending to
the seasonal variation in co-pollutant concentrations, variation that itself differs from region to
region across the country. A different approach to addressing potential confounding by gaseous
pollutants is exemplified by the multi-city hospitalization studies, including NMMAPS n
(Schwartz 2000, Zanobetti 2000). Firstly, the description of these methods is difficult to follow
in the CD narrative (6.223-225). Descriptions of these alternative approaches to accounting for
co-pollutant effects are difficult to follow. I still cannot figure out the rationale behind some of
these approaches from reading this section, which may mean that others cannot either. Clearer
rationale for the specific approaches taken is needed. Parenthetically, I wonder whether the
correct correlation (r) between PM and the co-pollutants should be the correlation after adjusting
for long-term trends and meteorology (that is, correlations between the effect estimates rather
than raw correlations). Secondly, we have much less confidence in the success of this approach
given the much smaller number of cities (and often smaller size of cities [e.g., Boulder,
Youngstown) used for these analyses. The CD seems to uncritically accept this approach to
controlling confounding by the gaseous pollutants (6-126, line 4, for example).
There has been discussion of the potential for the gaseous pollutants to confound the
association between PM and health effects from the perspective of the definition of confounding.
It is argued that some of the co-pollutants cannot be viewed as confounders since, based on
biomedical knowledge, they should not affect the outcomes of interest. Neither SO2, sulfate nor
CO can reasonably be argued to cause many of the effects with which they are often associated.
It would be true that these pollutants could not confound if in fact the ambient co-pollutant
concentrations were truly measuring exposure to these specific pollutants. Realistically,
however, they do not. The co-pollutants are likely measuring various aspects of the pollution-
meteorology mix and acting as surrogate measures of important exposures that we do not now
A-12
-------
understand. Further, PM is qualitatively no different than the gaseous pollutants in this regard.
Therefore, it still makes sense to consider the gaseous pollutants as potential confounders of PM.
Similarly, attempts to stop considering some of the co-pollutants as confounders, arguing, as is
done in the CD, that they are merely steps in the mechanistic causal chain (see below), are not
valid.
4. Regional heterogeneity
The emphasis in this CD on NMMAPS n is justified. The heterogeneity in the estimates
of PM effect across US cities is obvious (28 of 88 cities having non-positive estimates of effect).
This is the first good impression available of the degreee of heterogeneity that is present. The
heterogeneity might be due to random variation in estimates of effect (presumably supported by
the CD in developing the argument that in general the absence of effect is observed for cities
with fewer observations, i.e., less power), or it may represent true regional differences. It should
be noted that the heterogeneity observed in the NMMAPS II study of 88 cities is present for
single-pollutant models. PM effects from two-pollutant, or multipollutant models, would be
expected to have shown even more heterogeneity among the cities. Further, interpreting the
meaning of an estimate of overall effect (0.5% increase for a 10 mcg/m3 increase at lag 1)
assumes that the effects across city come from a single distribution of effects, which might not
be the case if heterogeneity of effects is real and due to some as yet to be identified factor(s) that
distinguishes cities in which effects are detected from those in which they are not. The cause(s)
of this apparent heterogeneity (random variation or "real") clearly has implications for setting
US-wide, health-based standards.
An attempt is made to explain part of the observed heterogeneity of effect by noting that
negative or absent effects were more likely to be seen in cities with the lower concentrations of
PM (6-263, line 30-). This is unjustified given that the NMMAPS investigators explicitly tested
that hypothesis and found no support for it.
5. Chronic effects
Although the cohort studies are invariably referred to as studies evaluating the effects of
"long-term", or "chronic", exposure, this is an assumption. The title of section 6.2.3 ("Mortality
Effects of Long-Term Exposure to Ambient Particulate Matter") already makes this assumption.
Merely because exposure in these studies is specified in terms of long-term averages, this does
not imply that the observed associations are in fact due to these long-term averages. An
alternative is that these effects are simply a cumulation of acute effects. It is argued that simple
accumulation of acute effects cannot account for the size of effects estimated in the cohort
studies. However, these estimates are somewhat sensitive to covariates and analytic approach
(e.g., adjustment for population mobility, spatial correlation and control for SO2 as demonstrated
in the ACS Reanalysis Study). Therefore, confidence in the size of these reported effect
estimates is not great. Reflecting even more confusion is the statement in the CD (p.6-80, line
31) that chronic effects must be present since effect estimates for chronic PM exposure are much
higher than those for the time series studies; this point is irrelevant and in no way argues that a
chronic effect above that observed in the time series studies must be present (this is an "apples
and oranges" comparison). The comparison of the spatial features of effects from NMMAPS n
and the Cohort Reanalysis Project (6-265, line 24-31) does not necessarily enhance the argument
for consistency, given the above.
Effects of acute exposures can theoretically be approximated by calculating a cumulation
of acute effects, something which has been attempted previously. I recommend revisiting the
issue of cumulating time series effects (incorporating the impact of multiple days) to compare to
the range of estimates of PM effect from the cohort studies (esp. the ranges of effects estimated
in the ACS reanalyses based on different models). If this argument is convincing in showing
that acute effects could not conceivably reproduce findings from the cohort studies, then the
above points become moot. The lung cancer findings, if valid, would provide a strong argument
for chronic effects, but this discussion is largely lacking from this version of the CD.
In my opinion, given the above, the conclusion regarding "long-term exposure to PM'
(6-94) needs to be qualified.
A-13
-------
6. Susceptible sub-populations
It is surprising that the most important study to date on identifying susceptibility of
populations subgroups based on pre-existing medical disorders is discussed so little (Goldberg
2000), being presented last in a discussion of previous studies that, because of design, are limited
in the information that they provide in this regard. This study confirms many of the findings of
studies that attempt to address the issue by stratifying on cause of death. However, it is
interesting that no increased risk was identified for the subgroup of subjects with chronic
obstructive lung disease, a group considered to be at high risk based on cause-of-death
stratifications. Parenthetically, I believe the description of the Goldberg study gets it wrong.
PM pollutant measures were associated with mortality, not with acute respiratory disease, etc.
(p.6-74, line 24-30) as stated. The latter were the susceptibility subgroups (that is, for assessing
interaction effects, essentially).
There is legitimate concern that the stratification of by cause of death is fraught with
problems misclassifying
7. Miscellaneous "large" issues
Gaseous pollutant effects: The summaries provide relatively balanced syntheses of
recent gaseous findings (p.6-75, line 14-23 & p. 6-76, line 24-). This balance is sometimes
lacking in the descriptions of specific studies in which a "particle-centric" perspective is
maintained (see point 2 above). For example, the conclusion that fine PM effects on
cardiovascular hospitalizations are most important (6-235, line 1) ignores the important findings
by Burnett and Moolgavkar on the role of gases in affecting estimates of PM effects.
Threshold concentrations: The discussion of thresholds is unconvincing. The argument
attributed to Schwartz that a threshold is mathematically impossible in the face of population
differences in sensitivity (p.6-246, line 3-5) holds only if the most sensitive members of a
population are sensitive to very low concentrations, which may not be the case. Further, the CD
is not consistent in its support of a no-threshold concentration-response relationship. The
argument that heterogeneity in studies of PM composition is due to variable concentrations of
PM components (with studies showing no effects having concentrations too low to show effects,
6-78, line 1) is not consistent with the absence of thresholds. The same point can be made if
heterogeneity of effects in NMMAPS is argued to be due to variability in PM concentrations
across city (see point 4 above).
Measurement error: The description of the Zeger (2000) work on measurement error
(6.249-252) is just about comprehensible. "Dumbing" this section down, if possible, would be
allow it to have the impact that it deserves.
Smaller issues for chapter 6 (by page number):
Introduction
6-2 I don't believe Rothman would assign more inferential strength to case-control studies
than cohort studies (line 14).
6-3 The prospective cohort studies in this setting do not make use of "individual exposure"
(line 2). The subjects in a cohort study do not need to be recruited independent of
exposure (line 4), and in fact were not (e.g., 6-Cities Study).
6-4 Line 19-25. The discussion of causal pathways, although correct, is not relevant in this
context. Because SO2 contributes to sulfate formation does not imply that SO2 effects
cannot be separated from sulfate effects, if correlations are not too strong. But, this is an
issue of collinearity.
6-5 One gets the impression here that meteorology is acting solely as an effect modifier (line
1), when in fact the more important issue is its role as a potential confounder (see
"Confounding" section above).
A- 14
-------
Mortality
6-6 Line 25-26. "Statistically independent" is unclear here. Like for effects due to infectious
illnesses, respiratory and cardiovascular causes can be difficult to entangle, and can be
caused by the same insult.
6-9 Line 26. It would seem that harvesting could be more readily addressed in the context of
"identifiable PM episodes" than the typical time-series studies in which only day-to-day
variability in concentration is studied. Is it being suggested that low level exposure may
have a different lag profile?
6-40 Line 14. Should be "88" cities.
6-42 Line 5-6. On p.6-7, the reported reasonable range of effects from the 1996 CD is
estimated to be 2.5-5.0% increase per 50 mcg/m3. The 2.3% increase estimated in
NMMAPS n is outside this range and therefore, strictly, not consistent. As noted, there
were studies by 1996 showing statistically significant effects that were smaller than 2.5%
(e.g., 1.8%), but that is not relevant. See also p.6-49 (line 19) and p.6-76 (line 3).
6-43 Line 6-7. We do not know from NMMAPS n whether a different tack to trying to
account for gaseous pollutant effects (see above) on the PM estimates would have
reached different conclusions.
6-44 Line 20-26. In the 10-Cities studies, the attempt to control for gaseous pollutant effects
is severely hampered by lack of power (see point 3 above).
6-59 Line 26. This should be Table 6-3 rather than 6-1.
6-61 Table 6.3. The "single pollutant models" heading is confusing, since this includes 2-
pollutant model findings.
6-62 Line 6. This should be Table 6-4 rather than 6-3.
6-78 The threshold discussion here is premature, since no studies presenting data on thresholds
have been presented up to this point.
6-92 Why is a 20 mcg/m3 increment used for both PM2 5 and PMl5?
6-101 It is not clear why these studies of mortality in children and on development (IUGR)
(6.2.3.4) are included in a section on purported long-term effects on mortality, nor why a
time-series study (Loomis p.6-104) is included. I would argue with the descriptor
"likely" (p.6-103, line 5).
Morbidity
6-125 Line 14. This is not strictly a subset of the 88 cities, I believe (e.g., Boulder?).
6-139 The Seaton study (line 26) also found a reduction in hemoglobin concentration in
association with PM.
6-141 I would not consider the data on blood viscosity as "highly suggestive" (line ), given the
negative findings of several other studies including the more recent Seaton study (1999).
6-141 Line 9. It should be noted that effects at longer lags are often not investigated.
4-143 Line 25 to p.6-172. The point that the single "best" lag represents the most valid effect
estimate, based on it being biased high but countered by not reflecting the full impact of
multiple lags, is ingenious but nevertheless nonsense.
6-173 Line 8. Power for Edinburgh for hospitalization counts (but not for mortality) should be
adequate.
6-175. Line 16. Why is the Sheppard study particularly "unique"?
6-176. Line 13-17. The effects of acid aerosol and PM should not be compared on a meg per
meg basis.
6-176 Line 3. Asthma ER visit studies that complicate the argument about fine PM here
include Lipsett 1997 from Coachella Valley on coarse PM and Chapela (year?) from
Mexico City in which no PM effects were detected.
Interpretation
6-216 Line 25. Have some studies really looked at cardiac symptoms?
6-226 The argument that SO2 cannot be a confounder of PM because it is part of the causal
pathway is wrong in most settings. If the point that is being made that we have effects of
the host of pollutants together from similar sources (for example, summer haze), then this
A-15
-------
is OK. But in most settings it still makes sense to speak of effects being due to PM or
S02.
6-238 Line 26-27. The "best" lag approach again.
6-256 The case-crossover study is first introduced here, but should be included in the
intoduction to study designs on p.6-3.
6-258 Line 27. I think "mortality" should be "morbidity".
6-266 Line 17. This is a misuse of "strong" in this setting. These are all weak associations.
The associations may be consistent, coherent, and have large public health impacts, but
they are nevertheless relatively weak associations.
Chapter 8. Toxicology.
In general, this chapter does an excellent job of presenting a great deal of new, and often
apparently conflicting, data. The summaries, especially the final summary, is well-reasoned and
balanced, and makes conclusions with appropriate qualifiers.
As noted below, one important purpose of the lexicological work is to enhance the
plausibility of the epidemiological findings. Much of the work done using easily studied
particles such as ROFA, and work using extremely high concentrations of particles, although
arguably useful when negative findings are obtained, are less relevant when attempting to
interpret positive findings. This strongly motivates the use of CAPs studies where real-world
particles are used. It is argued that because of day-to-day variability in the particle composition
of CAPs, that experimental studies will not have the statistical power to use a factor analysis to
successfully identify the components of CAPs that are particularly toxic. If valid, this would
limit the usefulness of CAPS. However, recent work by Koutrakis's group (EHP 2000, see
comments by Koutrakis on this chapter) in which factor analysis was successfully used for this
purpose suggests that this is not the case. Although the findings from the multitude of
lexicological studies are difficult to interpret, they have contributed to enhancing the plausibility
of the epidemiological findings. It is anticipated, especially with further work using CAPs, that
the picture will become clearer as work progresses.
Introduction
8-3 I like the notion of enhancing biological plausibility, rather than assessing dose-response,
at this stage.
Respiratory effects
8-20 The paragraphs beginning on line 11, as well as those on 8-37 and 8-43, present useful
perspectives on ROFA. The later discussion on the differences between tracheal
instillation and inhalational exposures is also helpful.
Systemic effects
8-31 Line 27. Why does an increase in t-alternans suggest an anti-arrhythmic effect of PM?
8-34 This summary in para 1 is excellent.
Compromised
8-41 What is the paragraph starting on linelS doing here? There seems little place for the
reference to the Nel paper, since the statement attributed to it (the increase in allergic
rhinitis being due to diesel exhaust) is pure conjecture, unless it serves as a starting point
to present findings to support or refute it, which it does not.
8-43 The paragraph starting on Iinel6 on the Goldsmith paper is repetition of a previous
paragraph on 8-37.
Mechanisms
8-65 Line 28. Need to justify the significance of alkaline phosphatase production, particularly
since it seems, based on the preceding sentences, that silicon dioxide was potent.
8-68 The section on ultrafmes beginning here presents an artificial motivation for interest in
ultrafmes. Yes, surface area will increase dramatically as particle size decreases, fora
A- 16
-------
given mass, as demonstrated in Table 8-9. However, it is clear, given the distributions of
particle mass, particle surface area and particle number by particle size in ambient air.
that particle mass falls dramatically as one enters the ultrafme size range: so much so, in
fact, that surface area begins to fall before getting down to the ultrafme range, and then
falls dramatically once the ultrafme range is reached (as shown in the Whitby plot on p.2-
7, Fig. 2-1). I would drop Table 8-9. It is conceivable that the Whitby plot is outdated,
since it is at least 23 years old. It might be useful to have similar work repeated for
modern-day atmospheres.
8-82 The first paragraph describes a clinical study that has no place here. One could
alternatively note that, "Clinical studies have observed....". Here is what the
lexicological studies show:
Summary
This summary, as I noted above, is excellent.
Chapter 9. Integrative synthesis.
I would recommend that this chapter, as the title suggests, serve the primary purpose of
an integrative synthesis rather than a stand-alone summary of the CD. Of course it is possible
for it to partially serve as a summary, while still serving primarily as a synthesis, but this would
take considerable thoughtfulness.
In terms of organization, I would like to see the chapter focus on the big issues that were
largely unresolved at the time of the 1996 CD, and then proceed to addressing how much
progress has been made in answering these questions. This should take the form of integrating
new findings across discipline. For example, in addressing the issue of coarse fraction effects,
the epidemiological findings in isolation would suggest that there are effects of the coarse
fraction, apart from those due to the fine fraction, and that strong consideration should be given
to setting a coarse particle standard of some form. However, this course is tempered somewhat
when information on coarse PM measurement issues and exposures are also considered.
The chapter could end with posing questions that remain unanswered, and which still
need further work.
Barbara Zielinska, PhD
Chapter 2: Physics, Chemistry, and Measurement of Particulate Matter
In my opinion, this chapter requires more work. At present, the chapter makes the
impression on the reader that it was written by several independent authors, without any attempt
to integrate it into one consistent document. Following are the specific examples:
1. On page 2-47, line 19-21 (Section 2.2.3), the authors state discussing the experiments
with two quartz fiber filters deployed in series in order to examine the artifacts connected
with SVOC partitioning: "Unless the individual compounds are identified, the
investigator does not know what to do with the loading value on the second filter (i.e. to
add or subtract from the first filter loading value)". I agree with this statement -
moreover, even if the individual compounds were identified on back-up filter, the
decision concerning adding or subtracting back-up filter loading would not be
straightforward. However, the authors discuss subsequently in detail (page 2-51 to 2-62)
in several places throughout the Section 2.2.3 several experiments with Teflon-quartz or
quartz-quartz back-up filters that produced conflicting results. The references of Turpin
et al., 2000, and Kirchsteller et al, 2000, are discussed on p. 2-52 - 2-53 and again on p.
2-61 - 2-62 (in addition, the reference of Turpin et al., 2000, is missing). This would be
confusing to the reader who is not very familiar with the problem of positive and
negative sampling artifacts. It would be desirable to organize the discussion in more
consistent manner, shorten it significantly, and not scatter it throughout the whole
Section 2.2.3
A-17
-------
2. There are repetitions of the same statements throughout the chapter. For example, the
discussion of sulfate and nitrate in western and eastern U.S. on page 2-21 (line 12-22) is
repeated on page 2-51 (line 1-7).
3. The discussion of the various denuder techniques and their limitations (Sections 2.2.3.2
and 2.2.3.3) is certainly important, especially since the popularity of these techniques has
increased greatly recently. The selection of the correct denuder type, its dimensions, flow
rate, etc., greatly influence the results and incorrect conclusions could be drawn if the
user is not familiar with the denuder technique. It would be desirable if authors put more
emphasis on discussing these factors and organize them in more logical manner (instead
of the extensive discussion of the front-back-up filters collection methods, which produce
doubtful results anyway).
Some statements or opinions express by authors are not accurate, for example:
1. Page 2-19, line 18-19: "... some primary organic compounds ... are found.. .in the fine
particle mode." As a matter of fact, most of the combustion-generated organic
compounds are found in the fine particle mode.
2. Page 2-24, line 13:"... adsorption of organic gases... (e.g. poly cyclic aromatic
hydrocarbons)". Only 2 ring PAH are gaseous at ambient temperature, with 3 and 4 ring
PAH distributed between the gas- and particle-phases.
3. Page 2-53, line 3-12: this discussion is impossible to follow, is there part of the sentence
missing?
4. The PC-BOSS and RAMS denuders are discussed extensively throughout the chapter
(page 2-55, 2-58, 2-89, 2-103, 2-105). However, both devices use a virtual impactor
upstream of the denuder that removes not only a majority of the gases from the aerosol
flow, but also particles smaller than 0.1 um. Thus, the gas-particle distribution is
changed even before the aerosol enters the denuder! In addition, particulate OC
estimates have to be corrected for particle losses in the inlet of 46 to 48%. Is this 46 to
48% factor independent of temperature, pressure and other factors? How accurate are the
measurements, taking into account these corrections? It would be desirable if authors
discuss the limitations of these denuders as well as put the results obtained with these
devices in proper perspective.
5. Page 2-95: The discussion of the commercially available automated carbon analyzer
seems to be a little premature in this document, since no comparison data with other
established techniques is available yet. There is no clear understanding what the
instrument really measures in comparison with TOR and TOT techniques.
6. For completeness, a newly developed continuous photoacoustic technique for black
carbon measurement should be included in Section 2.2.5. The technique and its
applications are described by Moosmuller et al. (1998) and Arnott et al. (1999; 2000).
7. Page 2-103, line 18-23: One has to be careful when expressing the opinion that the
denuder technique is an improvement over the filter/adsorbent collection method. It
should be followed by the caveat that this is not an "out of the shelf technique, it is not
straightforward and requires thorough understanding by the user. If not used properly, it
is subject to numerous artifacts and may lead to erroneous conclusions. Also, one
doesn't have to use a charcoal impregnated glass-fiber filter for SVOC collection
(especially that it is not readily available commercially); other solid adsorbents (such as
PUF/XAD plugs) are used as well.
The minor problems that require corrections are as follows:
1. Page 2-10, line 4-5: missing word, "the term ultrafine", "the term nanoparticle"
2. Page 2-13, line 13: prior to 1987
3. Page 2-20, line 22: ".. .or on or in."?
4. Page 2-21, line 7: "in" before SO4 not necessary
5. Page 2-25, line 1: "are" is missing
6. Page 2-33, line 29: what is "PNA organic compounds"?
7. Page 2-56, line 19-21: an awkward sentence, instead of which method?
A- 18
-------
8. Page 2-57, line 23-25: this sentence is a repetition of the line 16-17
9. Page 2-62, line 15: absorbent?
10. Page 2-73, line 19: The instrument operated by the Desert Research Institute was not a
"high-volume carbon sampler", but the medium-volume (113 L/min flow rate) fine
particles (PM25) and semi-volatile organic compounds (i.e. filter followed by a solid
adsorbent) sampler.
11. Page 2-77, line 13-14: an awkward sentence, I'm not sure what it means
12. Page 2-83, line 21-27: either "it is important" or "its importance"
13. Page 2-91, line 8: remove "because"
14. Page 2-105, line 23-25: not all ATOFMS instrument can measure particles ranging in
size from 10 nm to 2 um (see page 2-94).
There are several missing references, mostly recent ones (Turpin et al., 2000; Casimiro et al.,
2001) but also older, such as Turpin et al., 1991. I didn't check them all - it would be desirable
if authors make sure that the references are in order.
References:
Arnott et al., 1999: Atmospheric Environment, 33, 2845-2852;
Arnott et al., 2000: Rev. Sci. Instrum., 71, 4545-4552;
Moosmuller et al., 1998: J. Geophys. Res., 103, 28,149 - 28,157.
Chapter 3: Concentrations, Sources, and Emissions of Atmospheric Particulate Matter
I would recommend several minor revisions for this chapter, as follows:
1. Page 3-5, line 1-3: Figure 3-2 shows that although the nationwide PM10 concentration
trend shows the clear decline from 1989 to 1995, it seems to level-out for the last 3 years,
especially for urban-suburban sites.
2. Page 3-6, Figure 3-3 is not clear. The reader may have troubles with assigning the EPA
regions to the graphs.
3. Page 3-22, line 4: the main reason of heated inlets in continuous PM mass measurement
instruments is to remove water (as discussed in Chapter 2), so the removal of water is not
a sampling artifact.
4. Page 3-26, line 26-30, the discussion of Table 3-3: it is not apparent from the data
presented in this table that water and cations associated with sulfate are the most
abundant species in Philadelphia. Also, sulfate concentrations is not listed, just the total
sulfur.
5. Page 3-28, line 7-9: not only trace metals concentrations are highly uncertain; Al shows
very high uncertainty as well.
6. Page 3-30, line 18 to the end of the paragraph, the discussion of Table 3-5. The selection
of marker species for individual source categories seems to influence greatly the results.
In particular, Pb, Br and Mn as the only tracers do not seem to adequately represent
motor vehicle emissions.
7. Page 3-35, Table 3-7: EC sources for anthropogenic PM>2.5 include tire and asphalt
wear as well.
8. Page 3-42, line 13-15. Table 3-1 doesn't show that water, sulfate and cations associated
with sulfate are the major components of PM in the eastern U.S. Also, the newer studies
listed in Table 3-8 showed that not only diesel but also gasoline vehicle exhausts are
important sources of PM.
9. Page 3-45, line 8-10: an awkward sentence
10. Page 3-45, line 30: 1998, not 1988
11. Page 3-46, line 5-6: "However... but..."?
12. Figure 3-23, page 3-50: the figure caption says "... principal source categories for
nonfugitive dust sources...", but the figure shows 44.2% of fugitive dust contribution.
13. Page 3-56, line 11-13: This is not a valid argument, since PM25 which are discussed here,
have longer residence time.
14. Page 3-56, line 28-30: an awkward sentence
A- 19
-------
15. Page 3-57 and 3-58, line 29-31 and 1-3: please clarify
16. Page 3-59, line 7-9: the discussion on page 3-55 and 3-56 states that the reasons for this
apparent discrepancy between emission inventory and receptor modeling results are not
clear.
17. Page 3-59, line 21: what PM2.5PM10 refers to?
18. Appendix 3A: Table 3 A-2 should include some data from more recent Northern Front
Range Air Quality Study (NFRAQS), carried out in winter 1997. Ambient data are
presented in volume A (Chow et al., 1998) of the final report (Watson et al, 1998) and
are available on the web (http://www.nfraqs.colostate.edu/index2.html)
19. Appendix 3A, Table 3A-2: Are organic compound concentrations really in ngC/m3 (C =
carbon) or rather in ng/m3?
20. Appendix 3B, page 3B-12, line 13-15: fuel type?
21. Page 3B-13, line 1-10: are "diesels" mentioned here light- or heavy-duty vehicles?
22. Page 3B-18, line 1-17: PAH were also reported in volume B (Zielinska et al., 1998) of
the NFRAQS final report (Watson et al., 1998)
23. Page 3B-18, line 7-10: at atmospheric conditions, PAH with mw 228 (BaA, chrysene and
triphenylene) are predominantly particle-associated, with only traces of these PAH in the
gas-phase (see, for example, Arey et al., 1987).
References:
Arey et al., 1987: Atmospheric Environment, 21, 1437-1444 (page 1439)
Joe L. Mauderly, DVM
Chapter 7: Dosimetry of Particulate Matter- General Comments:
This chapter covers a reasonable range of topics, but needs some editing. There are several
places where terms are used incorrectly, or where uncommon terms are not defined.
