vvEPA
820-R-10-014
COMMENT-RESPONSE SUMMARY REPORT
for the
PEER REVIEW
of the
Fluoride:
Exposure and Relative Source Contribution Analysis
DOCUMENT
November 2010
Office of Water
Office of Science and Technology
Health and Ecological Criteria Division
U.S. Environmental Protection Agency
Washington, D.C. 20004
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
TABLE OF CONTENTS
ACKNOWLEDGMENTS ii
I. INTRODUCTION 1
II. CHARGE TO PEER REVIEWERS 2
III. PEER REVIEW COMMENTS AND EPA RESPONSES 3
IV. EXTERNAL PEER REVIEW REPORT 25
November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
ACKNOWLEDGMENTS
This document was prepared with the technical assistance of the Toxicology and Hazard
Assessment Group of the Environmental Sciences Division of Oak Ridge National Laboratory,
Oak Ridge, Tennessee, under EPA IA No. DW-89-92209701 andDOEIAGNo. 1824-S881-A1.
Principal EPA Scientists are Joyce Morrissey Donohue, PhD, and Tina Duke, MPH, Health and
Ecological Criteria Division, Office of Science and Technology, Office of Water, U.S.
Environmental Protection Agency, Washington, DC.
The Oak Ridge National Laboratory is managed and operated by UT-Battelle, LLC., for the U.S.
Department of Energy under Contract No. DE-AC05-OOOR22725.
11 November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
I. INTRODUCTION
The United States Environmental Protection Agency (EPA), Office of Water is charged with
protecting public health and the environment from adverse exposure to chemicals and microbials
in water media, such as ambient and drinking waters, waste water/sewage sludge and sediments.
In support of this mission, the Office of Water/Office of Science and Technology (OST)
develops health standards, health criteria, health advisories, and technical guidance documents
for water and water-related media. Under this work assignment, documents prepared by OST are
to undergo external peer review.
Peer review is an important component of the scientific process. It provides a focused, objective
evaluation of a research proposal, publication, risk assessment, health advisory, guidance or
other document submitted for review. The criticisms, suggestions and new ideas provided by the
peer reviewers ensure objectivity, stimulate creative thought, strengthen the reviewed document
and confer scientific credibility on the product. Comprehensive, objective peer review leads to
good science and product acceptance within the scientific community.
The Peer Review for "Fluoride: Exposure and Relative Source Contribution Analysis" was
conducted on May 14, 2010, in Washington, DC, to allow the external peer reviewers to discuss
their evaluations of the EPA/OW document. The Peer Review was conducted under EPA
Contract Number EP-C-07-059 (Work Assignment 2-02) to Eastern Research Group, Inc., 110
Hartwell Avenue, Lexington, MA.
The list of external peer reviewers and their affiliations are shown below:
Linda C. Abbott, Ph.D., Regulatory Risk Analyst, Office of Risk Assessment and Cost-Benefit
Analysis, U.S. Department of Agriculture
Mary A. Fox, Ph.D., Assistant Professor, Department of Health Policy and Management,
Johns Hopkins Bloomberg School of Public Health.
E. Angeles Martinez Mier, DDS, MSD, Ph.D., Associate Professor, Department of Preventive
and Community Dentistry, Indiana University School of Dentistry
David L. Ozsvath, Ph.D., Professor of Geology and Water Science, Department of
Geography/Geology, University of Wisconsin-Stevens Point
November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
II. CHARGE TO THE PEER REVIEWERS
1. Please describe any suggestions you have for improving the clarity, organization,
and/or transparency of the draft document.
2. Have the uncertainties associated with the analysis been adequately characterized?
Are there any important uncertainties in the data that are not discussed adequately in
the document, especially in the synthesis sections? Please describe any concerns you
have and any specific suggestions for improving or enhancing the uncertainty
discussion.
3. Please consider the studies that have been selected as representative of exposures for
the specific age groups and/or exposure media. Have these studies been adequately
summarized and interpreted? Indicate any deficiencies in the descriptions of the
studies and any suggestions you have for improvement. Describe any concerns you
have about the selection of these studies, as well as any recommendations you may
have for alternative studies that you believe are more representative of exposures.
4. Please comment on EPA's rationale for selection of specific data elements to
represent average exposures for each of the age groups. Has the selection been
scientifically justified and clearly and objectively described? What changes or
improvements would you suggest?
5. Please comment on the validity of basing the food intake estimate for the 1940's on
the McClure (1943) publication, as supported by the concentrations found in various
food groups from more recent analytical data. Do you agree with this approach? If
not, what approach would you suggest for estimating food intake for the 1940's?
6. Provide citations (and, where possible, pdfs or hard copies) for any references you
suggest EPA should consider adding to the document, and describe where you
suggest these references be added.
7. Please provide any additional comments and/or further suggestions you may have for
improving the document.
November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
III. PEER REVIEW COMMENTS AND EPA RESPONSES
Charge Question 1. Please describe any suggestions you have for improving the clarity,
organization, and/or transparency of the draft document.
Reviewers noted agreement in their written comments that the purpose of the draft RSC
Analysis was not clearly stated and that adding an overview or "road map," including a statement
of objectives, in the introductory section of the document would help address this concern.
EPA Response: Dr. Donohue noted that the Agency agreed with these pre-meeting comments,
and had decided to add forwards to both the RSC Analysis and the Dose-Response Report in
order to outline the study objectives and present a road map. She said that EPA planned to post
both documents on the Internet at the same time. Dr. Donohue also noted that the external peer
reviewers, as well as EPA internal reviewers, had clearly indicated confusion around
presentation of how the Agency used the McClure (1943)l data to estimate food contributions in
1942. She proposed that EPA would remove the description of the McClure (1943) data from the
draft RSC Analysis and add it as an appendix to the Dose-Response Report. This seemed
appropriate since those data are used in the Dose-Response Report rather than the RSC Analysis.
She asked reviewers for feedback on whether they thought this change would address their
concern about this section of the report.
Reviewers supported the idea of a forward and agreed to talk about the best location for the
McClure (1943) data when they discussed Charge Question 5.
Dr. Ozsvath remarked that it was difficult to determine what the critical steps of the analysis
were and why they were taken. He had alluded to this concern in his written response to
Question 1 and elaborated on it in his response to Question 7 (see Appendix F). He noted that
this concern was also echoed in Dr. Abbott's detailed written comments. He recommended that
the document include a road map, a list of objectives, and a statement at the end of the document
clarifying how the objectives were met.
EPA Response: EPA has prepared a preface to the exposure and RSC document which states the
objectives of the effort and the over arching factors that governed the EPA analytical approach.
The requested map of the assessment is included in the preface.
The last chapter provides a statement on how the objectives of the document have been met.
Dr. Martinez Mier noted that some of the terminology was not consistently used throughout the
document. Also, some of the steps in the fluoride analysis are not consistent with those found in
the current dental literature for fluoride analysis. Though the document is not intended for the
1 McClure, FJ. 1943. Ingestion of fluoride and dental caries. Quantitative relations based on food and water
requirements of children 1-12 years old. Amer. J. Dis. Child. 66:362. [Republished in Publication 825, U.S. Public
Health Service, 1962].
3 November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
dental community, it will inevitably be circulated within that community because it impacts
dental fluorosis and caries prevention. Dr. Martinez Mier referred EPA to her written response to
Charge Question 1 (see Appendix F) for specific suggestions about reorganizing the discussion
of analytical methods and changing how things are referenced in order to bring the discussion
more in line with contemporary analytical approaches used by the dental community.
EPA Response: The information on analytical methods was prepared by an analytical chemist.
It was included to provide background on the methods that were used by the researchers who
conducted the published studies included in the report. This limitation is now included in the
preface to the report. EPA acknowledges that there have been improvements in the methods
used to measure fluoride in different media, however the methods used more recently are, for the
most part, not those that were used for the analyses cited in this report.
Dr. Fox said she agreed with recommendations to improve clarity and transparency by stating
the questions the document will address and the approach EPA used to address those questions.
She also had concerns about the discussion of public health implications in Chapter 8. She had
provided a detailed comment about this in her written response to Charge Question 7 and noted
that she would talk about this in detail when Question 7 was discussed.
EPA Response: The Preface mentioned in the response to the first comment lays out the
objectives for the exposure and RSC document and the factors governing the approach that EPA
used in its development. The absence of discussion of public health implications in Chapter 8
was a deliberate omission because the public health implications extend beyond the purview of
the EPA and impact fluoridation guidelines (Centers for Disease Control) as well as the role of
the Food and Drug Administration with its oversight of toothpaste, bottled water, and food
labeling. EPA has since entered into discussions with Health and Human Services about the
findings of the exposure and RSC analysis. The outgrowth of those discussions may promote
inclusion of more information on public health implications of the EPA findings. EPA did
expand the original closing paragraphs of Chapter 8 to provide statements on the major findings
of the document relative to risk.
Dr. Abbott referenced all of her written comments in response to Charge Question 1 (see pages
F-6 to F-8 of Appendix F). She also noted one additional comment. She recommended that, at
the beginning of the document, EPA acknowledge the spectrum of populations obtaining
drinking water from the full variety of sources, including community water supplies, spring
water, bottled water, and well water, and clarify whether these sources factor into the MCL. The
document seems to combine these at different levels in the RSC calculation because food intake
is considered across the population as a whole, not just in those who drink community water.
This is a particular problem when considering beverages. For example, beverages prepared by
adding water and made away from home, like tea or beverages made from dry powders with
extra indirect water, may not come from the same water supply used at home. In the Continuing
Survey of Food Intakes by Individuals (CSFII), all these different types of water sources are
associated with the different types of food and beverages. The document should acknowledge
from the outset that the analysis combines these different populations for different parts of the
estimate. Without this acknowledgment, readers will be confused when they learn that the
document only considers populations served by community water systems and not those who are
November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
not served by those systems but who are exposed to fluoride in their diet by ingesting food or
beverages made with fluoridated water (e.g., at a processing plant).
EPA Response: In response to this comment, Dr. Donohue clarified that EPA used data from the
Food and Drug Administration's (FDA's) total diet study, which clearly identified in the market
basket which beverages were made from public water and which at home. In the analysis, EPA
subtracted out the portion of beverages made with public water.
Dr. Abbott responded that the CSFII would provide a more refined estimate of the source of
indirect water used in beverage preparation.
EPA Response: Dr. Donohue clarified that EPA relied on the published literature for the RSC
Analysis and did not conduct any independent analyses.
In light of that clarification, a reviewer suggested that EPA state at the beginning of the
document that the Agency had not conducted any new analyses and instead had relied
exclusively on the literature. Other reviewers agreed it was important for EPA to more clearly
explain what they did and did not do in the analysis.
EPA Response: As mentioned earlier, EPA added a preface to the document which explicitly
states the bounding conditions for the EPA analysis. The Introduction lays out the Agency
policies that govern the RSC analysis. Some revisions have been made to that section in order to
increase its transparency.
EPA's age-specific drinking water intake data consider both direct and indirect drinking water
intake as reported in the most recent Continuing Survey of Food Intake by Individuals (1994-
1998). When EPA selected the studies that represented the exposures for each age group of
concern, studies where home-prepared beverages were reconstituted with deionized-distilled
water were used if they were available, and plain tap water intake was not included in the
beverage category. If the only available data came from market basket studies that did not
require home-prepared beverages to be reconstituted with deionized-distilled water, the locality
with the lowest level of fluoride was selected to represent the age group and the fluoride
concentration of the local drinking water was provided in the data summary Table. The studies
selected did not include plain drinking water in the beverage category.
EPA obtained information on the home-prepared beverages in the present and prior FDA market
baskets and presents that information in the report, thereby providing perspective on the
uncertainties that might result from the use of market baskets that could have used tap water for
beverage preparation. The uncertainties in the beverage data from the key studies are discussed
in the text.
November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
Charge Question 2. Have the uncertainties associated with the analysis been adequately
characterized? Are there any important uncertainties in the data that are not discussed
adequately in the document, especially in the synthesis sections? Please describe any
concerns you have and any specific suggestions for improving or enhancing the
uncertainty discussion.
Reviewers agreed that the document should provide a better discussion of uncertainty.
Individual comments included:
• The draft RSC Analysis describes sources of uncertainty, but does not attempt to show
how the RSC would be affected if different assumptions were made. A sensitivity
analysis would help show how different assumptions would affect the RSC. Page 113 of
the draft document states that the Agency believes these are reasonable estimates, but the
Agency does not justify this statement; this part needs more scientific rigor.
• The draft RSC Analysis recognizes some sources of uncertainty (e.g., "technological
limitations"), but should be more explicit in discussing the uncertainties associated with
specific values from various studies related to fluoride analysis. The fact that uncertainty
in fluoride analysis can go either way (leading to much lower or much higher values)
should be discussed. Additionally, there is evidence (e.g., policy statements by the
National Institutes of Health and some regulatory bodies) that the methodology has a
strong influence on results. The current document "tiptoes" around this reality; EPA
needs to provide stronger statements about this, for example by stating that the
methodology "does" as opposed to "may" have influence. The lack of strong
acknowledgment of the limitations of fluoride analysis carries over into the McClure
discussion later in the document, where these limitations are important. Finally, multiple
studies recognize the large individual variation in the sources for water intake; this
uncertainty is not well addressed.
• Dr. Fox referenced her written comments in response to this charge question (see page F-
19 of Appendix F), in which she listed a series of questions, which she had made from
the perspective of a risk assessor: "Does the available data overestimate, underestimate or
both? Can the sources of uncertainty be prioritized? Are the different analytical methods
contributing to more uncertainty than changes in diet? How does an understanding of the
largest sources of uncertainty help us understand the data or help us characterize the
analyses?" She felt that the uncertainty discussion would be much clearer if the document
would address these questions. Agreeing with the prior comment on the role of
methodological uncertainty, she recommended that the document more adequately
characterize the methodological difficulties, and include any definitive quantitative
information on uncertainty, whether leading to under- or overestimates.
• Dr. Abbott referenced her written comments in response to this charge question (see pages
F-8 to F-9 of Appendix F). She emphasized that quantitative assessment of how
influential the food intake studies were in calculating the RSC was both possible and
important. She also questioned whether combining the 90* percentile value for drinking
6 November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
water intake with all the other values would yield a valid estimate of total drinking water
intake.
EPA Response: Dr. Donohue clarified that use of the 90th percentile value for drinking water
intake and use of the average body weight were Agency policy. In response to a reviewer
question, she agreed that a sensitivity analysis would improve understanding of the range of
uncertainty around the data points selected for the analysis.
The reviewer responded that choosing one number was not erring on the side of being
conservative; a sensitivity analysis would illuminate how the selected number compared with
other possibilities that might be more protective of public health.
EPA Response: The EPA RSC analysis is governed by EPA policy rather than the
methodologies employed by the researchers that provided the data. The experimental
methodologies described in the text are those provided in the published studies that were used by
EPA. The EPA policy limitations are acknowledged in the introduction and are now included in
the preface as well. EPA has increased the discussion of uncertainty in response to the peer
reviewer's suggestions. The methodological variables inherent in the different critical studies
are not included in the added text. However, the added text does demonstrate the impact of EPA
study selection criteria on the outcome of the analysis.
As suggested by the peer reviewers, EPA has evaluated the impact of choosing a different data
point as representative of specific media on the output from the analysis. It has also included
information on the change that would result if there were an allowance for tooth brushing twice a
day. There already was discussion of the confidence bounds around many of the averages
selected as representative of the solid foods, beverages, and toothpaste intakes for the age groups
in the peer review draft. The fact that average values were used in the analysis is a matter of
EPA policy as are the use of the average drinking water F concentration and the 90* percentile
drinking water intake. These limitations are now mentioned in several locations in the report.
The analysis of the dietary contribution of fluoride to exposure at the time of the Dean (1942)
study (the basis of the RfD) was a confusing feature of the peer review draft of the exposure and
RSC analysis. That analysis and the data that support it have been removed from the document
and are now included as an appendix to the noncancer dose-response assessment. That analysis
provides the dietary contribution for the RfD derivation based on the Dean (1942) drinking water
concentration data. The peer reviewers were fully supportive of this change.
November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
Charge Question 3. Please consider the studies that have been selected as representative
of exposures for the specific age groups and/or exposure media. Have these studies been
adequately summarized and interpreted? Indicate any deficiencies in the descriptions of
the studies and any suggestions you have for improvement. Describe any concerns you
have about the selection of these studies, as well as any recommendations you may have
for alternative studies that you believe are more representative of exposures.
Reviewers held differing opinions about the adequacy of study descriptions and whether the
selected studies were representative of exposures.
• Dr. Abbott referred EPA to the written comments she had provided in response to this
question (see pages F-10 to F-l 1 of Appendix F) and asked that the Agency consider all
these comments. She noted in particular that there appeared to be a mistake in the "0.5 to
> 1 year" row of Table 6-1, since it seemed that the value for the addition of powdered
formula should have been higher than stated. A complete explanation of this concern is
provided in her written comments in response to Charge Question 7.2
• A reviewer recommended that the document provide more discussion about the quality of
the literature on the various exposure assessment methodologies, including whether any
methodologies are better suited than others for the analysis and how this factored into
choosing the key studies.
• Dr. Martinez Mier said she found the selected studies to be representative and thought
they would be recognized as valuable by others with dental expertise. She suggested that
dental professionals with experience in this area would probably understand the studies,
as described in the document, but she was concerned whether readers outside this area of
expertise would be able to understand them. She felt that some studies that reflect more
contemporary analytical methodologies should be included, and referred EPA to her
written comments for more details on this comment and for suggested references (see
pages F-26 to F-27 of Appendix F).
• Speaking from a soil and groundwater perspective, Dr. Ozsvath clarified that there is no
database for fluoride in soils comparable to that used to analyze fluoride concentrations
in public drinking water supplies. There are studies supporting a range, but there is no
way to characterize those data statistically. He noted that a 2009 document prepared for
2 This response reads as follows: "Table 6-1 states the overall mean from Ophaug (1985) (0.17 mg/d) was adjusted
by subtracting the milk/formula intake form the earlier Ophaug study and then adding the 0.14 mg/d estimate from
the powdered formula study by Van Winkle et al., 1995 using tap water to make the formula. This almost doubles
the original Ophaug estimate and appears to be an incorrect combination of a fluoride concentration in food (Van
Winkle et al. 1995) with an exposure estimate (Ophaug, 1985). To add the Van Winkle study to the Ophaug study,
you would have to know how much formula was consumed in the Ophaug study. If the results from Van Winkle
really are concentrations and are reported in mg/L (and not in mg/kg or some mass-mass basis) you would have to
know the amount of liters of formula consumed in Ophaug to estimate the mg F/d contributed to the diet."
8 November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
Health Canada3 used 100 ppm as the average fluoride concentration in soil to estimate
daily fluoride intake from soil ingestion, but this value might not apply to the U.S. He
suggested that EPA provide more justification for its approach, including the use of water
and soil data that do not have the same level of precision. A reviewer asked if this should
be included in a sensitivity analysis. Dr. Ozsvath responded that the drinking water
database is large enough to characterize probabilistically, so it is possible to know how
the representative number relates to the whole database; however, this is not possible
with the soil data.
EPA Response: The document discusses the strengths and weaknesses of the dietary
methodologies for estimating populations and individual exposures (Section 2.5.1). However,
the reviewer is correct in that there was no statement about which methods are best suited to the
exposure and RSC analysis. A paragraph has been added at the end of Section 2.5.1 that
provides the EPA's opinion on which of the dietary methodologies were most appropriate for
their analysis.
EPA examined the estimate for the formula fed children and did find an error which has been
corrected. The revised value is lower than the original number by 0.01 mg.
Table 6-1 of the peer review draft explained the EPA calculation for formula fed children as
follows:
• The Overall mean (0.17 mg/day) from 22 market baskets, and national food intake data
(see Table 2-24), was the starting point for the calculation. The Ophaug et al. (1985) data
apply to 6-month-old infants.
• The mean was adjusted by subtracting the milk/other dairy/formula intake of 0.06 mg/day
from (Ophaug, 1980a Table 2-23) and replacing it with 0.14 mg/day from powdered
formula (Van Winkle et al., 1995) [0.17 mg/day - 0.06 mg/day + 0.14 mg/day = 0.25
mg/day]. The Van Winkle intake from 1 L of formula was 14 mg. Children in the age
range of interest drink 0.971 L/day. Thus, the formula contributed 0.14 mg/L x 0.971 L =
0.136 mg (rounded to 0.14 mg).
• The fluoride from the water added to reconstitute the formula is included in the drinking
water compartment [0.87 mg/L x 0.971 L = 0.84477 mg (rounded to 0.84 mg)].
The major reason why the value corrected to reflect children fed from reconstituted formula is
higher than the Ophaug et al. (1985) value is because the Ophaug et al. number is based on a diet
that includes both milk and formula intakes while the EPA number is based on a diet that
replaces the milk and formula from Ophaug with formula prepared from powdered concentrate
yielding a higher daily intake of fluoride (0.14 mg/day) then the milk and formula from Ophaug
et al (1985) of 0.06 mg/day. Milk fluoride levels are low (0.02 mg/L; see Table 2-10) while the
powdered formula results in a higher contribution to the formula as served (0.14 mg/L).
3 Federal-Provincial-Territorial Committee on Drinking Water. 2009. Fluoride in Drinking Water
http://fluoridealert.org/canada.2009.report.pdf
9 November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
Dr. Marlines Mier submitted a list of studies of fluoride exposure for EPA's consideration.
These studies were evaluated by EPA; those that represented populations from outside of the
United States were not considered suitable for inclusion in the exposure and RSC report.
EPA contacted Health Canada to discuss the basis for their 100 ppm soil estimate. They were
not able to verify that the cited source was the origin for the number they used. The author of the
Health Canada document indicated that she will inform EPA when she finds a good reference for
a soil number.
10 November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
Charge Question 4. Please comment on EPA's rationale for selection of specific data
elements to represent average exposures for each of the age groups. Has the selection
been scientifically justified and clearly and objectively described? What changes or
improvements would you suggest?
Dr. Martinez Mier noted that, in their written comments, all reviewers had recommended that
EPA needed to provide further discussion on the selection of age groups. Reviewers had also
provided several comments that the water and food exposure for infants appeared to be high and
that the report needed to provide a better description of the variation in food concentrations. A
number of reviewers also commented on the limited geographic area considered in the studies
and the number of subjects involved in the studies. She briefly listed other individual written
comments (see Appendix H) and then opened the floor for discussion.
• Dr. Abbott commented that EPA's rationale for selection of the age groups selection was
not clear, which made this portion of the document very confusing, particularly since the
review document appeared to use different age groups than the underlying data. It was
also puzzling why the draft RSC Analysis utilized historical data for this analysis since it
seemed to her that calculations using more current data could have been made. She
emphasized that this part of the document was very confusing and she referred EPA to all
of her written comments in response to this question (see page F-l 1 in Appendix F).
• Dr. Fox commented that EPA should provide a better explanation of the criteria the
Agency used to identify key data and studies and the considerations used to determine
what was representative. She referred the Agency to her written comments for details (see
page F-l 9 of Appendix F).
• Dr. Martinez Mier said she understood that use of data from the Dean study (1942) was
necessary due to the lack of recent fluorosis prevalence data, but the RSC Analysis
needed to better explain why those data elements were selected. She was particularly
concerned that the dental community would question why EPA used the Dean (1942)
data rather than more recent data. She recommended that EPA better clarify its rationale
for including this study.
• A reviewer noted that while she understood the benchmark dose issue and that the Dose-
Response Report determined the choice of Dean (1942) as the best study for addressing
dose-response, she still questioned why EPA did not use current dietary exposure
information in the RSC Analysis. It was unclear whether the RSC would be based on the
Dean (1942) data or more recent data.
Noting their confusion about use of the Dean (1942) data, reviewers revisited the idea, proposed
by Dr. Donohue during the Question 1 discussion, that EPA remove the description of the
McClure (1943) data from the draft RSC Analysis and add it as an appendix to the Dose-
Response Report. They agreed this was a good idea.
EPA Response: The data that were included in the exposure and RSC report for the purpose of
estimating exposure at the time the critical study (Dean, 1942) was conducted was the source of
11 November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
much of the confusion reported by the reviewers in response to this question. After the peer
review meeting, EPA removed the early studies from the report and included them as an
Appendix to the dose-response document. This enabled EPA to also remove the dose-response
discussion in the exposure and RSC report that were linked to the application of the McClure
data and contributed to the reviewer's confusion. The reviewers accepted the EPA's proposed
approach to remedy the situation.
As a result of these changes all of the dose-response information that supports the EPA RfD is
now in the appropriate document. EPA believes that this change removes the origin of most of
the reviewers concerns expressed in their responses to the charge question.
EPA has added a statement to the preface which states the reason why the age groups were
partitioned as they were. The criteria applied when selecting the critical studies to represent each
age group were added to the introduction to Chapter 6 and the text dedicated to identifying the
preferred study was expanded so that the reasons for not selecting a different representative study
are provided in cases where it had not been included in the peer review draft. Chapter 7 provides
information on RSC values that would had resulted from the use of a different critical study as
part of the sensitivity analysis suggested by the peer reviewers.
12 November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
Charge Question 5. Please comment on the validity of basing the food intake estimate for
the 1940's on the McClure (1943) publication, as supported by the concentrations found
in various food groups from more recent analytical data. Do you agree with this
approach? If not, what approach would you suggest for estimating food intake for the
1940's?
Reviewers noted that Charge Question 5 probably should have referred to McClure (1949)4
rather than (1943). All reviewers agreed that the report should better clarify the food intake
estimates, especially in Table 2-25, where the McClure data are compared to more contemporary
data.
Dr. Donohue clarified that the McClure food data were used to determine the dose-response and
not the RSC.
Reviewers emphasized that EPA needed to add a forward in both the RSC Analysis and the
Dose-Response Document to clarify the purpose of the two documents and their relationship to
one another, since it cannot be assumed both documents will be read together.
Reviewers also recommended that discussion of adding 0.01 to get the POD be moved to the
Dose-Response Report and that EPA clarify how the 0.5 ppm value was selected from the four
values presented by McClure in the 1940s studies.
Dr. Donohue affirmed that, in response to reviewers' comments, the Agency will move
discussion of the McClure data to an appendix in the Dose-Response Report, and will respond to
reviewers' comments on that section even though it will be part of a different document.
However, Dr. Donohue noted that EPA will need to keep some of the historical data in the RSC
Analysis to show that the analytical methods have given different results (e.g., that values for
fluoride in meat and poultry decreased over time since the 1940s as the analytical methods
changed).
Additional reviewer comments included the following:
• A reviewer noted that the document did explain the importance of determining the food
contribution, but the introduction was confusing, as was the section comparing the RfD
(based on McClure data) and current RSC values.
• Another reviewer noted that, at one point in the draft RSC Analysis, the McClure data are
considered a reasonable basis for estimations, but Table 2-25 states that the McClure
results differ from current studies. The analytical limitations of the McClure study
compromise the results, and this needs to be clearly stated in RSC Analysis
4 McClure, FJ. 1949. Fluoride in foods. Public Health Reports 64:No. 34, pp 1061-1074.
13 November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
• Dr. Martinez Mier said she wanted to amend a sentence discussing fluoride
concentrations in her pre-meeting comments to add the word "or higher" so that the
sentence would read: "Interferences most likely resulted in reported fluoride
concentration in foods lower or higher than actual concentrations." She added that
several reports in the dental literature have compared food and beverage and analysis of
dental tissue where different methods have been used, and the interferences from these
methods, particularly less sensitive methods, resulted in both higher and lower fluoride
concentrations. This should be clarified the RSC Analysis.
• The approach for the food intake estimate is reasonable, but the need to estimate fluoride
from 1940s food intake data, and the selection of 0.5 ppm from McClure (1943) as the
estimate for fluoride in solid foods, is not clearly explained or supported. This
explanation is very important and belongs up front in the purpose/problem statement. The
discussion of Table 2-25 in the analysis offers possible explanations for why there are
differences between McClure and more contemporary data, but the document should
clarify how the differences inform the RSC Analysis.
• It is unclear why data from 1943 (McClure) are being used to set today's RSC. The data
seem irrelevant to the RSC Analysis. This needs to be clarified. There is a discrepancy
between Table 2-12, which reports data from McClure (1949) and Table 2-25, which
reports data from McClure (1943, 1949).
• Dr. Ozsvath referenced his written comments in response to this charge question, in
which he noted a possible contradiction (see pages F-30 and F-31 of Appendix F).
Specifically, on page 34 the authors state that the differences between McClure's (1943,
1949) data and the more recent USDA (2005)5 data shown in Table 2-25 "cast doubt on
the results of exposure assessments derived from some of the early food data." However,
in the third paragraph on page 122, the authors cite the same data (in Table 2-25) and
state that "McClure's (1943) estimate for dietary intake based on a diet where solid foods
had an average of 0.5 ppm fluoride appears to provide a reasonable basis for the
contribution of solid foods to total exposure in the 1940s." Dr. Ozsvath noted that these
statements appear to be contradictory, but that could perhaps be because the authors
intended to cite Table 2-41, not 2-25 on page 122. This issue needs clarification.
EPA Response: As indicated in some of the prior responses and in the peer review report, the
inclusion and application of the early dietary fluoride data was confusing to the peer reviewers.
EPA internal reviewers had similar difficulties and it was clear to the EPA that the revisions that
had been made based on the internal peer review comments had not resolved the problem.
EPA believes that their proposal to remove the early dietary data and place it in an appendix that
would be added to the dose-response document is a good solution to the problem. The EPA
suggestion was endorsed by the peer reviewers and has been executed.
In developing the Appendix for the dose-response document, EPA removed all dietary data
published before 1980 unless there was a specific reason to retain it. In the new Appendix the
EPA expanded the comparison between the analytical results from McClure (1943, 1949) and
5 USDA (U.S. Department of Agriculture). 2005. USDA National Fluoride Database of Selected Foods and
Beverages, Release 2. Nutrient Data Laboratory, Agricultural Research Services, U.S. Department of Agriculture.
Beltsville, MD.
14 November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
those from the USDA (2005) fluoride database to include a much greater assortment of foods
likely to have been part of the pre World War II diet.
The comparison of the early analytical results to the current values support the EPA conclusion
that, for some food groups, the colorimetric analysis utilized by McClure led to fluoride
concentration values that were high and that there was no biological reason why those food items
would have higher levels of fluoride in 1930 to 1940 than they do now.
The more current fluoride concentration values for foods that had high fluoride concentrations in
the McClure data set supported the selection of 0.5 ppm as a reasonable estimate for total dietary
fluoride in solid foods during the 1930 to 1940, time period when the children in the Dean (1942)
study were exposed.
It was mentioned by the peer reviewers that Health Canada used a different value for food in the
era of the Dean (1942) study. EPA contacted Health Canada to identify the approach used. In
one respect, the approach was similar to that used by EPA. The concentrations in foods from
current analyses (Dabeka et al., 1995) were utilized in place of those from McClure. The
Canadian approach differed from that used by the EPA because they based their value on an
assumed 1940 era diet rather than the caloric allotments and average ppm concentrations in foods
that were the basis of the McClure (1949) estimates used by EPA.
15 November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
Charge Question 6. Provide citations (and, where possible, pdfs or hard copies) for any
references you suggest EPA should consider adding to the document, and describe where
you suggest these references be added.
Dr. Donohue clarified that EPA had limited the RSC Analysis to studies from the U.S. and
Canada initially, eventually eliminating many of the Canadian studies. Reviewers made the
following comments:
• Dr. Abbott drew EPA's attention to the dietary intake references she had included in her
written comments (see page F-13 of Appendix F). She was unsure where they should be
added. She said that, because the RSC Analysis included so much historical data, she had
been unsure what type of references to provide.
• Dr. Martinez Mier referred EPA to the many references she had provided in her written
comments (see pages F-22 to F-28 of Appendix F). She strongly recommended that EPA
considering including studies from other countries and not just U.S. data when describing
methodologies. For example, studies by Jackson, Dunipace, and Levy use more recent
methodologies and not, for example, ashing, a method that has not been used in many
years.
• Dr. Ozsvath suggested two references in his written comments that he recommended be
added to the RSC Analysis (see page F-31 of Appendix F). The first reference (Beltran et
al., 2002)6 is relevant to increases in fluorosis cases, a central focus of the RSC Analysis;
this study addresses how fluorosis prevalence changed over time within populations
exposed to drinking water with different fluoride content. The second reference (Federal-
Provincial-Territorial Committee on Drinking Water, 2009)7 is significant (even though
the RSC Analysis does not consider Canadian studies), because the Canadian equivalent
found a very different RSC for drinking water. He suggested that EPA add an explanation
of this difference to the RSC Analysis.
EPA Response: In response to the peer review comments, EPA did the following:
• Added the suggested studies that applied to the U.S. population.
• Requested information from Health Canada regarding the estimate of the 1930-1940
dietary fluoride level for children.
6 Beltran, E.D., S.O. Griffin, and S. A. Lockwood. 2002. Prevalence and trends in enamel fluorosis in the United
States from the 1930s to the 1980s. Journal of American Dental Association 133(2): 157-165.
http://iada.ada.Org/cgi/content/full/133/2/157
7 Federal-Provincial-Territorial Committee on Drinking Water. 2009. Fluoride in Drinking Water
http://fluoridealert.org/canada.2009.report.pdf
16 November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
• Expanded the discussion of the analytical methodologies used by Levy and Jackson as
found in the published papers.
The Health Canada estimate for the exposure of children under age 12 during the 1930-1940
time period, when it is rounded to two decimal places, is the same value used by EPA (0.01
mg/kg/day) even though the approach differed (Table B-5 of the Canadian Report). This
number is the Canadian value that applies to the situation where there is no fluoride in the
drinking water supply.
In the Health Canada dietary study all foods and beverages were prepared with drinking water
containing 1 mg/L water. The studies selected by the EPA either used distilled water for
beverage preparation or water with considerably lower fluoride content. The Canadian dietary
estimate for young children is similar to the EPA value. In the case of adults, the value is higher
but use of water containing 1 mg/L for beverage preparation should account for some of that
difference.
17 November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
Charge Question 7. Please provide any additional comments and/or further suggestions
you may have for improving the document.
Dr. Martinez Mier noted the wide variety of written comments reviewers had provided in
response to this question. She read through the list summarizing these comments (see Appendix
H) and invited discussion. Reviewers agreed that Section 8 of the document was confusing and
that the draft RSC Analysis ends abruptly. They recommended that a concluding paragraph or
section be added summarizing the outcomes and conclusions of the RSC Analysis. Discussion
included the following comments:
• Dr. Abbott referenced her written comments in response to Charge Question 7 (see pages
F-13 to F-18 of Appendix F). She noted that many of those comments also pertained to
other charge questions, and clarified that she did not have additional comments.
• Dr. Martinez Mier commented that the dental literature has made an effort to distinguish
between the terms "intake" and "exposure" for fluoride specifically. A document
published by CDC on the safe use of fluoride provides definitions for these terms, one
being systemic, the other topical. For fluoride via the oral route, "intake" pertains to
ingestion, whereas "exposure" pertains to what is put in the mouth but not ingested.
Toothpaste is an example of "exposure," but toothpaste that is swallowed becomes
"intake." Dr. Martinez Mier recommended that EPA clarify its use of these terms.
Another reviewer agreed that use of the term "intake" in the RSC Analysis was
confusing; in particular, it was unclear whether it referred to fluoride intake or also
connected with food consumption.
• In reference to p. 79 of the draft RSC Analysis, two reviewers suggested it would be
useful to have an idea of how many children under 14 years of age are served by
community water systems and potentially impacted by fluoride in these systems. This
context should be added in one of the documents resulting from this process, perhaps the
final management decision document. It may be possible to estimate this from NHANES
data by using the survey weights to estimate how many children were exposed. Also, it
would important to look at actual beverage sources and not just assume that the liquid in
all beverages came from the reported drinking water system. The CSFII provides a great
amount of detail about beverage sources.
• Dr. Martinez Mier noted that there is some dental literature on the "halo effect," which
means that individuals who are not in a community with fluoridated drinking water can
still be exposed to fluoride by consuming products made in other communities that do
have fluoridated water. The literature does not quantify this amount, but mentions that it
plays a role.
• A reviewer noted that water fluoridation is considered one of the top ten public health
achievements in the last century, yet the draft RSC Analysis discusses potential health
effects associated with this achievement. She thought it important to consider the
potential public health implications associated with release of the RSC Analysis to the
18 November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
extent that it could generate concern over fluoride in drinking water. Even though they
are important, the public health benefits of drinking water fluoridation are not within the
document's scope. She wondered whether the document should specify what percentage
of the population served by public water systems would potentially be affected by dental
fluorosis. Dr. Donohue clarified that this issue is acknowledged in the Dose-Response
Report and that, legally, EPA cannot endorse dental fluoridation.
• Section 8.2 (Estimates of Tolerable Upper Limit Exposures) is important, but very
confusing as written and needs to be clarified. In particular, the use of the term "margin
of exposure" in the draft RSC Analysis (see page 121) is potentially confusing to risk
assessors, who use this term in a different way. The document should avoid, or at least
clarify, terms such as this that can be construed differently by different specialists.
• Dr. Fox noted that interpretation of exposures over the RfD (as discussed on page 125)
will inevitably vary; however, in assessments of anthropogenic chemicals (where
exposures are involuntary and not nutritionally essential), interpretations vary across
different parts of EPA, and increased risk is not proportionate to dose or exposure
increases. She noted that she had provided EPA Office of Air definitions of "hazard
index" and "hazard quotient" in her written comments (see page F-21 of Appendix F).
• Dr. Ozsvath emphasized that Table 8-2 and Figures 8-1 through 8-3 need clarification
and re-wording and referred EPA to his written comments for details (see pages F-32 and
F-33 of Appendix F).
• Noting a written comment that use of mixed units was confusing, a reviewer clarified that
standardization of units, though desirable, may not always be possible, because the
choice of units is often driven by the analytical method used. Conversion is not possible,
for example, unless the specific homogenization process used by labs for each specific
food is reported in the original study. Use of mixed units is to some degree an outcome of
the lack of standardization of units in the food-fluoride literature. Another reviewer
responded that conversion of beverage units from mg/L to mg/kg (the units used for food)
may be possible by making an assumption about the specific gravity of the beverage.
This would inject another source of uncertainty, but would standardize the units, which
would benefit comparison.
• A reviewer found it difficult to follow whether the document was discussing daily
exposure (mg fluoride/day) or the amount of fluoride intake from food (mg
fluoride/kg/day); the latter should not be used as representative of what people are eating
because it is not clear what they have consumed. EPA should make clear that the
document reports the units provided in the original study.
EPA Response: Several suggestions made by the peer reviewers were implemented during the
post peer review revisions to the document as follows:
• All usages of the term intake and exposure were checked to make sure they were applied
appropriately. EPA is in agreement that intake refers to what is actually ingested via the
oral exposure route. However, in many cases EPA believes that "exposure" or "oral
19 November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
exposure" is the appropriate term because the value presented was not a measurement of
ingestion by a group of subjects in a controlled study, but an estimate of a population's
oral exposure made on the basis of what the general population is likely to have consumed
in their diets, not necessarily what was consumed, swallowed and absorbed.
• The description of the IOM (1997) Tolerable Upper Intake Level was derived from the
IOM publication. However it was revised in an attempt to strengthen the discussion with
out diverting from the IOM definition of the UL values. The term "margin of exposure"
was removed from the document.
• EPA has attempted to clarify the fact that the beverage exposure values are not based on
the assumption that the water in the beverage comes from the local public water system.
In fact the opposite is true, EPA choose studies where reconstituted beverages were made
with distilled deionized water if possible. If no study was available that met that
condition, a study from a market basket collected in the area with the lowest fluoride
concentration was chosen and the fluoride concentration in the water provided for the
reader.
• EPA reworded the text associated with Table 8-1 and Figures 8-1 to 8-3.
• EPA added a more extensive summary and conclusion to chapter 8 than was in the peer
review draft.
EPA did not make any changes to the units reported by the researchers in their published studies.
Tables report the units [mg/L or mg/kg food, or mg/kg/day] as they were reported by the
published papers. All EPA estimates in Chapter 8 are in units of mg/day.
The hazard index (HI) terminology is not generally used by the EPA OW and was not introduced
into the document. The EPA authors feel that HI approach is more relevant to anthropogenic
chemicals that lack benefit than to chemicals with established benefits at appropriate intakes.
EPA did not introduce any direct discussion of the halo effect into the document. The authors
feel that the impact of the halo effect is captured in the studies that included commercial water.
The only cases where tap water from the consumers tap is reflected in a beverage value were
acknowledged, and the fluoride concentration of the tap water was provided.
EPA did not add any data on the number of children age 14 or younger that are served by public
water systems. Those data are collected and utilized by the OGWDW during the second six-year
review of the fluoride MCL. The data submitted by the states does not identify the population
served by age categories. The dose-response document and this exposure and RSC document are
resources that will be used in the process of evaluating whether or not the current MCLG and/or
MCL will be revised.
The EPA dose-response and exposure reports do not negate the benefits associated with
fluoridation. Those benefits are clearly evident in the dose-response analysis which also clearly
demonstrates a leveling off of the benefit as fluoride levels in drinking water increase above the
20 November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
fluoridation range. Given the current media that lead to fluoride intake by children, the
challenge is to find the best approach to use for insuring that intakes do not exceed the levels that
confer a lasting benefit.
The EPA monitoring data from public water systems clearly show that there are many public
water systems that provide their customers with drinking water that routinely contains more than
1.2 mg/L fluoride. The EPA analysis was initiated to determine if there is a meaningful
opportunity to reduce the risk of children in those systems from developing teeth with severe
dental fluorosis.
The fact that the average fluoride levels from public water systems that report the detection of
fluoride falls in the fluoridation range is a reflection of the number of systems nationwide that
fluoridate their water supply.
Closing Discussion and Remarks
Dr. Donohue summarized the key messages the Agency had heard from reviewers during the
meeting:
• A reader's expertise affects how the RSC Analysis is read and understood. EPA should
therefore provide more clarifications (e.g., of terminology, studies, etc.), so readers with a
variety of backgrounds can better understand the document.
• Both the draft RSC Analysis and the Dose-Response Report currently lack an overview
or roadmap. The Agency will add forwards to both documents that explain the "big
picture." The forward for the RSC Analysis will clearly:
o State that the analysis relied exclusively on published studies from the U.S. for
the dietary data. A few of the key studies for toothpaste ingestion and brushing
frequency were Canadian.
o Describe the limitations of the analysis
o State the objectives of the analysis
o Outline the critical steps in the analysis
o Describe how the objectives were met
• EPA understands the reviewers' concern about including a sensitivity analysis; however,
a sensitivity analysis of a 0.02 to 0.03 mg/kg/day difference between beneficial and
apparently adverse amounts of fluoride would likely simply show that the uncertainty
values overlap. This makes it difficult to define where beneficial becomes adverse and
adverse becomes beneficial. The Agency can look at how choosing a different study
would affect the results of the RSC Analysis.
• Because people do not consume the same type or quantity of foods and beverages every
day, any number in a consumption study will not be fully representative. Drinking water
consumption is probably going to be more consistent than food consumption, but this is
an inherent difficulty in dealing with essential nutrients and predicting how much people
need or how much is too much. The Agency can try to improve the presentation in this
21 November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
part of the report, but will always encounter that difficulty. Dietary risk assessment
practices differ from toxicology risk assessment practices.
• EPA will use consistent terminology or at least recognize differences in terminology
among disciplines.
• The Agency will rework the concluding section to describe what the next steps will be.
• EPA will move the discussion of the McClure data out of the RSC Analysis and into
Dose-Response Report.
o
Reviewers agreed this was generally a good reflection of key points they had made , and added
the following comments and recommendations:
• The document needs to provide a clear justification for the selection of specific data sets
and critical studies, perhaps in the forward/overview that EPA will be adding at the
beginning of the RSC Analysis.
EPA Response: The text explaining the reasons for the selection of the key studies was
expanded in the final report.
• The uncertainty discussion should be expanded.
EPA Response: The uncertainty discussion was expanded as suggested by the peer
reviewers.
• In Table 7-2, EPA should use different values in the boxes representing fluoride intake
from different sources to see how that affects the RSC.
EPA Response: EPA examined the impact of selecting alternative studies on the Total
Exposure and RSC and reported the impact of the changes in the revised report.
• EPA should discuss how the characterization of beverages in a separate category from
food affects the use of the estimates of indirect drinking water consumption implicitly
included in the daily drinking water consumption reported in EPA's 2004 document9 and
how these estimates are used in the RSC calculation. The document should better clarify
how the Pang (1992)10 data were used, in particular: What constitutes a store-bought
beverage or a beverage prepared with de-ionized water? For example, is iced tea brewed
in a restaurant or store considered store-bought? This matters because the drinking water
figure from 2004 in the RSC Analysis includes indirect drinking water. However,
beverages such as brewed iced tea, that are made from water in a store, provide another
source of fluoride from community water supplies. In summary, EPA should clarify how
the market basket approach relates to commercially purchased beverages and whether it
could include any beverages that might have fluoride from a community water source.
8 In a post-meeting comment, Dr. Abbott noted that she did not agree that dietary risk assessment practices
necessarily differ from toxicology risk assessment practices, especially when the chemical in question is being
regulated as a contaminant and not as a dietary supplement.
9 U.S. EPA. 2004. Estimated Per Capita Water Ingestion and Body Weight in the United States - An Update. U.S.
EPA, Washington, D.C.
10 Pang, D.T.Y., C.L. Phillips, and J.W. Bawden. 1992. Fluoride intake from beverage consumption in a sample of
North Carolina children. J. Dent. Res. 71:1382-1388.
22 November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
EPA Response: The Pang report did not provide the information to answer the question
posed. EPA has reported the steps that were takes to minimize any impact of any indirect
water that was included in the drinking water compartment of the total exposure section
from being double counted. For most of the age categories all reconstituted beverages
were prepared with distilled water. If there was no study where that was the case, EPA
selected the data where the local water had the lowest fluoride concentration which was
in all cases well below the average tap water concentration from public drinking water
systems.
• A reviewer asked why beverage intake for infants was listed as zero in the draft RSC
Analysis.
EPA Response: EPA clarified that the market basket survey defined "juice" as a "fruit."
The reviewer recommended that EPA clarify this in the document. EPA expanded the
description of the liquids included in the beverage component of a market basked survey.
• Dr. Martinez Mier noted that the University of Minnesota has developed a software
program to assess individual fluoride intake based on what, in an interview, an individual
reports having eaten. The software uses mean fluoride values from peer-reviewed
laboratory analysis, so could be helpful to EPA as a source of data on mean values of
fluoride in various types of food. Values are added to the database on an ongoing basis as
new data become available. Dr. Martinez Mier agreed to send a reference for this
database to EPA.11
EPA Response: Dr Martinez Maier provided a link to the database to EPA. However, it
was not used for the Exposure RSC report because the EPA report is based on the
findings of peer reviewed published studies.
• A reviewer suggested that EPA examine the USDA fluoride database to see what it might
yield in terms of current consumption values and to get an idea of how much uncertainty
was associated with relying on older studies with few groups of foods versus data
available in the current USDA database.12
EPA Response: EPA used the fluoride data from the USDA database in order to gauge
the accuracy of some of the published data and provides a number of Tables based on that
data in Section 2 of the Exposure and RSC report.
• A reviewer suggested that the document add the uncertainties associated with the likely
future increase of fluoride in food due to increased used of fluoridated water in preparing
food products.
11 After the meeting, Dr. Martinez Mier sent the following links to this database and software:
http://www.ars.usda.Qov/SP2UserFiles/Place/12354500/Data/Fluoride/F02.pdf
http://www.ncc.umn.edu/products/databasenutrientsratioscomponents.html
12 In a post-meeting comment, this reviewer added that EPA should investigate using current fluoride concentration
in food data from the USDA's fluoride database in combination with more recent consumption data (from the CSFII
or NHANES databases) to provide a quantitative estimate of the difference between the values from the older
studies and a more current exposure estimate, at least for those food groups where both types of concentration data
exist. This could be accomplished by using a modeling approach like that used by EPA's Office of Pesticide
Programs for sulfuryl fluoride (use of the DEEM model and use of historical concentration data versus more current
concentration data).
23 November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
EPA Response: The EPA hypothesizes that the increase in the number of localities that
fluoridate is responsible for the increased fluoride in commercial beverages and foods.
Data on the mean fluoride concentrations in carbonated beverages, tea, and fruit flavored
drinks are used to support the EPA hypothesis.
• A table such as Table 7-2 could be incorporated into a sensitivity analysis to get an idea
of which of the numbers that have been estimated are most critical. Lacking that analysis,
it is hard to have perspective on how sensitive some of the possible changes in values the
reviewers have discussed will be to the RSC. For example, the document ignores the
amount of fluoride in the atmosphere and this is probably appropriate. The importance of
the soil intake is not certain, but it may be relatively low compared to food.
EPA Response: The revised report included expanded discussion of the relative
contribution of the media quantified to total exposure. The impact of selection of
alternatives to the key studies was also added to the revised document.
Final Note: All editorial comments made by the reviewers were addressed by EPA.
24 November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
IV. EXTERNAL PEER REVIEW REPORT
25 November 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
Summary Report of the Meeting to
Peer Review EPA's Draft Document
Fluoride: Exposure and Relative Source Contribution
Analysis
Arlington, VA
May 14, 2010
Submitted to:
Office of Water
U.S. Environmental Protection Agency (EPA)
Washington, DC 20460
Submitted by:
Eastern Research Group, Inc.
110 Hartwell Avenue
Lexington, MA 02421
Final Report: July 28, 2010
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
Notice
This report was prepared by Eastern Research Group, Inc. (ERG), a U.S. Environmental
Protection Agency (EPA) contractor, as a general record of discussion during the Peer Review
Meeting on EPA's draft document Fluoride: Exposure and Relative Source Contribution
Analysis, held May 14, 2010, in Arlington, Virginia. This report captures the main points and
highlights of the meeting. It is not a complete record of all details discussed, nor does it
embellish, interpret, or enlarge upon matters that were incomplete or unclear. Statements
represent the individual views of meeting participants.