Throughout the chapter, it should be stated whether the exposures of humans were nasal, oral, or
both. The difference affects deposition, as the author notes, and the results from individual
studies can't be placed in context by the reader without the information.
Throughout the chapter, it should be stated whether the models and their predictions have been
validated by comparison of results to those from actual measurements. More models have not
been validated than have. This is an important point for the reader to understand.
The chapter could benefit from the addition of a few more figures and tables showing
comparative data that illustrate the points being made. A reader well-informed on
deposition/retention issues can understand the points being made, but many readers will have
difficulty envisioning the relationships described. A simple graph of particle size vs. regional
and total deposition taken from any of the several sources cited would help. Figure 7-1 is not
inappropriate, but it falls unnecessarily short of illustrating both total and regional deposition. A
table listing some representative values for comparative (between species) amounts of deposited
and retained PM of a few discrete sizes would also help. Other than the figure on Page 7-8 and
the flurry on pages 7-30-31, the chapter makes no use of tabular or visual material to illustrate
key points.
Specific comments:
P 7-3, L 12: Don't confuse "aerosols" with "particles". It's the particles that have a
polydisperse size distribution. The "size" of an aerosol is the size of its container.
P 7-6, L 1: All deposition is "by physical contact". What we are talking about are the
mechanisms that cause physical contact. A material is deposited when contact is made,
regardless of the cause.
A-20
-------
P 7- 6, L 15: Are particles charged either negatively or positively? If so, are there charges that
reduce deposition as well as those that enhance it?
P 7- 7, L 10: By definition, if a particle is in the "inspired volume" it is inhalable. Conversely,
if a particle is not inhalable, it won't be in the inspired volume. This sentence should read
"—particle present in the ambient air".
P 7-9, L 1-13: For these citations, state whether the exposure is nasal, oral, or both. That makes
a big difference for ultrafmes, and the smaller the particle, the greater difference it makes.
P 7-14, L 24 - P 7-15, L 3: You need to state that these are estimates from models, not actual
measurements, and you also need to state the type of model used.
P 7-15, L 11-12: The sentence implies that there geographical areas where coarse PM are not
present. Where would such an area be?
P 7-15, L 29: Again, do not use the word "aerosol" for "particle".
P 7- L 17: Once again, it's "particle" not "aerosol".
P 7-19, L 5: Give the geometric standard deviation for the ROFA.
P 7-19, L 18: Throughout the chapter, you should state whether the exposures were nasal, oral,
or both. This is an important variable, and deposition really can't be understood without this
information.
P 7- 22, L 3: This study measured total deposition, not "lung" deposition.
P 7- 22, L: It is not clear how a tumor would increase diffusion deposition.
P 7-24, L 13: It is not clear what the "shallow region of the lungs" would be. Would this be the
central airways?
P 7-25, L 14: Of course inhalability can be important for humans. It's important in a dust storm.
It's important if you are riding a motorcycle (remember the old joke about bugs in the teeth).
P 7-25, L 25-26: What does "upper and lower airway bifurcations" mean?
P 7-26, L 6-7: Just say "—generation is constant" rather than "adopts a constant value". It's
hard to see how an airway generation can adopt anything.
P 7- 26, L 14-20: A figure would help the reader understand what you are saying about
deposition minima and maxima. A simple line graph showing fractional deposition with particle
size for humans and rats, for example, would be useful.
P 7- 27, L 9: Mention whether or not these model predictions have been validated.
P 7- L 14: First, it's the MMAD of the particle size distribution, not the "aerosol" distribution.
Second, give the geometric standard deviation of the size distribution.
A-21
-------
P 7- 27, L 15: What does "comparable respiratory intensity levels" mean? I don't know what
"intensity level" might imply.
P 7- 27, L 22: Again, has there been any validation? It is important throughout the chapter to
indicate whether or not models have been validated against actual measurements.
P 7-28, L 9: The statement is incorrect. The study did not measure the "volume density of
deposition", whatever that might be. The study measured, using a morphometric technique
based on volume density, the retained material. A post hoc study of tissue cannot evaluate
deposition, but only the amount and location of retained material.
P 7- 28, L 12-14: The statement is incorrect. It is not true that "different cells contact retained
particles" in the two species. The difference was not absolute. There was relatively more
material in the interstitium in one species and relatively more in the alveolar lumen in the other,
but there was some material in both compartments in both species.
P 7- 28, L 21: The point is that there can be greater differences between abnormal humans and
normal rats. The present wording doesn't convey this; it suggests that the greater difference you
are talking about is between humans and rats.
P 7-28, L 23-27: This section inappropriately brings response into the dosimetric picture. Dose
is dose regardless of response - these are related, but separate, issues. Interspecies dose
extrapolation per se has nothing to do with interspecies differences in response or dose-response
relationships. Comparative response has to do with both differences in both dose and response,
but comparative dose has nothing to do with differences in response.
P 7.29, L 3: In summary, this section could greatly benefit from some tables or figures showing
example results and comparisons. It also needs attention to which model predictions have been
validated.
P 7-32, L 23-24: The magnitude of response also has to do with PM composition, not just with
particle number.
P 7- 33, L 1-11: Lymphatics should be mentioned in this paragraph.
P 7- 33, L 14: Do you mean 5% by mass or number?
P 7- 33, L 17-18: Alveolar surface fluid is also transported, at least in some in part, up the
airways. Surfactant of alveolar origin has been reported in the surface fluid of conducting
airways. If this is true, then you should mention this path rather than implying that all PM-
derived material solubilized in alveolar fluid is absorbed through the epithelium.
P 7-34, L 8: What do you mean by "nonuniform"? Do you mean spatially or temporally non-
uniform within individuals, or are you referring to variability among individuals?
P 7- 35, L 5: You need to clarify throughout this chapter whether the statements about
deposition site are derived from measurements or whether these are assumptions from deposition
models. Most, if not all, are from the latter, which assume plug flows that are not likely to be
absolute.
P 7-35, L 22: Deposition was "estimated", not "calculated". The latter term implies a certainty,
or direct measurement, that doesn't exist here.
P 7-37, L 25-26: The phagocytic activity need not necessarily be decreased, it could be simply
overwhelmed. More particles could reach the interstitium because of either or both effects.
A-22
-------
P 7- 40, L 18: You need to explain what "mechanisms such as two-phase gas-liquid interaction"
means.
P 7- 40, L 20: Do you mean that transport is more effective (ie, more rapid)?
P 7-41, L 13: It should read "—those obtained".
P 7- 41, L 21: I doubt this statement. I'd wager that more coughs occur in the U.S. annually
because of internal reasons (viral infections, chronic bronchitis, etc.) than from an "inhaled
stimulus".
P 7- 42, L 29: Again, there is confusion between deposition and retention. The 1 mg value is an
amount of retained PM. not deposited PM. If you deposit that amount slowly enough, there will
be no overload from the deposition.
P 7- 44, L 16: Do you really mean "random" here, or do you mean "uniform"? I think the latter
would be a better term.
P 7-46, L 18: It should read "The model results were in good agreement", not that the "model"
was in good agreement. "Models" don't agree with anything, but good ones produce "results"
that do.
P 7- 47, L 7: Any results or validation here?
P 7- 47, L 15: Again, any validation?
P 7- 47, L 27: Once again, any validation?
P 7- 47, L 29: Please explain what "general dynamic equation for size evolution" is. I don't
understand this, and there may be others like me.
P 7- 48, L 9-10: I think you are saying that the combined effects yield a narrower size
distribution. If so, why not just say that, instead of saying "decrease the size nonuniformity" and
"variance"?
P 7-50, L 16: It should read "—data are". Data is a plural word.
P 7-50, L 21: Define "acinar airways". That's a new term for this chapter.
P 7- 52, L 25: It should read "—rats and monkeys exposed—". The statement talks about two
species, but you only name one.
Chapter 8: Toxicology of Particulate Matter - General Comments:
The chapter is a good draft, but needs considerable editorial clean-up of both text and
tables, and some additional attention to content and conclusions. The former is addressed by
numerous of the following specific comments. The latter pertains to the several places where
sentences that portray conclusions (although not necessarily marked as such) that are unclear,
misleading, or in conflict with one another. These are also addressed in the specific comments
below.
The chapter could be better balanced in its treatment of the types of PM that are emphasized. As
one example, it contains greater emphasis on ROFA than is warranted. Granted, there has been a
tremendous investment in ROFA research, but aside from demonstrating the importance of
soluble transition metals (which is important), the extension of this work to other ambient PM is
limited. As one contrast, very little attention is given to "bioaerosols", and what information
A-23
-------
there is pertains almost solely to endotoxin. As another example, no convincing rationale is
given for excluding the considerable database from engine emissions studies from this chapter.
Diesel PM is cited for its potential adjuvant effects, but no mention is made of the several other
potential effects of either diesel or other combustion PM and co-pollutants. Therein lies our
greatest body of information on PM and co-pollutants, and some studies have explored the
absolute and relative roles of different constituents of the mixture. It is especially astonishing
that, while the emissions studies are ignored, studies of animals housed in urban and rural air,
with no characterization of exposure, are cited. The latter have provided almost no useful
information to date on the additive or interactive effects of PM and co-pollutants.
Regarding endotoxin, it is noted in one paragraph that ambient particles may have been
contaminated by endotoxin - presumably during handling and storage. If this is a concern, and it
may certainly be, why not note the concern more broadly with regard to many, if indeed not all,
of the studies using collected particles? This surely is not a concern only for those studies to
which endotoxin effects are central.
The exposures cited in the text (and in some cases, in the tables) need to be more uniformly and
more completely described. There are numerous instances in which studies are cited for which
either the PM exposure concentration, time, or pattern are not given. Noting an effect, for
example, of an exposure and only listing the concentration does not give the reader adequate
information to place the findings in context.
The text and tables need to be screened to ensure that all abbreviations are defined. Some are
apparently not defined.
The discussion of ultrafme particles seems to be ignorant of the portion of ambient ultrafme PM
population that is in droplet, rather than solid, form. The discussion follows the classical
ultrafme litany of greater penetration and surface per unit mass, but never mentions the ultrafme
particles that are likely to spread, disperse, or dissolve after contact with liquid surface layers,
and thus are probably never apparent to cells as "particles" per se. The points to be made are: 1)
an acknowledgement that such PM exist, are ubiquitous, and need to be studied; and 2) there has
been little or no research on this class of material.
Finally, the chapter does not do an adequate job of summarizing the key changes in our
understanding of the toxicity between this and the last PM Criteria Document. The last section
gets at this issue, but needs to be bolstered. As just one example, the Mechanisms of Action
section (8.7.2) is a single paragraph that states that there may be more than one mechanism and
that we don't know the mechanisms "unequivocally". While those are both true and
understatements, there is not an indication of whether we know more about the plausbility of any
mechanisms (ie, have more evidence) than we did last time. We do.
Specific Comments:
P 8-1, L 15: It should read "ambient PM', not "ambient air".
P 8-2, L 23: It is not clear what "total" means in "total exposure".
P 8-3, L 4-5: The distinction here is not clear. Presumably, both "low" and "high" toxicity PM
cause effects because of size and composition. Are PM of low toxicity neither ambient or
surrogate?
P 8-3, L 8-11: The selective treatment of diesel particles (DPM) is not clear and is of
questionable logic. DPM can cause a range of non-cancer effects. They are an integral
component of PM nearly everywhere, and can predominate in some microenvironments. The
fact that EPA developed a separate hazard assessment for diesel emissions should not preclude
the inclusion of DPM in this document. The selection of only the potential immunological
A-24
-------
effects of DPM for discussion in this document doesn't seem logical. At a minimum, this
document should summarize the conclusions from the diesel hazard assessment.
P 8-3, L 14-16: There is something wrong with this sentence. First, it seems to mix the issues of
inhalation and instillation. Second, it probably isn't true that most studies have used inhalation.
Probably more have used instillation. The points that 1) both methods have been used, and 2)
most doses have been high, are valid, but the sentence is confusing.
P 8-4, L 5-14: This paragraph needs attention. First, the only study in healthy volunteers in
Table 8-1 uses a concentration of 1000 • g/m3, yet the text notes 2000 • g/m3. Second, the text
discusses clearance, but there is no report in the table about clearance. Third, if you are going to
cite studies or results in the text that are not in the table, give the references.
P 8-4, L 17: If this is a 1997 reference, why isn't it in the table?
P 8-5, Table 8-1: First, give the exposure days/wk for the studies (first two) that use repeated
exposures. Second, if the first study used only neutral sulfites, why is it in an "acid" table?
Third, shouldn't the units in the Lee study be • g/m3 and notmg/m3?
P 8-6, L 9: How do you get "up to 6400" mg/m3 if the exposures were for either 100 or 200
mg/m3 for 45 min, as listed in Table 8-2?
P 8-6, L 22: References for the first statement?
P 8-6, L 25: Was it the vanadium or the responses that were elevated 9-fold? How do we know
that the effects were due to vanadium in these subjects?
P 8-7, Table 8-2: For the Lay et al. Studies, why not give mass doses like the rest of the listings
in the table? Did the paper not report mass doses (I think it did).
P 8-8, Table 8-2: In the last listing, was all of the ROFA vanadium pentoxide? Shouldn't the
"particle" listing be ROFA?
P 8-9, L 13: It is not clear what a "host generated decrease in the availability -" means. Does
this mean that reactive iron was removed after deposition?
P 8-11, Table 8-3: First, why list the concentrator type for the first study if you don't for the rest
of the CAPS studies? Second, "CAPS" is not a sufficient descriptor. The location and time of
concentration (at least something like "Boston, fall 1999") should be given. This document
should avoid perpetuating the common, but naive, notion that CAPS is some standardized or
consistent material. Third, the age of the subjects is given for some studies and not others. If
age is important (and it probably is), it should be given for all. The same for gender. Fourth, for
the Kennedy et al. Study, give the dose administered. Fifth, what is the distinction between
"instillation" in the Kodavanti et al. Study and "intratracheal instillation" in the Li et al. study?
Finally, how could "instillation" in the Kodavanti et al. study be administered "6 hr/day - 2-3
days"?
P 8-13, Table 8-4: First, in the Brain et al. study, the time and location of sample collection
should be in the "Particle" column, not the "Size" column. Second, the age and gender of the
subjects should be listed. Third, where are "CFA, CMP, WC, and MCT" defined (Broeckaert et
al. study, Costa & Dreher study)? Fourth, what does "emission source" mean in the Costa &
Dreher study? What emission, what source? Fifth, in the Gardner et al. study, why note that the
material was instilled in saline? Does this mean that none of the other studies used saline as the
vehicle if it wasn't listed? Are the "0.3 and 1.7" ml, mg, or what? Sixth, why is "exposure
duration" listed as "N/A" for the Gavett et al. study. "Duration" is given for other instillation
A-25
-------
studies, and is presented as observation time after instillation. Seventh, no PM size is listed for
the Hamada study. Eighth, what word is "alveotis" supposed to be in the description of the
Kodavanti et al. 2000b study? Finally, were the deposited doses the same for instillation and
inhalation in the Watkinson et al. study?
P 8-18, Table 8-5: First, give age and gender of subjects for each study. Second, in the
Creutzenberg et al. study, does "retention increased" mean that clearance slowed, or simply that
the lung burden increased? If that is the only reported effect, why bother to list the study?
P 8-19, L 1010: What were the lengths of the exposures cited in the paragraph. As a general
principal, exposures need to be described by concentration, pattern, and length in order to be
placed in context by the reader. Concentration alone isn't an adequate description of an
exposure.
P 8-23, L 7: If by "injected" you mean instilled, then use "instilled" as is done elsewhere.
P 8-23, L 19: The important issue is not whether biologicals can "account" for the PM effects,
the important issue is whether they might contribute to the effects. It's not a credible proposition
that any single PM feature or type can "account" for the effects.
P 8-24, Table 8-6: First, if the PM concentration and size aren't known in the Cormier et al.
study, and the only particle description is "swine building", what is the study doing in the table?
We apparently have no idea what the exposure was or what part particles might have played in
the effects. Second, in the Elder et al. study, does the 100 • g/rrf refer to the carbon, the
endotoxin, or both? Third, was there no estimate of PM concentration in the Rose et al. study?
Overall, the poor characterization of exposures in the studies in this table renders most of them
pretty useless for understanding the respiratory effects of bioaerosols. Aren't there any reports
of effects of airborne pollen? Those are also bioaerosols.
P 8-26, Table 8-7: First, are "OTT" "MSH" defined somewhere? Second, why give the
monocrotaline dose in the Costa & Dreher study - that isn't given for other monocrotaline
references. Third, the location & time of collection of the CAPs should be given. Fourth, is
"FOFA" something different than "ROFA"? Fourth, the gender & age of subjects should be
given. Finally, the Minami et al. paper is a ridiculous citation. Both the experimental design and
the interpretation are absurd. They injected undefined material into the jugular vein until the
animals died, and noted that the heart acted up before death. You could do the same with tap
water! This is an excellent example of the fact that not all published papers are worth including
in this document. You can publish almost anything, but that doesn't mean that all publications
contain meaningful information.
P 8-29, L 6: Here and elsewhere, the author's name is "Muggenburg", not "Muggenberg".
P 8-31, L 15-19: It is noted that there was little pulmonary effect in the dogs, but also that
lavage neutrophils were doubled. That apparent conflict needs more explanation.
P 31, L 21: "Indice" should be "index".
P 8-31, L 26: "Suggests" should be "suggested".
P 8-31, L 26-28: This sentence doesn't make sense. Why do you call an increase in T-wave
altemans an "anti-arrhythmic" effect?
P 8-32, L 6-19: This paragraph is confusing, and suggests that the author must be confused
about these dog studies. It notes that Muggenberg (sic) found results in dogs exposed to ROFA
that contrast with Godleski's results in dogs exposed to CAPs. That's an "apples and oranges"
A-26
-------
comparison. Later, it notes that the Muggenburg ROFA was collected at a different time than
that used by Godleski, but never cites any Godleski ROFA study. What happened was the
Godleski did studies with ROFA, then proceeded to work with CAPs. Muggenburg did studies
with ROFA provided by Godleski, got different results than Godleski's ROFA results, and then
found that the ROFA provided by Godleski wasn't the same as Godleski had used before. There
isn't any connection between the ROFA studies and the CAPs studies. The point that the
findings of little (Godleski) or no (Muggenburg) effect of ROFA suggests that the typically
small amount of metals in CAPs may not be driving the effects of CAPs has some validity. In
order to make that point, however, you need to clean up the paragraph.
The fact that different animal studies yielded different results doesn't reflect the problem
of interspecies extrapolation, as stated. It reflects the difficulty of extrapolating among any
differently-designed studies (animal or human). The animal studies quoted did not use the same
exposure materials, and the results differed. That's understandable, but doesn't have much to do
with interspecies extrapolation.
P 8-34, L 4-14: Another hypothesis that is not mentioned here is the direct transfer of PM from
the lung to the heart. That has been shown to occur, although it's poorly documented and
understood.
P 8-34, L 20: Has an effect of nutritional status on individual susceptibility to PM been
demonstrated? If so, cite a reference. If not, don't imply that it has.
P 8-36, L 27-28: The difference in rat responses between the labs is more likely due to the
difference in CAPs than to differences between rats or labs. This possibility is not even
mentioned. As in other places, the wording here suggests the very naive view that "CAPs is
CAPs". You can hardly calibrate one response against another unless you show that the
exposure material was identical.
P 8-37, L 7-8: I guess it depends on what you call a "limited number". There have been quite a
few real-time exposures to CAPs now, and several to actual urban air.
P 8-37, L 15: I think you mean "no difference in lung volumes" rather than "no difference in
lung volume measurements". The two are not the same.
P 8-38, L 20: "Organisms" should be "mice".
P 8-40, L 5: What kind of particles were acid coated?
P 8-40, L 15: The two "loci" should be "locus".
P 8-40, L 27: Why note that "replication of this study is necessary"? Why any more necessary
for this particular study than for others?
P 8-42, L 13: Greater than additive to what, or in comparison to what?
P 8-43, L 18: This sentence says "daily exposure", but the preceding sentence says "single
exposure". What kind of exposure are you really talking about?
P 8-45, L 14-15: How do the two quoted studies of BAL show that DPM cause an increased
antigenic response in the nose?
P8-46, L 1: "Antimicrobial defenses against microbes" is redundant.
P 8- 46, L 16: What exposure level of DPM?
A-27
-------
P 8-47, L 23-27: These two sentences are redundant.
P 8-48, L 10: There ought to be a paragraph in this section, perhaps here at the end, describing
the different cell types used in the in vitro studies, and their relevance to cells in the human
respiratory tract.
P 8-55, L 24-29: The point is made here that endotoxin might be a confounding factor in the
response to ambient PM. It is good to note that endotoxin might be an important factor in some
ambient PM. On the other hand, if there is concern that endotoxin contamination after the fact
might have confounded this study, why would the same concern not be expressed for every other
study that used collected and stored samples of not only ambient, but also other types of PM?
P 8-60, L 24: "Correlated" should be "correlate".
P8-62, L 3: Do you really mean a "combination of several components" as the sentence says, or
do you mean a combination of metals? The subsequent sentence continues talking about
multiple metals. "Components" includes both metals and lots of other constituents.
P 8-62, L 12-13: The statement suggests that all biological responses of ambient PM and ROFA
depend on metals. Certainly, metals have been shown to play a key role in some responses, but
you surely don't mean to imply that metals are the key to all biological responses to all PM.
P 8- 62, L 16-17: It should be "hours" and "sides".
P 8- 63, L 9-10: The last statement in the paragraph is correct, but the paragraph only deals with
metals. The section is on reactive oxygen species. The material in the section tends to leave the
reader with two false impressions: 1) that all reactive oxygen species are mediated by metals,
and 2) all biological effects are due to metals, and by extension, to reactive oxygen species. Do
you really intend to make these claims? If not, the paragraph ought to mention mediation of
reactive oxygen species by other PM constituents, and make clear that you don't intend to imply
that all biological effects are caused by this pathway.
P 8-70, L 23: "Time" should be "times".
P 8-71, L 5-7: There is evidence to support this statement for slowly-soluble, solid ultrafme
particles, but that is only a part of the ultrafme PM population. This statement, like the entire
section, seems to be ignorant of the existence of the portion of ultrafme PM that is not solid, but
consists of droplets, mostly organic material and often condensed on nuclei of sulfur compounds.
For example, this type of material makes up a sizable portion of the number count of ultrafme
particles in engine emissions. To the extent that these particles are miscible in the liquid layer
covering the epithelium, they would cease to exist as "particles" per se, and would not penetrate
cells as such. While it is true that there has been almost no research on this class of PM, it is
also true that we know it exists, and can't be ignored in the CD.
P 8-77, L 20-21: The type and ratios of pollutants are key factors that are missing from this
recitation of factors affecting interactions.
P 8-78-79, Table 8-10: This table and the text seem to ignore the most common studies of
combined PM-gas mixtures, studies of whole combustion emissions. Emissions studies are all
studies of PM and co-pollutants, and several have tested the importance of different components.
It is inappropriate to only cite studies of simple combinations of two or a few components and
ignore studies of complex mixtures.
P 8-80, L 18: Again, what about the many emissions studies?. It is not true that the toxicology
database is quite sparse in this regard.
A-28
-------
P 8-81, L 9: "Interaction" should be "interactions".
P 8-82, L 16 to P 8-83, L 8: It is astonishing that these field studies of whole air (urban and
otherwise) are cited as contributing to our understanding of the co-pollutant issue, while well-
characterized combustion emission studies are not cited at all! These studies provide very little
useful information. With regard to the topic of the section, they are basically ecological
epidemiology studies with very few subjects of the wrong species. In line 26-27, it is stated that
"extrapolation is hampered" by a lack of exposure characterization. What an understatement!
Considering all the problems with these studies, it is questionable whether they merit inclusion at
all. As in all air pollution studies, but especially true for studies of co-pollutant interactions, if
you don't know the exposure, you don't know anything.
P 8-83, L 21-22: I disagree with this statement. The key to plausibility is not knowing the
components and the individuals at risk. The key is to plausibility is understanding the linkage
between the two (ie, a plausible mechanism).
P 8-85, L 13-14: This sentence contrasts with the earlier statement on page 8-63 that metals
have been established as a key (it actually implied metals were the only key) contributor to
health impacts of PM via reactive oxygen species. It is stated that the ROFA studies have
important implications, but it doesn't state what the implications are.
P 8-86, L 5-14: This section on "bioaerosols" only talks about endotoxin. What about all the
other bioaerosols? Endotoxin is seldom, if ever, actually a "bioaerosol". It is a contaminant of
airborne PM. Pollen proteins, plant debris, and many other airborne materials of biological
origin are not mentioned.
P 8-86, L 20: First, "PM is responsible" should be PM are the responsible". Second, there other
health effects of concern for diesel PM in addition to the adjuvant effect. Why not mention them
in this chapter?
8-87, L 29: It should say "animals with certain types of compromised health", or "animals with
compromised cardiorespiratory health" or some such wording. Not all types of compromised
health would be expected to affect susceptibility to inhaled PM (a broken toe, as an extreme, but
illustrative example).
P 8-88, L 3-6: This closing statement needs work. First, validation of animal models is as
important as identification, and this important point is overlooked in the section, and too often
overlooked by researchers. Second, what is the connection between making "solid progress" and
the fact that large numbers of people are needed for epidemiology studies? Would our progress
be less solid if fewer numbers of people sufficed for epidemiologists? The author probably has a
couple of good thoughts here, but it's not clear that they belong in the same sentence.
P 8-88, L 12-13: This sentence is trite. I think we can go beyond saying that there "may be"
multiple mechanisms to state that research to date clearly indicates that there "are" multiple
mechanisms.