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
Contents
1. INTRODUCTION 1
Background 1
Development of the Draft RSC Analysis 2
2010 External Peer Review 2
2. OPENING REMARKS 4
ERG Remarks 4
EPA Remarks 4
Discussion Notes 6
3. REVIEWER DISCUSSION 3
Charge Question 1 8
Charge Question 2 6
Charge Question 3 11
Charge Question 4 12
Charge Question 5 12
Charge Question 6 14
Charge Question 7 15
Closing Discussion and Remarks 21
Appendix A: Peer Reviewers
Appendix B: Charge to Peer Reviewers
Appendix C: EPA Slides Presented During Teleconference
Appendix D: Meeting Agenda
Appendix E: Meeting Observers
Appendix F: Reviewer Post-Meeting Comments
Appendix G: EPA Slides Presented at the Peer Review Meeting
Appendix H: Panel Chair's Summary of Reviewer Pre-Meeting Comments
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis "
1. Introduction
As part of a series of activities, described below, to evaluate the drinking water standards for
fluoride, the U.S. Environmental Protection Agency (EPA) conducted a relative source contribution
analysis of fluoride in drinking water and developed the draft document, Fluoride: Exposure and
Relative Source Contribution Analysis (RSC Analysis). This report summarizes discussion at a
workshop, held in Arlington, Virginia, on May 14, 2010, to peer review the draft RSC Analysis.
Background
EPA established the current drinking water standards for fluoride in 1986. The Maximum
Contaminant Level Goal (MCLG) and the Maximum Contaminant Level (MCL) were set at 4.0
milligrams per liter (mg/L) of water to protect against stage three skeletal fluorosis, a disorder
characterized by calcification of ligaments, immobility, muscle wasting, and neurological problems
related to spinal cord compression. Exposure to fluoride from other sources, including food,
beverages, dental products, supplements, industrial emissions, pharmaceuticals, and pesticides, was
not considered at that time. EPA also established a non-enforceable secondary standard (the
Secondary Maximum Contaminant Level, or SMCL) of 2.0 mg/L to protect against moderate/severe
dental fluorosis (considered a cosmetic effect at the time). The MCLG/MCL were set assuming all
exposure to fluoride would come from drinking water—a 100% relative source contribution (RSC).
The 1996 Safe Drinking Water Act Amendments require 6-year reviews of drinking water
contaminant standards. EPA performed the first such review of the fluoride standards in 2002-2003.
The Agency concluded that no revision was appropriate at that time because it had asked the
National Academies of Science (NAS) National Research Council (NRC) to evaluate the health
effects and occurrence of fluoride in public water supplies and to examine the current standards
considering new data published after the 1986 regulation and a 1993 NRC review of fluoride that
EPA had also requested. Fluoride would be reexamined during the next 6-year review.
In response to this charge, the NRC released a March 2006 report, Fluoride in Drinking Water: A
Scientific Review of EPA 's Standards11*, and concluded that the current MCLG of 4.0 mg/L does not
adequately protect against severe dental fluorosis. The report also recommended that EPA update the
dose-response assessment for severe dental fluorosis, consider susceptible populations, characterize
uncertainty and variability, and update exposure assessment (i.e., the RSC). In response to these
recommendations, EPA developed a three-part action plan, which included: 1) conducting a dose-
response analysis for severe dental fluorosis and skeletal effects; 2) examining the relative source
contribution of fluoride in drinking water; and 3) studying the relationship of fluoride to cancer.
The first part of the action plan was addressed with the development and March 2008 peer review of
a report on Dose-Response Analysis for Severe Dental Fluorosis and Skeletal Effects (Dose-
Response Report). This document established a point of departure (POD) for severe dental fluorosis,
but was unable to complete a dose-response analysis of the skeletal effects of fluorosis due to
inadequate data. EPA will initiate the third part of the action plan, examining the relationship of
fluoride to cancer, once an ongoing study, recommended by the NRC as a starting point, is
13 NRC (National Research Council). 2006. Fluoride in Drinking Water. A Scientific Review of EPA's Standards.
Washington, DC: National Academy Press.
1
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis "
published. Development, peer review, and fmalization of Fluoride: Exposure and Relative Source
Contribution Analysis (RSC Analysis) constitute the second part of the action plan. This report
summarizes the results of the peer review of the draft RSC Analysis.
Development of the Draft RSC Analysis
EPA prepared the draft RSC Analysis in order to determine current fluoride exposures, as well as
exposures at the time (193Os-1940s) when the dose-response data used for the 1986 SMCL were
collected. The SMCL is based on severe dental fluorosis and is non-enforceable, but does require
public notification when the average fluoride concentration in drinking water from a public drinking
water system exceeds 2 mg/L in order to alert families of the risk for severe dental fluorosis in
children.
Calculation of the health-based MCLG for non-cancer endpoints generally requires application of an
RSC to adjust for sources of exposure other than drinking water that are not reflected in the POD for
MCLG derivation. In the case of fluoride, the endpoint of concern was identified by the NRC (2006)
as severe dental fluorosis, characterized as displaying confluent pitting in the enamel of at least two
permanent teeth. EPA selected Dean (1942)l as the critical study and used the 95th percentile
lower bound concentration in drinking water associated with a 0.5% prevalence of severe dental
fluorosis in that study as their POD.
The draft RSC Analysis report uses data published in the peer-reviewed literature to estimate dietary
fluoride in the 1930s-1940s, as well as current total exposures from plain (direct and indirect)
drinking water; solid foods; commercial beverages; toothpaste; soils; and use of the recently
registered pesticide, sulfuryl fluoride. The recent exposure data are used to determine the current
relative source contribution of fluoride in drinking water to total exposure for the 90* percentile
drinking water intake, in accordance with EPA policy.
The draft RSC Analysis was internally peer-reviewed by representatives from EPA's Office of
Children's Health, Office of Congressional and Intergovernmental Relations, and Office of Research
and Development. That review was completed in January 2010.
2010 External Peer Review
In spring 2010, Eastern Research Group, Inc. (ERG), an EPA contractor, organized an independent
peer review of the draft RSC Analysis to assess its scientific quality and utility. Four experts
(Appendix A) conducted this review:
• Linda C. Abbott, U.S. Department of Agriculture
• Mary A. Fox, John Hopkins Bloomberg School of Public Health
• E. Angeles Martinez Mier, Indiana University School of Dentistry
• David L. Ozsvath, University of Wisconsin-Stevens Point
14 Dean, H.T. 1942. The investigation of physiological effects by the epidemiology method. In: Fluoride and Dental
Health. Publ. Amer. Assoc Advanc. Sci., no. 19. pp 23-31.
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis "
ERG provided the reviewers with the draft RSC Analysis and a charge (Appendix B), which asked
for reviewer comments on the various aspects of the document.
In the first stage of the review, the experts worked individually to prepare individual written pre-
meeting comments. To kick off this stage of the review, ERG organized and facilitated a conference
call in which EPA presented background on the context for developing the document (see slides in
Appendix C) and reviewers were invited to ask questions of clarification to understand the context
for the review and the charge questions. (The background provided on this call is summarized in the
"Background" section, above.) Reviewers then prepared and submitted their pre-meeting comments,
hereafter referred to as "written comments," which ERG then provided to all reviewers and EPA
prior to the meeting.
In the second stage, ERG convened a one-day peer review meeting, on May 14, 2010, in Arlington,
Virginia. Appendix D provides the meeting agenda and Appendix E provides the list of meeting
observers. After the meeting, ERG prepared a draft summary report of the proceedings. Reviewers
were asked to check the report for accuracy and completeness, and also to submit post-meeting
comments by amending their pre-meeting comments in light of the discussions. Reviewers post-
meeting comments are included as Appendix F.
This report summarizes the meeting proceedings as follows:
• Section 2 presents ERG and EPA opening remarks.
• Section 3 summarizes the reviewers' discussions organized by charge question.
• The appendices provide the list of peer reviewers (Appendix A), charge to peer reviewers
(Appendix B), EPA slides presented during the background teleconference (Appendix C), the
meeting agenda (Appendix D), meeting observers (Appendix E), reviewer post-meeting
comments (Appendix F), EPA slides presented at the beginning of the peer review meeting
(Appendix G), and Dr. Martinez Mier's summary of reviewer pre-meeting comments
(Appendix H).
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis "
2. Opening Remarks
ERG Remarks
Jan Connery (ERG), the meeting facilitator, opened the meeting by welcoming the reviewers and
asking the reviewers and EPA document authors to introduce themselves. Ms. Connery reviewed the
meeting agenda (Appendix D). She noted that the pre-meeting comments were developed by
reviewers working individually prior to the meeting, and that reviewers may express different
opinions during the meeting. Ms. Connery made clear that all discussions were to be among
reviewers only; however, reviewers could request, and EPA could offer, clarifications where
relevant.
Ms. Connery then introduced Eric Burneson, Chief of the Targeting & Analysis Branch, Standards
and Risk Management Division, EPA Office of Ground Water and Drinking Water, and Joyce
Donohue from EPA's Office of Water (OW), to provide opening remarks. Dr. Donohue served as a
co-author of the draft RSC Analysis.
EPA Remarks
Mr. Burneson welcomed the reviewers and thanked them for their participation. He noted the
purpose and importance of the review to EPA as part of the RSC Analysis development process, and
provided brief background on the fluoride drinking water standard. He emphasized that the RSC
Analysis would inform future decision-making by OW's Standards and Risk Management Division,
particularly concerning the future review and potential revision of the national primary drinking
water regulation for fluoride. EPA is required to review the drinking water standards at least every 6
years, and has recently concluded its most recent round of national primary drinking water standard
reviews. However, the Agency may initiate a review and rulemaking at any time when the science
suggests there is a meaningful opportunity for public health protection by revising a drinking water
standard. This peer review is an important part of the process to ensure that EPA has the best
available science assembled for decision-making with respect to the fluoride drinking water
regulation.
Dr. Donohue then summarized the history of the fluoride drinking water standards, the 2006
recommendations by NRC, and the results of those recommendations (see slides in Appendix G).
When EPA completed its first 6-year review of the fluoride drinking water standard in 2003, the
Agency asked the National Academy of Sciences to examine the hazards associated with fluoride in
drinking water and the relative sources of exposure to fluoride in light of new data since the NAS'
prior review in the 1990s. The resulting 2006 NRC report15 recommended that EPA conduct a dose-
response analysis for several dental fluorosis, which the NAS for the first time considered to be an
adverse health effect (as opposed to a cosmetic effect) because data indicated increased
susceptibility to cavities as a result of pitting and thinning of enamel. The NRC also asked EPA to
do a dose-response analysis for skeletal effects, including skeletal fluorosis and fractures. They
recommended an extended assessment of exposure to sensitive populations, including children
15 NRC (National Research Council). 2006. Fluoride in Drinking Water. A Scientific Review of EPA's Standards.
Washington, DC: National Academy Press.
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis "
during the period of tooth formation and adults for skeletal effects. NAS recognized that the issue of
whether fluoride causes osteosarcomas is highly controversial and recommended that EPA wait until
the results of an ongoing study at Harvard University were published to make a decision on this
issue. The Agency intends to prepare a report on this potential effect once the Harvard study is
published.
EPA developed a draft Dose-Response Report, which was externally peer-reviewed in March 2008.
The reviewers agreed that the Agency chose the correct critical study, but suggested that EPA model
to the 0.5 percent level, if possible, rather than the 1 percent level in the report. The Agency did
model to the 0.5 percent level, and that changed the point of departure for the critical endpoint from
2.1 mg/Lto 1.87mg/L.
EPA began to develop the RSC Analysis after the March 2008 peer review. An initial draft of the
analysis was internally reviewed within EPA during the past year and then revised in response to
internal comments. The resulting draft is the one the current external reviewers received to review.
Dr. Donohue described the key steps that would follow this external peer review, including revising
the RSC Analysis in response to reviewer comments and posting the revised document
simultaneously with Dose-Response Report on the Internet. At that point, EPA's Office of Ground
Water and Drinking Water will consider whether to revise the fluoride MCLG and/or MCL, which
are currently both 4.0 mg/L based on severe skeletal fluorosis. Any revision of the standards should
maintain or improve health protection, but the cost of achieving the revised standard must be
justified by the health benefits of doing so.
Dr. Donohue then explained the differences between the MCL and MCLG, as well as EPA's general
policies in calculating the MCLG. The MCLG, which is not enforceable, will be based on the NRC
report as the description of hazard and dose-response assessment. The MCL, which is enforceable,
considers not only human health data but also analytical and technological limitations, as well as the
cost involved in requiring compliance. She noted that it is difficult to remove fluoride from water.
In response to a reviewer's question, Dr. Donohue confirmed that the MCL applies only to public
water systems (including non-transient, non-community systems) and that EPA has no regulatory
authority over private water supplies, including wells, cisterns, and bottled water. The reviewer
pointed out that the RSC would need to include consideration of the contribution that populations
served by public water systems would also receive from water sources other than public water
systems.
Dr. Donohue then described how EPA generally approaches developing the MCLG. First EPA
determines a Reference Dose (one of the outputs from the Dose-Response Report). The Agency uses
the Reference Dose (RfD) to calculate the Drinking Water Equivalent Level (DWEL), which
assumes that the total exposure comes from drinking water. EPA determines the DWEL by
multiplying the RfD by the weight of the sensitive population (this could be done for each age group
identified), and then dividing by the drinking water intake. The Agency uses OW's 2004 survey of
individual food intake, in part to stay consistent with the age groupings the Agency has been using.
EPA then calculates an MCLG, protective for populations served by public water systems, by
multiplying the DWEL by the RSC. Dr. Donohue explained that the Agency cannot use its
subtraction option because of the tolerances for pesticides containing fluoride. When another office
within EPA has a regulation governing a contaminant, EPA policy requires that a percentage
5
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis "
approach be used. EPA policy is to use average body weight and the 90th percentile drinking water
intake when determining the RSC.
In the case of fluoride, the 1.87 mg/L from the benchmark dose analysis could be used as the point
of departure to which an RSC would be applied in order to avoid the uncertainties involved in
converting the concentration to dose (mg/kg/day) using body weight and water intake estimates that
would apply to the 193Os-1940s time period when the exposures from the critical study occurred.
These are options, but no decisions have been made about what approach will be used. Each age
group could be used independently by applying its RSC to develop age-specific MCLGs and select
the limiting value, or the Agency could normalize across the 0.5- to 14-year age range that is
susceptible to severe dental fluorosis and get one value. EPA will be considering these options when
the Office of Water considers potential changes to the MCLG after the RSC Analysis is finalized.
Dr. Donohue shared EPA's assessment of findings in the draft RSC Analysis and the Dose-Response
Report. In the draft RSC Analysis, the exposure estimate indicates that the fluoride RfD is being
exceeded for some public water system consumer age groups at an average fluoride concentration of
0.87 mg/L. The prevalence of dental fluorosis has increased from approximately 10 percent (at the
time of the critical study) to 23 percent in 1986-87, to 32 percent in 2000; unpublished data from
CDC in 2004 show another substantial increase (based on the unpublished National Health and
Nutrition Examination Surveys [NHANES] data).
The dose-response data considered by NAS are supportive of an increased risk of cavities when
dental fluorosis is severe. Dental fluorosis, once acquired, cannot be reversed; the consequences are
lifelong. Cavities meet the definition of an adverse health effect, especially when untreated, and the
Agency cannot assume all cavities are treated. Over 4,000 public water systems in the U.S. deliver
water supplies with naturally occurring fluoride at average concentrations above 0.87 mg/L. Most of
these systems are groundwater systems, and many are small.
Dr. Donohue then shared the results of the Iowa Fluoride Study conducted by Dr. Steven Levy. The
study involved 622 children of moderate to high socioeconomic status. Intakes of water, beverages,
selected food supplements, and toothpaste were reported by the parents at 3- to 4-month intervals
from birth to 48 months. When the children's teeth were examined between ages 8 and 10, eight
cases of severe fluorosis were detected. The study used a different scale from the Dean (1942)
study16, and did not limit the severe designation only to cases where there was pitting of the enamel.
Therefore, EPA asked Dr. Levy to provide information on whether any of the eight severe cases
exhibited pitting. Photographs of seven of the severe fluorosis cases (photographs for the eighth case
could not be located) showed that one of the seven children had pits in the enamel of the affected
teeth. This child's average intake from 9 to 36 months was 0.075 mg/kg/day. The larger data set
showed no severe cases of dental fluorosis when exposure was equal to or less than 0.06 mg/kg/day.
Dr. Donohue noted that EPA and the Centers for Disease Control (CDC) had agreed to work
together to more thoroughly evaluate the drinking water concentrations that may have been
associated with cases of severe dental fluorosis identified in recent data collections from the
NHANES.
Discussion Notes
16 Dean, H.T. 1942. The investigation of physiological effects by the epidemiology method. In: Fluoride and Dental
Health. Publ. Amer. Assoc Advanc. Sci., no. 19. pp 23-31.
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis "
Ms. Connery thanked Dr. Donohue for her opening remarks and turned the meeting over to E.
Angeles Martinez Mier, the panel chair, to begin the reviewer discussions. Dr. Martinez Mier had
developed a powerpoint presentation (Appendix H) summarizing the reviewers' pre-meeting
comments for each charge question, which she displayed when initiating the consideration of each
charge question to remind reviewers of their preliminary comments.
Early in the discussion, one of the reviewers, Linda Abbott, read a statement17 she had prepared and
asked that it be included in this report. She said she had provided as many comments as she could in
the time available for the review, and was concerned there could be additional issues in the
document she would not have had time to identify. Other reviewers commented that they
nevertheless found Dr. Abbott's comments very valuable. Ms. Connery asked Dr. Abbott whether
she would have time after the report to add comments she had not had a chance to make earlier, and
Dr. Abbott replied that her ongoing job responsibilities would preclude that possibility.
Also during subsequent discussion, two reviewers noted that they had interpreted the requirement
not to make copies of the report during the review as meaning they should not print a copy of the
report. They reviewed the document on screen, but having a print version would have been helpful.
17"I originally declined the request to participate in this peer review. My schedule precluded me from devoting more than
20 hours to the project and I thought that was insufficient time to conduct a comprehensive review of the scientific
information supporting the determination of a maximum contaminant level or maximum contaminant level goal. The
letter of commitment estimates the level of effort for this task to be 28 to 30 hours including the eight-hour workshop. I
estimate that I have already spent over 30 hours reviewing the document prior to the workshop. Even 30 hours was not
an adequate amount of time to review such a complex document that relies so heavily on historical exposure levels,
especially given the added effort to obtain the underlying references. Thus my assessment of the document should not be
considered to be complete - there may be issues that I was unable to identify due to lack of time. The absence of
comments on the use of an exposure assessment method, dataset or historical study should not necessarily be interpreted
as agreement with the peer review draft."
7
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis "
3. Reviewer Discussion
Charge Question 1: Please describe any suggestions you have for improving the clarity,
organization, and/or transparency of the draft document.
Reviewers noted agreement in their written comments that the purpose of the draft RSC Analysis
was not clearly stated and that adding an overview or "road map," including a statement of
objectives, in the introductory section of the document would help address this concern.
Dr. Donohue noted that the Agency agreed with these pre-meeting comments, and had decided to
add forwards to both the RSC Analysis and the Dose-Response Report in order to outline the study
objectives and present a road map. She said that EPA planned to post both documents on the Internet
at the same time. Dr. Donohue also noted that the external peer reviewers, as well as EPA internal
reviewers, had clearly indicated confusion around presentation of how the Agency used the McClure
(1943)18 data to estimate food contributions in 1942. She proposed that EPA would remove the
description of the McClure (1943) data from the draft RSC Analysis and add it as an appendix to the
Dose-Response Report. This seemed appropriate since those data are used in the Dose-Response
Report rather than the RSC Analysis. She asked reviewers for feedback on whether they thought this
change would address their concern about this section of the report.
Reviewers supported the idea of a forward and agreed to talk about the best location for the McClure
(1943) data when they discussed Charge Question 5.
Dr. Ozsvath remarked that it was difficult to determine what the critical steps of the analysis were
and why they were taken. He had alluded to this concern in his written response to Question 1 and
elaborated on it in his response to Question 7 (see Appendix F). He noted that this concern was also
echoed in Dr. Abbott's detailed written comments. He recommended that the document include a
road map, a list of objectives, and a statement at the end of the document clarifying how the
objectives were met.
Dr. Martinez Mier noted that some of the terminology was not consistently used throughout the
document. Also, some of the steps in the fluoride analysis are not consistent with those found in the
current dental literature for fluoride analysis. Though the document is not intended for the dental
community, it will inevitably be circulated within that community because it impacts dental fluorosis
and caries prevention. Dr. Martinez Mier referred EPA to her written response to Charge Question 1
(see Appendix F) for specific suggestions about reorganizing the discussion of analytical methods
and changing how things are referenced in order to bring the discussion more in line with
contemporary analytical approaches used by the dental community.
Dr. Fox said she agreed with recommendations to improve clarity and transparency by stating the
questions the document will address and the approach EPA used to address those questions. She also
had concerns about the discussion of public health implications in Chapter 8. She had provided a
detailed comment about this in her written response to Charge Question 7 and noted that she would
talk about this in detail when Question 7 was discussed.
18 McClure, F.J. 1943. Ingestion of fluoride and dental caries. Quantitative relations based on food and water
requirements of children 1-12 years old. Amer. J. Dis. Child. 66:362. [Republished in Publication 825, U.S. Public
Health Service, 1962].
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis "
Dr. Abbott referenced all of her written comments in response to Charge Question 1 (see pages F-6
to F-8 of Appendix F). She also noted one additional comment. She recommended that, at the
beginning of the document, EPA acknowledge the spectrum of populations obtaining drinking water
from the full variety of sources, including community water supplies, spring water, bottled water,
and well water, and clarify whether these sources factor into the MCL. The document seems to
combine these at different levels in the RSC calculation because food intake is considered across the
population as a whole, not just in those who drink community water. This is a particular problem
when considering beverages. For example, beverages prepared by adding water and made away from
home, like tea or beverages made from dry powders with extra indirect water, may not come from
the same water supply used at home. In the Continuing Survey of Food Intakes by Individuals
(CSFII), all these different types of water sources are associated with the different types of food and
beverages. The document should acknowledge from the outset that the analysis combines these
different populations for different parts of the estimate. Without this acknowledgment, readers will
be confused when they learn that the document only considers populations served by community
water systems and not those who are not served by those systems but who are exposed to fluoride in
their diet by ingesting food or beverages made with fluoridated water (e.g., at a processing plant).
In response to this comment, Dr. Donohue clarified that EPA used data from the Food and Drug
Administration's (FDA's) total diet study, which clearly identified in the market basket which
beverages were made from public water and which at home. In the analysis, EPA subtracted out the
portion of beverages made with public water. Dr. Abbott responded that the CSFII would provide a
more refined estimate of the source of indirect water used in beverage preparation. Dr. Donohue
clarified that EPA relied on the published literature for the RSC Analysis and did not conduct any
independent analyses. In light of that clarification, a reviewer suggested that EPA state at the
beginning of the document that the Agency had not conducted any new analyses and instead had
relied exclusively on the literature. Other reviewers agreed it was important for EPA to more clearly
explain what they did and did not do in the analysis.
Charge Question 2: Have the uncertainties associated with the analysis been adequately
characterized? Are there any important uncertainties in the data that are not discussed
adequately in the document, especially in the synthesis sections? Please describe any concerns
you have and any specific suggestions for improving or enhancing the uncertainty discussion.
Reviewers agreed that the document should provide a better discussion of uncertainty. Individual
comments included:
• The draft RSC Analysis describes sources of uncertainty, but does not attempt to show how
the RSC would be affected if different assumptions were made. A sensitivity analysis would
help show how different assumptions would affect the RSC. Page 113 of
the draft document states that the Agency believes these are reasonable estimates, but the
Agency does not justify this statement; this part needs more scientific rigor.
• The draft RSC Analysis recognizes some sources of uncertainty (e.g., "technological
limitations"), but should be more explicit in discussing the uncertainties associated with
specific values from various studies related to fluoride analysis. The fact that uncertainty in
fluoride analysis can go either way (leading to much lower or much higher values) should be
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis "
discussed. Additionally, there is evidence (e.g., policy statements by the National Institutes
of Health and some regulatory bodies) that the methodology has a strong influence on results.
The current document "tiptoes" around this reality; EPA needs to provide stronger statements
about this, for example by stating that the methodology "does" as opposed to "may" have
influence. The lack of strong acknowledgment of the limitations of fluoride analysis carries
over into the McClure discussion later in the document, where these limitations are
important. Finally, multiple studies recognize the large individual variation in the sources for
water intake; this uncertainty is not well addressed.
• Dr. Fox referenced her written comments in response to this charge question (see page F-19 of
Appendix F), in which she listed a series of questions, which she had made from the
perspective of a risk assessor: "Does the available data overestimate, underestimate or both?
Can the sources of uncertainty be prioritized? Are the different analytical methods
contributing to more uncertainty than changes in diet? How does an understanding of the
largest sources of uncertainty help us understand the data or help us characterize the
analyses?" She felt that the uncertainty discussion would be much clearer if the document
would address these questions. Agreeing with the prior comment on the role of
methodological uncertainty, she recommended that the document more adequately
characterize the methodological difficulties, and include any definitive quantitative
information on uncertainty, whether leading to under- or overestimates.