Chapter 9 Integrative Synthesis - General Comments:
In general, the chapter is well-developed, and with some modest editing, will serve well as an
integrated synthesis. With minor editing, it will hit approximately the right level of detail, and
give appropriate attention to making the major points and drawing conclusions.
Some additional attention needs to be given to this chapter to accommodate the fact that many
people will read only this chapter. It proposes to be a synthesis of all of the Criteria Document
except the environmental effects. First, one wonders why the environmental effects couldn't
A-29
-------
also be summarized. Second, the chapter needs some additional definitions, attention to
terminology, and figures in order to better serve as a stand-alone summary.
There are inaccuracies in this chapter that carry over from the same problems in preceding
chapters. There are also sentences scattered throughout the chapter that don't make sense as
written. This may have resulted from attempts to condense more expanded information in
preceding chapters, but it needs to be corrected.
Specific comments:
P 9-3, L 14-15: While it is true that the term "aerosol" is often used incorrectly, why not use the
correct terminology in the CD? "Aerosol" and "particle" are not the same thing. This chapter
perpetuates the error.
P 9-4, L 16-18: It is stated that the nuclei mode is only distinguishable in remote areas or near
sources. Elsewhere, it is stated that the nuclei mode is not observed in remote areas. Because
the nuclei mode is short-lived, it presumably would be found only near sources; thus, if it is in
remote areas, there must be sources there. These facts need to be reconciled so the chapter
presents a consistent story.
P 9-4, L 20-21: I have heard emission scientists distinguish "nanoparticles" as being in the 50
nm or less size range. Does the Agency care to set forth any criteria for these terms? That
would be a useful service.
P 9-9, L 1 and 5: Wouldn't PM formed by condensation also be called "secondary"? That is,
not all secondary PM is formed by "chemical reactions", right (or do you call condensation a
chemical reaction)?
P 9-14, L 11-12: It is not clear if you are saying that these species exist, or should exist, or
possibly exist, or what.
P 9-15, L 6: This statement conflicts with P 9-10, L 19-20 that states that nuclei mode particles
are not found in rural areas. Let's settle on one story and stick to it.
P 9-15, L 28-29: The meaning of this sentence is not clear. The point about not being able to
characterize particles because of lack of reference standards is not clear.
P9-16, L3: It should be "data—are needed". Data is a plural word.
P 9-16, L 31: The point about particle-bound water is not clear. In fact, the whole issue of
particle-bound water is not clear. Presumably, water is associated with some PM in the
atmosphere. If so, then water is part of the particle, and you want to know the mass and number
of particles, and their health effects, with water, not without. I can see how you would want to
avoid data that include the accumulation of water by particles after collection, but why would
you only want to know the mass of particles with no water?
P 9-18, L 1-2: It would provide useful perspective to give a typical portion of PM mass that
cannot be speciated at present. It is often the majority of mass, not a tiny portion. That would be
a surprise to most people.
P 9-20, L 3-4: State the time period of the children's health study, or the information here is not
useful.
P 9-21, L 5-6: It is not clear what you mean by saying that the amplitude of the peaks is smaller
than the daily means. That is not intuitive, and the reader (eg, me can't understand your
statement.
A-30
-------
P 9-24, L 4-5: It is not clear what you mean by "not influenced by exhaled breath" If exhaled
breath actually influences the nature or concentration of materials in the breathing zone, then
why would you exclude that effect? Another example of how you need a bit more explanation
for this summary chapter.
P 9-26, L 8-24: This entire paragraph is difficult to follow. If the "attenuation factor" is worth
mentioning (which I don't doubt), then you need to explain it and its application more clearly. It
can't be understood from this section alone.
P 9-27, L 10-14: This information is repetitive of earlier sections.
P 9-31, L 6: It should be "breathe", not "breath".
P 9-32, L 4: Is should be "alveolar", not "alveoli".
P 9-32, L 9-13: These sentences repeat errors that were noted in Chapter 8. First, the study did
not evaluate deposition at all. It evaluated the location of retained material, and that could differ
from the deposition site. Second, it is not true that different cells were exposed in the two
species. The site of predominant retention differed between the species, but there was overlap.
The same cells were exposed -just to a different degree, or with a different prevalence, in the
two species.
P 9-34, L 22: Where are the data supporting this statement? I don't know of data showing that
"overload" affects clearance differently in rats and humans. You would have to measure
clearance rates in rats and humans having the same degree of "overload", and that hasn't been
done.
P 9-36, L 12: What is a "biomedical" coherence? Do you mean "biological"?
P 9-37, L 3: Ambient PM exposure is always, not "usually", accompanied by exposure to other
pollutants. Why be tenuous about this?
P 9-43, L 2-3: This sentence is not clear. What is the point about "identifiable" PM episodes?
P 9-60, L 26: This is the first time I've heard PM charged with affecting "morality"! I think you
mean "mortality".
P 9-66, L 23-29: First, this 7-line sentence need broken up. Second, what is meant by "semi-
individual"? Third, eliminate "studies" in line 26.
P 9-70, L 4: It should be "admissions of persons".
P 9-72, L 22: It should be "there are some data".
P 9-73, L 13-17: The sentence is confusing. It appears as though you are saying that CO could
be a better surrogate for PM than PM itself. If that's not what you are saying, what are you
saying?
P 9-75, L 15: "Suffers" should be "sufferers".
P 9-75, L 18-24: This paragraph is not clear. It is especially not clear what you mean by the
sentence on lines 23 and 24.
P 9-76, L 11: It should be "these data were".
A-31
-------
P 9-76, L 30: It should be "these data".
P 9-77, Figure 9-11: The label of the horizontal axis should be "Change in Peak Flow", not
"pulmonary function". Peak flow is what was measured, and that's only one of myriad indices
of pulmonary function.
P 9-81, L 1: It should be "relation to season".
P 9-82, Figure 9-13: First, in this summary chapter, you need to explain "posterior distribution".
Second, there is no value in the inset box in the upper right hand corner of the figure because the
numbers are all the same. What's the point?
P 9-83, L 14-15: If the advance is so noteworthy, it is worth explaining in this summary chapter.
From this chapter, the reader doesn't know what a "distributed lag model" might be. The
chapter explains lags, but not distributed lag models.
P 9-84, L 13: Again, what are "posterior mean effects"? When you first talk about the
"posterior" terms on earlier pages, you need to explain what you mean.
P 9-84, L 23: What are "secular" components? Are they defined in this chapter?
P 9-85, L 2: Again, you need to explain the attenuation factor. This parameter and its
significance are not adequately described in the chapter.
P 9-85, L 12-14: It is not clear what you mean by saying that correlations are not correlated.
The sentence needs re-writing.
P 9-85, L 24: "Statical" should be "statistical".
P 9-86, L 15: It should be "correlations".
P 9-87, L 29: Use the term "48 contiguous states", as you do later.
P 9-88, L 6-26: It would help make your points if you included example figures from the
Krewski et al. paper. Unless the reader is familiar with the figures, it is hard to envision the
points you are making from them.
P 9-89, L 8-11: This sentence is not clear.
P 9-89, L 14: "Materials" should be "information".
P 9-94, L 8: You should just state that the material was ROFA, instead of "combustion
particles". You talk about ROFA elsewhere, and using a different term implies that this was
something different.
P 9-95, L 1: The statement is incorrect. It is clear that particles enter the blood. There is lots of
evidence for that, unless you envision transport to other organs via some other mechanism.
What we don't know are the mechanisms and transport rates. We certainly know that transport
occurs.
P 9-96, L 20-22: Perhaps this sentence was intended to start the next section. It doesn't belong
where it is.
P 9-97, L 22: Gee, I thought the review draft diesel HAD was marked "do not cite or quote".
A-32
-------
P 9-98, L 4-11: This section purports to refer to "bioaerosols", but like the bioaerosols section in
Chapter *, it only refers to endotoxin. That's far too narrow a view of bioaerosols, and misleads
a poorly-informed reader.
P 9- 98, L 13-20: The criticality of analyzing CAPs composition should be mentioned. Such
studies place a premium on knowing composition, and are nearly useless without that
information, yet CAPs studies often to not. This is an issue sufficiently important to mention.
P 9-98, L 22-31: It is not clear why this section is included under links between PM components
and health. It is a related, but different subject, and warrants its own heading. In fact, it fits
better under the next major heading.
P 9-101, L 26: Has "COH" been defined?
Allan Legge, PhD
OVERALL COMMENTS:
These comments are restricted to Chapter 1 Introduction and Chapter 4 Environmental
Effects of Particulate Matter found in Volume I. The authors of Chapter 9 are to be commended
for all of their efforts in revising this chapter. The text is significantly improved and expanded in
important areas over the first draft of October, 1999. Much more effort has been made by the
authors to tell the readers what the science 'says'. This will greatly help in the 'risk assessment'
analysis from the welfare perspective. One very important point emerges a number of times in
the text and that is that welfare responses are very much driven by the history of exposure of the
various environmental receptors. While there is some repetition of material in the text, it does
not distract the reader.
SPECIFIC COMMENTS:
A. Chapter I. Introduction
1. Page l-2,line 11. '—sulfate' should read '—sulfur'.
B. Chapter 4. Environmental Effects of Particulate Matter.
1. Page 4-4,line 4. The term 'runoff should be replaced by 'washoff.
2. Page 4-5,line 17. Should read "Neither nitrate nor sulfate".
3. Pages 4-6 and 4-7,Section 4.2.1.1 Effects of Coarse Particles.
The issue of 'saline aerosol' due to either road salt or cooling tower drift is missing from
this section. The following references are suggested:
Grattan,S.R.,Maas,M.A. and Ogata,G. 1981. Foliar uptake and injury from saline aerosol.
J. Environmental Quality 10(3): 406-409.
Hofstra,G. And Hall,R. 1971. Injury on roadside trees: leaf injury on pine and cedar in
relation to foliar levels of sodium and chloride. Canadian Journal of Botany
49:613-622.
McCune,D.C,Silberman, D.H.Mandl,R.H.,Weinstein,L.H.,Frudenthal,P.C. and
Giardina,P.A. 1977. Studies on the effects of saline aerosols of cooling tower
origin on plants. J. Air Pollution Control Association 27(4):319-324.
PiattJ.R. and Krause,P.D. 1974. Road and site characteristics that influence road salt
distribution and damage to roadside aspen trees. Water,Air and Soil Pollution
3:301-304.
TalbotJJ. 1979. A review of the potential biological impacts of cooling tower salt drift.
Atmospheric Environment 13: 395-405.
Viskari,E-L. And Karenlampi,L. 2000. Roadside Scots pine as an indicator of deicing salt
use - a comparative study from two consecutive winters. Water, Air and Soil
Pollution 122:405-419.
A-33
-------
4. Page 4-7,lines 14-18. Similar thoughts expressed. Suggest that the two sentences
be combined.
5. Page 4-10,lines 2-3. What is 'tail water'?
6. Page 4-14, line 12. Should read "concluded that her studies—"
7. Page 4-15, lines 27-28. It is not the 'particles' that may be taken up through the leaf surface
but rather some or all of the chemical constituents of the particle.
8. Page 4-17, line 27. Should read 'saprophytes' not 'parasites'.
9. Page 4-18, line 22. Should read 'benzaldehyde'
10. , lines 27 and 28. Define 'POPs' and 'SOCs'. While it is true they are defined later
in the text, this is the first time they are mentioned.
1 l.Page 4-19,line 4. Should read "controls the vapor-particle partitioning)-"
12. Page 4-19, lines 14-17. A better reference than Smith 1990d is as follows:
Geron,C.,Rasmussen,R.,Arnts,R.R. and Guenther,A. 2000. A review and synthesis of
monoterpene speciation from forests in the United States. Atmospheric
Environment 34:1761-1781.
13. Page 4-24,line 23. Suggest that this read "stressed ecosystems do not recover readily,and
may be further —"
14. Page 4-26, line 3. Should this read "-particulate matter" rather "-particulate dust"?
15. Page 4-32, line 5. Suggest this read "Persistent organic pollutants (POPs) which are
chlorinated such as PCBs,PCDFs,and PCDDs,can be"
16. Pages 4-39 to 4-41. The following is an additional reference re SUVB and crop plants:
Krupa,S.V.,Kickert,RN. and Jager,H-J. 1998. Elevated Ultraviolet (UV)-B Radiation and
Agriculture.Springer-Verlag,Berlin,Germany. 296pp.
17. Page 4-39, lines 19-21. This needs to be rewritten. The sentence suggests that "CFC
production is at a peak level now". CFC production was halted as a result of the signing
of the Montreal Protocol. Perhaps what the author meant to say was that CFC levels in
the stratosphere have reach peak levels and are beginning to fall as a result of the signing
of the Montreal Protocol. Refer to text on Pages 4-132 and 4-133.
18. Page 4-41, line 28. What is meant by "-informed-"?
19. Page 4-46, line 22. Should read "- in field-" not -"infield-".
20. Page 4-49, line 10. Should read " - nitrogen saturated-" not "-nitration saturated-".
21. Page 4-50, lines 16-18. Unclear as worded. Something is missing.
22. Page 4-51, line 1. Has "Paerl et al, in press" been published yet?
23. Page 4-52, line 11. Should read "-Johnson and Mitchell (1998)-" not (1988). Also change in
reference Page 4-174,line 14.
24. , lines 20-22. Needs to be rephrased. The following is suggested. "This
vegetation had been exposed to chronic low concentrations of sulfur dioxide (SO2) and
hydrogen sulfide (H2S) for more than twenty years and then was additionally exposed to
fugitive elemental sulfur aerosol."
25. Page 4-79, line 14. Should read " -(e.g., Astragalus is an —".
26. Page 4-81, line 31. Should read "—"bottom line" that is driven by an".
27. Page 4-84, Section 4.2.3. Ecosystem Goods and Services and Their Economic Valuation,
lines 12-25.
Some mention should be made of 'organics' and food chains. It is mentioned in
the Summary on Page 4-158, lines 8-11.
28. Pages 4-85 to 4-86,Section 4.3.2.1 Anthropogenic Pollutants. The 'arctic haze'
issue is not mentioned. The following reference is suggested:
Barrie,L. 1986. Arctic air chemistry:an overview. In: Arctic Air Pollution, B. Stonehouse
(Editor),Cambridge University Press,Cambridge,Great Britain, pp.5-23.
29. Page 4-100, lines 22-23. It is noted that there are presently over 70 sites employing the
IMPROVE program monitoring methods and that it is anticipated that an additional 80
sites will be added in 2000. Since it is now 2001, how many sites are there currently
employing the IMPROVE program monitoring methodology
30. Page 4-115, line 3. Should read "Metals undergo natural—"
31. Page 4-136, line 17. Should read "—and stratospheric ozone depletion."
A-34
-------
32. Page 4-147, lines 10 -14.1 believe that "the range of Rocky Mountain spotted tick vectors"
already extends into the northern US states and southern Canada in the west.
33. Page 4-148, line 27. Should read "-with A. aegypti or -".
Paul J. Lioy, PhD
Chapter 3
Most of the information and analyses presented in Chapter 3 are typical of those presented in
previous criteria documents on Particulate Matter (PM). Further, the analyses completed for the
PM2 5 concentrations collected with the new standard reference method are valuable as an initial
assessment of annual or daily exceedences.
My major concerns are with the emissions and source apportionment sections. The focus of the
emissions section is on sources of primary particulate matter. This is a good start, but is
deficient with respect to sources of secondary particulate matter. The source apportionment
assessment also provides more information on the nature of primary particle sources. At the
same time the source apportionment analyses also point out the significant contributions of
secondary particulate matter to the mass of PM25, known as accumulate mode particles.
The source apportionment analyses can do an effective job investigating the percentage of
contributions of secondary particles to the mass. They do not, however, provide quantitative
information on the levels and types of precursor emissions which contribute to the formation of
the mass.
In addition, there is no discussion on the chemistry that leads to the formation of secondary
particles, and the residence time for fresh or aged secondary particles in the atmosphere. The
only statement made that comes close to discussing secondary particles is on chapter 3, page 51.
However, it states on line 26, that gaseous emissions "cannot be translated directly into
production rates for PM." Based upon the many years of particle formation modeling that has
been completed by many laboratories, this statement is not accurate.
The lack of information or predictions for secondary particle formation is serious. This is based
on the information presented in the current criteria document, and many papers published since
1976, which indicate that a large quantity of the mass of PM25 in many urban suburban areas
includes secondary particles.
The above deficiency requires that a section be added to the chapter that specifically addresses
particle formation by photochemical smog or wintertime reducing smog processes. Modeling
activities that include assessments of emissions inventories and a number of chemical processes,
e.g., developed by Caltech, EOHSI, and other investigators, need to be described in the section.
They are necessary to establish the types and levels of precursors that lead to the formation of
secondary aerosol. The section could also provide a context for coupling the efforts for
controlling ozone and other pollutants, to reducing formation and accumulation of particles.
Thus, I recommend that a section be added that focuses specifically on particle formation in
photochemical smog by dark phase and sunlight phase processes. It should be developed to
provide the proper context for evaluating the peak concentrations observed in the summertime.
Condensation and heterogeneous chemical processes and aerosol production will assist in
understanding wintertime chemistry. The section should also have a discussion on products,
lifetimes, concentrations, and neutralization.
The new section will provide a framework for discussion about the significance of both "soot"
and "secondary particles" in causing PM air pollution. It is essential that during the
development of the SIP, we do not focus on sources that will provide marginal gains in particle
A-35
-------
control when it may be possible to benefit from ozone control strategies required to achieve the
new 8-hour standard.
Some References:
Georgopoulos, P.G., Purushothaman, V., and Chiou, R. Comparative evaluation of methods for
estimating potential human exposure to ozone: Photochemical modeling and ambient
monitoring. J. Exp. Analy and Enviro. Epid.. 7. 191-215, 1997.
Georgopoulos, P.G., Arunachalam, S., and Wang, S. Alternative metrics for assessing the
relative effectiveness of NOX and VOC emission reductions in controlling ground-level ozone. L
of the Air & Waste Management Assn.. 47. 838-850, 1997.
Georgopoulos, P.G., Walia, A., Roy, A., and Lioy, P.J. Integrated exposure and dose modeling
and analysis system. 1. Formulation and testing of microenvironmental and pharmacokinetic
components. Env. Science & Tech.. 31. 17-27, 1997.
Georgopoulos, P.G and Seinfeld, J.H. Nonlocal description of turbulent dispersion. Chem Eng.
ScL,44, 1995-2016, 1989.
Kerminen, V.M. and Wexler, A.S. The occurrence of sulfuric acid-water nucleation in plumes:
urban environment. Tellus. 48B, 65-82, 1996.
Korhonen, P., Kulmala, M., Laaksonen, A., Viisanen, Y., McGraw, R. and Seinfeld, J.H.
Ternary nuclation of H2SO4, NH3, H2O in the atmosphere. J. Geoph. Res.. 104. 26349-26353,
1999.
Lazaridis, M., Isukapalli, S., Georgopoulos, P.G. Modelling of aerosol processes in plumes.
Mus, 538,83-93,2001.
Lazaridis, M. Gas-particle partitioning of organic compounds in the atmosphere. J.Geoph. Res..
30, 1165-1170, 1999.
Lazaridis, M. and Skouloudis A. Computer simulation of the transport, formation and dynamics
of atmospheric particles. Water Air and Soil Pollutioa 112. 171-185, 1999.
Lazaridis, M. and Koutrakis, P. Simulation of formation and growth of atmospheric sulfate
particles. J. of Aerosol Sci.. 28. 107-119, 1997.
Lurmann, F.W., Wexler, A.S., Pandis, S.N., Musarra, S., Kumar, N. and Seinfeld, J.H.
Modeling urban and regional aerosols - JJ. Application to California's south Coast air basin.
Atmos. Environ.. 3J_, 2695-2715, 1997.
Pandis, S.N., Harley, R.A., Cass, G.R. and Seinfeld, J.H. Secondary organic aerosol formation
and transport. Atmos. Environ.. 26. 2269-2282. 1992.
Pilinis C. and Seinfeld, J.H. Continued development of a general equilibrium model for
inorganic multicomponent atmospheric aerosols. Atmos. Environ.. 21. 2453-2466, 1987.
Rao, S.T. and Sistla, G Efficacy of nitrogen oxides and hydrocarbons emissions controls in
ozone attainment strategies as predicted by the Urban Airshed Model. Water. Air, and Soil
Pollution, 67, 95-116, 1993.
Roselle, S.J. and Schere, K.L. Modeled response of photochemical oxidants to systematic
reductions in anthropogenic NOX and VOC emissions. J. ofGeo. Res.. 100. 22929-22941, 1995.
A-36
-------
Wexler, A.S., Lurmann, F.W. and Seinfeld, J.H. Modeling urban and regional aerosols: I. Model
development. Atmos. Environ.. 28. 531-546, 1994.
Chapter 5 - General:
1. The chapter on exposures is a vast improvement over the previous version.
2. The chapter provides a reasonable summary of all recent studies on exposure, and
interpretative analyses of previous work.
3. Unfortunately in the attempt to be current, the authors have forgotten to put some major
concepts and results into a historical context. Some of the recent studies look as if they are
presenting the first set of results on a particular issue. They clearly build upon previous
research. This should be acknowledged by referring to previous criteria document (AQCD,
1996) for further information on specific concepts.
4. There is still an over-emphasis on correlations. I have stated before, an "association
(correlation) makes the poison" is not a valid concept. Every particle that deposits in the lung
becomes part of a dose delivered to the individual. Although the variability is very relevant to
results obtained in many epidemiological studies that support PM health effects, no one has yet
shown that a constant or "quasi-constant" baseline level of PM from indoor or personal sources
is irrelevant in causing health effects. This point is mentioned in the integration chapter (9), but
not in chapter 5. The variable portion may provide the final stress to individuals who has had
sustained contact and deposition of particles from all sources. So, both Eag and E^ may have
partial influence on the ultimate dose affecting an individual at risk for one or more disease
endpoints, especially potential acute effects.
5. The chapter needs another E descriptor, Eov_rxn_lv or E(loRn). This is PM exposure derived from
outdoor vapor (ov) reacting (rxn) with indoor vapors (iv). This is a source that could also vary
with outdoor PM when the (ov) is ozone.
6. The range and distribution of many variables that affect PM penetration and deposition are
nicely presented in the discussion. However, these are never integrated and placed into a final
context for the uncertainties about the conclusions. The entire discussion is still attempting to
steer us to a mean value for exposure used in epidemiological studies, a point that is well
established. Unfortunately, the current approach ignores the distributional aspects of exposure to
outdoor and other sources. It precludes further efforts in the staff paper to mention the
uncertainties about the dose of specific agents or the entire mixture of PM from indoor and
outdoor air, which could be relevant to acute or chronic outcomes. It precludes any discussion in
the staff paper on the variety of exposures and sources, which may cause health effects. I do not
believe the major ion contributing to the mean PM (e.g., SO4"2) is necessarily the chemical of
concern. It may be an indicator, but we still need to define what it is an indicator of — ambient
PM2 5 mass or toxic sub-fractions.
7. Last conclusion is a working hypothesis, but it is not the sole reason for understanding
exposure. We need to eventually determine which dose or doses contribute to acute or chronic
effects. The statement needs to be modified accordingly.
Detailed Comments:
P. 5.6, Table 5.1 Very good summary.
P. 5.7, Line 6 We have no definitive "outer limit" it is still a guess, and/or convenient
location on the person. It is usually found somewhere within the personal
envelop for inhalation.
P. 5.8, Line 21 Integral referenced to, NRC 1991. It was published previously by Lioy,
A-37
-------
1990. Reference Lioy, PJ. "The Analysis of Total Human Exposure for
Exposure Assessment: Multi-Discipline Science for Examining Human
Contact with Contaminants" Environmental Science & Technology, 24.
938-945, 1990.
P. 5.11
Good summary of published activity pattern data.
P. 5.13 to 5.14, 5.3.2.2.2
5.3.2.3
5.3.2.3.1
P. 5.19
P. 5.30
P. 5.31 to 5.35
P. 5.37, Lines 9-10
P. 5.39, Lines 29-30
P. 5.41
P. 5.41
P. 5.41, Lines 26-27
Very simple explanation of mass balance model. Authors need to
remind readers that all variables have ranges, and in some cases
may change in value by a factor of 5 to 10. Therefore, sensitivity
and uncertainty analysis are necessary when attempting to explain
results.
The equation is a linear simplification of exposure and ignores possible
synergisms. The authors need to provide qualifiers here!
Need to state that equilibrium is a simplification of indoor systems that are
occupied by residents. Thus, equilibrium may only represent a "virtual"
set of individuals or populations at potential risk. The alpha in Equation
5-9 can, and will, vary based upon lifestyle, meteorology, etc.
Also, need qualifiers because of personal activities, housing
characteristics, and particle size and composition.
Very good introduction, and Table 5.4 is well done. There are others, but
most are still work in progress (e.g., RIOPA study by Weisel et al; COPD
by Koutrakis, et al.). Table 5.5 good summary table.
Mage - Qualify to "average person" in PTEAM.
The net result is that there are many different types of correlations and you
can get many different results. Conclusion, we still need and more work
on which variable(s) is (are) needed to represent personal ambient
exposure. This is essential for assessing which compounds and which
exposures cause the observed effects.
A low correlation doesn't mean much, r2 < 0.05!
Is "tracked" the right term? This only explains 25% of variability.
Subjects in Baltimore were very sedentary!! Could these individuals be
described as stationary personal monitors?
Sulfate is an indicator of ammonium sulfate, and not even the dominant
acid species (sulfuric acid, ammonia bisulfate). In areas where there are
large organic, or nitrate loadings, the SO4"2 ion may not be an indicator of
those portions of the mass. I think SO4"2 is an indicator of the variability
of aged secondary aerosol in the fine fraction.
Confusing. SO4"2 is a strong indicator of neutralized sulfur paniculate
exposure, where there are no indoor sources. In contrast, PM2 5 has many
sources besides SO/2.
P. 5.43, Lines 6-8 Is this the appropriate way to interpret these data?
A-38
-------
P. 5.43, Lines 21-29
P. 5.45
P. 5.45, Lines 21-30
P. 5-47, Lines 1-10
P. 5-48
P. 5-49, Line 10
P. 5-51 to 5-56
P. 5-59
P. 5-61
P. 5-61 to 5-63
P. 5-67
P. 5-73
P. 5-78
P. 5-79 (5.5.4)
P. 5-80(5.5.5)
P. 5-81 (5.6.1), Lines
Please eliminate, the section does not add anything to discussion.
There is an assumption that there is no cross linkage between
accumulation due to chemistry outdoors, and chemistry indoors. Ozone is
present indoors and outdoors. Thus part of the PM assumed to penetrate
indoor could be a mischaracterization of new particle accumulation
indoors, due to reactions between ozone and VOC. The reason: ozone
usually varies with PM25, in the summertime.
Agree with statement.
However, the baseline PM from primary indoor PM sources may still
account for the mass burden to the lung that is built upon by the variable
portion caused by the outdoor concentration and exposure.
These analyses are consistent with other previous studies. Need a
reference to previous document, AQCD (1996).
Need to add the BaP data in THEES. Outdoor BaP was the same at all
outdoor sites across 3 sampling periods. (See attached article by
Waldman et al). Is a good study of BaP indoor/outdoor/personal
exposure. It indicates seasonal differences due to sources and activities.
These are very good sections. However, the results are discounted or
ignored when the authors try to construct mean linear relationships
between Eog, and E,g, etc.
Indoor air chemistry is discounted and/or ignored. If we were to put it
into an appropriate context for exposure there would be an Eov_rxn_lv or
E(loRn) exposure variable for particles generated by gases outdoors, reacting
with gases indoors to produce fresh particles.
Good section.
Ignored in mass balance representations. The chapter authors lean toward
averaging everything to point estimates. I would recommend sensitivity
analyses to begin understanding and presenting a distribution of exposure.
Lines 18-19 need to be at beginning of the paragraph.
Need to add the BaP exposure results from THEES (see attached article,
pg. 211-215). A very comprehensive analysis, which shows a lot about
seasonal variability of indoor/outdoor sources and resultant changes in
personal exposure to BaP.
Oglesby et al 2000, lines 11-14 is a very good analysis, and is an honest
"qualitative" discussion about the uncertainties. But still ignores the fact
that "association does not make the poison."
Ignores freshly generated aerosol indoors.
Good except for the lack of Eov.rxn.lv or
E(ioRn) •
I-15 Should bring to beginning of the chapter. All of page 81 is
excellent, and should be moved closer to the front of the
A-39
-------
document.
P. 5-82, Lines 15-30 Need more research and not just hypotheses to explain "paradox". In the
end, there may be complex synergisms, which preclude simple decoupling
of indoor and outdoor particles. Again, this does not discount the strong
epidemiological "association" established and summarized in volume 2.
The comment tries to direct attention to the ultimate goal of the dose to
the lung and other systems.
P. 5-82, Line 28 Add - Co-generation of fresh fine and ultra fine PM from outdoor air and
indoor gaseous air pollutants.
P. 5-84, Lines 6-19 The E^g may not provide the variability, but will add to the daily
baseline dose received by the lung.
P. 5-84, Lines 20-27 Good point, needs to be highlighted in conclusions.
P. 5-85 Need to include Eov.rxn.lv.
P. 5-89 to 5-92 Good analysis of the problem. The uncertainties around the various mean
values or at least the variability of each variable must be part of any
presentation in the staff paper.
P. 5-90, Line 30, to 5-91, Line 1-3 Still does not discount the need to consider the presence
and addition of the quasi-constant non-ambient mass.
Exposures will yield a dose from indoor, outdoor, and
personal PM.
P. 5-91, Lines 11-14 Good point, but lines 15-19 are just as important.
P. 5-93, Lines 21-25 Very important. Should be part of conclusions.
P. 5-95, Lines 5-7 It is a working hypothesis. Needs to be stated as such here and on page
101.
P. 5-95, Lines 29-31 Point about describing a single individual needs to be made earlier. The
assumption in the text is that it represents the mean, and this has to be
couched by a statement on distribution functions for all variables and the
need to establish a probabilistic distribution of exposure, including
95%tile.
Missing - How will exposure data be used to address causality issues. A dose from
indoor/outdoor/personal exposures to fine and coarse particles will be delivered to the lung. Do
we need research that looks at the incremental toxicity of each for specific endpoints, or the
synergisms that can occur among various toxic compounds of each fraction?
Mort Lippmann, PhD
CHAPTER 7
Page Line(s) Comments
7-1 12 after "aerodynamic" replace "a" with a "comma", and after
"thermodynamic", insert ", and/or electrostatic".
A-40
-------
7-1 15-22 change "translocated" to "clearance" and vice-versa. The
usage of these terms is in error, and is inconsistent with
usage later in the chapter.
7-3 1 insert "components of before "aerosols".
7-3 14 delete "a", and insert an "s" after "parameter".
7-3 16 insert "from specific sources" after "aerosols". The ambient
aerosol is generally composed of multiple log-normal
distributions of aerosols from specific sources.
7-3 18 change "ag" to "ag".
7-3 19 change "(or 16th % particle size to the 50th % size" to "%
particle size to the 50th % size, or the 50th % to the 16th %
size"".
7-3 20 delete "aerosol", and insert "of a specific aerosol" after "sizes"
7-4 21 delete "cellular", and insert "cells of airway surfaces in the"
before "ET".
7-5 11 change "1 • m" to "2 • m".
7-5 13 change ">0.5 • m" to ">1 • m".
7-5 19 change "lower" to "smaller" and delete "largest".
7-5 20 change ", which" to "that".
7-5 28 change "0.3 to 0.5" to "0.2 to 1.0".
7-6 4 insert ", but their length is the factor that determines
interception deposition" after "length".
7-6 6 delete "when it is electrically neutral". This is an entirely
redundant statement.
7-6 9 insert "generally" before "lose".
7-6 10 delete "slowly"
7-6 14 insert "positive and negative" before "charges".
7-6 15 change "some particles may result in an" to "particles
will result in".
7-6 20 change "probably" to "often".
7-7 23 insert "ET" before "deposition".
7-7 30 change "0.3 to 0.5" to "0.2 to 1.0".
A-41
-------
7-12 8 insert "that are either very large or very small" after
"particles".
7-12 19-26 The data that are cited here should be described in greater
detail and/or presented here in terms of a graph or table.
7-13 8 Reference should be made here to the work of Brody et al.
(ARRD 123:670-699, 1981); Brody and Roe (ARRD 128:724-
729, 1983); and Warheit et al. (Exp. Lung Res. 16:83-99, 1990)
indicating that particles also deposit preferentially at
bifurcations of alveolar ducts in small animals.
7-13 23 insert "distal to the larynx" after "volume".
7-14 16 insert "average" before "surface".
7-14 19 insert ", and furthermore do not take the concentration of
deposition on carinal ridges into account" after "effects".
7-14 28 insert "The thoracic fraction of the" before "coarse".
7-15 3,5,6,14 change "NP" to "ET" for consistency with previous text in
this chapter.
7-15 14 change "lungs" to "respiratory tract".
7-16 20 change "differ in" to "have different", and insert
"distributions" after "parameter".
7-17 25 insert "large airway" after "increased".
7-28 9 change "deposition" to "retention".
7-28 12 insert "at the respiratory acini" after "tissue". The
importance of the existence of respiratory bronchioles in
humans, but not in rodents, should be discussed at this
point.
7-28 30 insert "for specific surface regions" before "that".
7.34 4-5 The sentence is incomplete.
7-37 2 insert "toxicant" before "exposure".
7.44 21 This discussion is incomplete without a further elaboration
of the fact that inhalation exposure results in concentrations
of deposited particles on the bifurcations of both large and
small airways.
7-52 31 This discussion is incomplete without a further reference to
Nikola et al. (2000), which compared retention sites in lab
animals (surficial) to humans (interstitial).
A-42
-------
7-52 31 This chapter is incomplete without a summation indicating
the most critical dosimetric unknowns and those amenable
to resolution by further research.
CHAPTER 8
Page Line(s^ Comments
8-3 13-14 The cited references refer to silica. Where can the reader go
for an update on asbestos? The most recent ATSDR
Toxicological Profile, or Lippmann (Environ. Toxicants, 2nd
Edition, 2000) could be cited.
8-4 7,8 This sentence is redundant, and should be deleted.
8-6 4,5 This sentence is a real reach. The least that is needed here is
a citation to the chapter section that attempts to justify this
conclusion.
8-6 11-14 A reference citation should be provided to indicate where
these data come from.
8-6 27 This discussion should be a separate paragraph.
8-6 31 Change "deposition" to "retention".
8-9 2 insert "is" before "present".
8-10 8 insert "some of' after "investigating", and "may" before
"cause".
8-10 22-24 This sentence is far too definite a statement!
8-19 4-10 There should be a citation here to the later discussion of the
"overload" issue in this chapter.
8-21 23 This discussion beginning here and extending to p. 8-23, line
11 provides strong evidence that transition metals may not
be as important as repeatedly stated elsewhere in this
chapter, and should signal a more general reassessment of
many of the statements made elsewhere in this chapter.
8-25 19 insert ", but growing," before "number".
8-29 5 change "human" to "humans with".
8-29 26 change "health" to "healthy".
8-30 28 The statement".... and that PM metal content was a better
indicator than PM mass" is clearly not supported by the
preceding discussion! There must have been more transition
metal content in the ROFA than in the Ottawa ambient PM.
8-32 13,14 The preceding discussion of Godleski's research was
restricted to concentrated ambient PM, not to ROFA.
A-43
-------
8-62 10,11 The preceding discussion does not provide an adequate
basis for such a firm conclusion.
8-62 13 change "subject" to "subjects".
8-62 17 change "side" to "sides".
8-65 29,30 How does the preceding discussion provide a basis for this
conclusion? It could be made in any case without citing the
preceding discussion.
8-67 5 If, in fact, the 94 mg/m3 was not an erroneous value, it is
difficult to understand why such an outrageous and
irrelevant exposure was worth citing in the CD.
8-70 23 change "time" to "times".
8-70 29 change "scrutinization" to "scrutiny".
8-72 29 change "to" to "that was".
8-73 7 insert "some of before "the pulmonary".
8-73 8-10 If a contrast is to be drawn, then the concentrations at issue
should be cited. If the work of Amdur and colleagues were
included, the conclusion drawn would be quite different.
8-73 20-22 What does the 10,000 • g/m3 refer to? It clearly was not to
acid. Was it to carbon?
8-75 1 What relevance can an exposure at 15,000 • g/m3 have to the
discussion? Inclusion of citations to such ridiculous
exposures do not belong in this CD.
8-75 10-13 What exactly are the authors saying here? Is there a serious
intent here? If so, it should be justified and elaborated.
8-85 14 What implications? We, the readers, are at least entitled to
some elaboration on what the implications in the authors'
minds may be.
8-86 1 delete "However," insert "low concentrations of sulfuric
acid on" before "ultrafine", and insert "metal oxide" before
"particles".
8-86 2 change "focussed largely on" to "demonstrated"; change
". and" to "However,".
8-86 3 insert "also" before "have".
8-86 25 Add the following: "However, ambient diesel particle
concentrations have decreased during the time of increasing
asthma prevalence."
8-87 12 change "has" to "can have".
A-44
-------
8-87 20 delete "however,".
8-88 Section 8.7 SUMMARY ignored the discussion in Section
8.5.3 on "Potential Cellular and Molecular Mechanisms"
(pp. 8-58 through 8-68). Was it because it had no apparent
relevance to the issues at hand?... or because the results
cited were too various and confusing to show how further
research on biological mechanisms can be structured to
advance the understandings needed to guide the
identification of the physical and chemical properties of
ambient PM that lead to adverse health effects. This
summary section is incomplete without a reasoned summary
of what previous research on biological mechanisms of PM
health effects has determined, and how strategic planning
for further research efforts can best be structured to resolve
the unknowns in this important area.
CHAPTER 9 - INTERACTIVE SYNTHESIS - General Comment
In general, this chapter is well organized and provides a clear summary statement and synthesis
of the PM literature described in the preceding chapters. It will, of course, need some fine
tuning, updating, and more definitive conclusions following receipt of CASAC and public
comments. It is well on its way to serving its intended purpose and represents a welcome
evolution from earlier PM criteria documents.
Specific Comments
Page(s) Line(s) Comments
9-3 3 insert "for regulatory purposes" after "pollutants".
9-4 4 change "enter" to "penetrate".
9-4 5 change "excluded" to "retained".
9-4 11 insert "or trimodal" after "bimodal" and "minimum between
1.0 and 3.0 • m" to "minima at about 0.06 and 2.0 • m". The
figure referred to (Figure 9-1) is clearly trimodal, even
though it represents the special case of near major roadways.
9-4 13 change "the" to "that".
9-7 10 insert "and PM10 includes only those coarse mode particles
that can penetrate into the human thorax" after "equivalent".
9-7 28 insert ", which are predominantly in the fine mode" after
"compounds", and insert ", which is predominantly in the
coarse mode" after "material".
9-9 15 insert "relatively" after "only".
9-26 1 The authors should know better than to give credence to the
notion of "some exposure analysts feel that ambient
concentrations represent a surrogate for total personal
exposure". This is a place where what we know should take
precedence over ill-considered conjecture!
A-45
-------
9-27 17 insert "source and/or" after "each".
9-27 27 change "several" to "many (-16)".
9-28 15 change "lower" to "smaller".
9-28 22 insert "directly proportional to the number of charges"
before "inversely".
9-28 23 change "likely" to "generally".
9-30 5 change "and through segmental bronchi" to ", bronchi and
bronchioles". There are "hot spots" on deposition on
bifurcations at all branching levels, as I noted in my review
of the Dosimetry chapter.
9-30 8-10 This statement is flat-out wrong, and needs to be
reconsidered. Deposition peaks in the segmental bronchi.
9-32 29 "mucociliary" is misspelled.
9-33 24 change "(< 24 h)" to "(< 10 days)". The clearance via
alveolar macrophages is minimal during the first 24 hours.
9-33 26 insert "moderately" before "soluble". Highly soluble
materials do not retain their particulate form long enough to
be translocated.
9-35 11 change "particles" to "deposits".
9-39 15 for consistency, insert "(SOx)" after "sulfur oxides", "(NOx)"
after "nitrogen oxides", and "(O3)" after "ozone".
9-66 26 The "McConnell et al" reference is to one of the papers from
the CARB sponsored children's health study at USC. The
reference here should be to a paper by McDonnell et al on
the AHSMOG data.
9-69 Figure 9-9 There is no translation given for the "HF" and " 1 HD"
caption designations in the figure. They refer to congestive heart
failure and ischemic heart disease respectively. This also applies
to Figure 6-6.
9-74 Figure 9-10 The hospital admissions data for Detroit reported by
Lippmann et al. (2000) should be included in this summary
presentation data. This also applies to Figure 6-7.
9-79 and Section 9.6.2.3.3 This section is incomplete without discussion of a recent
series of important papers from the Children's Health
Study in Southern California. In particular, discussion
needs to be added for the following:
9-80 A. Papers that were cited in Chapter 6: 1) McConnell et al.,
EHP, 1999; 2) Peters, J.M. et al., Am. J. Resp. Crit. Care
Med., 1999b and c; 3) Gauderman et al., Am. J. Resp. Crit. Care
A-46
-------
Med., 2000.
B. Papers not previously cited:
1. Gilliland, F.D. et al. (2001). The effects of ambient air
pollution on school absenteeism due to respiratory illnesses.
Epidemiol. 12:45-54.
2. Avol, E.L. et al. (submitted). Respiratory effects of
relocating to areas of differing air pollution levels.
3. McConnell et al. (in preparation). Childhood asthma
exacerbation and fine paniculate air pollution in Southern
California.
Contact Dr. John M. Peters at USC for copies of these papers.
p. 9-90 11-17 The section on the ROFA studies needs to acknowledge that
the effects observed were attributed to much higher
concentrations than those that occur in ambient air.
p. 9-104 1-4 This discussion needs to distinguish between infants and
children. Premature mortality occurs among infants
(< 1 year of age) but not in children over one year of age.
Excess morbidity and functional decrements are seen in
children, especially those active out-of-doors. Lumping the
two groups together is misleading and incorrect.
CHAPTER 6 EPIDEMIOLOGY - General Comment
The authors of Chapter 6 are to be commended for an outstanding scholarly summary and
synthesis of an enormous and highly complex literature on PM epidemiology. It
comprehensively reviews the peer reviewed literature and systematically addresses what is
known, what is uncertain, and what issues need to be resolved by further research.
One background topic not specifically addressed is the role that past regulatory decisions
on the selection of PM indices have played in the evolution of the PM epidemiologic literature
base. The adoption of PM10 in 1987, and of PM2.5 in 1997, have generated ambient air
concentration databases that made it possible for epidemiologic researchers to address and
resolve many of the previously unresolved linkages between airborne PM and human health, and
the newly authorized network of speciation samples holds promise for further advances in the
near future on the identification of the more influential components of the ambient pollution
mixture.
While there must, of necessity, be an end to the inclusion of newly accepted peer
reviewed literature, the authors should make every attempt possible to include more of the
emerging research findings as possible. In this regard, I call the attention of the authors to some
of the potentially most important papers of which this reviewer is aware. In this regard, the text
of this section should be expanded to reflect some recent relevant research reports, such as:
1. The report by Laden et al. on the follow-up study of the 6-cities cohort (Abstract
ISEE-437, in: Epidemiol. 12(4): S81, July 2001), and the one by Pope et al. on the follow-up
study of the ACS cohort (Abstract ISEE-205 in the same issue of Epidemiol.). The paper by
Pope et al. (ISEE-205) describes a follow-up analysis of the American Cancer Society cohort in
51 U.S. cities for 16 years of mortality experience will report significant associations between
PM2.5 and both cardiopulmonary and lung cancer mortality. (The Abstract that appears in
Epidemiol., July 2001 does not describe the recently completed analyses.) There were no
associations of mortality with the coarse thoracic mass (PM10-2.5).
A-47
-------
2. The paper by Kiinzli et al. on the justification for relying on the cohort mortality
studies for the best estimates of PM-related premature mortality (Am. J. Epidemiol. 153(11):
1050-1055,2001).
3. Research reporting significant PM-related infant mortality to supplement the previous
paper by Woodruff et al. (1997). These include an 8-city study (in the U.S.) by Kaiser, Kiinzli,
and Schwartz (Am. J. Respir. Crit. Care Med. 163(5): 881, Apr. 2001) as well as 2001ISEE
Abstracts (Epidemiol. 12(4), July 2001). One, by Ha et al. (ISEE-134) describes PMlO-related
mortality in Seoul, Korea. Two others describe PMlO-related reductions in birthweight, which
provide coherence support for premature mortality. Bobak (ISEE-209) provides data for the
Czech Republic, and Wojtyniak et al. (ISEE-331) provide data for Poland.
4. Research on the effect of PM on the health of children in Southern California beyond
those reported in the PM CD draft. These include:
a. Gilliland, F.D. et al. (2001). The effects of ambient pollution on school absenteeism due to
respiratory illnesses. Epidemiol. 12:45-54.
b. Avol, E.L. et al. (submitted). Respiratory effects of relocating to areas of differing air
pollution levels.
c. McConnell et al. (in preparation). Childhood asthma exacerbation and
fine particulate air pollution in Southern California.
Contact Dr. John M. Peters at USC for copies of these papers.
Specific Comments on Text
page line(s^ Comments
6-4 12 add to end "while NO2 contributes to the formation of
organic aerosols during photochemical transformations.
6-6 11 The generally accepted abbreviation for coefficient of haze
is "CoH", not "COH".
6-7 7 insert "annual average" before "commmunity".
6-7 15 insert "short-term" before "mortality".
6-7 22 insert "than average" before "relative".
6-11 12 insert "Short-Term" before "Information".
6-39 1 change "most" to "nearly".
6-39 5 insert "are" before "generally", and change "comport" to
"consistent".
6-80 14 insert the following sentence after "mortality". "On the
other hand, the ACS cohort was largely Caucasian and above
average in a socioeconomic sense, and its mortality RR would be
expected to be lower than a more representative U.S. population".
6-83 1 delete "out".
A-48
-------
6-105 7 change "newly" to "later".
6-108 26 change "constituent" to "index".
6-132 8 change "which" to "that" (also p. 6-184, line 26; 6-205,
line 10).
6-138 7 change "which" to "that" (also p. 6-269, line 24).
6-140 18 change "is" to "are".
6-141 18 insert ", the variability of pollutant concentrations
within the community," after "sites".
6-172 8 after "associations", insert the following words from line
9: "have been
reported by several investigators".
6-172 31 insert "those" after "than".
6-175 15 transpose "U.S." and "various".
6-175 19 delete either "Both" or "jointly".
6-175 27 delete "Turning to non-U.S. studies". This study involved
a mixture originating, at least in part, in the U.S., and it was based
on the same kinds of measurements and models used in U.S.
studies.
6-180 13 insert "hospital" after "asthma".
6-183 29 insert "in one second" after "volume" and change "FEV" to
"FEV1".
6-184 10 change "PF" to "PEF".
6-184 16 change "PF" to "PEF".
6-205 20 delete "As" and "other".
6-218 3 change "that" to "which".
6-225 28 insert "to be" before "expected".
6-228 4 This section (6.4.2.3.) should not end without some interpretive
statement andVor identification of what additional investigation
is needed to make this alternative approach more useful for
analyses of PM source impacts on human health.
6-228 25 insert "cohort" before "study".
6-229 11 insert "large" before "U.S.".
6-230 12 transpose "as the exposure metric" with "a three-day
running average".
A-49
-------
6-243 12 This section (6.4.4.) should not end without a discussion
of which approaches might resolve this important issue.
6-267 2,10 insert "thoracic" before "fraction".
6-267 15 insert "well" before "beyond".
6-268 20 insert "thoracic" before "fraction".
6-268 28 change "may not yet be" to "are not yet".
Jane Q. Koenig, PhD
Chapter 6
I complement the authors on an ambitious and generally successful job of summarizing
recent studies in the field of epidemiology. I do have some major concerns.
Major
1) In my opinion, this chapter includes an unacceptable amount of editorial comment. It is
my understanding that the purpose of the CD is to summarize the scientific literature and
that comments and critiques of that literature are reserved for the Staff paper.
2) I know of at least two important papers that were not included in the document. This is
of concern as there may also be others that I didn't notice. What was the process for
inclusion of studies?
3) It is disturbing that the health effects of exposure to PM from wood smoke or other
vegetative combustion sources are not mentioned. Wood smoke health effects should
have been included in section 6.5. I believe this is a major oversight that should be
corrected.
4) Apparently there is no discussion of potential associations between PM exposure and
cancer. This may be an oversight.
Other general comments
Table 6-1 contains too much text. I think it detracts from the usefulness of the table (which is to
provide an easily read comparison of data). This problem is present in the other large tables in
the chapter as well. Would Table 6-1 be more useful if there were columns for lag times, RR,
etc that are easy to scan? A table of significant associations between gaseous pollutants and
mortality would be useful. I suggest notation of effects seen at concentrations below the current
PM10 and proposed PM2.5 standards throughout the chapter.
5-1 2nd sentence, I think cardiac dysfunction should be mentioned right up front
5-45 Mar et al. gases were more highly correlated with PM2.5. PM2.5 and CO corr =0.85, with
NO2 corr = 0.79 than noted in the CD
5-45 bad idea to use county for the unit. Certainly in King co people in gold Bar are not
exposed to what Beacon Hill measures!! This is an example of using quick and easy to
obtain data sets. Maricopa county appears to give very different outcomes than Phoenix.
5-46 -recommend that composition comments here be moved to 6.2.2.4
Table 6-16 This table would be more useful if the Emergency Dept studies were separated from
Hospital Admissions. Also in general the tables in the Morbidity section are much easier to use
than those in the Mortality sections.
A-50
-------
Table 6-23 Respiratory Sx, lung function and biomarker effects.. What biomarkers are
investigated? I didn't find any. Table 6-22 (asthmatic subjects) is entitled just Sx and lung
function.
6-216 6.4.1 This section appears to belong in ch 9??
5-225-227 Is it commonly accepted that SO2 cannot be a confounder for PM???
5-226 Discussion of the use of factor analysis is a good addition.
5-238 Mention of the Lipsett (1997) study is an opportunity to mention the role of wwod smoke
as a constituent of PM. This should have been emphasized. In general there is not
enough use of the role of geographical differences in PM composition as a means of
understanding the toxic components.
5-246 Discussion of thresholds. If individual responses to PM prevent establishment of a
threshold, how does that fit with the language of the CAA that requires setting a NAAQS
for the most sensitive members of society??
5-266 6.5 Conclusions
# 2. Would it be more useful to describe heterogeneity as geographic differences in the
composition of PM?
#3 I think short term v long term exposures need to be considered very, very carefully. We do
not know to what extent prior exposure to air pollution is involved in the premature death cases
in the short-term time series studies.
#4 The CF data may be telling us that there are geographic differences in PM
#5 This conclusion highlights effects during early pregnancy and post-natal periods. However
these data are not presented forcefully in the prior text of the CD.
#9 As I mentioned earlier, I suggest a systematic description and summary of effects of co-
pollutants.
#12 this paragraph (or a separate one) could include a discussion of the fact that there are likely
different mechanisms for different PM-induced health effects. For instance, the mechanisms
underlying air pollution aggravation of asthma will be entirely different from those underlying
death from congestive heart failure.
#13 Should this paragraph be merged with # 4?
Comparison with the November 1999 draft CD
1) CAS AC deemed that draft to be too encyclopedic and yet I don't see that the current
draft is any less so.