• Dr. Abbott referenced her written comments in response to this charge question (see pages F-8
to F-9 of Appendix F). She emphasized that quantitative assessment of how influential the
food intake studies were in calculating the RSC was both possible and important. She also
questioned whether combining the 90th percentile value for drinking water intake with all the
other values would yield a valid estimate of total drinking water intake.
Dr. Donohue clarified that use of the 90th percentile value for drinking water intake and use of the
average body weight were Agency policy. In response to a reviewer question, she agreed that a
sensitivity analysis would improve understanding of the range of uncertainty around the data points
selected for the analysis. The reviewer responded that choosing one number was not erring on the
side of being conservative; a sensitivity analysis would illuminate how the selected number
compared with other possibilities that might be more protective of public health.
10
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis "
Charge Question 3: Please consider the studies that have been selected as representative of
exposures for the specific age groups and/or exposure media. Have these studies been
adequately summarized and interpreted? Indicate any deficiencies in the descriptions of the
studies and any suggestions you have for improvement. Describe any concerns you have about
the selection of these studies, as well as any recommendations you may for alternative studies
that you believe are more representative of exposures.
Reviewers held differing opinions about the adequacy of study descriptions and whether the selected
studies were representative of exposures.
• Dr. Abbott referred EPA to the written comments she had provided in response to this
question (see pages F-10 to F-l 1 of Appendix F) and asked that the Agency consider all these
comments. She noted in particular that there appeared to be a mistake in the "0.5 to > 1 year"
row of Table 6-1, since it seemed that the value for the addition of powdered formula should
have been higher than stated. A complete explanation of this concern is provided in her
written comments in response to Charge Question 7.19
• A reviewer recommended that the document provide more discussion about the quality of the
literature on the various exposure assessment methodologies, including whether any
methodologies are better suited than others for the analysis and how this factored into
choosing the key studies.
• Dr. Martinez Mier said she found the selected studies to be representative and thought they
would be recognized as valuable by others with dental expertise. She suggested that dental
professionals with experience in this area would probably understand the studies, as
described in the document, but she was concerned whether readers outside this area of
expertise would be able to understand them. She felt that some studies that reflect more
contemporary analytical methodologies should be included, and referred EPA to her written
comments for more details on this comment and for suggested references (see pages F-26 to
F-27 of Appendix F).
• Speaking from a soil and groundwater perspective, Dr. Ozsvath clarified that there is no
database for fluoride in soils comparable to that used to analyze fluoride concentrations in
public drinking water supplies. There are studies supporting a range, but there is no way to
characterize those data statistically. He noted that a 2009 document prepared for Health
Canada20 used 100 ppm as the average fluoride concentration in soil to estimate daily fluoride
intake from soil ingestion, but this value might not apply to the U.S. He suggested that EPA
provide more justification for its approach, including the use of water and soil data that do not
have the same level of precision. A reviewer asked if this should
19 This response reads as follows: "Table 6-1 states the overall mean from Ophaug (1985) (0.17 mg/d) was adjusted by
subtracting the milk/formula intake form the earlier Ophaug study and then adding the 0.14 mg/d estimate from the
powdered formula study by Van Winkle et al, 1995 using tap water to make the formula. This almost doubles the
original Ophaug estimate and appears to be an incorrect combination of a fluoride concentration in food (Van Winkle et
al. 1995) with an exposure estimate (Ophaug, 1985). To add the Van Winkle study to the Ophaug study, you would have
to know how much formula was consumed in the Ophaug study. If the results from Van Winkle really are
concentrations and are reported in mg/L (and not in mg/kg or some mass-mass basis) you would have to know the
amount of liters of formula consumed in Ophaug to estimate the mg F/d contributed to the diet."
20 Federal-Provincial-Territorial Committee on Drinking Water. 2009. Fluoride in Drinking Water
http://fluoridealert.org/canada.2009.report.pdf
11
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis "
be included in a sensitivity analysis. Dr. Ozsvath responded that the drinking water database is
large enough to characterize probabilistically, so it is possible to know how the representative
number relates to the whole database; however, this is not possible with the soil data.
Charge Question 4: Please comment on EPA's rationale for selection of specific data elements
to represent average exposures for each of the age groups. Has the selection been scientifically
justified and clearly and objectively described? What changes or improvements would you
suggest?
Dr. Martinez Mier noted that, in their written comments, all reviewers had recommended that EPA
needed to provide further discussion on the selection of age groups. Reviewers had also provided
several comments that the water and food exposure for infants appeared to be high and that the
report needed to provide a better description of the variation in food concentrations. A number of
reviewers also commented on the limited geographic area considered in the studies and the number
of subjects involved in the studies. She briefly listed other individual written comments (see
Appendix H) and then opened the floor for discussion.
• Dr. Abbott commented that EPA's rationale for selection of the age groups selection was not
clear, which made this portion of the document very confusing, particularly since the review
document appeared to use different age groups than the underlying data. It was also puzzling
why the draft RSC Analysis utilized historical data for this analysis since it seemed to her
that calculations using more current data could have been made. She emphasized that this
part of the document was very confusing and she referred EPA to all of her written comments
in response to this question (see page F-l 1 in Appendix F).
• Dr. Fox commented that EPA should provide a better explanation of the criteria the Agency
used to identify key data and studies and the considerations used to determine what was
representative. She referred the Agency to her written comments for details (see page F-19 of
Appendix F).
• Dr. Martinez Mier said she understood that use of data from the Dean study (1942) was
necessary due to the lack of recent fluorosis prevalence data, but the RSC Analysis needed to
better explain why those data elements were selected. She was particularly concerned that the
dental community would question why EPA used the Dean (1942) data rather than more
recent data. She recommended that EPA better clarify its rationale for including this study.
• A reviewer noted that while she understood the benchmark dose issue and that the Dose-
Response Report determined the choice of Dean (1942) as the best study for addressing dose-
response, she still questioned why EPA did not use current dietary exposure information in
the RSC Analysis. It was unclear whether the RSC would be based on the Dean (1942) data
or more recent data.
Noting their confusion about use of the Dean (1942) data, reviewers revisited the idea, proposed by
Dr. Donohue during the Question 1 discussion, that EPA remove the description of the McClure
(1943) data from the draft RSC Analysis and add it as an appendix to the Dose-Response Report.
They agreed this was a good idea.
Charge Question 5: Please comment on the validity of basing the food intake estimate for the
1940s on the McClure (1943) publication, as supported by the concentrations found in various
12
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis "
food groups from more recent analytical data. Do you agree with this approach? If not, what
approach would you suggest for estimating food intake for the 1940s?
Reviewers noted that Charge Question 5 probably should have referred to McClure (1949)21 rather
than (1943). All reviewers agreed that the report should better clarify the food intake estimates,
especially in Table 2-25, where the McClure data are compared to more contemporary data. Dr.
Donohue clarified that the McClure food data were used to determine the dose-response and not the
RSC. Reviewers emphasized that EPA needed to add a forward in both the RSC Analysis and the
Dose-Response Document to clarify the purpose of the two documents and their relationship to one
another, since it cannot be assumed both documents will be read together.
Reviewers also recommended that discussion of adding 0.01 to get the POD be moved to the Dose-
Response Report and that EPA clarify how the 0.5 ppm value was selected from the four values
presented by McClure in the 1940s studies. Dr. Donohue affirmed that, in response to reviewers'
comments, the Agency will move discussion of the McClure data to an appendix in the Dose-
Response Report, and will respond to reviewers' comments on that section even though it will be
part of a different document. However, Dr. Donohue noted that EPA will need to keep some of the
historical data in the RSC Analysis to show that the analytical methods have given different results
(e.g., that values for fluoride in meat and poultry decreased over time since the 1940s as the
analytical methods changed).
Additional reviewer comments included the following:
• A reviewer noted that the document did explain the importance of determining the food
contribution, but the introduction was confusing, as was the section comparing the RfD
(based on McClure data) and current RSC values.
• Another reviewer noted that, at one point in the draft RSC Analysis, the McClure data are
considered a reasonable basis for estimations, but Table 2-25 states that the McClure results
differ from current studies. The analytical limitations of the McClure study compromise the
results, and this needs to be clearly stated in RSC Analysis
• Dr. Martinez Mier said she wanted to amend a sentence discussing fluoride concentrations in
her pre-meeting comments to add the word "or higher" so that the sentence would read:
"Interferences most likely resulted in reported fluoride concentration in foods lower or
higher than actual concentrations." She added that several reports in the dental literature
have compared food and beverage and analysis of dental tissue where different methods have
been used, and the interferences from these methods, particularly less sensitive methods,
resulted in both higher and lower fluoride concentrations. This should be clarified the RSC
Analysis.
• The approach for the food intake estimate is reasonable, but the need to estimate fluoride
from 1940s food intake data, and the selection of 0.5 ppm from McClure (1943) as the
estimate for fluoride in solid foods, is not clearly explained or supported. This explanation is
very important and belongs up front in the purpose/problem statement. The discussion of
Table 2-25 in the analysis offers possible explanations for why there are differences between
McClure and more contemporary data, but the document should clarify how the differences
inform the RSC Analysis.
21 McClure, F.J. 1949. Fluoride in foods. Public Health Reports 64:No. 34, pp 1061-1074.
13
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis "
• It is unclear why data from 1943 (McClure) are being used to set today's RSC. The data
seem irrelevant to the RSC Analysis. This needs to be clarified. There is a discrepancy
between Table 2-12, which reports data from McClure (1949) and Table 2-25, which reports
data from McClure (1943, 1949).
• Dr. Ozsvath referenced his written comments in response to this charge question, in which he
noted a possible contradiction (see pages F-30 and F-31 of Appendix F). Specifically, on
page 34 the authors state that the differences between McClure's (1943, 1949) data and the
more recent USDA (2005)22 data shown in Table 2-25 "cast doubt on the results of exposure
assessments derived from some of the early food data." However, in the third paragraph on
page 122, the authors cite the same data (in Table 2-25) and state that "McClure's (1943)
estimate for dietary intake based on a diet where solid foods had an average of 0.5 ppm
fluoride appears to provide a reasonable basis for the contribution of solid foods to total
exposure in the 1940s." Dr. Ozsvath noted that these statements appear to be contradictory,
but that could perhaps be because the authors intended to cite Table 2-41, not 2-25 on page
122. This issue needs clarification.
Charge Question 6: Provide citations (and, where possible, pdfs or hard copies) for any
references you suggest EPA should consider adding to the document, and describe where you
suggest these references be added.
Dr. Donohue clarified that EPA had limited the RSC Analysis to studies from the U.S. and Canada
initially, eventually eliminating many of the Canadian studies. Reviewers made the following
comments:
• Dr. Abbott drew EPA's attention to the dietary intake references she had included in her
written comments (see page F-13 of Appendix F). She was unsure where they should be
added. She said that, because the RSC Analysis included so much historical data, she had
been unsure what type of references to provide.
• Dr. Martinez Mier referred EPA to the many references she had provided in her written
comments (see pages F-22 to F-28 of Appendix F). She strongly recommended that EPA
considering including studies from other countries and not just U.S. data when describing
methodologies. For example, studies by Jackson, Dunipace, and Levy use more recent
methodologies and not, for example, ashing, a method that has not been used in many years.
• Dr. Ozsvath suggested three references in his written comments that he recommended be
added to the RSC Analysis (see page F-31 of Appendix F). The first reference (Beltran et al.,
2002)23 is relevant to increases in fluorosis cases, a central focus of the RSC Analysis; this
study addresses how fluorosis prevalence changed over time within populations exposed to
drinking water with different fluoride content. The second reference (Federal-Provincial-
22 USDA (U.S. Department of Agriculture). 2005. USDA National Fluoride Database of Selected Foods and Beverages,
Release 2. Nutrient Data Laboratory, Agricultural Research Services, U.S. Department of Agriculture. Beltsville, MD.
23 Beltran, E.D., S.O. Griffin, and S.A. Lockwood. 2002. Prevalence and trends in enamel fluorosis in the United States
from the 1930s to the 1980s. Journal of American Dental Association 133(2): 157-165.
http://iada.ada.Org/cgi/content/full/133/2/157
14
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis "
Territorial Committee on Drinking Water, 2009)24 is significant (even though the RSC
Analysis does not consider Canadian studies), because the Canadian equivalent found a very
different RSC for drinking water. He suggested that EPA add an explanation of this
difference to the RSC Analysis.
Charge Question 7: Please provide any additional comments and/or further suggestions you
may have for improving the document.
Dr. Martinez Mier noted the wide variety of written comments reviewers had provided in response to
this question. She read through the list summarizing these comments (see Appendix H) and invited
discussion. Reviewers agreed that Section 8 of the document was confusing and that the draft RSC
Analysis ends abruptly. They recommended that a concluding paragraph or section be added
summarizing the outcomes and conclusions of the RSC Analysis. Discussion included the following
comments:
• Dr. Abbott referenced her written comments in response to Charge Question 7 (see pages F-
13 to F-18 of Appendix F). She noted that many of those comments also pertained to other
charge questions, and clarified that she did not have additional comments.
• Dr. Martinez Mier commented that the dental literature has made an effort to distinguish
between the terms "intake" and "exposure" for fluoride specifically. A document published
by CDC on the safe use of fluoride provides definitions for these terms, one being systemic,
the other topical. For fluoride via the oral route, "intake" pertains to ingestion, whereas
"exposure" pertains to what is put in the mouth but not ingested. Toothpaste is an example of
"exposure," but toothpaste that is swallowed becomes "intake." Dr. Martinez Mier
recommended that EPA clarify its use of these terms. Another reviewer agreed that use of the
term "intake" in the RSC Analysis was confusing; in particular, it was unclear whether it
referred to fluoride intake or also connected with food consumption.
• In reference to p. 79 of the draft RSC Analysis, two reviewers suggested it would be useful to
have an idea of how many children under 14 years of age are served by community water
systems and potentially impacted by fluoride in these systems. This context should be added
in one of the documents resulting from this process, perhaps the final management decision
document. It may be possible to estimate this from NHANES data by using the survey
weights to estimate how many children were exposed. Also, it would important to look at
actual beverage sources and not just assume that the liquid in all beverages came from the
reported drinking water system. The CSFII provides a great amount of detail about beverage
sources.
• Dr. Martinez Mier noted that there is some dental literature on the "halo effect," which
means that individuals who are not in a community with fluoridated drinking water can still
be exposed to fluoride by consuming products made in other communities that do have
fluoridated water. The literature does not quantify this amount, but mentions that it plays a
role.
24 Federal-Provincial-Territorial Committee on Drinking Water. 2009. Fluoride in Drinking Water
http://fluoridealert.org/canada.2009.report.pdf
15
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis "
• A reviewer noted that water fluoridation is considered one of the top ten public health
achievements in the last century, yet the draft RSC Analysis discusses potential health effects
associated with this achievement. She thought it important to consider the potential public
health implications associated with release of the RSC Analysis to the extent that it could
generate concern over fluoride in drinking water. Even though they are important, the public
health benefits of drinking water fluoridation are not within the document's scope. She
wondered whether the document should specify what percentage of the population served by
public water systems would potentially be affected by dental fluorosis. Dr. Donohue clarified
that this issue is acknowledged in the Dose-Response Report and that, legally, EPA cannot
endorse dental fluoridation.
• Section 8.2 (Estimates of Tolerable Upper Limit Exposures) is important, but very confusing
as written and needs to be clarified. In particular, the use of the term "margin of exposure" in
the draft RSC Analysis (see page 121) is potentially confusing to risk assessors, who use this
term in a different way. The document should avoid, or at least clarify, terms such as this that
can be construed differently by different specialists.
• Dr. Fox noted that interpretation of exposures over the RfD (as discussed on page 125) will
inevitably vary; however, in assessments of anthropogenic chemicals (where exposures are
involuntary and not nutritionally essential), interpretations vary across different parts of EPA,
and increased risk is not proportionate to dose or exposure increases. She noted that she had
provided EPA Office of Air definitions of "hazard index" and "hazard quotient" in her
written comments (see page F-21 of Appendix F).
• Dr. Ozsvath emphasized that Table 8-2 and Figures 8-1 through 8-3 need clarification and re-
wording and referred EPA to his written comments for details (see pages F-32 and F-33 of
Appendix F).
• Noting a written comment that use of mixed units was confusing, a reviewer clarified that
standardization of units, though desirable, may not always be possible, because the choice of
units is often driven by the analytical method used. Conversion is not possible, for example,
unless the specific homogenization process used by labs for each specific food is reported in
the original study. Use of mixed units is to some degree an outcome of the lack of
standardization of units in the food-fluoride literature. Another reviewer responded that
conversion of beverage units from mg/L to mg/kg (the units used for food) may be possible
by making an assumption about the specific gravity of the beverage.
This would inject another source of uncertainty, but would standardize the units, which
would benefit comparison.
• A reviewer found it difficult to follow whether the document was discussing daily exposure
(mg fluoride/day) or the amount of fluoride intake from food (mg fluoride/kg/day); the latter
should not be used as representative of what people are eating because it is not clear what
they have consumed. EPA should make clear that the document reports the units provided in
the original study.
Closing Discussion and Remarks
Dr. Donohue summarized the key messages the Agency had heard from reviewers during the
meeting:
16
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis "
• A reader's expertise affects how the RSC Analysis is read and understood. EPA should
therefore provide more clarifications (e.g., of terminology, studies, etc.), so readers with a
variety of backgrounds can better understand the document.
• Both the draft RSC Analysis and the Dose-Response Report currently lack an overview or
roadmap. The Agency will add forwards to both documents that explain the "big picture."
The forward for the RSC Analysis will clearly:
o State that the analysis relied exclusively on published studies from the U.S.
o Describe the limitations of the analysis
o State the objectives of the analysis
o Outline the critical steps in the analysis
o Describe how the objectives were met
• EPA understands the reviewers' concern about including a sensitivity analysis; however, a
sensitivity analysis of a 0.02 to 0.03 mg/kg/day difference between beneficial and apparently
adverse amounts of fluoride would likely simply show that the uncertainty values overlap.
This makes it difficult to define where beneficial becomes adverse and adverse becomes
beneficial. But the Agency can look at how choosing a different study would affect the
results of the RSC Analysis.
• Because people do not consume the same type or quantity of foods and beverages every day,
any number in a consumption study will not be fully representative. Drinking water
consumption is probably going to be more consistent than food consumption, but this is an
inherent difficulty in dealing with essential nutrients and predicting how much people need
or how much is too much. The Agency can try to improve the presentation in this part of the
report, but will always encounter that difficulty. Dietary risk assessment practices differ from
toxicology risk assessment practices.
• EPA will use consistent terminology or at least recognize differences in terminology among
disciplines.
• The Agency will rework the concluding section to describe what the next steps will be.
• EPA will move the discussion McClure data out of the RSC Analysis and into Dose-
Response Report.
Reviewers agreed this was generally a good reflection of key points they had made25, and added the
following comments and recommendations:
• The document needs to provide a clear justification for the selection of specific data sets and
critical studies, perhaps in the forward/overview that EPA will be adding at the beginning of
the RSC Analysis.
• The uncertainty discussion should be expanded.
25 In a post-meeting comment, Dr. Abbott noted that she did not agree that dietary risk assessment practices necessarily
differ from toxicology risk assessment practices, especially when the chemical in question is being regulated as a
contaminant and not as a dietary supplement.
17
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis "
• In Table 7-2, EPA should use different values in the boxes representing fluoride intake from
different sources to see how that affects the RSC.
• EPA should discuss how the characterization of beverages in a separate category from food
affects the use of the estimates of indirect drinking water consumption implicitly included in
the daily drinking water consumption reported in EPA's 2004 document26 and how these
estimates are used in the RSC calculation. The document should better clarify how the Pang
(1992)27 data were used, in particular: What constitutes a store-bought beverage or a
beverage prepared with de-ionized water? For example, is iced tea brewed in a restaurant or
store considered store-bought? This matters because the drinking water figure from 2004 in
the RSC Analysis includes indirect drinking water. However, beverages such as brewed iced
tea, that are made from water in a store, provide another source of fluoride from community
water supplies. In summary, EPA should clarify how the market basket approach relates to
commercially purchased beverages and whether it could include any beverages that might
have fluoride from a community water source.
• A reviewer asked why beverage intake for infants was listed as zero in the draft RSC
Analysis. EPA clarified that the market basket survey defined "juice" as a "fruit." The
reviewer recommended that EPA clarify this in the document.
• Dr. Martinez Mier noted that the University of Minnesota has developed a software program
to assess individual fluoride intake based on what, in an interview, an individual reports
having eaten. The software uses mean fluoride values from peer-reviewed laboratory
analysis, so could be helpful to EPA as a source of data on mean values of fluoride in various
types of food. Values are added to the database on an ongoing basis as new data become
available. Dr. Martinez Mier agreed to send a reference for this database to EPA.28
• A reviewer suggested that EPA examine the USDA fluoride database to see what it might
yield in terms of current consumption values and to get an idea of how much uncertainty was
associated with relying on older studies with few groups of foods versus data available in the
current USDA database.29
• A reviewer suggested that the document add the uncertainties associated with the likely
future increase of fluoride in food due to increased used of fluoridated water in preparing
food products.
• A table such as Table 7-2 could be incorporated into a sensitivity analysis to get an idea of
which of the numbers that have been estimated are most critical. Lacking that analysis, it is
hard to have perspective on how sensitive some of the possible changes in values the
26 U.S. EPA. 2004. Estimated Per Capita Water Ingestion and Body Weight in the United States - An Update. U.S.
EPA, Washington, D.C.
27 Pang, D.T.Y., C.L. Phillips, and J.W. Bawden. 1992. Fluoride intake from beverage consumption in a sample of North
Carolina children. J. Dent. Res. 71:1382-1388.
28 After the meeting, Dr. Martinez Mier send the following links to this database and software:
www.ars.usda.qov/SP2UserFiles/Place/12354500/Data/Fluoride/F02.pdf
http://www.ncc.umn.edu/products/databasenutrientsratioscomponents.html
29 In a post-meeting comment, this reviewer added that EPA should investigate using current fluoride concentration in
food data from the USDA's fluoride database in combination with more recent consumption data (from the CSFII or
NHANES databases) to provide a quantitative estimate of the difference between the values from the older studies and a
more current exposure estimate, at least for those food groups where both types of concentration data exist. This could
be accomplished by using a modeling approach like that used by EPA's Office of Pesticide Programs for sulfuryl
fluoride (use of the DEEM model and use of historical concentration data versus more current concentration data).
18
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis "
reviewers have discussed will be to the RSC. For example, the document ignores the amount
of fluoride in the atmosphere and this is probably appropriate. The importance of the soil
intake is not certain, but it may be relatively low compared to food.
Ms. Connery noted that reviewers would receive a copy of the draft summary report to check for
accuracy and completeness before it was finalized. She thanked the reviewers for their participation.
Dr. Donohue thanked reviewers for the diversity of their comments, noting that they had opened the
Agency's eyes to different perspectives and that had been very valuable.
19
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
Appendix A:
Peer Reviewers
A-l
-------
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis "
United States
Environmental Protection Agency
Office of Water
Peer Review Workshop of EPA's Draft Document
Fluoride: Exposure and Relative Source Contribution Analysis
Navy League Building
Arlington, VA
May 14, 2010
Reviewers
Linda C. Abbott, Ph.D.
Regulatory Risk Analyst
Office of Risk Assessment and
Cost-Benefit Analysis
U.S. Department of Agriculture
1400 Independence Avenue, SW
Washington, DC 20250
202-720-8022
labbott@oce.usda.gov
Mary A. Fox, Ph.D.
Assistant Professor
Department of Health Policy
and Management
Johns Hopkins Bloomberg
School of Public Health
624 N. Broadway - Room 455
Baltimore, MD 21205
443-287-0778
mfox@jhsph.edu
E. Angeles Martinez Mier, DDS, Ph.D. (Chair)
Associate Professer
Department of Preventive and
Community Dentistry
Indiana University School of Dentistry
Indianapolis, IN
317-274-8822
esmartintgiiupui.edu
David Ozsvath
Professor of Geology and Water Science
Department of Geography/Geology
University of Wisconsin-Stevens Point
Stevens Point, Wl 54481
715-346-2287
dozsvath@uwsp.edu
A-2
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
Appendix B:
Charge to Peer Reviewers
B-l
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
Technical Charge to External Peer Reviewers
Fluoride Exposure and Relative Source Contribution Analysis
BACKGROUND
Calculation of the health-based Maximum Contaminant Level Goal (MCLG) for noncancer endpoints
generally requires application of a Relative Source Contribution (RSC) to adjust for sources of exposure
other than drinking water that are not reflected in the Point of Departure (POD) for the MCLG derivation.
In the case of fluoride, the POD is a drinking water concentration from 1942 for a 0.5% incidence of
severe dental fluorosis, characterized as displaying confluent pitting in the enamel of at least two
permanent teeth.
This document uses data published in the peer-reviewed literature to estimate dietary fluoride in the
1940's as well as current total exposures from plain (direct and indirect) drinking water, solid foods,
beverages, toothpaste, and soils. The recent exposure data are used to determine the current relative
source contribution of fluoride in drinking water to total exposure for the 90th percentile drinking water
intake (EPA policy).
CHARGE QUESTIONS
1. Please describe any suggestions you have for improving the clarity, organization, and/or
transparency of the draft document.
2. Have the uncertainties associated with the analysis been adequately characterized? Are there any
important uncertainties in the data that are not discussed adequately in the document, especially
in the synthesis sections? Please describe any concerns you have and any specific suggestions for
improving or enhancing the uncertainty discussion.