2) CASAC recommended emphasis on cardiovascular effects and on infant mortality. I
expected to see a separate table for these outcomes—certainly for infant mortality as
there are only a few studies.
3) Is there really any more risk assessment in this draft than in the 1999 draft?
4) I believe that the strategy used to select the articles cited in the CD is still lacking in
spite of a specific request following the last meeting of CASAC.
A-51
-------
Chapter 5. Human Exposure to PM and its Constituents
I am not by any means an expert in the filed of exposure assessment. That said here are
my impressions on this chapter.
My overall impression of this chapter is that it is very different in scope from chapter 6
and 8. The emphasis appears to be a description of models available for describing exposure.
As with chapter 6, this chapter would benefit greatly from a short paragraph at the beginning
describing the goals and intent of the chapter. As with Ch 61 am disturbed that the data on
wood smoke have not be considered. The indoor/outdoor studies of fine PM from wood smoke
may offer some useful information on penetration of PM indoors.
Another impression is that the chapter listed individual papers published since 1996 but
did not compare and contrast these studies.
Specific comments
4-1 The second sentence should state that the lung AND HEART are the targets of concern.
4-4 Is the nomenclature |j,e accepted in the field. I don't like it—micro environments have
nothing to due with scientific measures of micrometers etc.
at) In all figures the authors need to be very clear not are measured data and what are
deduced from the models.
Should there be some description of exposure assessment to co-pollutants?
Petros Koutrakis, PhD
Executive Summary:
This has not yet been provided by EPA.
Introduction:
The introduction is very informative and concise. I agree with the approach to build upon the
recent 1996 Criteria Document, which made it possible to focus on recent information.
Considering that the number of publications on particle exposures and effects has increased
exponentially over the last few years, I was afraid that the new Criteria Document would be
larger than the yellow pages. I am very pleased to see that this is not the case and the EPA staff
have provided, for most of the Document, a critical review of the existing information rather
than merely a catalogue of papers. It is clear that the particle health effects field has
significantly matured and that the continuing interactions between EPA and the research
community have helped to build a scientific consensus. This is very gratifying and will enable
us to address this serious public health issue in a cost effective way.
Chapter 2: Physics, Chemistry, and Measurement of Particulate Matter
Overall, this chapter is very well written. As indicated in my previous review, I had only a few
comments which now have been addressed by the authors. In addition, the authors have
referenced many recent papers relevant to this chapter.
To obtain organic carbon total mass concentration it is necessary to multiple the carbon
concentration (detected as OC) by 1.4. This conversion factor corresponds to an average
molecular weight ratio of ambient air organic compounds to carbon. Because this conversion
factor can depend on aerosol composition, which can vary by time and location, there may be
A-52
-------
substantial uncertainty in the organic carbon measurements. I do not remember whether this
issue was addressed in this chapter. If not, this is an important topic that should be discussed
extensively. These inherent organic carbon measurement uncertainties may make it more
difficult to reach particle mass closure.
This chapter would be better if the discussion about carbon was a little more concise. In
contrast, the discussion on the elemental analysis, pages 77-80 is only rudimentary. If the
authors do not want to give many details about the methods, this is fine, but at least they should
discuss the advantages and shortcomings of these methods. A more critical discussion is
appropriate because elemental tracers are of paramount importance to source apportionment
studies. The problem (which is good news from the health effects perspective) is that trace
element concentrations have decreased and XRF may not be the most adequate method, although
is the most practical one. Reading this section one cannot find a take-home message about the
state-of-science in this area.
Below are some minor specific comments:
Section 2.2.1.5; It may be worthwhile to report some results from the compliance network. I
know OAQPS has results from the FRM audits. I think it is important to mention something
about the precision of the FRM sampler under real ambient conditions as operated throughout
the country by different states.
Page 2-64, lines 7-8; I am not sure if this statement is correct.
Page 2-73, line 10; Particle/particle interactions also are very important. For example, the
reaction between ammonium nitrate and sulfuric acid or other acid sulfate particles can result in
the formation of nitric acid vapor that can be lost from the filter. Of course, it is not
thermodynamically possible that ammonium nitrate coexists with acid sulfates in the
atmosphere. However, during the collection of a multi-hour sample, e.g. 24 hours, these
compounds can be present in the atmosphere at different times and thus can be collected on the
same collection medium
Section 2.2.5; There was in inter-comparison study of continuous methods supported by CARB
in Bakersfield. Some of the findings from this study should be presented here, since this is one
of the most comprehensive continuous particle methods inter-comparison studies.
Page 2-91, line 22-until the end of the sections; Too many details are presented here and I do not
understand the point being made.
Page 2-94; A section on continuous sulfate measurements should be included. Discuss the old
FID method and the new Fluorescence method which will be used by most of the supersites (by
Allen et al, Harvard University).
Page 2-97; The discussions on data quality is very superficial and incomplete. I suggest to fix it
or drop it.
Page 2-102, lines 24-26; Need to eliminate. The authors have been made a big deal of this
throughout the chapter.
Page 2-105, lines 4-5; This sentence does not mean anything, therefore I suggest to eliminate it.
Chapter 3: Concentration, Sources, and Emissions of Atmospheric Particulate Matter
The revised chapter has been considerably improved. The authors considered most of our
previous comments. They have used more recent information and have expanded the scope of
A-53
-------
the chapter. Many of the poor quality figures have now been replaced. The addition of the
appendices has strengthened the chapter. I really enjoyed reading them. Although they were
long, I found them very informative and am sure many people will use this information.
The summary section needs some improvement. It should be more comprehensive and contain a
good synthesis of the presented information. It does not read like a good summary, rather it only
presents some highlights of chapter 3. The summary section should include a concluding
paragraph on the background concentrations. In the main text the authors presented a nice
discussion on this topic, but it was not clear what was the bottom line on this issue. Therefore,
one would also expect some mention of this in the summary.
The weakest part of the chapter is the discussion about emissions and their trends. I know this is
a very difficult topic and the existing information is very limited. For this reason the NRC
committee on particle research has recommended that EPA investigates particle source emission
in a comprehensive manner. The chapter does not acknowledge this lack of information, rather
it tries to make a good story which is not there. The introduction of the emission section is
confusing and the discussion on uncertainties is rudimentary. Some discussion on methods to
measure emissions may be worthwhile to include. Also some discussion about the importance of
biogenic sources would be worth including.
Below are some minor specific comments:
Table 3-1; Do the percent contributions correspond to sulfate and nitrate or to ammonium nitrate
and ammonium sulfate?
Figure 3-1; This figure is not clear. It is hard to distinguish the solid circles. Same for Figures
3-4a and 3-4b. It is really hard to read these figures.
Figure 3-2; How were the nationwide trends calculated? Were the lines in the figure interpolated
between the two successive years or are they moving averages?
Page 3-10, line 7; ..acids. Define which acids.
Page 3-10, lines 24-27; This sentence is not clear.
Page 3-12, lines 6-8; This sentence is not clear, needs editing.
Table 3-8; These studies should be sorted: alphabetically, chronologically, or geographically.
Page 3-49, section 3-4; This section is not well written. What is the message here? This section
is confusing.
Figure 3-23; The title of figure should read "PM2.5 Total Primary Emissions....". It should be
clear that this table presents only primary emissions.
Page 3-53; In lines 21-24 you mentioned that sulfate concentrations decreased less than the
corresponding sulfur dioxide (I agree with this statement). However, in Table 3-10 sulfate
decrease is 39% and sulfur dioxide is 16%. Something is wrong here?
Table 3A-1; This is a very useful table. It would be nice (if it can be done easily) to include the
OC/EC ratios for the different sites.
Page 3B-1; In the first three lines you use three times the word "discuss/discussions".
Page 3B-24, line 19; Use = instead of- for sulfate.
A-54
-------
Chapter 5: Human Exposure to Particulate and its Constituents Introduction
This chapter has now been substantially improved and the authors should be commended for
their efforts. A comprehensive review of the most recent exposure studies has been included in
this chapter. My remaining criticism for this chapter is that although it presented a very
comprehensive review of personal and indoor particle studies, it failed to critically synthesize
this information. What are the most important conclusions that this chapter should highlight? In
my opinion the following points need to be clearly made:
a) Personal exposures are associated with both indoor as well as outdoor sources; b) the personal
exposure/outdoor concentration ratios present substantial intra- and inter-personal variability; c)
Although we originally thought that this variability was mainly due to the presence of personal
and micro-environmental sources, the results from recent exposure studies suggest that it is the
varying impact of the outdoor particles on indoor environments that is mainly responsible for the
observed intra- and inter- variability in personal exposure/outdoor concentration ratios and; d) It
appears that home characteristics may be the most important factor that affects the relationship
between the average population exposures and ambient concentrations. Air exchange rate seems
to be an important home characteristic surrogate that can explain a large fraction of the observed
inter- and intra-personal variability.
These findings explain why longitudinal studies (many repeated measurements per person)
provide stronger correlations between personal exposure and outdoor concentrations than cross-
sectional studies (few repeated measurements per individual). Since home characteristics is the
most important factor affecting personal exposures then one would expect that correlations
between average population exposures and outdoor concentrations will vary by season and
geography. To test this hypothesis, Janssen et al. 2001 (Environmental Health Perspectives, in
press) examined the relationship between Hospital admissions (for cardiovascular and
respiratory diseases) in a large number of US cities (NMMAPS study) and found that central air
conditioning use explains a large fraction of the variability among cities.
Also one important issue that should be stressed in this chapter is that multi-pollutant personal
exposure studies have suggested that ambient concentrations of gaseous co-pollutants are
surrogates of personal exposures to particles rather than confounders (Sarnat et al. 2001,
Environmental Health Perspectives, in press).
However, because the authors have provided a reasonable and objective interpretation of the
findings of the existing exposure studies, it will not be difficult to fix this chapter. This can be
easily done by revising the summary section and by providing some critical discussions
throughout the chapter.
Throughout the chapter the authors discuss the distinction between outdoor and indoor sources.
Although I agree with their approach and that this should be presented, I disagree with their
decision to make this the central issue of the chapter. I think there is some exaggeration here.
Finally, the discussion on the exposure error is an important one, but I think it needs to be
concise and straightforward. Many people do not have the background to understand this
discussion which is very important.
Specific minor comments:
Page 5-2, lines 1 and 14-15; need editing.
Page 5-14, line 5; P and k are also function of home characteristics, not only particle size and air
exchange rate. Same comment for Pint, see page 5-16, line 4.
A-55
-------
Page 5-17, equation 5-10; the coefficient a in this equation is not constant and presents
substantial intra- and inter-personal variability.
Page 5-18, line 4; This statement is wrong. The chapter contradicts itself, see Figure 5-2 on page
5-44.
Same page, line 7-9; This is not fully correct. It is not just the physical and chemical properties
of particles, house characteristics are also important.
Title 5.4.1; change to: Types of Particulate Matter Personal Measurement Studies.
Page 5-19, line 24; I do not understand what is the daily average? I know you describe this on
page 5-31, but I still find it confusing.
Page 5-22, line 14; "many studies..." This is not true.
Section 5.4.2.3 on page 5-24; short and not-well written interpretation of particulate matter
exposure data.
Figure 5-46; If I remember well they used sulfur to calculate the fraction of particles associated
with outdoor sources. But we know that the S may not be a good tracer for ultrafmes and coarse
particles, there fore, the results presented at this figure should be presented with caution.
Page 5-47, lines 17-19; if personal activities include closing or opening the door and windows,
then these activities will impact the non-ambient levels.
Page 5-24, line 19; fix nitrate and ammonium, same thing for table 5-13.
Page 5-86, lines 1-2; There is a recent paper by Long et al. 2001 (Environmental Health
Perspectives, published) that compares the toxicity of ambient and indoor-generated particles.
Page 5-98, lines 11-12; please see my previous comment on the variability of sulfate
personal/outdoor concentrations.
Chapter 9: Integrative Synthesis: Particulate Matter Atmospheric Science, Air Quality,
Human Exposure, Dosimetry, and Health Risks
The first 23 pages is "the best of chapters 2 and 3". It is nicely done but I do not see the
synthesis.
Section 9.4, summarizes the entire human exposure chapter 4. This is relatively short compared
to the presentation of chapters 2 and 3. This is fine because I think that it is the first 23 pages
which need to be substantially truncated. Again the authors failed to deliver the synthesis of the
exposure studies to date. Please see above my main comment for chapter 4.
The dosimetry section, 9.5, was very concise and informative.
Page 9-44, lines 30-31 and next page lines 1-2; Janssen et al found that the % of PM10
associated with vehicular emissions and the fraction of homes using central air conditioning per
city explained most of the heterogeneity among NMMAPS cities (Janssen et al. 2001,
Environmental Health Perspectives, in press).
The section on epidemiology is too long. Again this reads like the best of the epidemiology
chapter.
A-56
-------
The rest of the sections are fine.
Overall, I think the different chapter sections should be balanced and use the same approach.
Sometimes this chapter reads like a very long summary rather than an integrative synthesis and
definitely there is very little connective tissue among the different sections. In other words, one
could place each of the sections at the end of the corresponding chapter as a summary. I hope
the executive summary (which I have not seen yet) will provide some real synthesis and a
discussion which expands across disciplines.
Roger McClellan, DVM
OVERALL COMMENTS
The present draft represents a significant step forward in summarizing the current status
of knowledge on the health effects of ambient particulate matter (PM). However, the present
version is not an adequate review and synthesis of the information on PM required for
establishing the indicators, level, averaging time and statistical form of National Ambient Air
Quality Standards for PM.
The document will be improved by using the "source to health responses" paradigm
shown in the NAS/NRC PM report as an integrating structure for the Criteria Document.
In my opinion, the document tends to overstate positive associations between increased
levels of ambient PM and increased rates of mortality and morbidity and does not always convey
the high degree of uncertainty in the date. While the NMMAP study represents a substantial
advance in our identification of PM in some locales as having hazardous properties, the high
degree of variability in effects estimates across the U.S. with lack of statistical significance in
many cities suggests caution in interpreting relative risks of less than 1.1 and certainly for
relative risks of less than 1.05. The use of normalized values of 50 • g/m3 for PM10 and 25
• g/m3 for PM25 and PM10_2 5 tend to exaggerate the actual findings. This could be illustrated by
constructing a table presenting the actual estimated relative risk in percentage relative to the 10th
to 90th percentile (or 25th to 75th percentile) range of the PM measurements.
The CD needs, in multiple places, to offer an admonishment that the quantitative
statement of effects estimators, while useful for comparing and interpreting data, should not be
used to make "body count" estimates or predictions for any city or region and certainly not for
the U.S.
CHAPTER 6 - EPIDEMIOLOGY - GENERAL COMMENTS
In general, this chapter provides a comprehensive survey of the epidemiological studies
that have analyzed for PM associated health effects. However, the chapter can and should be
improved to provide a more balanced presentation of the current information available on the
human health effects of PM.
The chapter could be improved by development of an expanded introduction. Three key
elements of an expanded version would be sections on (a) baseline health statistics, (b) the issue
of inter-city and intra-city (temporal) variations in air quality and (c) analytical methods and
statistical considerations. All three of these issues become critical to the conduct and
interpretation of epidemiological studies. The baseline health statistics data are covered in a
cursory manner in Chapter 9. That information should be presented at the beginning of Chapter
6 in an expanded format. To help the reader appreciate inter-city variability in health, a
distribution histogram might be developed of the cardiovascular and respiratory death rates for
the 88 cities in the NMMAP study. It would be preferable to show the rates for cardiovascular
and respiratory deaths separately, rather than combining them as done in Table 6A. To illustrate
intra-city temporal trends, the figure from Kelsall et al (1997) showing mortality data (1974-
1988) for Philadelphia should be included.
For air quality data, distribution histograms could be developed for PM10 from the
NMMAP study data to illustrate inter-city variability. The intra-city (temporal) trends could be
A-57
-------
illustrated using a figure from Kelsall et al (1997) for multiple pollutants for Philadelphia (1974-
1988). The inclusion of these figures will help to illustrate the challenge faced in "teasing out"
air pollution impacts from other factors that influence mortality (and morbidity) from respiratory
and cardiovascular disease.
The above discussion would lay the general groundwork for the section on analyStical
and statistical considerations. In this reviewer's opinion, the most significant advances since the
1996 CD are derived from the NMMAP study. This study benefited from the use of a common
database and a common analytical methodology as well as increased statistical power related to
analysis of data from 88 cities over a relatively long time period (1987 - 1994). A brief
discussion of the analytical methodology used in conducting the NMMAP study and related
studies would be helpful to the general reader who is not an expert in epidemiology.
The balance in the chapter could be improved by giving more attention to issues of
statistical certainty/uncertainty. The authors have tended to call attention to statistically
significant results while tending to avoid calling attention to the lack of statistical significance in
other studies. The authors need to do everything possible to ensure that all studies are reviewed
and reported in an even-handed manner. If certain studies are given "special weight," the basis
for doing so should be clearly articulated.
The present draft does not adequately treat the issue of co-pollutants and conveys a view
that the authors are zealous in putting PM center stage and pushing other pollutants into the
background. For each study, the CD should clearly state whether the analytical methodology
considered only some indicator of PM or also considered co-pollutants. For each of the various
endpoints, tables should be created that would include more detailed information from the
studies that considered co-pollutants. This would include presentation of the relative risks for the
other pollutants when they were determined.
The present draft does not adequately treat the issue of heterogeneity of effects estimates,
especially as reported in the NMMAP study. While it is correct to say that the basis for city to
city differences in effects estimates is unknown, more attention should be given to elaborating on
potential explanations for the differences. This would include the possibility that the differences
are real and the levels of PM10 present in certain cities did not yield statistically significant
effects estimates for PM10 for the period studied (1987-1994).
CHAPTER 6 - EPIDEMIOLOGY - SPECIFIC COMMENTS
Page 6-3, line 18: "Confounding and Effect Modification." This section addresses a very
important point when it notes that "the health outcomes attributed to particles are not very
specific." Indeed, the modifier "very" could be dropped to make the statement more accurate. It
would be helpful to the reader to illustrate the extent to which the majority of the typical health
outcomes are attributable to other factors. Indeed, the terms - confounders and effects modifiers
- do not adequately relate the extent to which the health outcomes are attributable to factor
others than the identified modifiers and effects modifiers.
Page 6-5, lines 28-30 and page 6-6, lines 1-2: It would be useful to add a paragraph or two here
placing the pollutant increments in perspective. For example, for most of the U.S. increments of
50 • g/m3 for PM10 or 25 • g/m3 for PM25 are not at all representative. The use of these
increments tend to present an exaggerated view of PM effects. I suspect that is why the
NMMAPS authors elected to normalize their results to 10 • gm3 of PM
10-
Pages 6-6 and 6-7: The approach used through the document of discussing the 1996 CD
findings and then the post 1996 CD finding is confusing. I would prefer to see all of the
evidence "weighed" to reach a current conclusion. The integrated finding could then be
compared to the 1996 CD findings.
Table 6-1. The table should be expanded to include information on the effects estimators for
pollutants other than PM when the individual study has evaluated other pollutants.
Alternatively, this could be done in a separate table for those studies which have looked at
A-58
-------
multiple pollutants. In presenting the results, it would also be useful to complement information
on pollutant effects estimators with information on actual pollutant levels so that the role of the
individual pollutants would be more apparent.
Page 6-42, line 7 and page 6-43, line 6. It would be useful for the CD to include an expanded
discussion of the handling of county-specific variables and co-pollutants in the NMMAP studies.
Specifically, it would be useful to include one or more tables that present specific data on the
effects estimators used for other pollutants such as NO2, O3, SO2, and CO and for temperature
(both elevated and reduced). This would be helpful in understanding the total air pollution effect
and the relative importance of PM. It is not sufficient (as in page 6-44, line 2-3) to relate that the
PM10 effect on mortality "did not appear to be affected by other pollutants in the model."
In presenting the NMMAPS results it would be useful to include a graphical display that conveys
the slope of the effects estimators for the 90 cities. At a minimum, the regional data could be
plotted relative to the measured range of PM10 values used to derive the effects estimators. The
latter values might be the 25th to 75th or 10th to 90th percentile of the PM10 values that were used
in the analyses plotted on the horizontal and the mortality rate on the vertical.
Page 6-49, section 6.2.2.4 (The Role of Particulate Matter Components) This section should
either begin with or end with a discussion of the challenge of characterizing the role of specific
paniculate matter components. Two major issues should be covered. First, epidemiological
analyses can only be carried out on the components that have been measured. In that regard, a
major problem relates to the past excessive domination of monitoring by concern for regulatory
compliance, with a progression in the U.S. from TSP to PM10 and most recently to PM25
measurements and with measurements of PM indicators made only every 6th day. The ability to
test for the role of other components that may be significant will continue to be dependent upon
having long-term measurements of these components. The second issue is the challenge of
teasing out very small relative risks. It is apparent, and especially from the staff paper, that large
study sizes are needed to obtain relatively stable and statistically significant results—studies for
which the product of number of mortality/morbidity events per day multiplied by the number of
days monitored is at least 10,000.
Page 6-58, lines 19-20: The statement indicating that wind-blown endotoxins and molds are
contributing to PM10-2.5 fraction effects in the Phoenix area needs to be supported by
references or omitted if it is mere speculation.
Page 6-58, line 2.7. In view of the role of SO2 in the Wichmann, et al (2000) study, it would be
appropriate to give an indication of the SO2 levels measured and how they compare to levels
measured in the eastern U.S.
Page 6-67, Source-Disputed Evaluation: It would be useful to review the analyses done by the
NMMAPS investigators (perhaps even unpublished analyses) to determine if any of the results
provide any insights into source-oriented impacts. For example, did the NMMAPS investigators
explore any weekday versus weekend effects that might give insights into mobile source related
effects?
Page 6-72, line 1: Show the Confidence Interval for excess other deaths; i.e., 1.3% increase per
50 • g/m3 PM10. It would also be appropriate to expand the discussion of other deaths to consider
regional differences.
Page 6-73, lines 28-30: It would be useful to expand the discussion of sample size issues for
sub-categories of disease. This could be done using the study size calculations in the staff paper
for the NMMAP study showing how the study size decreases progressing from total mortality to
cardiac to respiratory causes because of decreases in number of events. This discussion could be
A-59
-------
tied back to the base-line health statistics presented in Chapter 9 (tables 9-9 and 9-10).
Page 6-77, lines 23-26: The summary statement on biogenically-derived particles in the PM10_2 5
fraction in this reviewer's opinion is over-stated relative to the evidence. If the statement is
retained, it must be backed up with specific evidence.
Page 6-80, lines 5-6: In view of the emphasis given to the relative risks for PM2 5 derived from
the ACS study, it would be useful to briefly describe the methodology used in the ACS study to
arrive at PM2 5 values.
Pages 6-86 and 6-91 were missing from all copies of the CD provided to me.
Page 6-102, line 17 to page 6-103, line 4: It would be useful to give the low, medium, and high
PM10 levels studied as an aid to relating the research to contemporary PM10 levels in the U.S.
Page 6-133, Individual-Level Studies of Cardiovascular Physiology This section could be
strengthened by including a discussion on the statistical problems of detecting small increases in
"signals" for "low prevalence effects." This could be done by considering the minimum study
sizes needed to give statistically significant effects for cardio-respiratory mortality (per staff
paper) and then applying these minimum sizes to the individual level studies that sought to
identify more subtle morbidity indicators.
Page 6-175, line 15 to page 6-176, line 17: In discussing the association of increased levels of
PM and other pollutants with asthma, it would be useful to include information on the effects
estimators for the other pollutants used in the various analyses. This will place the PM effects in
perspective relative to other pollutants.
Page 6-177, line 27. This discussion needs to be expanded and integrated with data presented in
tables 9-9 and 9-10.
Page 6-222, line 3: This would be an appropriate place to discuss the effects estimators for
PM10, O?, NO2, SO2, and CO, provide an indication of typical levels, and discuss the relative
contribution of each of the indicators to the total air pollution effect.
Page 6-245, Section 6.4.6, New Assessment of Threshold in Concentration-Response
Relationships. The issues that should be discussed in this section go well beyond considering
thresholds. This reviewer suggests the section be re-titled - "Ambient Concentration - Response
Relationships for PM Indicators." This is not merely an issue of threshold versus linear
relationships. The discussion should start with presentation of information on background levels
of PM10 and PM25, discussed elsewhere in the CD, and how one bridges from background levels
to ambient concentrations that show excess risk.
The discussion could then proceed to consideration of the range of PM indicator concentrations
evaluated. This might include population-weighted data for some studies, such as the NMMAP
study. The section should include a summary statement concerning the calculation of population
impacts of PM exposure. In my opinion, this would include a statement concerning the
inclusion/exclusion of lowest quartile or lowest half of ambient levels of PM in calculating PM
impacts for populations.
Page 6-258, line 29, Heterogenicity of Particulate Matter Effects Estimates: The section
could be improved by providing additional baseline data, especially relative to the NMMAP 90-
city study. This could include inclusion of a table showing the average baseline rate (total
mortality, cardiac and respiratory) for each of the cities studied, along with total population size.
The baseline mortality for each cause might be shown for each city since this was the base
A-60
-------
against which changes associated with PM10 were evaluated. In presenting data on
heterogenicity, it would be of interest to include data on cigarette smoking for each city and/or
region, recognizing that cigarette smoking is the largest factor driving cardio-respiratory baseline
rates.
Page 6-268, lines 3-6: This statement needs expanded discussion. If the effects estimates for
PM10 hospital admissions are higher than the effects estimates (percentage-wise) for PM10
mortality, does that imply that PM is more effective (than other underlying risk factors) in
causing hospital admissions as compared to mortality? If so, what is the potential explanation?
Page 6-269, line 3. Useful to add a sentence "However, the statistical association of health
effects with PM acting alone or with other pollutants should not be taken as an indicator of a
lack of effect of the other pollutants. Indeed, the effects of the other pollutants may be greater or
less than the effects attributed to PM."
Page 6-269, line 19: I suggest you omit reference to the APHEA study at this point in the
document. While being a useful study it should not have nearly the same influence as the
NMMAP study in terms of relevance to the U.S. The quality of the aerometric data was much
poorer than that used in the NMMAP study.
Page 6-270, lines 4-7: This broad statement sounds intuitively appropriate. However, I suspect
it is supported by very little data and the data were not reviewed in the CD.
Page 6A-2, Table 6A-1. For completeness, also present the data as rates; i.e., CVD deaths per
106/day. This will help in examining heterogenicity.
Page 6A-11: It would be useful in the interest of completeness to include the table shown as
Appendix A, Table 4 in the Staff Paper in the CD.
CHAPTER 7 - DOSIMETRY - GENERAL COMMENTS
This chapter is a useful summary of what is known concerning the dosimetry of inhaled
particles. However, the chapter does not have as strong a linkage to the rest of the CD and to the
issues of setting a NAAQS for PM as is needed. The chapter would be substantially improved
by providing a better linkage to aerosols characterized with PM10 and PM25 samplers at the
beginning of the chapter. At the end of the chapter, it would be useful to have a section
summarizing what can be predicted as the total deposition and regional deposition and retained
burden for various exposure conditions likely encountered in the ambient environment. This
should be done by using various PM indicators, i.e., PM10, PM10_25, and PM2j. In doing the
analysis, it is important to recall that the indicator measurement does not describe the total PM
size distribution and mass. For example, continuous exposure to ambient air characterized as
having either 3 0 • g/m3 of PMl 0, 15 • g/m3 of PMl 0_2 5, and 15 • g/m3 PM2 5 will yield the same
total deposition irrespective of which indicator was used assuming the size distribution was the
same in all three cases. It will also be important for the normalized calculations to be done for a
few key PM constituents.
Throughout the chapter, care should be taken to describe deposition relative to particle
size as probabilistic phenomena. This will help in conveying the correct view that particles from
0.5 to more than 10 • m can penetrate to and deposit in the nares, tracheo-bronchial region, small
airway, and alveoli—it is only the probability of doing so that changes.
CHAPTER 7 - DOSIMETRY - SPECIFIC COMMENTS
Page 7-2, line 28, 7.1.1 Size Characteristics of Inhaled Particles This section needs to be
expanded to provide a linkage to measurements of PM10 and PM2 5. In its present form, this
section is disconnected from the rest of the CD.
A-61
-------
Page 7-4, Structure of the Respiratory Tract. This section would be enhanced by including
one of the well-known figures illustrating the gross structure of the respiratory tract.
Page 7-9: The chapter would be enhanced by inclusion of a figure illustrating regional
deposition in the human as a function of particle size.
Page 7-24: The chapter would be enhanced by including one or more figures illustrating inter-
species patterns of total deposition and regional deposition for commonly used laboratory animal
species and the species of interest, humans.
Page 7-38: The chapter would be enhanced by including one or more figures illustrating inter-
species patterns of clearance and retained burden for commonly used laboratory animal species
and humans.
CHAPTER 8 - TOXICOLOGY - GENERAL COMMENTS
The introduction of the chapter could be strengthened with a better linkage to the epidemiology
chapter. The epidemiology chapter relates findings from multiple studies showing an increase in
health effects, primarily cardio-respiratory effects especially in susceptible populations
associated with various PM indicators when assessed in larger populations (usually a study size
of over 10,000 mortality or morbidity events times study days) with a relatively low prevalence
rate for the adverse events of concern. Restating this at the beginning of the Toxicology chapter
will help provide a setting for consideration of the lexicological findings on PM in humans and
laboratory animals under controlled exposure conditions. In my opinion, the lexicological
findings have generally not been very informative, as to how PM may be pathogenic in humans
or in identifying specific putative causative agents with PM. I suggest that the lack of progress
relates to the blunt "statistical" nature of current lexicological methods for tackling low
probability of added effects when the diseases of concern have low prevalence rate outcomes
even in susceptible populations.
It would also be useful if the introduction of the chapter could identify the challenge of moving
beyond characterizing whether a specific material is hazardous; i.e., capable of causing adverse
effects at any level of exposure, to the critical issue of the relevance of the findings at typical
ambient concentrations of PM.
The section of the chapter addressing susceptible populations should briefly consider the issue of
cigarette smoking as a risk factor. I submit that the vast majority of increased health effects
associated with PM in adult populations are observed in smokers or former smokers. These
populations contribute a disproportionate number of individuals with cardio-respiratory disease
and, thus, are the major susceptible population at risk from PM-related disease. It is noteworthy
that to date a well-defined animal model has not been found for cigarette smoking induced
cardio-respiratory disease. Smoking-related diseases develop slowly and are usually manifested
late in life. The absence of such models is also reflected in the lack of well-developed and
validated models of the common PM-related cardio-respiratory diseases. The minimal nature of
respiratory disease in young rats exposed for months to heavy doses of cigarette smoke may also
help rationalize the relatively refractory nature of rats exposed for modest lengths of time to PM
and constituents.
The section of Chapter 8 on in vitro exposures lacks information that would help place the in
vitro studies in perspective relative to in vivo exposures of humans to ambient PM. In comments
on Chapter 7,1 noted the need for calculations of deposition rates and steady state burdens of
PM in humans exposed to various levels of ambient PM. Such information presented in detail in
Chapter 7 could be summarized in Chapter 8 and provide a metric for comparison to the levels
used in in vitro studies. A review of these in vitro studies suggests that the concentrations of PM
and constituents studied are orders of magnitude in excess of any concentrations likely to be
A-62
-------
observed in humans at even the highest ambient concentrations encountered.
Chapter 8 also notes "there is growing lexicological evidence that diesel PM exacerbates the
allergic response to inhaled antigens." (Summary statement pages 80-86, lines 17-180.) This
statement and the supporting text needs to be qualified because of the high concentrations of
diesel PM or extracts used. The last published EPA Health Assessment for Diesel Exhaust
included a calculation of the quantity of diesel PM (and the organic fraction) inhaled and
deposited. That calculation should be referenced in this document in both the chapter on
dosimetry and in Chapter 8.
CHAPTER 9 - INTEGRATIVE SUMMARY - GENERAL COMMENTS
This chapter represents an excellent start toward providing an authoritative summary of
current knowledge of PM. It could be improved with some signification additions. As noted
earlier, the entire document from the introduction to this concluding chapter should build on the
sources-health responses paradigm recommended to the NAS/NRC PM Committee.
Section 9.3 on ambient paniculate matter could be enhanced by providing some summary
data on past and current PM levels. This could include information from the latest EPA "Trends
Report," the NMMAP study on 90 cities and the temporal trend for PM (as TSP) and other
pollutants for Philadelphia (from Kelsall et al, [1997]).
Section 9.4 on human exposures needs to be augmented with Figure 2-18 (Clayton, et al,
1993) from the Staff Paper.
Section 9.5 needs to be augmented with information on deposition rates and steady state
levels for various regions of the respiratory tract normalized to typical ambient PM
concentrations.
I suggest that a portion of Section 9.7 on Risk Factors be moved up after the present
Section 9.5. This new section, entitled "Baseline Health Statistics" could help set the stage for
the present Section 9.6 on Health Effects.
This new section should include the present tables 9-9 and 9-10 and additional
information on key health statistics. I suggest this include summary baseline data on inter-city
variability from the NMMAP study for 90 cities. It should also illustrate temporal variability
using the data for Philadelphia from Kelsall et al (1997).
At some point in the chapter it would be useful to include data, perhaps from the
NMMAP study on effects estimates for other key pollutants (O3, NO2, SO2, and CO), to help
provide perspective for the PM effects estimates.
Chapter 9 is seriously deficient in not providing a well-developed section on ambient
concentration-response relationships. This includes consideration of the threshold issue as well
as the relationship between ambient concentration-response as natural background levels are
approached. This is not merely a matter of thresholds versus linear responses.
A-63
-------
Giinter Oberdorster, PhD
Chapter 7 Dosimetry of Particulate Matter
Overall, this chapter summarizes well what has been presented in previous EPA
documents and gives additional useful new information. However, there are several rather
dogmatic statements which are unsupported and need either to be referenced or to be labelled as
speculative. Some sections are also rather simplistic by stating the obvious, a bit more depth
would help. This review summarizes on a page-by-page basis some suggestions for changes,
deletions, additions.
Page 7-7. line 7: In addition to defining the term "inhalability" it would also be useful to
define "respirability" since later on there appears to be some confusion as to which term should
be used.
Page 7-9. line 2: CMD is not necessary, it implies a size distribution whereas here the
upper limit is meant.
Line 4: The Frampton et al. study had both male and female subjects.
Line 9: Add after "diameter" the sentence: There was no gender difference.
Line 10: A statement could be added that this result compares favorably with the
ICRP 1994 model.
Line 13: A sentence should be added here listing some of the values of the Jaques
and Kim study, rather than giving the results only in relative terms.
Line 24: A sentence should be added here stating that at the same time, there is a
shift in deposition sites from more peripheral to central or extrathoracic regions.
Page 7-11. lines 18-20: 94 - 99 percent is not consistent with the result reported in the
previous paragraph (Yu et al.) where only 54% deposition was found for 1 nm particles, and
these have the highest deposition efficiency.
Page 7-12. lines 7-11: The efficiency of the nose as a filter for ultrafine particles has to
be seen in the context of the size within the ultrafine range. Whereas it can be very high for
nanoparticles below 10 nm, the filtering capacity becomes less for ultrafine particles of 20 nm
and greater.
Page 7-14. lines 10: Change "fine" to "ultrafine". In this paragraph again it would be
helpful to give some of the values that were found by Kim and Jacques in their studies in terms
of deposition efficiencies. A statement comparing their results with the ICRP model would also
be helpful, for example, the total deposition in the alveolar region found by Kim and Jacques for
40 and 60 nm particles of-33 and -27 percent, respectively, are in excellent agreement with the
ICRP model.
Line 30: To understand the modeling result it would be helpful to provide data on
the size distribution of the environmental aerosols in terms of MMADs and geometric standard
deviations.
Page 7-15. line 1: What kind of mathematical model was used? A brief descriptor
would be helpful.
Lines 4-6: If 36 of the inhaled coarse particles were deposited in the lung, that
doesn't add up if only 4 percent were in the tracheobronchial region and 2 percent in the alveolar
region. Please check. Likewise, 9 percent of the fine particles deposited in the lung is not
explained by 6 percent in the alveolar and a small fraction in the tracheobronchial region.
Lines 13-14: Here again 18 percent deposition in the lung is not explained by 2
percent in tracheobronchial and 3 percent in alveolar regions.
Line 23: I assume the cautionary note refers to the numbers (103, 105, etc) but
the general trend of differences between coarse and fine particle surface area and cell doses can
also be derived from other models, i.e., ICRP, MPP Dep model.
Page 7-17. lines 24-26: I suggest to add here also that exercising will cause a shift in
deposition sites from peripheral to more central airways as had been modeled by Martonen.
Page 7-18. lines 2: When differences in deposition between females and males are
described here, these results as well as those from other studies comparing the gender-related
deposition efficiencies should be critically evaluated: Both men and women breathed at the
A-64
-------
same tidal volume of 500 mL at 15 breaths/min, and this means for women, generally smaller
than men, an increased minute ventilation compared to their normal breathing condition.
Therefore, gender-related differences in deposition found here may be due to the fact that
women breathed at a relative larger minute ventilation and would not show if both men and
women would breath at their normal size-adjusted tidal volumes. A critical discussion along this
line should be added.
Line 13: It would be helpful to add here a summarizing paragraph since the
reviewed studies on gender differences show somewhat differing results and it would be
appropriate to have a summarizing concluding statement.
Page 7-20. lines 1-2: When comparing deposition efficiencies in the lungs of children
vs. adults, it should also be considered that children have a higher minute ventilation per unit
body weight compared to adults.
Line 26: Again, a summarizing paragraph would be helpful regarding age-related
deposition differences.
Page 7-25. line 17: ">5 jim" should be "<5 jim" since it is this lower range where
inhalability plays a role in deposition differences between rats and humans. Above 5 jim particle
size inhalability is no issue for rats as far as the lower respiratory tract deposition is concerned.
It would, however, be useful here to also discuss the importance of differences between rats and
humans with respect to respirability of particles, since differences here are more pronounced:
Particles >5 jim aerodynamic diameter are still well respirable in humans, but not in rats.
Page 7-26. lines 14-24: These model calculations by Hofmann and colleagues are not
easily understandable. For example, the statement that alveolar deposition in humans was lower
than in rats over the size range of 1 nm to 10 |im raises the question as to whether 10 pm
particles at all will reach the alveolar region in the rat? This is clearly beyond the respirability
range for rats. Did the model by Hofmann et al. consider the nasal filter in rats, or was it based
on particles entering the trachea? This needs some clarification. In addition, when comparing
deposition efficiencies between rats and humans, it should be mentioned here that to compare the
deposited fraction alone is not enough: What one needs to also compare is the deposited amount
per surface area which can give a quite different picture.
Page 7-27. line 8: Again, it is surprising that particle size-dependent deposition is
qualitatively similar in rats and humans for particles up to 10 jim, see comment on respirability
above.
Page 7-28. lines 3-14: This paragraph does not belong here, it is not dealing with
deposition but with retention pattern after chronic exposure to particles in rats and non-human
primates. In line 9 of this paragraph the term "deposition" should be replaced with "retention".
The whole paragraph should be moved to a later section where retention is addressed.
Lines 15-22: In lines 19 and 22, differences between rats and humans are
addressed without saying in which direction these differences go. This should be made clearer.
Moreover, this paragraph is rather vague, it needs to be a summarizing paragraph to point out the
major differences between rats and humans in a succinct way. The results by Hofmann et al.
summarized above are not easy to understand, and they certainly require a concluding, clarifying
summary.
Lines 23-31: This paragraph is a bit simplistic, and seems to have been written in
a hurry. I suggest in line 25 to replace "dose response" with "retention". In line 27, how is the
dose affected by species sensitivity? When different dosemetrics are addressed here in lines 28-
31, then all of them should be mentioned, i.e., number of particles, surface area of particles
(there are several studies showing the importance of particle surface area), the mass of particles
as well as the volume of particles. The dosemetric in terms of particle number vs. mass, etc.,
depends also on the physico-chemical characteristic of the particle, e.g., for soluble particles the
mass is probably still the more important parameter whereas any of the other parameters being
more important for poorly soluble particles. It is also not clear what is meant in line 30 with the
term "deposition": Is it deposition in terms of fractional deposition, deposition in terms of mass?
The deposition density in the rat is not necessarily higher than in humans because of the smaller
surface area of the rat lung, it depends very much on particle size and fractional deposition
A-65
-------
efficiencies as well as the ratio of rat to human lung surface areas. This paragraph needs to be
revised.
Page 7-29. lines 1 and 2: This concluding sentence stating that deposition density should
be considered when extrapolating health effects seen in rodent studies to the human situation
needs to be expanded in that other factors should be considered as well, such as dose in the
specific region, dose per unit surface area, dose per cell (e.g., alveolar macrophage), and also
particle parameters such as solubility, volume, surface area, size. Although deposition density is
very important, other factors should not be neglected.
In this section on interspecies differences, it would also be useful to mention the
availability of the Multiple Path Particle Deposition model (MPPDep) which allows the
calculation of particle deposition in human and rat respiratory tracts assuming different exposure
scenarios and breathing patterns and particle parameters.
In general, in this section on particle deposition efficiencies in the human respiratory
tract and in the rat, a figure would be useful so the reader would not have to consult other
publications for this purpose.
Page 7-31. Figure 7-3: If the size of the arrows in this figure indicates major vs. minor
clearance pathways, then the arrow from phagocytosis by alveolar macrophages to passage
through alveolar epithelium should clearly be a minor arrow since only a tiny fraction
phagocytosed by macrophages takes this route (studies by Harmsen et al.\ and the existence of
this route might even be questioned. However, under particle overload conditions the
translocation to interstitial sites via endocytosis by type I and type n alveolar cells becomes a
major pathway, but this does not occur via particle-laden alveolar macrophages.
The meaning of the double-headed arrow from pulmonary capillary endothelium to
phagocytosis by interstitial macrophages is not clear, does it mean that particles or interstitial
macrophages with particles are coming back from the endothelium? Also, the arrow from
phagocytosis by interstitial macrophages to pulmonary capillary endothelium is not clear: Is
there compelling evidence that, indeed, interstitial macrophages with phagocytized particles are
entering the pulmonary capillary endothelium?
Page 7-32. line 3: Not all solutes will be absorbed rapidly, it depends on the rate of
dissolution from a particle as well as on the molecular size of the solute and other parameters to
be discussed later.
Line 10: Probably meant here is that particles re-enter the airway lumen from
mucosal sites, is there any reference for that?
Line 23 and 27: I don't think that the general statement can be made that the
"magnitude of any increase in cell number (alveolar macrophages) is related to the number of
deposited particles rather than to total deposition by weight". This would result in a huge
increase in the case of deposition of ultrafme particles. Furthermore, cytotoxicity of a given
particle is certainly a big stimulus for inflammatory cell increase, and if particles are soluble then
the mass and not the number is the major determinant for eliciting cells. A better dosemetric to
relate cellular responses to deposited poorly soluble particles would be particle surface area, and
there are a number of studies which demonstrated that specifically for ultrafme and fine particles
- given that they are not chemically different - particle surface area correlates very well with the
increase in inflammatory cell numbers. Again, that applies only to poorly soluble particles and
not for soluble ones where mass is the more appropriate dosemetric.
Page 7-32. line 31: This describes the pathway in Figure 7-3 of macrophages traversing
the alveolar capillary endothelium directly entering the blood stream. Again, has this been
demonstrated for macrophages with phagocytized particles?
Page 7-33. lines 1-11: There are a number of statements in this paragraph which need to
be supported by appropriate references. For example, what is the evidence for macrophages with
phagocytized particles traveling to extrapulmonary organs? Are these new data? What is the
evidence of particles binding to macromolecules?
Lines 17-29: The clearance of solutes is a bit superficially treated here, it is not
that simple. It depends on the lipophilicity vs. hydrophilicity of solutes and the molecular
weight. There are also different solubilities depending on the intra- vs. extra-cellular localization
A-66
-------
of particles due to respective changes in local pH. After dissolution or leaching of some
components from a particle these can be binding of solutes (metals) to macromolecules; an
important pathway also is transport via caveolae across the epithelium as well as the
endothelium. The importance of differences between epithelial vs. endothelial pore sizes for
lower molecular weight solutes could also be addressed here.
Page 7-38. line 1: Snipes and Clem used 3, 9, and 15 jim particles and found only the 3
|im to be translocated, did Takahashi really see 5 and 9 jim particles being translocated?
Lines 4-6: One has to be very careful when drawing conclusions with respect to
lymphatic transport of particles based on intratracheal instillation studies: In such studies high
doses are instilled as a bolus leading to local overload which messes up the normal clearance
significantly and easily can result in lymphatic translocation which will not occur under normal
conditions. Also the statement that particles >5 jim have significant deposition within the
alveolar region is not correct for the rat. In the context of species differences related to
lymphatic clearance, studies by Thomas et al. (1971) could be cited here showing differences
between rodents and dogs, accumulation of particles in local lymph nodes being much greater in
dogs.
Page 7-42. line 29: A most important feature of Morrow's hypothesis is that a volumetric
overloading of alveolar macrophages occurs which eventually impairs its clearance function.
Page 7-43. line 11: I am not sure I understand why the slower alveolar macrophage-
mediated clearance in humans compared to rats (it is always slower in humans) would cloud the
overload relevance for humans: Humans also live about 25 times longer than rats.
Lines 14-15: It is hard to imagine how under normal environmental exposure
conditions, overload will occur in compromised lungs. What compromised lungs would that be?
Line 26: Although it is generally assumed that intratracheal instillation delivers
an "exact" dose to the lung, this does not mean that this dose is really found there shortly after
the instillation because some of the material is rapidly cleared out by the following exhalations.
The amount of this loss depends highly on the instilled volume as well as the instillation
technique, i.e., synchronizing with respiration or not.
Page 7-44. line 9: It is not clear what is said here, the amount that is deposited in the
lower airways by instillation can be adjusted, it is not due to by-passing the nose. Probably what
is meant is that the distribution of material is different between the two techniques.
Page 7-45. line 11: It is unclear what is meant by percentage retention of particles: Is
that the intercept of the retention curve with the ordinate, or is that the retention halftime? If the
retention halftime is meant here that would be explainable since normally by instillation high
doses are delivered which result in overloaded areas with retarded clearance. Thus, it might be
better to compare inhalation and instillation-associated retention kinetics by describing the
respective retention halftimes.
Line 18: The bulk of the instilled material certainly goes beyond the terminal
bronchioles, otherwise you would see all of it being cleared in a short time by the mucociliary
escalator. Of course, the very periphery of the lung is not well dosed, and as mentioned before,
the coverage depends also on the instillation technique, i.e., synchronization with breathing or
not.
Line 29: Disposition of particles is only one factor determining their biological
effects.
Page 7-50. line 1-6: For a discussion of "human equivalent concentration (HEC)" EPA's
RfC document should also be quoted here. Furthermore, earlier in this section, emphasis was on
the lung burden expressed as per unit lung surface area as being more appropriate, whereas here
the amount per gram of lung is indicated. This might be confusing for the reader.
Lines 13-19: The Asgharian 2000 reference is missing in the reference list, is that
a publication describing the MPPDep model which should be mentioned here as well?
As a general comment on this section, it should also be stated in a concluding
summarizing paragraph that all models are just that: models. They have inherent uncertainties,
which can be large and differences between model results can probably most of the time be
explained by these uncertainties.
A-67
-------
The title of this section is also somewhat misleading, both 7.6.1 and 7.6.2 deal with
deposition and some clearance and retention, but the disposition of particles in terms of where
particles move after deposition is not really addressed in this section on "Modeling of
disposition". Much of what is reviewed in this section is already described in prior sections of
this document and somewhat redundant.
Page 7-52. line 25-31: As we had discussed in the previous review, one has to be careful
with the interpretation of the results by Nikula et al. (1997) since it was derived from a one
timepoint post-exposure evaluation only: Rats with particle overload clear significant amounts
to the regional lymph nodes, which means that the particles have to become interstitialized first;
once in the interstitium, the rate of interstitial clearance to the lymph nodes may be much faster
in rats than in primates which cannot be evaluated from a result obtained from one timepoint
only. At this one timepoint, the interstitium in the rat could already be significantly cleared
which would incorrectly be interpreted as less interstitialization. Therefore, whether this reflects
truly a difference in retention pattern between rats and primates or a difference in interstitial
clearance rate cannot be decided from the analysis at one timepoint.
Chapter 8
Page 8-1. lines 5-10: Among the questions listed here should also be the most important
one, namely: Does PM at relevant ambient concentrations cause adverse effects?
Line 15: Change "air" to "PM'. Add at the end of the sentence in line 16: "or
suspension".
Page 8-6. lines 16-17: The study by Kuschner et al. used median concentrations of 133
mg/m3, at which concentrations the particles are no longer ultrafines, so one has to be careful
with their conclusion that there is no difference between fine and ultrafine particles. There is no
question that chemical composition, surface radicals, etc., play a role as well, which is not
disputed, just think about ultrafine PTFE vs. ultrafine TiO2. But to exclude size as an important
factor for toxicity is wrong. This comment has already been made by me for the previous
criteria document and obviously was not considered.
Page 8-7 and 8-8. Studies by Osier: The inhaled concentration for the TiO2 was 125
mg/m3 for 2 hrs. (not jig) in order to match the intratracheally instilled dose in terms of
pulmonary deposition.
Page 8-9. lines 19-22: The dose of 5 mg deposited in the human lung in this study is
certainly much more than can be deposited from ambient air.
Page 8-10. line 18: Change "Teflon polymer" to "PTFE".
Lines 22-23: Again, the study by Kuschner et al. is cited here as demonstrating
that composition and not particle size was responsible for health effects in this study. Given that
the median concentration of the particles was 133 mg/m3, these particles were no longer
ultrafines, but aggregates. Obviously, in addition to size, composition is also a very important
parameter and both need to be considered (see above).
Page 8-19. line 30: It would be useful to point out in this context that in general the
intratracheally instillation studies failed to include a benign particle such as TiO2 as a
comparison to show that the effects observed are more than just a general particle effect.
Page 8-22. line 24: I strongly suggest to include the word "high" when the ROFA doses
are addressed.
Page 8-23. line 30: The dose of LPS is given here as 5 or 50 jig. Is that the inhaled
dose? Is that the dose in the nebulizer, or an estimated deposited dose in the lung?
Page 8-24. Study by Elder et al.: The concentration of 100 |ig/m3 is for the particles, not
forLPS.
Page 8-30. lines 3-5: The effects observed here with ROFA inhalation should be viewed
in the context that the inhaled concentration was 15 mg/m3 and that inspite of this high
concentration there were much lower or no effects compared to instilled ROFA which caused
increased mortality.
Page 8-31. line 6: The concentrations of ROFA given were not only high, I suggest to
describe them as "very high".
A-68
-------
Line 19: Was the change in heart rate variability an increase rather than a
decrease? I think what should be stated here is that the ratio of low and high frequency band of
HRV decreased.
Page 8-32. lines 10-19: Here the two different dog studies by Godleski and Muggenberg
are compared, however, the studies are significantly different from each other in that Godleski
used CAPS and Muggenberg used ROFA, the particle size might also have been very different.
Thus, it is difficult to compare the different findings between the two studies given also that
storage of ROFA could have played an important role in altering its toxicity. It should also be
considered that the dogs in the study by Godleski were exposed via a tracheostomy tube.
Page 8-34. line 4: I suggest to change "high concentrations" to "only high
concentrations."
Page 8-37. lines 28-29: The exposure concentration of ROFA was 15 mg/m3?