3. Please consider the studies that have been selected as representative of exposures for the specific
age groups and/or exposure media. Have these studies been adequately summarized and
interpreted? Indicate any deficiencies in the descriptions of the studies and any suggestions you
have for improvement. Describe any concerns you have about the selection of these studies, as
well as any recommendations you may for alternative studies that you believe are more
representative of exposures.
4. Please comment on EPA's rationale for selection of specific data elements to represent average
exposures for each of the age groups. Has the selection been scientifically justified and clearly
and objectively described? What changes or improvements would you suggest?
5. Please comment on the validity of basing the food intake estimate for the 1940's on the McClure
(1943) publication, as supported by the concentrations found in various food groups from more
recent analytical data. Do you agree with this approach? If not, what approach would you suggest
for estimating food intake for the 1940's?
6. Provide citations (and, where possible, pdfs or hard copies) for any references you suggest EPA
should consider adding to the document, and describe where you suggest these references be
added.
B-2
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
1. Please provide any additional comments and/or further suggestions you may have for improving
the document.
B-3
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
Appendix C:
EPA Slides Presented
During Teleconference
c-i
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
Exposure and Relative Source
Contribution for Fluoride
Document Background
Peer Reviewer Presentation
April 20, 2010
Do Not Cite, Quote or Distribute - Internal EPA Deliberative Document
Regulatory History for Fluoride
« 1986 - Existing Drinking Water Standards
-i Established MCLG /MCL of 4.0 mg/L to protect against crippling (stage
3) skeletal fluorosis
-i Assumed all exposure comes from drinking water (i.e., a 100% relative
source contribution or RSC*)
-i Set Secondary MCL (SMCL) at 2.0 mg/L to protect against objectionable
(moderate/severe) dental fluorosis (then considered a cosmetic effect)
« 2003 - Finalized "first" review of drinking water standards
-i No revision appropriate at that time; requested National Academies of
Science National Research Council (NRQ to examine current standards in
the light of new data since 1986 regulation and a 1993 NRC report
« 2010 — Finalized "second" review of drinking water standards
_j Revisions to MCLG and MCL will be considered when the ongoing dose-
response and relative source contribution assessments are complete
*Note: As the exposure sources from other media increase, the RSC value for drinking water decreases. This lowers
the MCLG. The current MCLG assumes a 100% RCS for drinking water.
C-2
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
Six Year Review of Fluoride Regulations
(2002/2003)
• Required by the 1996 Safe Drinking Water Act Amendments
• Reviewed new scientific data related to the fluoride MCLG
and MCL and identified:
j Many studies of the effects of fluoride on bone published after the
EPA- requested 1993 NRC review of fluoride
_j Comprehensive recent reviews by WHO, England (2), ATSDR
_i Institute of Medicine (IOM) dietary intake guidelines and tolerable
upper limit values (2001)
• EPA recommended fluoride for a comprehensive review by
NRG
Do Not Cite, Quote or Distribute - Internal EPA Deliberative Document
NRC Charge and the 2006 Report
2003 EPA Charge to NRC
j Review recent health/sources of exposure data; Evaluate basis for MCLG/MCL and
SMCL; Advise EPA on adequacy of MCL and SMCL to protect children and others from
adverse effects; and identify data gaps/research needs
March 2006 NRC Report
j Dental Fluorosis - Most panel members concluded "severe dental fluorosis" is an
adverse effect (due to thinning/pitting of the tooth enamel which increases the risk for
cavities); MCL does not adequately protect against this effect
j Skeletal/Bone effects - MCL may not protect against bone fractures
j Other Effects - Human and animal data limited on endocrine and neurodevelopmental
effects; research needed
j Cancer - Evidence tentative and mixed (most studies look at bone cancer); Wait for
publication of two Harvard osteosarcoma (bone cancer) studies*
j Advice to EPA - update the dose-response assessment, consider susceptible populations,
characterize uncertainties/variability, and update the exposure assessment (i.e., the
relative source contribution)
Do Not Cite, Quote or Distribute - Internal EPA Deliberative Document
C-3
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
NRC (2006) Caries Conclusions
« Severe dental fluorosis is characterized by discrete and
confluent pitting which constitutes enamel loss
« Severe enamel fluorosis compromises the health protective
function of enamel and is consistent with the prevailing risk
assessment definition of an adverse health effect
j "Severe enamel fluorosis may increases carries risk by reducing the
thickness of the protective enamel layer and by allowing food and
plaque to become entrapped in the enamel defects"
« Data from 11 out of 14 comparisons of degree of fluorosis
with cavity incidence support the hypothesis that severe
enamel fluorosis is associated with an elevated incidence of
cavities when compared to lower levels of fluorosis
Do Not Cite, Quote or Distribute - Internal EPA Deliberative Document
U.S. EPA Action Plan
Three Documents
Dose-Response Analysis for Severe Dental Fluorosis and Skeletal Effects
j Fluoride in Drinking Water (NRC, 2006) provides hazard identification
j Peer reviewed - March 2008
j Post Peer Review Document completed
j Establishes a point of departure for severe dental fluorosis of 1.87 mg/L F
j Inadequate data to support a dose-response analysis of the skeletal effects.
» Protection against severe dental fluorosis likely to be protective for skeletal
effects
Relative Source Contribution
j U.S. EPA internal review completed in January 2010
j Peer Review May 14, 2010
Relationship of Fluoride to Cancer
j To be initiated after publication of Harvard study by Douglas as
recommended by NRC (2006) - Study not yet published
Do Not Cite, Quote or Distribute - Internal EPA Deliberative Document
C-4
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
Benchmark Modeling of the Dean Data
Dichotomous-Hill Model with 0.95 Confidence Level
13:21 08/13 2008
0.5% response: BMD = 2.14 mg/L; BMDL = 1.87mg/L
Revised from 1% based on peer review recommendations.
Do Not Cite, Quote or Distribute - Internal EPA Deliberative Document
Fluoride Concentration vs. DMFT /DMFS
I
FluorkJe (ppm)
[•olid circle* - DMFT: opffn bo KM = DMF&]
Do Not Cite, Quote or Distribute - Internal EPA Deliberative Document
C-5
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
EPA Analysis of the Fluoride Concentration and
Cavity Prevalence Relationship
« Data provide some support for the hypothesis that the enamel
defects in severe dental fluorosis increase the prevalence of cavities
compared to those with mild to moderate fluorosis
j Differences are minimal in some cases
« Cavity prevalence where there is no or minimal exposure to fluoride
is greater than that found in concert with severe dental fluorosis.
« School age children with drinking water concentrations of S4 ppm
F had a higher cavity prevalence than those in systems with 2 ppm F
but a lower prevalence than those in systems with 1 ppm F.
j The cavity data are not suitable for dose response modeling
» EPA analysis identified the increase in cavities as the adverse health
effects and used the dose-response for severe dental fluorosis as a
surrogate for the cavities
Do Not Cite, Quote or Distribute - Internal EPA Deliberative Document
Changes in Fluoride Exposure: 1940 vs.
2010
• There are many sources of exposure to fluoride that did not
exist at the time of the Dean (1942) analysis
_j Artificially fluoridated water
j Commercial baby formula (powdered and concentrate)
j Fluoridated toothpastes and mouthwashes
_j Sulfuryl fluoride pesticide (registered 2004)
• Dental fluorosis has increased
j 1940: ~ 10%
u 1987:23%
j 199902002: 32%
j Comparable data for severe dental fluorosis are not available
C-6
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
Categorical Regression Analysis
with Date of Study as the Variable
Figure 6-Fluorine Meta-analysis; ERC1 Line (SEV3); 90% 2-sided confidence, Link=Probit
Data demonstrate that the water concentration associated with a 1% increase in severe dental fluorosis decreases
with the date of the study. This is supportive of the conclusion that other sources of exposure have increased.
Do Not Cite, Quote or Distribute - Internal EPA Deliberative Document
Importance of the RSC
The RSC is applied to the Point of Departure for
severe dental fluorosis to identify the MCLG
The lower the RSC, the lower the MCLG
a Dental fluorosis only develops during pre-eruptive tooth
formation
a 0.5 years to 14 years is the period of vulnerability for
permanent teeth
a Includes development of third molars (wisdom teeth)
C-7
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
Appendix D:
Meeting Agenda
D-l
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis "
United States
Environmental Protection Agency
Office of Water
Peer Review Workshop of EPA's Draft Document
Fluoride: Exposure and Relative Source Contribution
Analysis
Navy League Building
Arlington, VA
May 14, 2010
Agenda
8:00 a.m. Registration/Check-in
8:30 a.m. Welcome, Introductions, Meeting Purpose & Agenda Jan Cannery, ERG
8:40 a.m. EPA Welcome Remarks Eric Burneson, Chief, Targeting & Analysis Branch, EPA/OW
8:45 a.m. Background Presentation Joyce Donahue, EPA/OW
9:05 a.m. Reviewer Discussions E. Angeles Martinez Mier (Chair) & Panel
1) Describe any suggestions you have for improving the clarity, organization, and/or
transparency of the draft document.
9:25 a.m. 2) Have the uncertainties associated with the analysis been adequately characterized?
Are there any important uncertainties in the data that are not discussed adequately in
the document, especially in the synthesis sections? Please describe any concerns you
have and any specific suggestions for improving or enhancing the uncertainty discussion.
10:00 a.m. BREAK
10:15 a.m. 3) Please consider the studies that have been selected as representative of exposures for
the specific age groups and/or exposure media. Have these studies been adequately
summarized and interpreted? Indicate any deficiencies in the descriptions of the studies
and any suggestions you have for improvement. Describe any concerns you have about
the selection of these studies, as well as any recommendations you may for alternative
studies that you believe are more representative of exposures.
10:50 a.m. 4) Please comment on EPA's rationale for selection of specific data elements to represent
average exposures for each of the age groups. Has the selection been scientifically
D-2
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis "
justified and clearly and objectively described? What changes or improvements would
you suggest?
Agenda (cont.)
11:30 a.m. 5) Please comment on the validity of basing the food intake estimate for the 1940's on
the McClure (1943) publication, as supported by the concentrations found in various food
groups from more recent analytical data. Do you agree with this approach? If not, what
approach would you suggest for estimating food intake for the 1940's?
12:15 p.m. LUNCH
1:15 p.m. 6) Provide citations (and, where possible, pdfs or hard copies) for any references you
suggest EPA should consider adding to the document, and describe where you suggest
these references be added.
1:30 p.m. 7) Please provide any additional comments and/or further suggestions you may have for
improving the document.
2:00 p.m. BREAK
2:15 p.m. Reviewer Conclusions & Recommendations E. Angeles Martinez Mier (Chair) & Panel
3:45 p.m. Closing Remarks Jan Cannery & EPA/OW
4:00 p.m. ADJOURN
D-3
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
Appendix E:
Meeting Observers
E-l
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis "
SEPA
United States
Environmental Protection Agency
Office of Water
Peer Review Workshop of EPA's Draft Document
Fluoride: Exposure and Relative Source Contribution Analysis
Navy League Building
Arlington, VA
May 14, 2010
Observers
Jan Connery (facilitator)
Eastern Research Group, Inc.
110 Hartwell Avenue
Lexington, MA 02421
781-674-7322
j an. connerv@erg. com
Michael Doherty
Office of Pesticide Programs
U.S. Environmental Protection Agency
Washington, DC 02460
703-305-1031
doherty.michael@epa.gov
Joyce Donohue
Office of Water, Office of Science and Technology
Health and Ecological Criteria Division
U.S. Environmental Protection Agnecy
Washington, DC 20460
202-566-1098
donohue.jovce(@,epa.gov
Tina Duke
Office of Science and Technology
Health and Ecological Criteria Division
U.S. Environmental Protection Agnecy
Washington, DC 20460
duke.tina@epa.gov
Martina McPherson (notetaker)
Eastern Research Group, Inc.
110 Hartwell Avenue
Lexington, MA 02421
781-674-7205
martina.mcpherson@erg.com
Dennis Opresko
Toxicology and Hazard Assessment
Oak Ridge National Laboratories
Oak Ridge, TN
Kathleen Souweine
Office of Ground Water and
Drinking Water
U.S. Environmental Protection Agnecy
Washington, DC 02460
202-564-4997
souweine.kathleen@epa.gov
E-2
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
Appendix F:
Reviewer Post-Meeting Comments
F-l
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
Reviewer Biographies
Linda C. Abbott, Ph.D.
Regulatory Risk Analyst
Office of Risk Assessment and Cost-Benefit Analysis
U.S. Department of Agriculture
Dr. Linda Abbott is a risk assessment modeler with USDA's Office of Risk Assessment and Cost-
Benefit Analysis (ORACBA) where she reviews regulatory risk assessments, provides advice on risk
assessments and collaborates with other USDA agencies to identify areas where risk assessment could
produce valuable information for regulatory decision makers. She reviews and provides guidance on
human health risk assessments, dietary exposure assessments, and ecological risk assessments on
conservation practices, agricultural chemicals and invasive species. She is also interested in the risks and
benefits associated with agricultural nanotechnology applications and recently served on the Joint Food
Agriculture Organization - World Health Organization Expert Meeting on Application of
Nanotechnologies in the Food and Agriculture Sectors: Potential Food Safety Implications. She served as
a Food Quality Protection Act Science Review Member of the Environmental Protection Agency's
Federal Insecticide, Fungicide, and Rodenticide Act Scientific Advisory Panel on Selected Issues
Associated with the Risk Assessment Process for Pesticides with Persistent, Bioaccumulative and Toxic
Characteristics. Prior to joining ORACBA, Dr. Abbott was an ecologist at USDA's Animal and Plant
Health Inspection Service, where she conducted ecological risk assessments on the application of
chemical or biological pesticides and analyzed the environmental impacts of various APHIS programs
involving the introduction of biocontrol agents, exclusion of foreign animal diseases and plant pests, and
eradication or suppression of introduced pests or disease. Dr. Abbott has a Ph.D. in Biology-Ecology
from Utah State University.
Mary A. Fox, Ph.D.
Assistant Professor
Department of Health Policy and Management
Johns Hopkins Bloomberg School of Public Health
Dr. Mary Fox is Assistant Professor in the Department of Health Policy and Management at the Johns
Hopkins Bloomberg School of Public health. She teaches Methods in Quantitative Risk Assessment
featuring probabilistic exposure assessment methods for the Johns Hopkins Risk Sciences and Public
Policy Institute's Certificate Program. Dr. Fox's research is focused on quantitative human health risk
assessment as a part of environmental policy making, particularly approaches to cumulative and chemical
mixtures risk assessment. Dr. Fox is currently serving on the IOM Committee on Long-term health
consequences of exposure to burn pits in Iraq and Afganistan and served on the NRC Committee on the
Health Risks of Phthalates. Dr. Fox received her MPH from the University of Rochester School of
Medicine and Dentistry and PhD from the Johns Hopkins Bloomberg School of Public Health. Dr. Fox
began her public health career conducting community health studies around hazardous waste sites as a
Research Scientist in the New York State Department of Health.
F-2
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
E. Angeles Martinez Mier, DBS, MSB, Ph.B.
Associate Professor
Department of Preventive and Community Dentistry
Indiana University School of Dentistry
E. Angeles Martinez Mier, DBS, MSB, PhB is an Associate Professor (tenured) in the Bepartment of
Preventive and Community Bentistry at Indiana University School of Bentistry. She serves as director of
the Fluoride Research Program within her department. Br. Martinez Mier previously held the positions of
Preventive Bentistry Bepartment Chair and Coordinator of undergraduate periodontics at the Universidad
Intercontinental in Mexico City, Mexico. After receiving her dental degree from the Universidad
Nacional Autonoma de Mexico, in Mexico City, Mexico, in 1989, Br. Martinez Mier completed a
Masters of Science in Bentistry, majoring in Preventive Bentistry at Indiana University School of
Bentistry, in 1994; and a three-year clinical fellowship in Periodontics, also at the Indiana University
School of Bentistry in 1995. She then obtained a PhB in Bental Sciences in 2000 from Indiana
University. Br. Martinez is the past president of the Indiana Chapter of the Hispanic Bental Association
and is a member of the International Association of Bental Research, the American Bental Education
Association, the American Association of Public Health Bentistry and the European Organization for
Caries Research. She has served as consultant for the Bivision of Oral Health of the Centers for Bisease
Control and Prevention. She was a member of the CBC Expert Panel on Methods to Analyze Fluoride for
the Surveillance of Total Fluoride Intake, and a member and organizer of the CBC Expert Panel on
Methods to Betect and Quantify Enamel Fluorosis for Surveillance Purposes. She currently serves as the
Chairperson for the American Association of Bental Research Science Information Committee.
Br. Martinez Mier's research has been funded by grants from the National institute of Bental and
Craniofacial Research, Clarian Health, Belta Bental Insurance, the Borrow Foundation and the West
Foundation. Her research projects have received multiple awards. Her research interests are mainly in two
areas: fluorides and community health. In the area of fluorides, she is particularly interested in refining
analytical methodologies to accurately determine fluoride presence and in the role of fluoride in the onset
and progression of demineralization and remineralization. She is also interested in developing
methodologies to diagnose dental fluorosis through clinical, photographic and digital imaging
evaluations, including dental fluorosis and incipient caries differential diagnosis and in better defining the
mechanisms by which fluoride affects the development of dental and skeletal fluorosis, including genetic
and environmental factors. Br. Martinez Mier is engaged in community-based research in designed to
address and identify disparities in dental caries and dental fluorosis in Latino/Hispanic patients.
F-3
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
David L. Ozsvath, Ph.D.
Professor of Geology and Water Science
Department of Geography/Geology
University of Wisconsin-Stevens Point
EDUCATION
Ph.D. Geology (1985), The State University of New York - Binghamton
M.S. Geochemistry (1979), The Pennsylvania State University
B.A. Geology (1976), The University of Vermont
PROFESSIONAL EXPERIENCE
1987 to present Department of Geography/Geology, University of Wisconsin-Stevens Point.
1986 to 1987 Visiting Professor, Department of Geology, Lafayette College, Easton, PA.
1983 to 1986 Geochemist / Hydrogeologist, O'Brien & Gere Engineers, Syracuse, NY.
1981 to 1983 Geologist / Hydrogeologist, Coates & Associates, Inc., Endicott, NY.
1979 to 1983 Teaching and Research Assistant, Department of Geological Sciences,
State University of New York-Binghamton, Binghamton, NY.
1978 to 1979 Geologist, Malcolm Pirnie Inc., Consulting Engineers, White Plains, NY.
RELEVANT PUBLICATIONS
2009 Ozsvath, D.L., Fluoride and Environmental Health: a Review; Reviews in Environmental
Science and Biotechnology, Vol. 8, No. 1, p. 59-79. (invitedpaper)
2007 Ozsvath, D.L., Natural Controls on Fluoride Concentrations in a Fractured, Crystalline
Bedrock Aquifer; 30th Annual Applied Geography Conferences, Indianapolis, IN.
2006 Ozsvath, D.L., Fluoride Concentrations in a Crystalline Bedrock Aquifer, Marathon
County, Wisconsin, Environmental Geology, vol. 50, no. 1, p. 132-138.
2006 Ozsvath, D.L., Fluoride in a Sole Source, Crystalline Bedrock Aquifer, Marathon County,
Wisconsin; Geological Society of America Annual Meeting, Philadelphia,
PA, http://gsa.confex.com/gsa/2006AM/fmalprogram/abstract 105521.htm.
F-4
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
Reviewer Post-Meeting Comments
Linda C. Abbott, Ph.D.
1. Please describe any suggestions you have for improving the clarity, organization, and/or
transparency of the draft document.
The document under review - Fluoride: Exposure and Relative Source Contribution Analysis -
should be reorganized to provide an overview of the question it seeks to address and the methodology
by which it does so. The impact of the Dean (1942), McClure (1943) and Ershow and Cantor (1989)
should be explained early in the document. Providing this overview at the beginning would put the
other material in context when it is encountered. As the document is currently organized, you don't
fully realize the importance of these older studies until after you have read about them. The paragraph
at the bottom of page 2 onto the top of page 3 is insufficient notice of the significance of the older
studies and the manner in which they must be used to reconstruct the exposure in the historical
period. The document does not fully explain why the calculation of the relative source contribution
(RSC) must rely on historical data rather than using more current exposure data for foods and
beverages.
The document would be strengthened by including a "road map" at the beginning describing the usual
process for establishing the relative source contribution (RSC) and the adaptations to that process that
were required in order to assess fluoride. Including the equation to calculate the RSC in the first section
would help set the stage for what follows. A flow chart depicting the types of information usually
required and the source of the information for atypical chemical versus fluoride would help the reader
appreciate the approach used and the information that follows. As currently organized, it is unclear why
there is an encyclopedic description of fluoride in foods from the 40's and 50's through the current time
until chapter 6.
The first section should also provide an overview of the unique challenges associated with setting the
RSC for fluoride. From my initial reading of the document these challenges appear to include: 1)
dose-response data used in the proposed RfD are from an era where the proportion of fluoride from
food, water, soil and dental products differs from the relative proportion attributed to these sources
today; 2) historical data estimating fluoride exposure through food and water may not reflect a
nationally representative estimate; 3) historical data do not separate fluoride exposure from food
from exposure from drinking water; 4) historical data frequently do not disclose actual food amounts.
The historical may not disclose the food items considered, the fluoride concentrations of individual
foods and cannot be evaluated to determine whether the historical exposures are representative of the
national population; 5) current residue data are lacking in many foods and beverages; 6) fluoridation
levels in community water supplies have varied over the historical data period; and 7) there has been
an increased use of fluoride in water and dental products during the period of historical data resulting
in potentially higher fluoride concentrations in 2010 foods and water than in historical data sets.
The first section should also provide an overview of how the data discussed in the document are to be
used in establishing a MCLG or MCL. In section 1.2, the types of data used to estimate the MCLG
are discussed. Units for each of the variables should be supplied. The relationship between the MCL
and the MCLG should also be discussed here. The equation in section 1.2 does not indicate that the
EPA will use body weight data other than that of the 70 kg adult. The MCLG or MCL is expressed in
mg/L while the RfD is expressed in mg/kg/d. Some discussion of the relationship between body
weight, drinking water intake and the resulting MCLG would help the reader interpret the potential
significance of selecting various data sources for age-specific drinking water intake, body weight or
F-5
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
food consumption. It is unclear how the age-specific fluoride exposure estimates in sections 7 and 8
will be used in establishing the MCL or MCLG. An overview of this process in section 1 would be
helpful. It is unclear whether in establishing the MCL/MCLG only water consumption from
community water supplies is considered. The USDA CSFII and the NHANES survey provide
information on the source of the drinking water for each respondent. In the CSFII, there are several
possibilities - community water supply, spring water, bottled water, well water or an unknown
source. Does the MCL address drinking water intake from only one, several or all of these sources?
This should be explained in section 1.
It is assumed that the DWI values reported in Table 6.3 include both direct and indirect community
water in the drinking water consumptions (mL/day) and do not reflect other water sources. The EPA
reference cited for Table 6.3 (EPA, 2004 - The Estimated Per Capita Water Ingestion and Body
Weight in the United States - An Update) estimates that for children less than two years, indirect
water consumption is greater than direct water consumption in the consumers only population that
consumes community water (Part IV, Table A2, p. E-144). For other age groups, the indirect water
consumed relative to the total direct plus indirect is not an insignificant amount. The EPA estimated
per capita drinking water document (EPA, 2004) defines indirect water as "water added to foods and
beverages during final preparation at home, or by food service establishments such as school
cafeterias and restaurants" and defines beverages to include milk, soft drinks, tea, coffee, juices as
well as many others listed in Figure 3.4 of EPA, 2004. Beverages include reconstituted juices, instant
coffee, coffee and tea brewed at home with water added at home or in a food service environment.
The indirect water used to reconstitute these beverages or create them from dry powders is considered
in the indirect water consumption reported in EPA, 2004. For example, the community water intake
for a subpopulation such as children from 6 months to less than 1 year in age includes the indirect
water used to reconstitute infant formula. The community water intake for children in the age range
from 1 to 3 years includes indirect water used to prepare tea, lemonade, and powdered drinks. The
discussion in section 6.1 of the document under review under the beverages subtitle relies heavily on
data from Pang et al (1992), Jackson et al (2002) and Singer et al (1985) to estimate fluoride
concentration in beverages. It is not clear from the presentation in the review document that double
counting of fluoride from indirect water has not occurred by relying on samples in these studies that
may have been reconstituted using tap water or some other water source that was not free from
fluoride. In that case, fluoride in the beverage source category that is the result of indirect water
addition would add to the fluoride from all sources other than drinking water and underestimate the
amount available for drinking water. Conversely, it is also unclear whether powdered beverages
which contain fluoride absent any addition of indirect drinking water are adequately considered by
the methodology used to derive Table 6.2. Further discussion of this would improve this part of the
document.
The document does not provide an estimate for the RSC. After reading the document, you do not
know what the estimate of the RSC is or how it might be applied to different age-classes. Many
different age-specific estimates are provided in section 7, but it is not clear how or whether these
estimates will be combined. It seems unlikely that there will be many different MCL values for the
same chemical. It may be that the estimate for the most sensitive population group as identified in the
dose-response document might inform this, but it is not possible to determine from the text of this
review document. Sections 7 and 8 should be expanded to address this issue.
A table summarizing the information given in sections 2 and 3 would allow the reader to quickly scan
the various sources of data. Relating these data to the calculation of the variables in the RSC equation
on page 114 would be especially useful. Moving this equation to the beginning of the document -
perhaps section 1- would provide the reader with a quick overview to the types of information that
will be needed to address the primary question posed by the document. Organizing the subsequent
F-6
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
chapters to begin with a reiteration of the equation - highlighting the data that chapter provides -
would provide a framework for discussing the material where the reader can immediately understand
the relevance of that chapter to the overall goal.
The use of a variety of different units creates an obstacle which makes easy comparison between
tables, the text and figures impossible. If possible, conversion of the values into the same unit would
highlight where differences may be found. The relationship between the RfD (expressed in mg/kg/d,)
MCL/MCLG (expressed in mL/d,) and fluoride exposure in mg/d is especially important to clarify.
The document presupposes knowledge of the contents of U.S. EPA (2009a). The discussion in
section 8 on the Dean and McClure studies and their relevance is particularly difficult to fully
understand without this underlying document. Perhaps the final relative source contribution analysis
will combine U.S. EPA (2009a) with this document and there is not a great need to repeat information
from that document here. But if this is to be a stand-alone document, much further discussion of the
contents and rationale used in EPA (2009a) is required. It would be useful for peer reviewers to have
copies of some of the critical foundational documents upon which the review document is based.
With an abbreviated time period in which to conduct a peer review, too much time was spent located
the underlying studies.
The Reference Dose (RfD) is not fully explained at an early point in the document. The OW
proposed RfD (first identified on page 120) is 0.08 mg/kg/d. The last paragraph on p. 120 describes
the derivation of the proposed RfD from "the 95th percentile lower bound drinking water (1.87 mg/L)
associated with a 0.5% prevalence of severe dental fluorosis in the population studied by Dean
(1942)." This information should appear much earlier in the document to alert the reader to its
significance when first presented with the data from the Dean study. The Dean study is mentioned
substantively for the first time on p 71, again on p. 74 before its mention on p. 120. Due to the
significance of the Dean study in the estimation of the RfD and the use of the scoring, it should be
described much earlier.
2. Have the uncertainties associated with the analysis been adequately characterized? Are
there any important uncertainties in the data that are not discussed adequately in the
document, especially in the synthesis sections? Please describe any concerns you have and
any specific suggestions for improving or enhancing the uncertainty discussion.
The section on uncertainty (§6.5) discusses many important sources of uncertainty but does not
attempt to quantify the effect of these sources of uncertainty on the estimates used in the RSC
estimation. The document states that in recognition of these uncertainties the values selected for the
various sources of fluoride intake were average to slightly above average. It is unclear how the
average fluoride intake was determined in order to select values that were average or above average
and whether average refers to the average intake value from across the available studies or some other
estimate of expected average exposure. Without actually determining the "average" intake, it is
unclear how the agency selected studies to ensure that at least the average intake is demonstrated by
the representative studies. Given the degree to which the analyses rely on older historical data from
studies of populations that may not be representative of the current food and water consumption rates
for the U.S. population, more should be done to assess the uncertainty of relying on these studies. A
quantitative estimate of uncertainty, at least for some of the individual studies used for food intake,
could be conducted and would demonstrate some of the uncertainty associated with the representative
studies. The uncertainties resulting from using the residue data without adjusting for bioavailability
of fluoride in certain types of foods (bones, seafood) and perhaps from the toothpaste-related fluoride
intake could be addressed quantitatively as well.
F-7
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
The uncertainty about the RSC values for each age class (Table 7.1) could be estimated by
incorporating the uncertainty about each intake source (e.g., food, drinking water, soil, etc.) into an
error analysis if the fluoride exposure for each intake is the average expected value and the
distributions are normal. Perhaps a sensitivity analysis for the variables in the RSC equation could
inform the agency of which variables most influence the RSC value and then conduct an uncertainty
analysis about the estimation of intake for those highly influential variables.
There should also be clarification of whether the use of sulfuryl fluoride is expected to increase, as a
relative share of the percent of the crop treated. With the reduction in methyl bromide use and the
restrictions on telone and iodomethane in some states and prospects for phosphine as a substitute for
methyl bromide, it could be possible that sulfuryl fluoride use may increase. Some discussion of the
dynamic nature of the fumigant market might provide additional information to qualitatively discuss
uncertainty with respect to future exposure levels due to sulfuryl fluoride.
There are uncertainties about the use of the USDA Continuing Survey of Food Intakes by Individuals
(CSFII) two-day water consumption values for the youngest age class considered. The EPA's
Estimated per Capita Water Ingestion and Body Weight in the United States—An Update is based on
data collected by the United States Department of Agriculture's 1994-1996 and 1998 Continuing
Survey of Food Intakes by Individuals and does not take into account the age of the respondent at
both dates in the CSFII. For many of the younger age classes, an error may be introduced if it is
assumed that the individual is in the same age class for both of the sampling dates. This is especially
important if the age classes are divided into months, reducing the number of respondents per age
class. For example, some of the respondents in the under one month category for the first day of the
survey are in the next category (one month old) or even the category after that for the second day.
Both survey dates are recorded in the CSFII and some dietary data can clearly be identified as being
associated with the next survey age class through subtraction of the early date (day 1) from the later
date (day 2). For example, seven respondents consuming community water in the less than one month
age category have more than 30 days (more than 1 month) between day 1 and day 2 survey dates. It
can be assumed that day 2 for these respondents does not represent the water consumption rate of an
under one month old. These seven drinking water consumptions are among the highest for that age
class. While for fluoride the youngest age class spans 6 months and probably contains more
respondent records than the less than one month category described above, there is still the chance
that some of the records included actually are records of respondents who are over one year in age.
The drinking water estimate for this youngest age class could be overestimated if those
inappropriately combined respondents consume more drinking water per day than those who belong
in this category. For at least this youngest age class, the total, direct and indirect drinking water
consumption estimates should be examined to determine whether inappropriately combined
respondents might be biasing the estimate.
The DWI (Table 7-1), based on the 90th percentile intake with average 0.87 mg F/L, reflects
consumers only. This intake level, while conservative, may overestimate exposure. The 90th
percentile consumption for the entire population and the mean consumption level should also be
examined in the RSC equation. Comparing the three calculated sets of RSC values will provide
some information on how much error may be incorporated by using a consumers-only 90th percentile
consumption to estimate drinking water exposure with the average general population exposure for
the other components of the RSC equation.
F-8
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
3. Please consider the studies that have been selected as representative of exposures for the
specific age groups and/or exposure media. Have these studies been adequately summarized
and interpreted? Indicate any deficiencies in the descriptions of the studies and any
suggestions you have for improvement. Describe any concerns you have about the selection
of these studies, as well as any recommendations you may for alternative studies that you
believe are more representative of exposures.
In general, there is inadequate information about the studies (for food) underlying the exposure
estimates to evaluate their adequacy. There is little or no discussion of the consumption rates for
foods (mg food per day or per body weight). Without further data, it is difficult to determine the
representativeness of a study that involves only a few individuals or one sex (Singer et al., 1980; San
Fillipo and Battistone (1971)) or few geographic areas (Pang et al, 1992; Jackson 2002; Singer et al,
1980; San Fillipo and Battistone (1971)) to use it as a proxy for the national population even for a
specific national subpopulation.
It is not clear why representative exposure data from past studies is used. An explanation of why it is
not possible to estimate the current fluoride exposure from food should be clearly stated. It may be
that the agency finds the current database of fluoride concentrations in food inadequate to estimate
fluoride intake through food for the U.S. population. Such an explanation is necessary because the
most of the studies selected to represent fluoride intake through food for the various age categories
required by the agency are not representative of the U.S. population, provide little information on
actual food intakes of the subjects and appear to conflate fluoride concentration in foods with food
intake rates by reporting only the fluoride exposure estimates (mg fluoride/day) for the small
populations they study.
As an alternative approach, it may be possible to construct a method of extrapolating from the foods
for which fluoride concentration data exist (USDA, 2009) to the U.S. population and various
subpopulations of interest using the CSFII average consumption rates for various foods or food
groups. The FDA Total Diet Study uses a similar approach to extrapolate from the 260+ foods
analyzed in the Total Diet Study (TDS) market baskets to the U.S. population. Using a similar
approach here would require comparing the existing data base for which fluoride concentrations in
foods exist and making some decisions about which other CSFII foods could be represented by the
foods that have been analyzed. This approach has advantages to relying on historical data to estimate
fluoride exposure through food. First, the consumption rates used are nationally representative of
recent trends in consumption. (A similar approach could be done using the most recent NHANES
dietary survey, but to my knowledge there is not already a 1:1 mapping between the various foods in
the survey as there is for the CSFII through the TDS tools.) Second, the most recent fluoride
concentrations in food items could be used. Third, this approach would not conflate the historical
fluoride concentrations in food (and possibly indirect water) with unknowable consumption rates to
arrive at an exposure estimate that may be more reflective of a unique subpopulation at an earlier time
in history than today's population. Fourth, because the underlying data (consumption rates and
fluoride concentrations in food) are transparent and available, a more sophisticated uncertainty
analysis for the food exposure for each age class could be conducted.
The Food Commodity Intake Database provides recipe translations which could be used to estimate
both indirect water consumption and direct water consumption by the U.S. population. If the current
measurements on fluoride in food could be mapped to those foods not measured as done in the TDS
survey, it may be possible to estimate exposure given current consumption patterns. The EPA Office
of Research and Development (ORD) is presenting their Stochastic Human Exposure Dose
Simulation model (SHEDS) dietary model before the FIFRA Scientific Advisory Panel this July.
F-9
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
This model will allow the use of residues in foods (like TDS residues in foods) as well as estimation
of residues in foods from residues in raw agricultural commodities in the same model run. Perhaps
the Office of Water could collaborate with ORD to use this new model to estimate the dietary (food
and beverage) and water exposure to fluoride.
4. Please comment on EPA's rationale for selection of specific data elements to represent
average exposures for each of the age groups. Has the selection been scientifically justified
and clearly and objectively described? What changes or improvements would you suggest?
First, the selection of the age groups needs further discussion. It is unclear why the specific age
groups need to be assessed. Section 2.5.5 states these age classes are equivalent to those used in the
dose-response document (EPA, 2009a) but it is not until the next page that the document discusses the
need to organize the data into age classes equivalent to those from Ershow and Cantor (1989) because
[Ershow and Cantor?] report drinking water intake data for the period closest to that of the Dean
study (1942). Even in section 2.5.5, the significance of the drinking water intake in Dean (1942) is
not explained. It is not until page 120 in section 8.2 that the significance of the Ershow and Cantor
study begins to be addressed. This needs to be explained much earlier in the document to prevent the
reader from asking some of the questions below which reflected my understanding the first several
times I read this section. Why do the age ranges differ from those used in the dietary assessment from
OPP for sulfuryl fluoride? Why are these particular ages of special interest? Is the rationale for the
various age groups due to differences in consumption or special sensitivity to the effects of fluoride?
This needs to be discussed in more depth. Some explanation is needed for the differing presentation
of the label for each age class. Four of the classes are bounded on the high end by a < X form, while
the other two age classes appear to be bounded by a particular age (e.g., 1-3). The same presentation
and analysis should be used for each unless there is a disclosed reason for the difference.
The average food exposure for infants is probably too high. (See discussion below on food intakes)
The average drinking water exposure for infants may be too high (see earlier discussion on
uncertainty about drinking water consumption values.)
The discussion of the contribution of sulfuryl fluoride to dietary exposure from food is difficult to
follow and complicated by the use of age ranges not used in the OPP assessment. There should be
clarification of whether the OPP estimates are based on tolerance level exposure or not. (See
discussion of sulfuryl fluoride below)
Combining average exposures for individual sources should not also include an upper level percentile
exposure value for one of the sources. Using the equation on page 114 to calculate the RSC assumes
that the denominator represents the total exposure to fluoride from all oral sources. Adding the mean
exposures from each of these sources will produce an estimate of the mean total exposure. Adding
the mean exposures from all but drinking water and then adding the 90th percentile drinking water
exposure will not produce an unbiased estimate of total exposure.
5. Please comment on the validity of basing the food intake estimate for the 1940's on the
McClure (1943) publication, as supported by the concentrations found in various food
groups from more recent analytical data. Do you agree with this approach? If not, what
approach would you suggest for estimating food intake for the 1940's?
This issue is especially difficult to evaluate without the dose-response analysis for fluoride (EPA,
2009a) or the Dean (1942), Ershow and Cantor (1989) and Maguire (1943) studies. It appears that
this issue relates to dose construction from various sources of fluoride exposure underlying the dental
F-10
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
fluorosis response in the Dean study. This needs to be discussed at the very outset of the document.
The first few times I considered this question it was unclear why it was necessary to estimate the
fluoride concentration from the 1940's. Without further explanation of the dose-response analysis, it
is not clear why the fluoride concentration in drinking water from the 1940's is not a dead letter.
More background on the need for such an approach is necessary.
Section 2.5.5 states the age classes used are equivalent to those used in the dose-response document
(EPA, 2009a) and goes on to assert that Ershow and Cantor (1989) report drinking water intake data
for the period closest to that of the Dean study (1942). The discussion of these drinking water intakes
in section 8 is complex and really requires the underlying dose-response analysis (EPA, 2009a) to
fully appreciate. Given that the dose-response analysis was not provided to the peer review
committee, it is difficult to follow what is probably one of the most significant chapters in the
document under review. While question 5 does not ask the peer review committee to opine on the use
of the Ershow and Cantor study, it appears to be fundamental to addressing the adequacy of the
McClure data for estimating the fluoride intake from food in the 1940's.
This question appears to flow from the discussion on page 122 to 123 in chapter 8.
The McClure (1943) data appear to overestimate the fluoride concentrations reported in the USDA
2005 data as shown in Table 2-25 in some cases. The USDA 2005 data appear to combine data from
multiple sources and are not constrained to estimation of food intakes in the 1940's. It is not clear
why comparison of concentration values in McClure to those in USDA 2005 provides confidence that
McClure is representative of the intake through food in the 1940's as stated on page 122. McClure
(Table 2-59) apparently only analyzed 10 different foods - milk, cheese, chicken, frankfurter, white
bread, apples, fish, spinach, tomatoes and carrots. This only describes fluoride concentration and not
consumption rates.
The issue of fluoride intake (mg/d) is addressed on page 53 in the market basket studies section. No
estimate of the consumption rate of the foods by various age groups is given - only the final product
of fluoride intake given various assumptions about the concentration in water and food. The
discussion of McClure's assessment of fluoride in dry foods is particularly intriguing as none of the
10 foods listed in Table 2-59 is described as dry or based on dry weight in McClure 1943. It is
unclear, without reading McClure, how "dry" is used in the analysis. Basing the exposure from food
on only ten food items without any explanation of how (or whether) these ten foods are extrapolated
to the rest of the diet would appear to incorporate a large amount of uncertainty about the estimate of
exposure through food. If these are the only available data to address the question the agency poses -
what is the contribution of food to the total fluoride exposure in 1942 - some type of quantitative
uncertainty analysis should be performed.
There are additional reservations in combining the age groups and body weights reported by Dean
(1942) with different (?) age groups and body weights (?) reported in McClure. The discussion on
page 122 of the McClure data assuming foods with 0.5 ppm fluoride should refer to Table 2-41 as the
data reported in Table 8-3 appear to rely on these data. One suspects this section of the report (§8.2)
contains the key findings of the document under review, yet this section does not describe in adequate
detail why the 1942 McClure data are important. These data do not appear to be used in this section
except to adjust the RfD by 0.01 mg/kg/d for the contribution of solid food. This issue - the
derivation of the RfD - would appear to be better addressed in the previous document on dose-
response analysis (EPA, 2009a).
Although not relevant to the question posed to the peer review committee, the rest of §8.2 does not
describe well how the figures were derived. The relative contribution of each of the sources to total
F-ll
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
fluoride exposure is shown in Fig. 8.1 but no clear explanation of how these values were calculated is
given. The other figures are equally puzzling with respect to the source of data.
6. Provide citations (and, where possible, pdfs or hard copies) for any references you suggest
EPA should consider adding to the document, and describe where you suggest these
references be added.
Consider adding these citations (for completeness more than specific information - they may not
actually be helpful. Time did not permit reading all of these):
RaoGS. 1984. Dietary intake and bioavailability of fluoride. Annu Rev Nutr. 1984;4:115-36.
Yakav et al. 2007. Determination of exposure and probable ingestion of fluoride through tea,
toothpaste, tobacco and pan masala. Journal of Hazardous Materials 142 (2007) 77-80.
Zohouri FV. Maguire A. 2006. Sources of dietary fluoride intake in 6-7-year-old English children
receiving optimally, sub-optimally, and non-fluoridated water. J Public Health Dent. Fall;66(4):227-
34.
7. Please provide any additional comments and/or further suggestions you may have for
improving the document.
Several comments do not fit neatly in the context of the charge to the panel. First, the question on the
adequacy of the studies selected and the representativeness of the values selected (Q. #3) and the
question on EPA's rationale for selecting certain data items (Q.#4) are related and rather than
repeating the answers for each of them, a combined answer appears below. Second, some issues on
the fumigant sulfuryl fluoride arise in the document. These are discussed in turn below. Finally some
page by page comments are provided.
Food intake
The representative food intake in Table 7-1 is based upon data from Table 6.1 which is in turn based on
several tables from section 2. The data from Table 6.1 need further support. First, it is unclear whether
the underlying studies cited in the rationale use the same age groups as Table 6.1. If not, further
explanation is needed to describe how the data from the original papers were categorized using the age
groups in Table 6-1. The use of the term "intake" in the second paragraph in §6.1 is confusing. The term
"intake" appears to be used differently in the phrase, "intakes for non-beverage food groups come from
USDA" than the title for Table 6-1, "Estimated Daily Dietary Fluoride intakes." In the first phrase it is
assumed "intake" means consumption rate of a particular food or food group, while in the second case
"intake" appears to mean "exposure estimate" of fluorine mass per day. The definition or use of the term
"intake" should be clarified when it is used.
Age 0.5 to < 1 year - The market basket study by Ophaug (1985) does not appear to contain the data in
Table 2-58 cited as a reference in Table 6-1. Table 2-58 reports results from a study of adult hospital
diets. Data from the Ophaug (1985) study are found in Table 2-34. There is insufficient data in the
description of this study to evaluate whether the data form an adequate basis for the F intake in the
youngest age class. While the study examined 44 market baskets in 20 cities, it is unclear how many
actual foods were included in those market baskets and how representative they may have been of the diet
F-12
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
for this age class. Reporting only the F intake in Table 2-34 does not indicate the consumption assumed
for the foods analyzed in the market baskets. Fourteen-day consumption records are rare - there is no
discussion of how the "average" fourteen-day consumption record was created. Dietary intake of a
possible contaminate through food is a function of both the concentration of the contaminant in the food
item and the amount consumed. Different types of uncertainty are associated with each of these
components of the equation. Here, in Table 2-34, the summary estimate of F intake is the product of both
consumption and concentration, but there is no way to evaluate the adequacy of either estimate. It is
unclear how representative the "average" fourteen day consumption may be for either 1985 or 2010.
Table 2-34 shows the daily fluoride exposure (mg/day) from food and total dietary. Back calculating the
body weight (kg) and water consumption (mL/day) results in a different body weight value and water
consumption value for each of the 22 market basket/year combinations. This implies that each market
basket/city combination involved a unique elicitation of the consumption values and that drinking water
consumption differs between the cities.
Table 6-1 states the overall mean from Ophaug (1985) (0.17 mg/d) was adjusted by subtracting the
milk/formula intake form the earlier Ophaug study and then adding the 0.14 mg/d estimate from the
powdered formula study by Van Winkle et al., 1995 using tap water to make the formula. This almost
doubles the original Ophaug estimate and appears to be an incorrect combination of a fluoride
concentration in food (Van Winkle et al. 1995) with an exposure estimate (Ophaug, 1985). To add the
Van Winkle study to the Ophaug study, you would have to know how much formula was consumed in the
Ophaug study. If the results from Van Winkle really are concentrations and are reported in mg/L (and not
in mg/kg or some mass-mass basis) you would have to know the amount of liters of formula consumed in
Ophaug to estimate the mg F/d contributed to the diet.
The document under review is particularly unclear about what these data from Van Winkle might
represent. On page 44, the Van Winkle data are reported to be the mean concentration of fluoride in
powdered concentrate infant formula (0.14 mg/L). Van Winkle et al. performed their analysis of the
powdered concentrate using distilled water. Later in Table 2-60, the 0.14 mg/L Van Winkle value is
presented as an "exposure estimate" in units of mg/day. This could only be correct if the average
consumption of powdered concentrate by infants in this age range is 1 L/d. There is no discussion of this
calculation and the data in the table column labeled "mg/d" is noted parenthetically for Van Winkle as
(mg/L). This appears to be a mistake, but without reading the original source whether it is a mistake or
not cannot be determined. The note in table 2-60 associated with the Van Winkle data raises another
issue. The note states that fluoride intakes will be based on formula intakes assuming the formula is
prepared with tap water at the average fluoride concentration (0.87 mg/L). This suggests that indirect
water - the amount of water used to reconstitute the formula - is added into the infant drinking water
estimate somehow. This is not well explained. Table 6.3 states that the 90th percentile intake of drinking
water (USEPA, 2004) is used to estimate fluoride exposure assuming the 0.87 mg/L F concentration. The
drinking water consumption data for this age group is not matched with consumption data used by
Ophaug in the 1985 study of 22 market baskets. First, the 90th percentile of drinking water consumption
is used but the average exposure from food is used. The food exposure value contains no information
about its relation to the distribution of drinking water consumption. It may be that the average food
consumption (and hence exposure as reported here) is not related to the 90th percentile water intake.
There is not enough information about the Ophaug study to evaluate the adequacy of this approach but it
seems optimistic that combining these two disparate data sources for fluoride exposure through food
(Ophaug) and exposure through water (EPA 2004) does not result in some under or over estimation for
the combined food and water exposure by this age class.
F-13
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
Age 1 to < 4 year
The Ophaug study is not well described enough to evaluate whether the consumption data are
representative. If the same methodology was used as for the infants - construction of a representative
fourteen day diet - the same reservations apply to this study as to the one for the six month olds. Using
this study for the food intake value from 2010 foods may not be reflective of today's fluoride exposure
from food. If the concentration of fluoride has increased in foods, this mean estimate from this study may
underestimate actual exposure. But there is little data provided on the composition of the two year old
diets used here. The EPA review document does not describe in enough detail the methodology used in
this study.
Age 4 to < 7
The Jackson study only addresses two localities - Richmond and Connersville, Indiana. These may not
be representative of national consumption estimates. The survey was apparently based on a Food
Frequency Questionnaire which would provide an estimate of the frequency of consumption but not the
amount consumed. The weighting of the 441 items sampled based on the food frequencies is not
described in enough detail to understand what was done. It unclear how the weighting factors may have
influenced how the 441 items sampled were used to create a composite USDA food groups. It is also
unclear what time period the exposure estimate was to represent (1 month? 1 year?). One description of
this study (pg 58) states that the 1998 CSFII upper intake data were not available at the time of the study
and that the authors estimated upper level intake using the 90th percentile of the fluoride concentration of
the food group samples. This would appear to be mixing an upper bound on consumption with an upper
bound on contaminant concentration. If this study is to be used by the Office of Water, the correct upper
bound consumption estimate should be used to estimate exposure.
Age 7-
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
The Singer studies are described as using FDA market basket surveys of 117 food items (Singer 1980,
1985). The market baskets were collected between 1975 and 1982. The FDA provides a mapping
between the foods sampled in their Total Diet Study (TDS) and the various foods reported in the CSFII
survey. Perhaps these data, if they include fluoride concentrations for foods in the CSFII or its
predecessors could form the basis of a dietary assessment using the current consumption rates for these
groups rather using the Singer data. It is assumed the market baskets actually measured fluoride
concentration in the 117 food items, although not disclosed in the document under review.
Beverage intake (non-water) average
Age 0.5 to < 1 year - Assuming this group ingests no beverages is incorrect. Juices make up an
important source of beverages for this group and not all of these juices are reconstituted. It is unclear how
powdered or frozen concentrate juices are considered by this analysis. Are they beverages or foods?
How is the indirect water used to reconstitute them accounted for in the analysis? A flow chart (see Q#l)
would greatly aid understanding here.
The treatment of beverages for the other age classes is confusing. Milk is excluded. Is milk included in
the food estimates in the previous section? The separation of beverages from foods is confusing. The use
of indirect water in reconstituting powdered or concentrated beverages would appear to be an important
exposure pathway for fluoride, yet it is not clear how this pathway is considered in the analysis. The
Food Commodity Intake Database recipes explicitly calculate the indirect water used to prepare foods.
Not only beverages are reconstituted with "indirect" water - soups and sauces and dehydrated foods are
also reconstituted with indirect water. It is unclear why the beverage category is separate from food. It is
unlikely that the Pang, Jackson or Singer studies provide a nationally representative estimate of beverage
consumption and fluoride concentration of those beverages to serve as reliable estimates in Table 6.2.
Drinking water intake
Age 0.5 to < 1 year - See comments under uncertainty (Q.#2).