Page 8-38. line 17: Change "Teflon particles" to "ultrafme PTFE fumes".
Page 8-39. line 9: In this section of age-related differences in PM effects, the studies by
Elder et al. should be included, they describe effects of inhaled carbonaceous model particles in
LPS-sensitized rats of old and young age (Elder, A.C.P., Gelein, Finkelstein, J.N., Cox, C. and Oberdorster,
G. Pulmonary inflammatory response to inhaled ultrafine particles is modified by age, ozone exposure, and bacterial
toxin. Inhalation Toxicology 12 (Suppl. 4): 227-246, 2000; Elder, A.C.P., Gelein, R., Finkelstein, J.N., Cox, C. and
Oberdorster, G. Endotoxin priming affects the lung response to ultrafine particles and ozone in young and old rats.
Inhalation Toxicology 12 (Suppl.): 85-98, 2000).
Page 8-40. line 2: Is a fibrotic response an important endpoint for ambient PM?
Page 8-39 thru 8-45: In this section on genetic susceptibility to inhaled particles, a
discussion on the dose levels used in the different types of studies would be useful to put them in
perspective to ambient levels and deposited doses.
Page 8-48. lines 7-9: Among the severe limitations of in vitro studies are the dose levels
which are generally orders of magnitude higher than experienced in vivo; and in addition the fact
that only acute effects and mechanisms can be evaluated in vitro which could be very different
from mechanisms causing chronic effects in vivo. These significant limitations should be added
onto the discussion in this section.
The title of Chapter 8.5 refers only to in vitro exposures, which gives the
impression that mechanisms can only be evaluated by doing in vitro studies. This is not correct,
mechanisms are also evaluated by in vivo studies, in fact, the in vivo studies may be more
important since they only can provide compelling evidence that any mechanistic pathway
explored in vitro, indeed, is also operating under in vivo conditions which are obviously much
more complex.
Page 8-57. lines 30-31: This two-line summary can be used for any type of particle and
is not very specific, and it may be useful here to also again point out that the high doses that are
used in these in vitro studies need to be considered. A sentence stating that detailed specific
mechanisms related to ambient PM still need to be uncovered should be included here.
Page 8-65. line 8: What does the study of i.p. injection of ROFA contribute to an
evaluation of mechanisms? This study doesn't seem to make much sense.
Lines 18-30: When comparing different dust materials in in vitro studies, it
becomes very difficult to rank the toxicity of the different dusts because it is not known as to
whether the different particles are internalized by the cells to the same degree, and also the
dosemetric in terms of mass vs. particle number or size can significantly influence the result.
The term "exposure-dose" used in line 30 is not clear, what does it mean?
Page 8-70. lines 15-16: This statement is only true if the chemical composition of the
ultrafine particle and larger particle is the same, which should be added here.
Lines 15-29: Lines 27 - 29 provide an explanation for the observation that high
doses of fine particles cause a greater effect than high doses of instilled ultrafine particles.
Indeed, results of our earlier study (Oberdorster et a/., 1992) demonstrated that the significant
amount of ultrafine particles being interstitialized when high doses are instilled causes a decrease
in the inflammatory cells in the alveolar space compared to inflammatory cell influx at lower
doses of instilled ultrafine particles.
Line 31: The studies by Oberdorster et al. (2000), which are alluded to here, in
A-69
-------
old and young rats and mice used only ultrafine carbon particles, see also the publications by
Elder et al. (2000, 2001) which were mentioned earlier in my comments.
Page 8-72. line 11: Replace "properties" with "area".
Page 8-73. lines 5-8: One has to be careful to characterize ambient PM as ROFA which
has been used in a number of animal and in vitro studies. The ROFA that was used was
collected from a bag house, and - as was pointed out earlier in this document - has a different
metal content than the fly ash which is actually released into the environment, also metal
solubilities are different. Furthermore, the high doses that were used in the ROFA studies need
to be mentioned here as well.
Page 8-74. Section 8.5.5.2: This section reiterates studies that have been described
before in this document. It should be remembered that the studies which are used here to
demonstrate a specific mechanism to cause systemic effects have been run at very high doses or
exposure concentrations, and thus, one needs to be very cautious to extrapolate these responses
to relevant ambient concentrations of PM. What the studies do is show that the concept of a
specific pathway or mechanism is valid in principle, but this needs to be validated and verified
by additional studies using relevant exposures.
Page 8-81. line 26: Include (Elder etal.,2001).
Page 8-83. Section 8.7 Summary: This section provides a good summary of our present
state of knowledge. There should be a few clarifications:
Page 8-85. line 14: Implications for what? The implication I see here is to conduct
further studies on the importance of metals, and that the ROFA studies have pointed out the
importance of the metal concept for PM toxicity in general.
Page 8-87. line 16: Another ultrafine ambient PM concentrator was developed by
Koutrakis and colleagues.
Section 8.7.1.2. Susceptibility: Among the susceptibility factors, not only
genetically or induced compromised health should be listed but also age as a factor.
Robert Rowe, PhD
Below are revised comments on the second draft CD and draft Staff Paper for the PM
NAAQS. The EPA staff are to be commended for the work to date, especially recognizing the
significant growth in literature relevant to the PM standard. My comments focus on economic
and visibility perception portions of the materials provided.
Visibility Impairment Assessment
The Staff Paper Section 5.2.5, and a supplemental paper, address a proposed approach to address
visibility impairment in terms of human judgement. I encourage EPA to pursue this perceptions,
preference and economic valuation work as important to setting the secondary standard.
However the work completed to date should be seen as being only very preliminary. It is
important that a more comprehensive workplan be developed, including how the results may be
used as input to decisions about the secondary national standard, and that the type and level of
work effort be consistent with the expected use of the work.
My key recommendations are that:
1. EPA should conduct several additional focus groups in the Washington, D.C. area to address
basic issues in the research.
2. EPA should then conduct similar preliminary analyses in another city that is diverse from
Washington, D.C.
3. Based on the above, EPA should develop a more comprehensive workplan on the issues and
objectives, data to be collected and its uses, steps to be performed, survey instruments, and
time schedule. I recommend a peer review at this stage.
4. EPA should develop a plan that will provide a sound sampling strategy, and not just one or
two focus groups in a variety of locations. This may require OMB approval.
A-70
-------
Below are additional detailed comments on the visibility focus group materials.
- Little confidence should be attributed to one focus group of 9 people in one location
(Washington, D.C), and this group should not be seen as sufficient to launch a multi-city
assessment. I advise repeated groups in the first location to obtain more data and to address
issues before proceeding to other locations, or to conclusions. Among the issues that could
be considered are (1) how do the types and kinds of locations presented in the vistas alter the
conclusions, if at all? (2) how much are perceived health concerns affecting the judgements,
and how can this be better addressed? (3) what does it means when people say the
impairment is acceptable or unacceptable? It is based on the view, the impact on their mood,
are there behavioral changes? Does this mean the identified threshold level is acceptable
every day or several days a year? Does this mean respondents are no longer impacted, or just
that they think the likely perceived costs of further control may not be worth it (and on what
basis do they make such a judgement), or that further improvements are not realistic. In this
rating, respondents are participating in a stated preference (SP) assessment, and more
attention should be given to the SP literature. (4) Which measure will be used? For example,
in the simple rating, the cross over point for unacceptable is 20 • g/m3, but with the "how
many hours a day" rating, 32.5 • g/m3 is acceptable for as many as 4 hours a day by two-
thirds of the respondents (and thus presumably a level of higher than 32 • g/m3 for 4 hours a
day would be acceptably on a simple 50% rule), and based on the economics data, there is
clear impairment below 20 • g/m3.
When moving to multiple locations, issues arise such as which vistas to present, what type of
impairment (which varies in some locations), and how correlated will the ratings across
locations be to existing conditions across locations (valuation literature would suggest status
quo bias leading to anchoring and some adjustment to improved conditions).
While the approach follows similar work at the state and local level, it is not clear that the
approach is sufficiently resolved for a national standard when the "impairment" threshold
may be highly variable across locations. How does EPA see using the results? How might the
results tie in to the PM NAAQS or other visibility rules?
The economic valuation questions are preliminary, yet highlight there may be meaningful
losses at visibility levels below the 50% rule for acceptable ratings. In the preliminary focus
group the switch from 50% acceptable to 50% unacceptable occurs at 20 • g/m3. However,
when provided a choice, 5 of 9 would choose 15* g/m3 and pay $50/year, as opposed to 22.5
• g/m3 and paying $10/year (2 were indifferent between 15* g/m3 and 22.5* g/m3, and 2 chose
22.5* g/m3 over the status quo of 32.5* g/m3). This suggests a significant value for visibility
conditions below the 50% rule level for either the simple ratings or hours per day ratings. I
support further investigation into the economic valuation approach, with much more
attention to survey design consistent with the stated preference valuation literature. To
address the joint product issue between visibility and health, one might revisit the Carson et
al. Cincinnati work performed for EPRI some years ago, which by the way showed losses
down to just a few days a year of visibility impairment (e.g., an indistinguishable change
when presented on an annual average basis).
There are important concerns with the proposed "focus group" approach to this assessment.
Generally a study consisting of a group of focus groups across different locations may not be
viewed as sufficiently rigorous for the intended policy application. EPA should see the focus
group approach only as a preliminary effort to a larger scale survey effort.
Staff Paper Visibility Section
This section is better than the corresponding section in the CD. The two sections should
be consistent. A few suggested editorial changes for the Staff paper (aside from continuing to
include but reduce the discussion of this work). On page 5-16,1 recommend active use and
A-71
-------
passive use values as opposed to use and non-use, to better identify that in some cases visibility
is actively enjoyed, while in other cases it is passively enjoyed, and realize that it is often
difficult to separate benefits by these categories (e.g., where does option value fall?). Page 5-23
of the staff paper was missing.
Criteria Document Chapter 4: Environmental Effects
General Notes Overall, this section is reasonably comprehensive. Two overriding
considerations are (1) can the presentation be more focused to key questions in the setting of
standards, rather than a litany of information and append!cies (this seems particularly true for the
global climate sections), and (2) can economics, if it is to be addressed at all, be addressed more
consistency in the various subsections.
Section 4.2.2: Natural Ecosystems
- Lines 7 through 15.1 recommend some terminology clean-up here, rather than propogating
terms inconsistent with the broader resource economics literature. All benefits from
ecosystems can be described as ecosystem services. I think this could use revision,
especially on page 4-20, to state something along the lines of "there are a wide range of
ecosystem services, including (1) some with readily recognized market value (e.g., fish,
timber, minerals,...) and (2) others services without current or readily identified market
values. For the purposes of this discussion only, we refer to the first group as "market
services" or "goods" and the second as "non-market services". Table 4.2 illustrates various
market and non-market services provided by ecosystems..." Then, I think Table 4-6 is much
more informative than Table 4-2 and could replace Table 4-2.
- Page 4-83 identifies economic literature to demonstrate the significance of ecologic
resources and services to mankind (Pimentel and Costanza). These numbers are presented,
perhaps, with too much credence. There is significant controversy in the economics literature
about the reliability of the specific estimates (See the Special Issue of Ecologic Economics,
April 1998, and Freeman, 1999), not the least of which is that economics is much better
suited to evaluate individual services, or better yet changes in service flows for an individual
ecologic service, than it is to evaluate the total value of all ecologic services. Economics
aside, most all agree that ecologic services are central to human life and obviously of
substantial value. Consequently, substantive impact on ecologic services have the potential to
have an important impact on human welfare.
Section 4.3.9 Visibility Economics. Generally, there should be more consistency to the Staff
Paper write-up. To the degree this is retained along its current lines, I note the following edits.
- Page 4-111 line 27. Replace "costs" with "losses" (here and generally throughout the
section).
Page 4-111, line 29, replace "cost/benefits" with "losses from visibility impairment".
- Page 4-111 line 31, and continuing to page 4-113, line 3. The avoided cost method, while
used as a market cost measure of materials damage, and sometimes in other application, is
not used in the visibility literature and should not even be discussed here. Just start with
something similar to line 4 "There are several methods...."
- Page 4-113, line 12, it would be useful to have a citation on visibility property value studies
(e.g., Chestnut and Dennis, or the NAPAP work from a few years earlier for summaries,
which is cited elsewhere in the CD and staff paper). There is quite a bit of property value
literature, with the difficulty of sorting out value differences into visibility and health
components. One could also cite some of the new property value applications (Thayer and
Murdoch).
A-72
-------
- Page 4-113, lines 13 through 19 focuses on CVM, but really is about stated preference
methods, including CVM applications. Some of the past economic studies are more like SP
choice studies than conventional CVM applications. Consequently, it may be appropriate to
merge the paragraphs starting on lines 13 and 20.
- Page 4-113, line 19 could use a citation, either NAPAP or Chestnut and Dennis, or Mitchell
and Carson.
- Does the Hanley and Spash reference discuss visibility applications in specific and in detail?
- Page 4-113, line 31. "Davis" should be "Dennis".
Health Risk Assessment (Staff paper Chapter 4 and separate paper).
- I support conducting the assessment in more than 2 locations, as discussed at the meeting.
Staff paper 4-13, lines 10-26 discusses assumptions about changes in ambient conditions to
meet standards, relying predominately on the rollback method. Using the rollback method is
reasonable, but EPA should give careful attention to the proposed sensitivity analysis of
alternative adjustments (lines 24-26). With increasing costs of compliance, episodic and
other control strategies that reduce the highest concentrations may receive increased
attention. Further, given that the population exposed is not uniform across concentration
levels, and many concentration-response functions are non-linear, differences in the
assumptions to reduce concentrations to achieve standards can have a significant impact on
the risk assessment.
- Deck et al, 2001 is cited several times, starting in the first paragraph, but is not available. It
may be useful to provide this paper for this review.
Criteria Document Chapter 9
This chapter is well done as a series of separate summaries, but it needs more integration
and needs to be reduced in length - not everything needs to be summarized. It appropriately
focuses on the larger questions of increasing consistency in the results of available health effects
literature and extensions to this literature. In terms of the important question of retaining or
revising the existing PM25 standard levels (15 ug/m3 annual average and 65 ug/m3 24 hours),
little is presented in this chapter on the strength of the evidence, shapes of the estimated C-R
functions around these levels, or effect thresholds (although this is touched on in Section 6.4.6).
Jonathan Samet, MD
Chapter 5 - General Comments:
In general, this is a cohesive and thorough chapter that carefully sets out concepts of exposure
assessment, measurement approaches, and findings. The literature review appears complete and
findings are well represented in tables and in the text. The chapter has a potentially key role in
setting a framework for interpreting the epidemiological data presented in Chapter 6. The
chapter does address the implications of the exposure literature for interpreting the
epidemiological evidence. Unfortunately, there is little linkage between the two chapters in this
regard; Chapter 6 almost reads as though Chapter 5 had not preceded it. There is a need for
better integration, a burden which clearly lies with the authors of Chapter 6.
This chapter also discusses issues related to confounding and measurement error that overlap
with Chapter 6. With regard to issues of confounding, it will be important to have a uniform
view throughout the CD. My comments for Chapter 6 should be considered in this regard.
A-73
-------
Specific Comments:
Page 5-19, lines 15-18: Generalizability (external validity) is not dealt with well here. A
"purposeful study" may give generalizable information; the extent of generalizability is a matter
of judgment, based on study participant characteristics. The term statistical inference is used
inappropriately here.
Page 5-41, lines 5-16: This paragraph is far too sweeping in its condemnation of the exposure
assessment literature. What are the "important questions" that are so poorly answered.
Chapter 6 - General Comments:
This lengthy chapter provides an exhaustive, descriptive summary of the most recent
epidemiological findings on paniculate matter and morbidity and mortality. The literature
review is comprehensive and the tables offer useful summaries of an extensive literature. There
are, however, weaknesses that should be addressed; these weaknesses may reflect the multi-
authored nature of the chapter which has resulted in an uneven approach in style, synthesis, and
interpretation. Key aspects of the chapter needing to be addressed include:
1. The chapter is not adequately connected to the remainder of the CD. There is a lack of
integration with Chapter 5, which should provide a foundation for exposure considerations
related to interpretation of the epidemiological literature. This foundation is not used, and far
less strong and competent text is provided.
2. The chapter fails to sharply set out key concepts—confounding, causal associations, and
causal pathways, in particular. Effect modification is also not handled well and the text related
to these key aspects of interpretation is often murky. In the chapter's introduction, it would be
useful to provide diagrams indicating the relationships that hold under confounding, direct causal
pathways, and indirect causal pathways. I have attached one possible set of diagrams.
Additionally considerations as to confounding, reflect biological understanding as to the
independent action of the confounder and not just patterns of association in data. Changes in
estimates are not a particularly useful gauge as to the presence of confounding in the presence of
measurement error, possible effect modification, and correlations among the independent
variables. The text in places offers some clear thinking on these difficult topics, but much of it is
not clear.
3. In interpreting data, there is excessive reliance on p values and attaining statistical
significance for effect estimates. This needs to be corrected; the p value should not be held as a
decision-making tool for data interpretation, as in the present chapter (see Epidemiology for a
recent set of perspectives on this issue. Emphasis should be placed instead on precision of effect
estimates (i.e., width of confidence intervals..
4. In fact, the chapter never clearly sets out how data will be interpreted. Summary judgments
are offered but without reference to any common framework Judgments are often couched
subjectively and there is a general failure to place the epidemiological evidence within the
broader context of lexicological and dosimetric understanding. The framework for interpretation
is badly needed. Throughout the text, there is variably use of significance level, precision of
effect estimates, and magnitude of effect, as the effects of PM are weighed against those of other
pollutants. The same problem is evident when the chapter interprets the literature on particle
characteristics and components.
Specific Comments
Page 6-2, lines 11-15: This proposed hierarchy of "inferential strength" is neither correct nor
useful.
Page 6-3, lines 19-21: This type of sweeping statement should be avoided. On close reading,
the sentence offers only a garbled pejorative comment.
A-74
-------
Page 6-3, lines 24-25: Around this point, the text needs to be very clear on causal and non-
causal pathways. Also, the term "effect" and not "effects" modification is in general usage.
Page 6-4, lines 6-13: An example of muddled text around the confounding/causality issue.
Lines 21-25, also exemplify this problem.
Page 6-9, lines 8-11: A not well developed fragment on measurement error that addresses its
consequences for effect estimates and for confounding. The second clause of the sentence raises
the complex issue of differential measurement error across independent variables with little
explanation.
Page 6-49, Section 6.2.2.4: This section initially needs to set out issues that arise in interpreting
the evidence on particulate matter components. Unfortunately, this has not been done well by
the authors of many of the reports and the authors of this report fall into some of the same traps,
particularly reliance on the p value (see, page 6-54, lines 1-8, for example).
Page 6-55, lines 25-30: These comments about PM2.5 sources need to be referenced.
Page 6-77, lines 1-5: Another example of very confused interpretation.
Page 6-96, lines 13-19: Basis of judgment not clear. Last sentence of paragraph needs
clarification.
Page 6-101, lines 1-16: Too speculative.
Page 6-136, lines 20-22: The statement concerning barometric pressure is far too strong, based
on a single study.
Page 6-126, lines 10-11: Multicity studies provide far more strengths than precision alone.
Page 6-217, lines 6-10: This sentence reads as though we have no prior knowledge on PM and
health and should give equal weight to all models. That is hardly the case.
Pages 6-217-218: The sweeping generalizations about modeling need to be toned down. This is
not the state-of-art.
Page 6-219, lines7-22: The discussion of lag structure, largely turning to statistical grounds for
choosing the appropriate lag, is off the mark. Certainly, we have some knowledge of the kinetics
of injury and substantial prior modeling work.
Page 6-225, lines 20-23: The conclusion may be correct, but its basis is not clear. The last
sentence is not clear.
Pages 6-226-6-227: This section would be much stronger with my suggested addition.
Page 6-239, lines 21-27: There is little basis to assume different relationships across locations.
Page 266, lines 11-15: This paragraph shows little understanding of how evidence is assessed to
determine causality of associations. What are "causal studies" from other disciplines.
Page 266, line 20-22: What is meaningful heterogeneity?
Page 6-269, lines 15-23: What is the careful evaluation that is needed? APHEA and NMMAPS
have been rigorously reviewed.
A-75
-------
Chapter 9 - General Comments
This chapter is offered with the general and needed purpose of providing an "integration of key
information". Unfortunately, it falls far short on this purpose, reading more as a summary, with
interspersed comments and indications of research gaps. Even these comments, are not
particularly penetrating. See, for example, lines 22-30, page 9-36, which overviews some issues
in interpreting the epidemiological evidence. A higher level of analysis should be adhered to,
particularly given the sophistication of the discussion since the last CD with regard to
interpretation of the epidemiological data.
One approach that could be taken in this chapter would be to follow the NRC Committee's
framework and to provide an "across the box" linking of what is known. The framework could
also be used to highlight what is known and the uncertainties, as well as systematically point to
research needs. This might be an appropriate way to conclude the chapter.
Follow-up Comments
These comments are intended to supplement the peer-review comments submitted in
advance of the July 23-24, 2001, meeting of the Clean Air Scientific Advisory Committee
(CAS AC). These comments are based on my spoken remarks concerning interpretation of the
epidemiological evidence in Chapter 6. Copies of transparencies used as the basis for these
remarks are attached.
My remarks strongly urged rigorous and standardized use of epidemiologic terminology
and concepts throughout the Criteria Document, particularly in Chapter 6. The Criteria
Document needs to define and then uniformly apply the concepts of confounding and effect
modification. As noted on the first transparency, confounding arises when a factor, associated in
its own right with the outcome of interest, is also associated with the exposure under
investigation. In this circumstance, a spurious association may arise between the exposure and
outcome because of the confounding factor. For a variable to be a "confounder" in a particular
data set, two conditions must be met: 1) the confounder needs to be associated with the
outcome factor independently (i.e., it is a risk factor for the outcome); and 2) the confounding
factor is associated with the exposure of interest in the data under consideration. A distinction
should be made between a confounding factor and a "potential" confounding factor, that is a
factor which would be a confounder if these two conditions were met in a data set of concern.
Frequently, critics of epidemiological findings raise the possibility of confounding, citing
numerous potential confounders, without attention as to whether these two conditions are, in
fact, met.
Effect modification is distinct from confounding. It refers to circumstances in which the
exposure/outcome relationship depends on the presence or absence (or level) of the modifying
factor. In such circumstances, there are a series of risks for the outcome of interest associated
with exposure, depending on level of the modifying factor.
The next two transparencies concern the theoretical example of particulate matter (PM), nitrogen
dioxide (NO2), and mortality. In many settings, PM and NO2, have common sources and there is
a potential for either confounding or effect modification. Of course, for NO2 to be a confounder,
it would need to be a predictor of mortality, an association that has not been consistently
demonstrated. Thus, on the basis of understanding of toxicology of NO2 and the epidemiological
data available, it would not be a candidate to be a confounder.
In the diagram labeled "causal", I have indicated that particulate matter is associated with
increased risk for mortality. NO2 and PM concentrations may be associated because of their
common sources, but the diagram does not link NO2 with increased mortality, reflecting
understanding of its toxicity.
Next on the page, the example is designated "causal pathway". In this example, NO2 is in the
causal pathway for the increased risk of mortality associated with PM. It contributes to the
A-76
-------
formation of secondary particles, which are the actual toxic agents. NO2 concentration (or its
sources) might be considered as a "surrogate" for the proximal causal agent, PM.
The third transparency provides diagrams for confounding and modification. As already
mentioned, NO2 is an unlikely confounder, given the lack of evidence of the increase in mortality
with rising NO2 concentrations. However, assuming that it were a risk factor for increased
mortality, the diagram represents the relationships for confounding. If NO2 level modified the
effect of PM, then a set of relative risks describing the association of PM with mortality would
be derived, corresponding to the strata of NO2.
There are a number of other epidemiological concepts to be considered in the Criteria
Document:
Confounding versus potential confounding: Throughout the document there should
be careful attention to whether conditions for confounding are met. As noted,
raising the possibility of confounding does not mean that confounding is actually
present.
Interaction: In places, the term "interaction" is used, generally in place of effect
modification. Interaction properly refers to the statistical terms used in a model to assess
effect modification.
The mixture problem: Admittedly, ambient air pollution is a complex mixture, of
which PM is one component. Nonetheless, the Clean Air Act has designated PM and
other "criteria" pollutants for regulation. Study designs and data analysis are directed at
attempting to characterize the effects of PM and these other pollutants, and not that of the
mixture itself. The criteria pollutants provide some indication of the characteristics of
these mixtures and consideration of effect modification represents an indirect approach to
understanding the toxicity of mixtures. The Criteria Document should acknowledge the
mixture issue and the related requirements of the Clean Air Act specifically.
Measurement Error: This is a key issue that should be addressed in Chapters 5 and 6.
Throughout the document, the concept of measurement error is considered but the
underlining formulations are variable and not necessarily accurate. The document should
be made uniform for this key issue. The consequences of measurement error are
complex and its potential consequences should be listed, at least in a general fashion.
Heterogeneity : The Criteria Document considers the heterogeneity of risk estimates
across the United States. This heterogeneity cannot be completely explained by
available, but crude, indicators. Heterogeneity does need to be explained, but its
presence is not a barrier to interpreting the findings on paniculate matter. Additionally,
summary estimates at a national level can be made in the face of heterogeneity as they
intrinsically weight the U.S. population's exposure by the underlying distributional
modifying factors.
Interpretation of epidemiological data: Chapter 6 offers a relatively literal interpretation of the
epidemiological evidence, absent a clear biological framework. In interpreting epidemiological
data, the need for a foundation in biological understanding is evident. However, Chapter 6 as
presently authored, makes little connection to the substantial literature that is reviewed in other
chapters. These connections should be made in Chapter 6 and then reinforced in Chapter 9.
A-77
-------
George Taylor, PhD
Air Quality Criteria for Particulate Matter: Chapter 4 (Environmental Effects)
General Comments
There are eight overarching comments on the issue of PM and ecological effects.