Toothpaste intake average
Time does not permit analysis of this
Soil intake average
Time does not permit analysis of this
Sulfuryl Fluoride Issues:
The Reference Dose (RfD) of 0.114 mg/kg/d included in the DEEM output (attachments 1, 2 &3 of
Appendix 1) is never discussed in the document. The 2006 "Human Health Risk Assessment for
Sulfuryl Fluoride and Fluoride Anion Addressing the Section 3 Registration of Sulfuryl Fluoride as a
Fumigant for Foods and Food Processing Facilities PP# 3F6573" provides some insight into this
value as it states that the toxicological assessment for fluoride by the Office of Water used the value
of 8 mg/day ([FR 51 (63)]). The 2006 OPP assessment goes on to state the 0.144 mg/kg/d value
corresponds to an 8 mg/day intake value (if consumed by a 70 kg adult). The May 28, 2009 EPA OPP
Sulfuryl Fluoride - Human Health Assessment Scoping Document in Support of Registration Review
(EPA-HQ-OPP-2009-0136-0006) for sulfuryl fluoride lists a chronic RfD of 0.008 mg fluoride/kg
bw/d and presents a dietary exposure assessment that differs from the one presented in attachments 2
& 3 of the document under review. While the scoping document (EPA-HQ-OPP-2009-0136-0006)
acknowledges there may be a change in the RfD if the EPA OW moves away from using skeletal
fluorosis as the toxicological concern, some explanation is needed for the differences between the
2009 OPP scoping document, the two OPP attachments from Appendix 1 and the discussion of the
proposed RfD of 0.08 mg/kg/d on page 120.
F-15
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
Beyond the differing RfDs used in the chronic dietary assessment (0.114 v 0.08 mg/kg bw/d) the
differences between the estimated exposure levels for the general population and various
subpopulations is also needed. In section E of the scoping document, the chronic dietary exposure to
fluoride differs from Table 6 (p. 11) of appendix 1. One explanation may be the apparent use of the
1989 -1992 CSFII data in the scoping document. While consistency with the OPP dietary
assessments may not be crucial, some explanation will be required for any differences between the
OW and other dietary assessments performed by the EPA.
It is not possible to determine how the daily intakes of fluoride (mg F/d) were calculated in Table
2.61. The body weights assumed for the age ranges are not given. The body weights appear to come
from the list given on page 123 attributed to U.S. EPA, 2004. More discussion (on page 73) is needed
as neither the body weight adjustment nor the combination of the original OPP age ranges into new
ranges is explained. Using the body weights from page 73 does appear to result in the mg per day
intakes reported for the structural or commodity fumigation as the exposures given for the U.S.
population from attachments 5 and 6 of Appendix 1.
While it is possible to replicate the mg F/d for the general population using the 70 kg body weight to
express the DEEM estimate expressed in mg/kg/d as mg/d, it is not possible to replicate the exposure
in terms of the new age categories required by the Office of Water. It is unclear how the DEEM
output was combined into the new age ranges. It is not clear why these particular age ranges are
required by the Office of Water. The explanation -that the projections were matched as in section
2.5.5. - is not helpful as it was equally unclear what was done in that section. It is unclear why the
Office of Water matched the age group projections from DEEM to the Office of Water defined age
groups - DEEM could explicitly produce the values that the Office of Water requires by defining
custom populations could be created to generate the chronic exposure for 1 to 3 year olds, 4 to 6 year
olds or any other age-sex combination required. That would seem to be the easiest way to obtain data
in the age class format required by the Office of Water and would also seem to introduce less error
than combining the original age classes in the DEEM output in the appendix.
Adding the fluoride exposure associated with structural and food fumigations (as in Table 2.61)
assumes both of these exposures are simultaneously occurring for the same individuals. It is entirely
possible that someone would be exposed to fluoride residues from food fumigation and not to
residues from structural fumigation or vice versa. Where only 0.4% of the crop is treated, it seems
unlikely that a person would consume both the structurally treated and the commodity fumigated
form of a commodity. As long as only the mean values are being added, the resulting total mean
exposure would be acceptable, but if some other value in the distribution is used (upper percentile
exposure) it would be incorrect to add these together. DEEM could be used to assess the aggregate
exposure to both sources of fluoride residues, but adding the two separate exposures is probably not
correct.
From the Flutriafol REVISED Human-Health Risk Assessment for Proposed Uses on Apple and
Soybean (EPA-HQ-OPP-2009-0184-0010)- "HED has previously conducted separate dietary
exposure analyses for fluoride resides from the insecticides cryolite and sulfuryl fluoride as well
as from naturally occurring fluoride residues in food and water." A summary of the HED
analysis would be helpful as the flutriafol analysis suggests that HED has assessed naturally occurring
fluoride in food and water. If this is true, perhaps the Office of Water could use the HED analysis for
some of the representative values needed in Tables 6-1, 6-2, 6-3. "It was noted that many pesticides
contain the fluorine atom, but it was assumed that only cryolite and sulfuryl fluoride would result in
meaningful increases in fluoride residues as compared to background levels (presumably due to the
lack of carbon-fluorine bonds in these two compounds)." It appears as though HED has conducted an
analysis of background levels of fluoride in foods and water. This analysis may have useful
F-16
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
information for OW - it would be useful for the reader to understand what HED did in its dietary
exposure analysis. Table 5.1.8.2. is a summary of the fluoride residue estimate in soybean from
flutriafol and the fluoride residue estimates incorporated into the sulfuryl fluoride and background
dietary exposure analyses (cryolite not registered for use on soybean). Based on this comparison,
flutriafol is not a significant contributor to fluoride residues in soybean.
Page by page comments:
p. 12- Many of the studies relied upon by this document involve human experiments. For example, the
Trautner and Siebert (1986) study exposed healthy adult volunteers to oral doses of fluoride. Has the use
of these data been evaluated by the Human Studies Review Board or some equivalent body?
p. 18 - Table 2-6 gives the fluoride concentration of fruit juices in mg/L while the concentration of other
foods are given in mg/kg. Could these data be converted to mg/kg using the specific gravity of the juice
to allow comparison between the different foods in the table?
p. 19 - Table 2-7 reports fluoride concentrations in mg/L for foods that normally would not be considered
to be liquids - strained meats and strained vegetables. Some explanation is necessary for use of this unit.
Is it used in the original reference?
p. 22 - The McClure (1949) study is described as reporting fluoride content on wet and dry weight. It is
unclear whether these foods were analyzed as prepared or not. If dry weights are given, can these be used
to explain the dry weights reported in Table 2-41 lines e through h attributed to McClure (1943). It is
unclear where the dry weights in McClure (1943) are derived.
p. 49 - Table 2-35 should use the same nomenclature for standard error as used in the preceding text. The
SEM abbreviation in the table is assumed to standard error of the mean - this should be defined as does
the SE used on the preceding page in the text. The choice of the three categories in Table 2-35 is unclear.
Were these selected by the author or reanalyzed by OW? Why are they included?
p. 63 - Table 2-50 is entitle "Fluoride Intake from foods." The table shows both fluoride intake in food
and the concentration of fluoride in drinking water. Due to the limited data on study methodology (no
information on food preparation techniques, unclear whether estimates include indirect water containing
fluoride level in drinking water, etc.) this study is of limited use.
p. 86 - Define the term SMCL in the text. This may be the first time it appears in the text. Although it is
in the definitions on page 10, it should be defined here.
p.l 15 - Table 7-1. Are the representative values based on average consumption and F concentrations?
F-17
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
Mary A. Fox, Ph.D.
1. Please describe any suggestions you have for improving the clarity, organization, and/or
transparency of the draft document.
To improve overall clarity and transparency the document needs a statement of purpose or problem
statement that articulates the questions to be answered. This section should lay out the approach that will
achieve the stated purpose. The approach will describe the data and analyses in each chapter and how
each chapter fits together to answer the questions. A figure or flow chart depicting the approach would
also be helpful. The figure should also reflect the connections to other documents in the Fluoride risk
assessment, particularly the dose-response document.
Chapter 8: The public health implications of the exposure assessment become clear in Chapter 8. It is
critical that the analyses and information are presented as clearly as possible. The analyses presented in
section 8.2 (from last paragraph of page 120 to Table 8-4 on page 123 and particularly the interpretation
of Table 8-2 in the last paragraph of page 121) are very hard to follow. The statement of purpose and a
chapter-specific approach (as described above) will help. (Further comments on Ch 8 in response to #7).
2. Have the uncertainties associated with the analysis been adequately characterized? Are
there any important uncertainties in the data that are not discussed adequately in the
document, especially in the synthesis sections? Please describe any concerns you have and
any specific suggestions for improving or enhancing the uncertainty discussion.
The uncertainty discussion (section 6.5) should include the direction of uncertainty related to the key data
used. Does the available data overestimate, underestimate or both? Can the sources of uncertainty be
prioritized? Are the different analytical methods contributing to more uncertainty than changes in diet?
How does an understanding of the largest sources of uncertainty help us understand the data or help us
characterize the analyses?
Trend of increasing F exposure can be mentioned again in section 6.5.
3. Please consider the studies that have been selected as representative of exposures for the
specific age groups and/or exposure media. Have these studies been adequately summarized
and interpreted? Indicate any deficiencies in the descriptions of the studies and any
suggestions you have for improvement. Describe any concerns you have about the selection
of these studies, as well as any recommendations you may for alternative studies that you
believe are more representative of exposures.
I would like more information on the quality of the various exposure assessment methodologies (Section
2.5.1) - are any of these methodologies preferable for the purpose of exposure characterization or were
they all seen as equally valuable?
I am not aware of other studies to add.
4. Please comment on EPA's rationale for selection of specific data elements to represent
average exposures for each of the age groups. Has the selection been scientifically justified
and clearly and objectively described? What changes or improvements would you suggest?
I would like to see text on the criteria used to identify the key data/studies. What were the important
considerations for being "representative"? Following on response to #3, were studies using a particular
methodology preferred ?
F-18
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
5. Please comment on the validity of basing the food intake estimate for the 1940's on the
McClure (1943) publication, as supported by the concentrations found in various food
groups from more recent analytical data. Do you agree with this approach? If not, what
approach would you suggest for estimating food intake for the 1940's?
I think the approach is reasonable but the need to estimate F from food intake for the 1940's is not clearly
explained (this belongs up front in the purpose/problem statement as described in response to #1).
The selection of 0.5 ppm from McClure (1943) as the estimate for F in solid foods is also not clearly
explained or supported in the document. The analysis of the McClure data must be very carefully and
completely spelled out and documented. The relationship between McClure's estimates and more recent
data is presented (Table 2-25) and possible explanations for differences are offered. More important is
how does the more recent data affect the analysis of the McClure data and the selection of 0.5 ppm as
representative of F in solid food in the 1940's?
6. Provide citations (and, where possible, pdfs or hard copies) for any references you suggest
EPA should consider adding to the document, and describe where you suggest these
references be added.
I do not have any suggestions for additional references.
7. Please provide any additional comments and/or further suggestions you may have for
improving the document.
Pg 79: Can get an estimate of percentage of kids under age 14 from US census
Pg 85, top: Is there a more recent reference (or personal communication) for statement about CDC's
temperature dependent F concentration recommendations. Reference cited is from 1995.
Page 121: Term "margin of exposure" is used incorrectly here. Margin of exposure is not the best way to
describe the estimated dose differences at issue. This section (bottom of page 121 to middle of 122) is
very confusing. What is the meaning and implication of "OW felt that a 0.01 mg/kg/day difference
between the IOM adequate intake estimate and a dose from drinking water that could cause severe dental
fluorosis was too small"? Why is a difference of 0.02 mg/kg/day "reasonable" but 0.01 mg/kg/day
difference too small? More explanation/justification is needed here.
Page 125: Interpretation of exposures over RfD - such exposures are of concern for increased risk but it is
not quantifiable risk information - increased risk is not proportionate to dose or exposure increases. See
definitions of hazard index and hazard quotient (attached as last page).
Typos:
Page 54, Table 2-41 last row 4-6 years column - extra decimal point/decimal point placement
Page 96, Table 4-7, col 5 last row, replace "/" with decimal point
Page 101, 2nd line, misplaced parenthesis should read "as reported by U.S. EPA (1988)"
Page 103, Table 5-2, missing parenthesis, column 2, 32 month column
F-19
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
NATA glossary HQ/HI:
Hazard index (HI):
The sum of hazard quotients for substances that affect the same target organ or organ system. Because
different pollutants may cause similar adverse health effects, it is often appropriate to combine hazard
quotients associated with different substances. EPA has drafted revisions to the national guidelines on
mixtures that support combining the effects of different substances in specific and limited ways. Ideally,
hazard quotients should be combined for pollutants that cause adverse effects by the same toxic
mechanism. However, because detailed information on toxic mechanisms was not available for most of
the substances in this assessment, EPA aggregated only the effects of different respiratory irritants. The
HI for respiratory irritation is only an approximation of the aggregate effect on the respiratory system
(i.e., lungs and air passages) because it is possible that some of the substances cause irritation by different
(i.e., non-additive) mechanisms. As with the hazard quotient, aggregate exposures below a HI of 1.0 will
likely not result in adverse noncancer health effects over a lifetime of exposure. However, an HI greater
than 1.0 does not necessarily suggest a likelihood of adverse effects. Furthermore, the HI cannot be
translated to a probability that adverse effects will occur, and is not likely to be proportional to
risk. A respiratory HI greater than 1.0 can be best described as indicating that a potential may exist for
adverse irritation to the respiratory system.
Hazard Quotient:
The ratio of the potential exposure to the substance and the level at which no adverse effects are expected.
If the Hazard Quotient is calculated to be less than 1, then no adverse health effects are expected as a
result of exposure. If the Hazard Quotient is greater than 1, then adverse health effects are possible. The
Hazard Quotient cannot be translated to a probability that adverse health effects will occur, and is
unlikely to be proportional to risk. It is especially important to note that a Hazard Quotient exceeding 1
does not necessarily mean that adverse effects will occur.
http://www.epa.gov/ttn/atw/nata/gloss.html
F-20
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
E. Angeles Martinez Mier, DBS, MSB, Ph.D.
1. Please describe any suggestions you have for improving the clarity, organization, and/or
transparency of the draft document.
Page 2, third paragraph, section 1. Introduction, subsection 1.1. Background, states that "... .based on data
relating fluoride concentrations in drinking water to the prevalence of severe dental fluorosis in selected
towns distributed across the United States (Dean, 1942). A clear rationale should be provided for the
choice of data. Although this information can be inferred later in the manuscript, mentioning it in this
section would improve clarity.
In page 2, fourth paragraph, section 1. Introduction, subsection 1.1. Background, fluoride gels and foams
should be added to dental products.
In page 6, first paragraph, section 2. Exposure from Foods and Beverages, mass spectrometry, gas
chromatography, electro-analysis, catalytic- enzymatic and radio-analytical methods should be added
when describing the newer methods employed for fluoride analysis.
In page 6, second paragraph, section 2. Exposure from Foods and Beverages, Subsection 2.1. Analytical
Methods, the different approaches currently employed to determine fluoride could be better described.
These may require more than the three distinct phases described and include pretreatment of samples,
separation and concentration of fluoride, actual measurement of fluoride ions, calculations of final
concentrations per unit of sample, and presentation of the data. Differences have been reported in all of
these steps.
In page 6, third paragraph, section 2. Exposure from Foods and Beverages, Subsection 2.1.1. Sample
preparation only describes preparation for solid foods, but later in the manuscript, beverages are included
as foods. This section should describe sample preparation for both solid foods and beverages.
In page 6, third paragraph, section 2. Exposure from Foods and Beverages, Subsection 2.1.1. Sample
preparation should mention that a trapping medium is required to ensure fluoride is not lost during both
diffusion and digestion processes.
Page 6, fourth paragraph, section 2. Exposure from Foods and Beverages, Subsection 2.1.1. Sample
preparation states that "..ashing of the food samples is used to remove the organic matrix." I suggest the
sentence is changed to say "has been used", since ashing of the samples is no longer a common practice.
A review of the literature shows that all food fluoride American studies described in the manuscript and
most international studies since 1990 (see attached table 1.- articles hyperlinked to Pub Med), use a
modification of the micro diffusion method of Taves (cited in the manuscript) with a pretreatment of
homogenization using a commercially available blender.
Fluoride content of powdered infant formula meets Australian Food Safety Standards.
Clifford H, Olszowy H, Young M, Hegarty J, Cross M.
AustN Z J Public Health. 2009 Dec;33(6):573-6.
Fluoride intake from meals served in daycare centres in municipalities with different fluoride
concentrations in the water supply.
Pagliari Tiano AV, Moimaz SA, Saliba O, SalibaNA, Sumida DH.
Oral Health Prev Dent. 2009;7(3):289-95.
F-21
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
Fluoride ingestion from food items and dentifrice in 2-6-year-old Brazilian children living in a
fluoridated area using a semiquantitative food frequency questionnaire.
Miziara AP, Philippi ST, Levy FM, Buzalaf MA.
Community Dent Oral Epidemiol. 2009 Aug;37(4):305-15. Epub 2009 Jun 9.
4. Dietary fluoride intake by children receiving different sources of systemic fluoride.
Rodrigues MH, Leite AL, Arana A, Villena RS, Forte FD, Sampaio FC, Buzalaf MA.
J Dent Res. 2009 Feb;88(2): 142-5.
Fluoride intake and urinary excretion in 6- to 7-year-old children living in optimally, sub-optimally
and non-fluoridated areas.
Maguire A, Zohouri FV, Fiindmarch PN, Harts J, Moynihan PJ.
Community Dent Oral Epidemiol. 2007 Dec;35(6):479-88.
[Fluoride concentrations in typical Brazilian foods and in infant foods]
Casarin RC, Fernandes DR, Lima-Arsati YB, Cury JA.
Rev Saude Publica. 2007 Aug;41(4):549-56. Portuguese.
Fluoride ingestion from toothpaste and diet in 1- to 3-year-old Brazilian children.
de Almeida BS, da Silva Cardoso VE, Buzalaf MA.
Community Dent Oral Epidemiol. 2007 Feb;35(l):53-63.
Sources of dietary fluoride intake in 6-7-year-old English children receiving optimally, sub-
optimally, and non-fluoridated water.
Zohouri FV, Maguire A, Moynihan PJ.
J Public Health Dent. 2006 Fall;66(4):227-34.
Fluoride intake from food and liquid in Japanese children living in two areas with different fluoride
concentrations in the water supply.
Nohno K, Sakuma S, Koga H, Nishimuta M, Yagi M, Miyazaki H.
Caries Res. 2006;40(6):487-93.
Availability of fluoride from meals given to kindergarten children in Brazil.
Buzalaf MA, Pinto CS, Rodrigues MH, Levy FM, Borges AS, Furlani TA, da Silva Cardoso VE.
Community Dent Oral Epidemiol. 2006 Apr;34(2): 87-92.
PMID: 16515672 [PubMed - indexed for MEDLINE]
Related citations
Fluoride intake and fractional urinary fluoride excretion of Colombian preschool children.
Franco AM, Saldarriaga A, Martignon S, Gonzalez MC, Villa AE.
Community Dent Health. 2005 Dec;22(4):272-8.
Total fluoride intake in children aged 22-35 months in four Colombian cities.
Franco AM, Martignon S, Saldarriaga A, Gonzalez MC, Arbelaez MI, Ocampo A, Luna LM,
Martinez-Mier EA, Villa AE.
Community Dent Oral Epidemiol. 2005 Feb;33(l):l-8.
Fluoride content in bottled waters, juices and carbonated soft drinks in Mexico City. Mexico.
Jimenez-Farfan MD, Hernandez-Guerrero JC, Loyola-Rodriguez JP, Ledesma-Montes C.
Int J Paediatr Dent. 2004 Jul;14(4):260-6.
F-22
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
Total and acid-soluble fluoride content of infant cereals, beverages and biscuits from Brazil.
Buzalaf MA, de Almeida BS, Cardoso VE, Olympic KP, Furlani Tde A.
Food Addit Contam. 2004 Mar;21(3):210-5.
Fluoride, calcium and magnesium intake in children living in a high-fluoride area in Ethiopia. Intake
through food.
Malde MK, Zerihun L, Julshamn K, Bjorvatn K.
Int J Paediatr Dent. 2004 May; 14(3): 167-74.
Fluoride intake from foods, beverages and dentifrice by children in Mexico.
Martinez-Mier EA, Soto-Rojas AE, Urena-Cirett JL, Stookey GK, Dunipace AJ.
Community Dent Oral Epidemiol. 2003 Jun;31(3):221-30.
Fluoride content of infant foods in Brazil and risk of dental fluorosis.
Buzalaf MA, Granjeiro JM, Duarte JL, Taga ML.
ASDC J Dent Child. 2002 May-Aug;69(2): 196-200, 125-6.
Fluoride intake in Japanese children aged 3-5 years by the duplicate-diet technique.
Murakami T, Narita N, Nakagaki H, Shibata T, Robinson C.
Caries Res. 2002 Nov-Dec;36(6):386-90.
[Fluoride intake by children from water and dentifrice]
Lima YB, Cury JA.
Rev Saude Publica. 2001 Dec;35(6):576-81. Portuguese.
Total fluoride intake and urinary excretion in German children aged 3-6 years.
Haftenberger M, Viergutz G, Neumeister V, Hetzer G.
Caries Res. 2001 Nov-Dec;35(6):451-7.
Fluoride content of foods made with mechanically separated chicken.
Fein NJ, Cerklewski FL.
J Agric Food Chem. 2001 Sep;49(9):4284-6.
Fluoride content of infant formulas and market milk in Turkey.
Atac A, Altay N, Olmez S.
Turk J Pediatr. 2001 Apr-Jun;43(2): 102-4.
[Fluoride intake and excretion in children of Hermosillo. Sonora. Mexico]
Grijalva-Haro MI, Barba-Leyva ME, Laborin-Alvarez A.
Salud Publica Mex. 2001 Mar-Apr;43(2): 127-34. Spanish.
Fluoride content of infant formulas prepared with deionized. bottled mineral and fluoridated drinking
water.
Buzalaf MA, Granjeiro JM, Damante CA, de Ornelas F.
ASDC J Dent Child. 2001 Jan-Feb;68(l):37-41, 10.
Page 7, third paragraph, section 2. Exposure from Foods and Beverages, Subsection 2.1.1. Sample
preparation states that "Several authors have described procedures for quantifying fluoride in food matter
that do not involve ashing, but these procedures are in a distinct minority". I suggest the reference to these
procedures being a minority is removed.
F-23
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
Page 120 ,first paragraph, subsection 8.2. Estimates of Tolerable Upper Limit Exposures. Should the
abbreviation UF be UL? If not, it needs to be defined, since this is the first mention
Page 127, last paragraph of the document states that very mild or mild fluorosis is associated with
decreases in tooth decay. I suggest this sentence is modified to say appears to decrease tooth decay.
2. Have the uncertainties associated with the analysis been adequately characterized? Are
there any important uncertainties in the data that are not discussed adequately in the
document, especially in the synthesis sections? Please describe any concerns you have and
any specific suggestions for improving or enhancing the uncertainty discussion.
Page 11, second paragraph, section 2. Exposure from Foods and Beverages, Subsection 2.1.4. Confidence
in Analytical Results states that "... investigators are keenly aware of technology limitation". I would add
that although they are aware not much progress had been made. I would cite the recommendations made
by the NIH to address this issue (see reference below) their subsequent funding of grant R21 DE 14716-1
Development of Gold Standard Methods for Fluoride Analysis, and the studies that have shown there is a
need for standardization (see reference below)
A comparison of results of fluoride determinations by different laboratories.
Mthethwa MT, du Plessis JB.
SADJ. 2005 Jul;60(6):238-40.
[ ' A I I i 11 i ' i i 11 I I i i 'i I , I < I i 'i i , III ill I I ii I I i I ' r '' I I , I 11 i I A I i I ' i , I ! I I i '' I i I I I 11
Clarkson JJ, Hardwick K, Barmes D.
J Dent Res 2000:79:893-904..
Page 11, fourth paragraph, section 2. Exposure from Foods and Beverages, Subsection 2.1.4. Confidence
in Analytical Results states that "... colorimetric procedures exhibited sufficient precision. Sufficient
precision needs to be defined.
Page 33, fourth paragraph, section 2.2 Natural Fluoride in Solid Foods, subsection 2.2.3. Summary of the
Data on Fluoride in Solid Foods states that "It is difficult to tell if changes in analytical methods over time
have influenced the results form studies of fluoride in foods." I would suggest changing it to say:
"Results of studies comparing different modifications to methods for fluoride analysis suggests that
changes n analytical methods most likely have influenced ...." The previously mentioned literature, plus
the following abstracts from the NIH funded grant previously mentioned support this statement:
E.A. Martinez-Mier, J. Cury, J. Heilman, S.M. Levy, Y. Li, A. Maguire, J. Margineda, D. O'Mullane, P.
Phantumvanit, A.E. Soto-Rojas O.K. Stookey, A. Villa, J.S. Wefel, G.M. Whitford, D.T. Zero and V.
Zohouri (2004). Development of Standard Fluoride Analytical Methods: Direct Analysis. Caries Res
38;372 (Abstract #45).