1. The consequences of particulate matter (PM) for welfare issues are largely relegated to
visibility. The effects on vegetation and ecosystems of ambient levels of PM are regarded as
being trivial and do not require substantive discussion. In contrast, the consequences of PM on
human health are highly significant, well characterized and easily quantifiable in economic and
human health dimensions. This (human health) is where the emphasis needs to be directed.
2. In light of the above, the CD is VERY excessive in its discussion of PM effects. The
excessiveness can be traced to several issues. The first is inclusion of topics that simply are not
relevant or are trivial. The second is the depth of discussion of issues that probably could be
succinctly presented in 50% or less space. The third is the "handle" applied to the issue of sulfur
and nitrogen inputs. This is a PM CD and sulfur and nitrogen are small contributors to the
nitrogen and sulfur inputs to landscapes. The breadth and depth of attention to nitrogen and
sulfur far exceeds the environmental concern as it is related to PM.
3. One of the major ecosystems affected by PM deposition and for which EPA has heavily
invested in R&D is deposition of particles to surface waters. The most notable studies are ones
from the Great Lakes and to a lesser degree the Chesapeake. It is important that these systems
be included.
4. By length alone, one might conclude that the nitrogen or sulfur issue is driven by PM.
This misinformation might be translated by policy makers into thinking that changes in PM will
affect significantly such issues as nitrification, etc. Since most (>80%) of the nitrogen and sulfur
that enters continental landscapes comes through processes other than PM, it is not appropriate
to present the information as currently presented in the CD.
5. The human health chapters do a creditable job of linking the sections on atmospheric
chemistry with the effects on human health. In the sense of a risk assessment, there is a tidy
linkage between exposure and effects. This linkage is missing altogether in the section on
environmental effects. There is no effort to relate the PM in the atmosphere to effects in
terrestrial or aquatic landscapes. The consequence is that the chapter fails one of the basic
premises of risk assessment. It is strongly recommended that the chapter better establish a
linkage between exposure and effects. Or, the other option is to simply delete the nitrogen and
sulfur topics from the PM CD.
In looking over the chapters on the atmospheric chemistry of PM, there is little quantitative
discussion of the magnitude of sulfur and nitrogen in PM. Although both are discussed, it is
difficult to see how the environmental chapter could be so "loaded" with nitrogen and sulfur
when the atmospheric chapter does not heavily present the same information.
6. The final overarching issue is a derivative of the above. The conclusions portray the
potential for PM to be a major stress on continental landscapes in the US. This is largely driven
by the obsessive discussion of nitrogen and sulfur and by the failure to effectively link exposure
in the atmosphere to effects. The conclusion is more alarmists than needs to be portrayed and
the data simply do not reflect that degree of concern. More realism is needed in the assessment.
7. Deposition is missing from this CD. For ecosystems, there is a critical linkage between
atmosphere concentration and effects and the vector is deposition. It is important to have a
section devoted to deposition so there is a frame of reference for know what the inputs to
A-78
-------
ecosystems are. There are a host of papers that address this issue and at least some should be
cited.
8. In looking at PM per se, it is interesting to note that the chapter fails to mention to one
type of ecosystem for which deposition of PM is likely to be very important - urban and
suburban forests (largely in parks). There is a great deal of literature on these systems. In fact, it
might be best to replace the current discussion of deposition to the IPS sites and replace that
material with the urban suburban forest analysis.
Specific Issues
1. There is little reason to consider in much depth the consequences of PM on vegetation
and ecosystems. In fact, most of the material in Chapter 4 characterizing the effects on
vegetation and ecosystems could be reduced by 50% or more. Much of the information
is appropriate to other documents (e.g., deposition of sulfur and nitrogen) but is only
tangentially (at best) related to PM and the standard setting process.
2. The discussion of wet and dry deposition on ecosystem processes is largely a function of
research conducted in the east where precipitation is the major mode of deposition. In
the western US, dry processes are far more important as a vector for deposition. It is
recommended that the research in the West be given some parity in the discussion
assuming that the discussion of deposition remains. In light of No. 1 (above), this issue
may be moot.
3. The discussion of direct effects of PM on vegetation (4.2.1) is appropriate to this
document but has no relevance to the standard setting process since effects are seen at
levels well above ambient rates of deposition. This section could be reduced in length by
75% or more.
4. The discussion of the consequences of nitrogen input to ecosystems (4.2.1.2) is hard to
justify in the depth presented. If it is important to include, it is recommended that the
dissimilarity between the eastern and western US be highlighted.
5. The same concern for sulfur is appropriate. The detail is only tangentially related to the
issue of PM and the deposition is unlikely to be of consequence.
6. On page 4-22, reference is made to the fact that ecosystem level responses to stress begin
at the population level. I am not quite sure that is accurate.
7. On page 4-24, the following statement is offered, "In contrast, anthropogenic stresses
usually are severe, debilitating stresses". I find it difficult to agree with this statement.
In the same paragraph, the four categories of stresses seem to be awkward. Where would
nitrogen deposition or CO2 increase fall in this scheme?
8. On page 4-25, reference is made to the concept of secondary succession and chronic
stresses. The concept of secondary succession as presented is not accurate and the syntax
of those sentences is not accurate. The entire process of secondary succession is a dated
concept in ecology and its relevance here is marginal.
9. On page 4-26, the comment is made that it is difficult to determine responses of
ecosystems to stress. As a blanket statement, this is simply not accurate. Maybe the
magnitude of the response is not known with certainty but the direction and many of the
changes are known with certainty.
10. The section on particulate matter, atmospheric turbidity and effects on vegetation
processes (page 4-34) is weak from a cause-effect perspective. This could be deleted.
11. Is the section on solar UV radiation (p4-39) needed in this document? The argument is
tenuous.
12. The conclusion paragraph (4-84) is too bold a statement regarding the effects. The lead
should be less alarmist and simply state that there is little reason to address secondary
effects of PM on vegetation and ecosystem processes. It is important to be accurate,
particularly in the summary sections.
13. On pages 4-113, the work of Chestnut and Davis is presented on the willingness to pay
for visibility. It is important that the results and conclusions of the authors be reported
rather than simply that they conducted a study.
14. If one is discussing nitrogen and sulfur in the PM document, then all of the other
A-79
-------
atmospheric stressors associated with PM should be included as well. These would
include base cations, hydrogen ions, heavy metals, pesticide residues, oxidants, etc. An
alternative would be to simply list these as part of the deposition process but not relevant
to the CD.
15. The processes discussed governing how PM affects vegetation are only a fragment of the
physics, chemistry and biology of PM. The concepts to be included should be effects of
velocity and particle size on deposition, solubilization, evapoconcentration, rainfall
events, wash off, re-suspension, transcuticular migration, etc.
16. Deposition to surface waters is entirely missing in the CD and yet this is a major issue for
understanding estuaries and lakes. There is a host of data for this topic for major
resources and the largest set of data is for the Great Lakes. Its omission in light of what
is included (e.g., nitrogen and sulfur at IPS sites) is a problem
Ronald H. White, M.S.T.
Chapter 6: Epidemiology of Human Health Effects from Ambient Particulate Matter -
General Comments
Overall, this chapter presents a comprehensive review of the extensive body of
epidemiological studies published since completion of the 1996 particulate matter criteria
document. The chapter properly interprets the studies discussed and appropriately emphasizes
the strengths and weaknesses of the current scientific evidence of the health effects of particulate
matter.
One key issue that requires further attention is the need for a consistent approach with
explicit criteria throughout the chapter for the selection of the analyses from the studies included
for summarization in the tables. For example, there are several criteria described (pg. 184; lines
8 -17) as providing the basis for selection of the analyses summarized in Table 6-19 and 6-20.
However other summary tables do not explicitly provide the criteria for the selection of analyses
summarized in the tables. Providing these criteria make the approaches used in selecting the
analyses included for summarization in these tables and avoid concerns regarding author bias in
the selection of analyses included for summarization.
The discussion of the infant mortality/related morbidity studies that have been published
since 1996 should be expanded. These data are important new findings that significantly
augment the more limited data available in the 1996 CD. A table summarizing these studies
should also be included in the chapter.
The discussion of lung cancer associated with PM exposure in the long-term prospective
studies should be expanded and receive additional attention in the text. Given the finding of a
statistically significant association of PM and lung cancer in the recent expanded ACS study
analysis by Pope, which I would presume will be included in the next revision of the CD, this
health endpoint deserves substantial further elaboration and emphasis. In addition, the entire
diesel particulate health effects literature regarding lung cancer is not referred to in this
discussion. Recognizing that the EPA Diesel Particulate Health Assessment document reviews
this literature in detail, the relevant science should be summarized in this chapter's discussion of
the lung cancer issue and the reader referred to the Diesel Health Assessment document for a
more complete discussion of this scientific literature.
Specific Comments
Pg. 6-226: This discussion regarding alternative methodological approaches to addressing
confounding omits reference to the selection of study areas where potentially confounding air
pollutant levels are relatively low (e.g. Vedal's 1998 study of asthmatic and nonasthmatic
children in Port Albeni, B.C.).
Appendix 6A and 6B: There is no explanation in Chapter 6 as to the rationale for the inclusion of
A-80
-------
these appendices. While the recent studies regarding the relationship of heart rate variability to
PM exposure provides one possible biological mechanism for the cardiac effects that may cause
morbidity and ultimately premature mortality, other potential mechanisms for cardiovascular
effects have also been identified (e.g. plasma viscosity, coagulation). The NMAPS data in
Appendix 6B should be integrated into the body of Chapter 6, with the daily deaths expressed as
an age adjusted rate as well as number of deaths.
Pg. 6-138: The use of the term "recent" in reference to the 1997 study by Peters et. al. is
inappropriate in a document that will be released in 2002. The use of this adjective with respect
to studies in this entire chapter should be reviewed to ensure that only studies published in the
last year or so are referred to as "recent", or alternatively the adjective should be eliminated from
the chapter's discussion of studies.
Chapter 9: Integrated Synthesis: Particluate Matter Atmospheric Science, Air Quality,
Human Exposure, Dosimetry, and Health Risks - General Comments
While this chapter is somewhat improved compared to the previous draft in terms of
writing style and providing some integration of information from different scientific disciplines,
the underlying flawed approach of providing sequential summaries of what has already been
summarized in previous chapters is retained. As such, this crucial chapter still does not provide
the reader with a true integration of the key information identified in the previous chapters as
being of major significance for the air quality standard-setting process.
In my December 1999 comments on the previous draft of this chapter, I had suggested an
approach that would structure the information provided in this chapter as responses to several
key questions regarding the health science information published since the previous Criteria
Document. In his written comments on this current chapter, Dr. David Bates has also suggested a
somewhat similar approach to structuring this chapter. As it currently is written, there is a
significant amount of repetition of information already provided and summarized in the previous
chapters. Key new information regarding PM exposure, toxicology, clinical studies and
epidemiology are not currently integrated in a manner that informs the standard-setting process.
Specific Comments
Pg. 9-65; lines 2-5: The data audit performed for the HEI Reanalysis Project was not conducted
by the study investigators as currently indicated in the text. The data audit was performed by an
independent team selected by HEI to perform this function for the study.
Warren White, PhD
4.3 Effects on Visibility - First impressions
The visibility portions of the March 2001 draft CD were prematurely circulated for external
review. Their inferiority relative to other parts of the document underscores the Agency's long-
standing disdain for this subject. I can think of no harsher criticism of the material than simply
reproducing a few of the highlights. Keep in mind that all come from fewer than two dozen
pages!
Some of the lines could have been written by Edward Lear:
"Light absorption by aggravated carbon at visible wavelengths is enhanced by no more than 30%
and diminishes if encapsulated by a nonabsorbing aerosol." (P4-90, L 19)
A-81
-------
"At the surface, a variable fraction of the solar radiation is reflected back upwards, referred to as
surface reflectance or the albedo, illuminating the atmosphere from above and below." (P 4-88,
L4)
"The increase was largest in the summer and decreased in the winter." (P 4-108, L 28)
"Some of the visibility impairment in northern California and Nevada, including Oregon,
southern Idaho and western Wyoming,..." (P 4-109, L 16)
"Horvath (1993) reported that measured light absorption efficiencies for light absorbing carbon
ranges from 3.8 to 17 m2/g. According to Horvath (1993), calculated absorption efficiencies are
too high, ranging from 8 to 12 m2/g for monodispersed carbon particles." (P 4-90, L 12)
"For most rural eastern sites, sulfates accounts for >60% of the annual average light extinction
on the best days .." (P4-108, L 23)
"However, several sites are not showing steady improvements in either visibility or PM25,
particularly in the number of worst visibility days (90th percentile)." (P 4-111, L 20) [In other
words, the number of days in a year is holding steady at about 365 per.]
There are tautologies and circular definitions of the sort associated with Lewis Carroll:
"Human vision is one of the factors that affects the way an object is viewed." (P 4-86, L 10)
"Discoloration may be used as a quantitative measurement of atmospheric color changes in
urban hazes." (P 4-98, L 2) [In much the same way as morbidity can be used as an indicator of
impaired health.]
"The light-extinction coefficient is the quantitative measure of haziness, defined as aext =
K/visual range, where K is the Koschmieder constant. The value of K is determined both by the
threshold sensitivity of the human eye and the initial contrast of the visible object against the
horizon sky. The visual range may be calculated from the light-extinction coefficient using the
Koschmieder equation .." (P 4-94, L 23)
There is simple technical ignorance:
"The cones, a receptor cell in the retina, govern visibility interpretations." (P 4-86, L12) [This
is why an eyeball can be offended by haze even after surgical removal from the head. And why
we see nothing after sundown.]
"Some of the light in the sight path is absorbed or scattered towards the observer. The remaining
light is absorbed or scattered in other directions." (P4-86, L 24) [Leaving the observer
searching in vain for any transmitted image.]
"The scattering and absorption efficiencies are determined by estimating the size distribution of
each particle." (P 4-89, L 20)
".. the extinction coefficient that is calculated from the visual range, corrected to 60% relative
humidity by the Koschmeider relationship." (P 4-109, L 29) [Versatile guy, that K.]
"Me scattering is the scattering of all visible wavelengths equally (Shodor Education
Foundation, Inc., 1996)." (P 4-87, L 1) [Which must be why Mie theory is computationally so
trivial. Distressingly, this claim is supported by the citation, which turns out to be on-line
training material developed for the Agency. The cited page also explains "how the shorter
wavelengths which our eyes detect as blue when mixed, are scattered at a right angle. If the sun
A-82
-------
is directly overhead, the sun and sky look almost white while the sky is blue off to the
sides in the direction of the scattered light." The student might wish to step outside some clear
day and check whether the horizon is indeed blue and the sky white.]
"The output of the Mie calculations includes efficiency factors for extinction, Qext, scattering,
Qscat, and absorption, Qabs. The Qext, Qscat, and Qabs give the fraction of the incident radiation
falling on a circle with the same diameter as the particle that is either scattered or absorbed. The
light scattering or absorption efficiency factor (in units of m2/g) is the change in the light
scattering or absorption efficiencies per unit change in mass of the fine particle constituent. ...
Multiplying the values of the light scattering efficiency factor by the aerosol volume
concentration (in units of |j,m3/cm3) gives the value of the light-scattering coefficient, asp, (in
units of Mm"1) for these particles." (P 4-89, L 15-26) [Students: find 3 different concepts of
'efficiency factor' in this paragraph. For extra credit, find 4 or more.]
".. over a 3 0-year period (1940 to 1990)" (P4-111,L3)
There are misstatements of the Agency's own key regulatory concepts:
"Visibility impairment is defined as any humanly perceptible change in visibility (light
extinction, visual range, contrast, or coloration)." (P 4-85, L 3) [The hypothetical observer in a
pure Rayleigh atmosphere thus experiences impaired visibility during each sunset and sunrise.
Will the Sierra Club have to sue before the Agency addresses the long-standing and pervasive
problem of twice-daily twilight?]
"dv = 10 log 10 (
-------
that point from this review! "Visibility impairment may be connected to air pollutant
properties... Human vision is one of the factors ... the appearance of a distant object is
determined by illumination of the sight path ... Visibility within a sight path longer than
approximately 100 km .. is affected by changes in the properties of the atmosphere over
the length of the sight path."
4.3.3 Optical properties of particles: Of the 23 different papers cited in this subsection, 17
were published by 1994 and 13 were reviewed in the 1996 CD. The technical discussion
is very confused, and diverse extinction efficiencies are jumbled together with no
context.
The Staff Paper includes a cross-plot (Figure 5-2) of ASOS airport visibility data versus
24-h PM2 5 concentrations at Fresno, CA. This is exactly the sort of analysis that is
needed to support a PM2 5 standard for visibility and is missing from the CD. But it is
only the first step: is the rest of the country just like Fresno? The CD instead gives us
indigestible factoids: "Richards et al. (1991) reported a scattering efficiency for fine
particles of ammonium sulfate of 1.2 nf/g .. Sulfate scattering efficiencies have been
reported to increase by a factor of two when the size distribution went from 0.15 to 0.5
|j,m .. The calculated scattering efficiencies for sulfates were 4.1 nf/g for 100% mass
removal and 3.4 and 5.6 nf/g for 25% mass removal. Calculated scattering efficiencies
for carbon particles ranged from 0.9 to 8.1 nf/g .."
4.3.4 Effect of relative humidity: This section cites a higher proportion of recent work and is
better written.
4.3.5 Measures of visibility: Of the 24 different papers cited in this subsection, 17 were
published by 1994 and 13 were reviewed in the 1996 CD. I don't see any new
information.
And including "fine paniculate matter concentrations" as a "measure of visibility" is
rather begging the whole question, is it not? The figure (4-22) supporting this subsection
simply assumes a relationship for which the previous subsections laid no theoretical or
empirical basis. (Note that the assumed Koschmieder coefficient in this figure differs
from that used in the next (4-23).)
4.3.6 Visibility monitoring methods and networks: The new ASOS and expanded
IMPROVE networks are appropriate topics for inclusion in this CD. The extinction
budgets in Table 4-7 are problematic, however, because the text has given no theoretical
or empirical basis for constructing and understanding them. It would better support a
visibility-based secondary standard to summarize the measured extinction/PM2 5 ratios
and regression relationships observed at those sites with optical data.
4.3.7 Visibility modeling: Modeling can't be credible until the science is, so I didn't bother
with this subsection.
4.3.8 Trends in visibility impairment: Much of this subsection (P 4-109, L 4-26) concerns
extinction budgeting rather than trends in space and time. As noted above at subsection
4.3.6, the text has laid no basis for such apportionment. Moreover, some of the
characterizations are a bit suspect - for example, the statement "In several areas of the
west, sulfates account for over 50% of the annual average aerosol extinction" is not
supported by Table 4-7.
The trend discussion is largely carried over from the 1996 CD; Figure 4-23 is an update
of Figure 6-112 by only three years and Figure 4-24 is a reprint of Figure 6-113.
Considering that this is supposed to be an incremental update of the 1996 CD, and that
A-84
-------
the data in Figure 4-24 end in 1992, it is hard to justify open-ended statements like "The
haziness over the Gulf states increased between 1960 and 1970 and remained virtually
unchanged since then."
4.3.9 Economics of PM visibility effects: Here, finally, is a subsection that does not just
rehash and garble the corresponding 1996 account. Unfortunately, the new account
seems inconsistent with the old, and the disagreement is nowhere acknowledged.
According to the 2001 review (P 4-114, L 2), "The results indicate a willingness to pay
per deciview improvement in visibility [in class I areas, capturing both use and nonuse
recreational values] of between $5 and $17 per household." According to the 1996
review (Table 8-6), the willingness to pay per deciview improvement in urban visibility
ranged from $8 to $231 per household (in older, more valuable dollars), with a median of
about $100. If visibility is really worth that much more in cities than in National Parks,
then why are almost all our visibility monitors in Parks? I couldn't find the $5 - $17
values in the cited reference, so I suspect that this is yet another instance of garbled
reporting.
The bottom line for section 4.3 is that no coherent attempt is made to connect visibility with the
health-based PM indicator.
A curious omission
The single most important visibility development since the 1996 CD has been the arrival of
Regional Haze Rules. These Rules establish a framework for regulating visibility that any
secondary PM standard will have to coexist with. Whereas any secondary standard will require
scientific review by CASAC, the Regional Haze Rules already in effect were developed largely
from an administrative/bookkeeping perspective. How does the Regional Haze bookkeeping
square with the science reviewed by the CD? This is a question the draft studiously ignores.
George T. Wolff, PhD
Chapter 1
1. p 1-8, lines 4 - 5 - Is this something new? CASAC has not had an opportunity to comment
collectively on the proposals in the past.
2. p 1-14, lines 1 - 2 - Does this mean that higher concentration studies that show no effect
were ignored?
Chapter 2
General - The chapter needs a glossary.
1. p 2-15, lines 2-6 - This appears to be worded too strongly given the conclusions reached in
chapter 6 (see page 6-266, lines 29-30).
2. p 2-18, line 23 - The photolysis of O3 is the major source of OH only in relatively pristine
atmospheres. The major source in urban atmospheres is likely organic gases.
3. P 2-19, lines 1-5 - This is also too strongly worded for the same reasons as 1.
4. P 2-20, lines 1-3 - While this statement is true for sulfates, it is not for nitrates. Because of
thermal decomposition at high ambient temperatures, nitrates particles tend to be higher in
the winter.
5. P 2-33, line 16-1 would remove the word "significantly" since droplet acidity is dominated
by in cloud formation and acid gas scavenging. Same comment for p 2-101, line 15.
A-85
-------
6. P 2-86, section 2.2.5.1 - A short description of the TEOM is needed.
7. Section 2.2.5 - Except for the TEOM, it is not clear if all of the devices mentioned in this
section are Class in FRMs.
8. Section 2.2.6 - "Data quality" section - Can anything be said about the magnitude of the
measurement error that would shed some light on the exposure errors associated with the
various epidemiology studies? It is mentioned that the coarse numbers are "inherently less
precise." Can that be quantified and put in perspective given that the epi studies have a
tendency to attribute a lower risk associated with the coarse mass relative to the fine and
PM10 mass?
9. p 2-104 lines 3-4 - The is something wrong with the sentence that begins "Fresh, submicron-
size...."
Chapter 3
1. p 3-10, lines 4-8 - This should be expanded to include more quantitative information on the
trends of specific constituents.
2. p 3-12, line 10 - Define FRM.
3. P 3-13, figure 3-7a - This figure needs more reference ticks on the y-axis and a better legend
explaining the meanings of the various symbols. What is the center bar and what is +?
4. P 3-16, figure 3-8 - Something is needed to distinguish between the PM2.5 and PM10 bars.
5. P 3-35, table 3-7, organic carbon row, anthropogenic column - delete "emitted by motor
vehicles" since there are other man-made sources of hydrocarbons.
Chapter 6
1. General Comment - When discussing the inclusion of gaseous pollutants in any analysis, it
is insufficient to merely say pollutant x was included without specifying which measure of
the pollutant was used. This is particularly important for O3. In time series studies, the 1-hr
or 8hr max are the appropriate measures to use not the 24-hr average which will introduce
unnecessary measurement error into the analysis and mask the true effect. In the cross-
sectional studies, the mean of the 1-hr daily max is the appropriate measure, not the annual
mean. The measure should be clearly indicated for each study, so the reader can make
judgements about the validity of the results. The same comments apply to meteorological
measurements.
2. General Comments - Some consistent rules need to be established about identifying the level
of statistical significance of results and their inclusion in subsequent discussions. As it
stands now, it appears that results are included regardless of significance level if they
support a desired conclusion.
3. P 6-1, lines 8-11 - Change "measurable excesses" to "statistical associations between," and
"being associated with" to "and."
4. Table 6-1 - The specific measure of the gaseous co-pollutants and meteorological variables
should be included in this and other summary tables.
5. Comments on the measures used in NMMAPS and the HEI Reanalysis Study - Since these 2
studies are highlighted in the CD to illustrate a number of points including the small or
nonexistent effect of ozone on the PM signal, a few comments on these studies are in order.
NMMAPS used the 24-hour average concentrations for gaseous pollutants including O3 and
CO. This averaging time, while consistent with the averaging time of PM, whose relevant
A-86
-------
health-response time at ambient levels remains unknown, has two major problems. First, it
is inconsistent with the known health-response times of both O3 and CO. As a result, the
impacts of O3 and CO on the regression are not properly characterized. Second, the 24-hour
averages of O3 and CO do not correlate perfectly with the peak 1-hour or 8-hour values that
are the measures associated with their health effect endpoints. Worse yet, urban sites tend to
have higher peak O3 than rural sites, but rural sites tend to have higher 24-hour O3
concentrations than urban sites. For the spatial analysis, NMMAPS used annual means
when it was appropriate to use the means of the 1-hour or 8-hour daily maxima. Cleaner
areas and rural areas generally have higher annual average ozone values than more polluted
urban areas. Consequently, the results involving O3 and CO are likely not meaningful.
6. The Reanalysis Team also used the same annual average O3 for the H6CS reanalysis.
Although the report states that they used the annual mean of the daily 1-hour maxima for the
ACS reanalysis, they used the annual average as well because the ozone values presented in
appendix G of the report are much too low to be the average of the 1-hour maxima. Another
problem with computing a valid annual mean for ozone is that many locations only measure
ozone during the ozone season, which has a different definition depending upon the local
climatology. In Michigan, which is typical of northern states, ozone is only measured from April
1 to September 31, whereas in Southern California, it is measured year round. Consequently, I
have the same similar concerns for the Reanalysis multi-pollutant results as I have for the
NMMAPS results.
Chapter 9
1. P 9-2, line 30 - Delete "in general."
2. Table 9-2, organic carbon row, anthropogenic column - delete "emitted by motor vehicles"
since there are other man-made sources of hydrocarbons.
A-87
------- |