A.E. Soto-Rojas, J. Cury, J. Heilman, S.M. Levy, Y. Li, A. Maguire, J. Margineda, E.A. Martinez-Mier,
D. O'Mullane, P. Phantumvanit, O.K. Stookey, A. Villa, J.S. Wefel, G.M. Whitford, D.T. Zero and V.
Zohouri (2004). Development of Standard Fluoride Analytical Methods: precision and trueness testing.
Caries Res 2004: 38;373 (Abstract # 46).
Soto-Rojas A. E, Martinez-Mier E. A., Margineda J. Development of Standard Fluoride Analytical
Methods: Analytical precision of diffusion analysis using different reagents. J Dent Res 83;Special Issue
A Abstract # 786 2004 www.dentalresearch.org .
F-24
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
Martinez-Mier EA, Margineda J and Zero DT (2003): Development of standard fluoride analytical
methods: calculations of standard curve raw data. J Dent Res 82;B211 (Abstract # 1595).
E.A. Martinez-Mier, J. Cury, A.J. Dunipace, J. Heilman, S.M. Levy, Y. Li, A. Maguire, D. O'Mullane, P.
Phantumvanit, O.K. Stookey, J.S. Wefel, G.M. Whitford, D.T. Zero and V. Zohouri (2003): Development
of Standard Fluoride Analytical Methods: pilot study. Caries Res 37;291 (Abstract # 70).
Page 113, second paragraph, subsection 6.5 Uncertainty. More detail is needed to explain why slightly
above average fluoride intakes are selected. It is implied that this is safety-based, but it needs to be
explicitly mentioned
Page 116, second paragraph. Individual variation regarding water intake individual variation should be
described. A recent study (see reference below) demonstrated that as much as 33% of infants were given
bottle water only for formula reconstitution.
Children's drinking water: parental preferences and implications for fluoride exposure.
Sriraman NK, Patrick PA, Hutton K, Edwards KS.
Pediatr Dent. 2009 Jul-Aug;31(4):310-5.
3. Please consider the studies that have been selected as representative of exposures for the
specific age groups and/or exposure media. Have these studies been adequately summarized
and interpreted? Indicate any deficiencies in the descriptions of the studies and any
suggestions you have for improvement. Describe any concerns you have about the selection
of these studies, as well as any recommendations you may for alternative studies that you
believe are more representative of exposures.
Studies are representative, well selected and described. A few additional mentions to references already
cited in manuscript are needed.
Page 9, figure 2-1 section 2. Exposure from Foods and Beverages, Subsection 2.1.2. Fluoride Recovery.
Comway cell have been substituted by standard Petri dishes as described by Whitford (already cited in
manuscript)
The Metabolism and Toxicity of Fluoride
Vaseline Port (sealed with Vaseline)
seal 50ulNaOHtrap for injection of H SO • HMDS
V N^ *
P
4ml 1.5/VH2SO4 - HMDS + F sample
i
Fig. 13. The essential features of the HMDS-facilitated diffusion method for the
isolation of ionic and acid-ionizable fluoride.
F-25
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
Page 9, first paragraph, section 2. Exposure from Foods and Beverages, Subsection 2.1.2. Fluoride
Recovery states that HMDS is presumed to accelerate diffusion of fluoride. I suggest the sentence is
changed to mention HMDS has been demonstrated to accelerate..." I also suggest a paragraph is added
after this describing the most recent modifications to Taves' method, such as those by Dunipace et al and
Rojas Sanchez et al (both already cited in the manuscript).
Page 44, fifth paragraph, section 2.5. Estimates of Dietary Fluoride Intake, subsection 2.5.1. Exposure
Assessment Methodologies. I suggest a paragraph is added describing the results of a study that compares
different exposure assessment methodologies as they specifically relate to estimation of fluoride in the
diet (see reference below)
Comparison of a dietary survey and the duplicate plate method for determining dietary fluoride
ingested by young children: a pilot study.
Martinez-Mier EA, Kelly SA, Eckert GJ, Jackson RD.
Int J Paediatr Dent. 2009 Mar; 19(2): 99-107. Epub 2008 Dec 11.
Section 4. Fluoride in dental products, subsection on toothpaste. I suggest results for the Iowa fluoride
study (already cited in manuscript) for dental fluorosis risk are cited here.
4. Please comment on EPA's rationale for selection of specific data elements to represent
average exposures for each of the age groups. Has the selection been scientifically justified
and clearly and objectively described? What changes or improvements would you suggest?
The selections are justified, no changes are suggested.
Page 19, second paragraph, section 2.2 Natural Fluoride in Solid Foods, subsection 2.2.1 Fluoride in
Infant Foods. The methods used by Singer and Ophaug for fluoride analysis (including preparation of
samples should be explained in more detail
Page 71, fifth paragraph, section 2.5. Estimates of Dietary Fluoride Intake, subsection 2.5.5. Combined
Estimated for Age Groups of Concern. This bullet point sates that "concentration of fluoride appear to be
related to food groups as follows: protein foods>grains and vegetables,>fruit,>beverages. I suggest a note
is included stating the concentrations in beverages vary widely.
F-26
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
5. Please comment on the validity of basing the food intake estimate for the 1940's on the
McClure (1943) publication, as supported by the concentrations found in various food
groups from more recent analytical data. Do you agree with this approach? If not, what
approach would you suggest for estimating food intake for the 1940's?
Page 122, third paragraphs subsection 8.2. Estimates of Tolerable Upper Limit Exposures. A clear
explanation need to be given why McClure's data was considered "reasonable basis", table 2-25 cited
here actually states how much McClure results differ from more recent ones. It is my opinion that the
results of this study are compromised by the deficiencies in the analytical techniques used at the time.
Interferences most likely resulted in reported fluoride concentration in foods lower or higher than actual
concentrations. My suggestion would be to compare food item by food item reported by McClure to more
recent studies that have utilized more precise methods and obtain some sort of correction factor to account
for analytical uncertainties.
6. Provide citations (and, where possible, pdfs or hard copies) for any references you suggest
EPA should consider adding to the document, and describe where you suggest these
references be added.
All cited references are hyperlinked to abstracts or full text in Pub Med
7. Please provide any additional comments and/or further suggestions you may have for
improving
General formatting comment: Description of results of studies should be kept in past tense.
Some other minor editorial comments in the abbreviations include: CDC should sand for Centers for
Disease Control and prevention and NIDR, should be changed to NIDCR - National Institute for Dental
and Craniofacial Research.
F-27
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
David L. Ozsvath, Ph.D.
1. Describe any suggestions you have for improving the clarity, organization, and/or transparency
of the draft document.
In my opinion, the organization of the draft document is fine, and for the most part I found it easy to
follow what the authors have written. However, I do have a suggestion for improving the clarity and
transparency of Section 8.2, where the 1940-era drinking water exposure estimate is adjusted to account
for the exposure from solid food using an "RfD-Equivalent" value. For those who are involved with
EPA's process to revise the MCLG for fluoride, this step is probably very clear; but for an outside reader,
it is not intuitive. Much of the confusion I experienced was a result of inconsistent terminology, but I also
found that the derivation of "RfD-Equivalents" is not entirely transparent. In my responses to Question 7
below, I provide details about where the text is confusing and how its clarity might be improved.
2. Have the uncertainties associated with the analysis been adequately characterized? Are there
any important uncertainties in the data that are not discussed adequately in the document,
especially in the synthesis sections? Please describe any concerns you have and any specific
suggestions for improving or enhancng the uncertainty discussion.
The uncertainties associated with the data presented in Sections 2 through 5 are adequately characterized,
and it is clear from reading these sections that the level of certainty varies between exposure media and
age categories. Section 6.5 acknowledges the uncertainty associated with data analysis, but the way in
which it is handled (to use "average to slightly above average fluoride intakes for the RSC analysis") is
not justified beyond the statement that the "EPA believes that these are reasonable estimates" (page 113).
It would be helpful if a sensitivity analysis was conducted in which RSC values were calculated using the
high and low estimates of fluoride intakes from the various exposure media for the different age
categories. This would provide a better picture of the effect that uncertainty has on the conclusions.
A secondary point relates to the way in which drinking water fluoride intake is treated as compared to
other exposure media. Although the authors have followed EPA protocol in using a 90th percentile water
intake to quantify exposure from drinking water, a similar quantification is not possible for other fluoride
sources, so estimates of the average intake values are used. This raises the question of whether it is
appropriate to compare 90th percentile drinking intake data to mean intake values for other media (Figure
8-1), or if mean intake values should be used for all exposure media (Figure 8-2). For the purposes of this
study, it seems that of the approaches which are scientifically defensible, the one that leads to the lowest
possible RSC for drinking water is the most appropriate to use.
3. Consider the studies that have been selected as representative of exposures for the specific age
groups and/or exposure media. Have these studies been adequately summarized and
interpreted? Indicate any deficiencies in the descriptions of the studies and any suggestions you
have for improvement. Describe any concerns you have about the selection of these studies, as
well as any recommendations you may for alternative studies that you believe are more
representative of exposures.
The only exposure media that I am qualified to comment on are drinking water and soil.
Section 3.2 is a rather brief treatment of the natural sources of fluoride in drinking water, and the
emphasis is on naturally high concentrations. If the intent had been to develop a representative fluoride
F-28
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
concentration for groundwater in the U.S., then I would expect a more detailed treatment of the subject.
However, given the fact that EPA is only addressing regulated drinking water and has a massive ICR
database available for analyzing fluoride concentrations in public drinking water supplies, Section 3.2 is
adequate as written.
With regards to the ICR data, the authors note on page 82 that the distribution of fluoride concentrations
is positively skewed, meaning that means are generally higher than medians for any given quarter or year.
In such cases, medians would normally be considered more representative of the dataset than means, but
the authors chose to use mean values in their RSC analysis. I assume this was done to avoid
underestimating the drinking water contribution; however, the net result is to increase the RSC for
drinking water, which also leads to a higher MCLG. Although the effect of choosing mean over median
values might not affect the final RSC percentages significantly, I think that some rationale for this
decision might be appropriate in the exposure assessment summary (Section 6.2 on page 110).
Unfortunately, there is no comparable database for fluoride concentrations in soil. Section 6.4 (pages
112) uses 400 ppm as an average concentration for fluoride in soil. Based on my understanding of the
literature, this is a reasonable number and at least does not underestimate the average concentration (400
ppm is near the high end of most ranges given for average soil concentrations). However, it should also
be noted that a 2009 document prepared by The Federal-Provincial-Territorial Committee on Drinking
Water (CDW) for Health Canada (link included below under Question 6) used 100 ppm as the average
fluoride concentration in soil to estimate daily fluoride intake from soil ingestion, although this value
might not apply to the U.S.
I do have one additional comment relative to the question of whether the studies selected are
representative. In large part, this document is an updated revision of two chapters from the NRC (2006)
report (Chapter 2: Measures of Exposure to Fluoride in the United States, and Chapter 11: Drinking
Water Standards for Fluoride). There are some additional studies included in this recent EPA effort, but I
noticed that the final relative source contribution (RSC) results are not very different from what is
reported in the NRC (2006) document. The fact that two separate teams of scientists came to essentially
the same conclusions lends a degree of confidence to the results (the slight differences are within the
range of uncertainty that is inherent in this process). However, it should also be noted that the 2009 CDW
document cited above (link included below under Question 6) presents daily fluoride intake values that
differ more substantially from the values presented by NRC (2006). Although the CDW document is
written for a different population, it is worth considering why their conclusions are not more similar to
those reported by NRC (2006) and EPA (2010).
4. Please comment on EPA's rationale for selection of specific data elements to represent average
exposures for each of the age groups. Has the selection been scientifically justified and clearly
and objectively described? What changes or improvements would you suggest?
This is outside my area of expertise.
5. Please comment on the validity of basing the food intake estimate for the 1940's on the McClure
(1943) publication, as supported by the concentrations found in various food groups from more
recent analytical data. Do you agree with this approach? If not, what approach would you
suggest for estimating food intake for the 1940's?
The answer to this question is outside my area of expertise, but I noticed on page 34 the authors state that
the differences between McClure's (1943; 1949) data and the more recent USDA (2005) data shown in
F-29
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
Table 2-25 "cast doubt on the results of exposure assessments derived from some of the early food data".
However, in the third paragraph on page 122 the authors cite the same data (in Table 2-25) and state that
"McClure's (1943) estimate for dietary intake based on a diet where solid foods had an average of 0.5
ppm fluoride appears to provide a reasonable basis for the contribution of solid foods to total exposure in
the 1940's." These statements appear contradictory, but I suspect that the authors intended to cite Table
2-41, not 2-25 on page 122? This issue could use some clarification.
6. Provide citations (and, if possible, pdfs or hard copies) for any references you suggest EPA
should consider adding to the document, and describe where you suggest these references be
added.
Beltran, E.D., Griffin, S.O., and Lockwood, S.A. 2002. Prevalence and trends in enamel fluorosis in the
United States from the 1930s to the 1980s; Journal of American Dental Association, 133(2): 157-
165. http://iada.ada.Org/cgi/content/full/133/2/157
Federal-Provincial-Territorial Committee on Drinking Water. 2009. Fluoride in Drinking Water
http://fluoridealert.org/canada.2009.report.pdf
Ismail, A.I, and Hasson, H. 2008. Fluoride supplements, dental caries and fluorosis: A systematic review;
J. Am. Dent. Assoc. 139:1457-1468. http://iada.ada.org/cgi/reprint/139/ll/1457
7. Please provide any additional comments and/or further suggestions you may have for improving
the document.
I have a number of comments/questions/suggestions, which are listed chronologically below.
(1) The first paragraph under Section 1.2 (on page 3) states that the fluoride MCLG was derived using an
estimated 20 mg/day chronic exposure. I think that it would be helpful to explain or mention where this
20 mg/day value comes from.
(2) Page 9 (in the third paragraph) is the first place I noticed the use of "mg F/kg" as a concentration, and
page 13 (in the first paragraph) is the first place I noticed the use of "mg F/L" as a concentration. There is
nothing wrong with using these units; however, in other places concentrations are expressed as "mg/kg"
and "mg/L" (the authors switch back and forth throughout the document). I think that consistency would
be preferable.
(3) A minor point regarding Section 2.1.3 (starting on page 9): Unless there is some reason for
organizing this section as they have, I think that it makes more sense to present the fluoride measurement
techniques in the order in which they were invented.
(4) The second sentence of the second paragraph on page 17: I don't believe that "optimum fluoride" has
been defined yet in this document, although most of its readers would probably know what this means.
(5) Fourth paragraph on page 32: Raisins are classified as an outlier because their fluoride content
reflects the use of cryolite as a pesticide. This issue comes up again later in the document, and I think that
it might be helpful to establish right away (on page 32) the fact that fluoride inputs from pesticides will be
F-30
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
treated separately. Otherwise it seems to the reader that this potentially important exposure pathway is
going to be ignored.
(6) Table 2-26 (on page 35): "Hargraeves" is misspelled in the footnote. The authors should check the
spelling on this name throughout the document, as I believe there are at least two other places where it is
misspelled (for example, see the second paragraph on page 38).
(7) I believe that the caption for Figure 3-1 on page 76 should cite Fleischer et al., 1974, not 1972.
(8) The last sentence in the first paragraph at the top of page 82 (i.e., "There were only 6 entries in the
data set that reported concentrations between 40 and 100 mg/L, and 420 greater than or equal to 100
mg/L.") is slightly ambiguous. Are the authors saying that 420 entries reported concentrations equal to or
greater than 100 mg/L? I assume that the purpose of this sentence is to state how many entries were
eliminated from the data set because their concentrations were considered anomalously high? Perhaps
this sentence could be written as: "A total of 426 entries were considered anomalously high and
eliminated from the database: 6 values between 40 and 100 mg/L, and 420 values greater than or equal to
100 mg/L"?
(9) The second sentence in Section 4.3 (on page 98) reads: "With few exceptions all of these studies were
published after in the early to mid-1990s and are likely to not reflect changes in guidance on the amounts
of toothpaste recommended for brushing (a pea-sized portion rather than a ribbon).". I believe that the
word "after" should be deleted from this sentence (to read: "... published in the early to mid-1990s...").
(10) The last paragraph on page 114 states that 90th percentile drinking water intakes are used for infants
and children up to age 14 and 2 L/day is used for individuals over age 14 because this is the 90th
percentile value for adults. I'm not sure why the authors do not simply state that the 90th percentile water
intake is used for all ages. By wording it as they do, it initially sounds as though the individuals above
age 14 are treated differently from the younger population. In fact, footnote "a" for Table 7-1 reinforces
this idea.
(11) The second sentence in the first paragraph on page 116 states: "It is apparent that, for most
individuals in the population, the contribution from drinking water has decreased considerably from the
100% assumed in the EPA 1986 derivation of the MCLG for crippling skeletal fluorosis." Actually, the
data in Figure 7-1 do not show that the contribution has necessarily decreased since 1986 but only that it
is less than the 100% assumed by EPA in 1986.
(12) The second paragraph on page 117 begins with the statement, "Geologically, one-third to one-half
of the U.S. has access to ground water containing less than 0.5 ppm fluoride...", but the next sentence
states that "69% of the U.S. population receives fluoridated water" (between 0.7 and 1.2 mg/L). I realize
that the percentages in the first sentence do not refer to the proportions of the population that actually
consume low fluoride water, but the juxtaposition of these two statements is somewhat confusing.
Perhaps the second sentence could begin with, "However," to help make the point that is then more
overtly stated in the third sentence.
(13) The value of 0.07 in Table 8-2 (on page 121) is shown in bold, perhaps because the EPA chose to
use this dose estimate as the contribution from drinking water in 1942? There should be a footnote to the
table that explains the bold font.
(14) The first sentence in the third paragraph on page 122 (which mentions that the OW increased the
0.07 mg/kg/day dose for drinking water by 0.01 mg/kg/day to account for exposures from food in the
1940's) is confusing to someone who has not read the EPA (2009a) document, because the implication
F-31
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis"
from Section 1.1 (especially the paragraph that begins at the bottom of page 2) is that no attempts have
been made in the past to account for fluoride exposure from sources other than drinking water. This
notion is then reinforced when one reads Section 1.2 and the first paragraph on page 116. It would be
helpful if Section 1.1 made it clear that EPA (2009a) has already made an estimate of the fluoride
exposure from food but that the current study will re-evaluate that estimate in light of additional
information (or in a more scientifically defensible manner, etc.).
(15) Figures 8-1 through 8-3 include a plot of the Pre-Peer Rev (or Proposed) RfD, but this is actually the
RfD multiplied by the body weight (also referred to as the "daily intake equivalent", the "RfD-based daily
intake limit", the "RfD-Equivalent" and the "OW RfD-equivalent" in the text). It would help clarify the
text, Table 8-4, and the graphs if only one term is used consistently in both text and graphs. Also, I
assume that the "Pre-Peer Rev Rfd" curve in Figures 8-1 through 8-3 is terminated at the 7 to <11 years
column to keep the vertical scale less than 3.5, but it would help to explain this in a footnote.
(16) In addition to the confusion created by using different terms to refer to the same variable, the text is
not clear why the "RfD-Equivalent" values in Table 8-4 on page 123 (which are referred to as "RfD-
based intake limit" in the equation used to calculate them) are being calculated. In the last paragraph on
page 122 the reader is told that the McClure (1943) data support the 0.01 mg/kg/day fluoride contribution
from solid food that was assumed by EPA (2009a), but there is no transition from this statement to the
derivation of "RfD-Equivalent" values. It would help if this process were explained more explicitly. For
example, the first paragraph on page 123 could be rewritten: "Therefore, an estimate of the total fluoride
exposure at the time of Dean's 1942 study was made by adjusting the reference dose (OW RfD) from
0.07 mg/kg/day (the estimated exposure from drinking water) to 0.08 mg/kg/day to include the exposure
from solid food during the 1940s. This OW RfD (or Rfd) was converted to a daily intake equivalent (or
whatever term the authors prefer) by multiplying the mean body weights for each of the age groups of
concern using the following equation:"
(17) In Table 8-4 (on page 123), right-hand parentheses are needed for the words "skeletal fluorosis" in
the rows for "Adult females" and "Adult males".
(18) I think that it would improve the document's clarity and transparency if Section 8 concluded with a
summary of how the study objectives were met. The way the current documents ends is somewhat
abrupt, and there are no recommendations or explicit statements about how the conclusions will be used
(although I am certain that the authors understand how their findings fit into the flow of what the EPA is
doing to address fluoride drinking water standards).
F-32
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
Appendix G:
EPA Slides Presented at the
Peer Review Meeting
G-l
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
Fluoride Exposure and Relative
Source Contribution (RSC)
Assessment
EPA Office of Water
External Peer Review Meeting
May 14, 2010
Background
Year
2003
2006
2008
2010
Event
EPA requested NRC to evaluate the health effects and occurrence of fluoride in
public water supplies
NRC Report Published: Recommended a dose- response analysis for severe
dental fluorosis and skeletal effects.
Recommended an extended evaluation of exposure to sensitive populations .
Advised EPA to wait for the publication of an ongoing study of osteosarcoma.
The study has not yet been published.
Completed and peer reviewed the dose-response analysis. Reviewers agreed with
Dean (1942) as the critical study. Suggested modeling severe dental fluorosis to a
lower response level (0.5% vs. 1%).
Completed the peer review draft of the Exposure and Relative Source Document
(June).
G-2
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
Next Steps
Address peer review comments and revise the Exposure
and RSC Report.
Post the Dose-Response and Exposure/RSC Reports on
the Web with peer review comments and EPA responses —
June 2010
OGWDW begins consideration of revisions to the
MCLG/MCL
j SDWA Statutory Mandate:
J The Administrator shall, not less often than every 6 years, review and revise, as
appropriate, each national primary drinking water regulation.
J Any revision to an NPDWR shall maintain, or provide for greater, protection of
the health of persons.
Differences between the MCL and MCLG
• MCLG is not enforceable
a Based on the NRC report as the description of hazard
and the EPA Dose-Response Assessment
.n MCL enforceable
a Considers analytical and technological limitations
a Considers relationship of the costs involved in requiring
systems to comply with the MCLG compared to the
benefits realized by doing so.
G-3
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
RSC and the MCLG: General Policies
i Determine a Reference Dose
i Determine the Drinking Water Equivalent Level (DWEL) from the RfD
j DWEL = RfD x body weight •=• drinking water intake
« Based on body weights and 90th percentile drinking water intakes for sensitive
populations (children in the case of fluoride)
« OW could potentially use the BMDL from the dose-response as the DWEL
i Calculate an MCLG that is protective for the persons served by public water
systems
j MCLG = DWEL x RSC
« Could use each age group and its RSC and choose the limiting value
« Could normalize body weight and drinking water intakes across the sensitive age
range and apply the RSC to the normalized value
Generalized Assessment Findings
« Exposure estimates indicate that the RfD is being exceeded for some
drinking water consumer age groups at an average drinking water fluoride
concentration of 0.9 mg/L
« Prevalence of dental fluorosis in the population among the population has
increased from -10% in the Dean Study to >32% in 2000
« Monitoring of the population demonstrates that there are individuals with
severe dental fluorosis in the United States
« The dose-response data are supportive of an increased risk of cavities if
dental fluorosis is severe (age independent)
J Cavities are an adverse health effect with consequences if left untreated
« >4,000 public drinking water systems have fluoride concentrations greater
than the average concentration of 0.9 mg/L (mostly ground water
systems).
G-4
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
Iowa Fluoride Study Data
622 children -mostly moderate to high socioeconomic status
Reporting of intakes of water beverages selected foods supplements and
dentifrice by parents from birth to 48 months (3-4 month intervals)
Teeth examined at ages 8-10 (mean 9.2 years).
Fluorosis prevalence
_i 8 cases categorized as severe fluorosis; one had enamel pits (central
incisors and first molars (0.2%)
« Affected child's average intake 9 though 36 months was 0.075 mg/kg/day and
0.079 mg/kg/day for 16 through 36 months. Apparently breast fed until
sometime between 6 and 9 months.
« Pictures for one other case could not be located — fluoride intake high at 6 and
9 months with limited reporting thereafter
« Fluorosis scale used categorized severe fluorosis as severe staining
and/or pitting of the enamel
_i No severe dental fluorosis cases for a Fluoride exposure less than 0.06
mg/kg/day
G-5
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
Appendix H:
Panel Chair's Summary of
Reviewer Pre-Meeting Comments
H-l
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
External Peer Review for Fluoride: Exposure
and Relative Source Contribution Analysis
Summary of Reviewers Comments
Provide suggestions for improving the clarity,
organization, and/or transparency of the draft document
> Reorganization and inclusion of an overview are needed
> Summary of how the study objectives were met is needed
^ Statement of purpose/problem and clear statements on the
impact of specific datasets
J> Road map, description of approach would be helpful
J> Overview of challenges is needed
> Clarification on MCLG, RSC and RfD is needed
Explanation of orig/n of 20 mglday values used to calculate MCLG is needed
^ Use of tables to summarize information and flowcharts to
explain approach
l» Reference to US EPA document
H-2
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
Provide suggestions for improving the clarity,
organization, and/or transparency of the draft document
> Rearrangement of information, earlier references to and
rationale for choice of critical data sets
> Chapter 8 particularly hard to follow / Chapter 2 would
benefit from updated information on analytical methods
> Standardize terminology and units
Explanations for choice of units and conversion are needed for
beverages and foods
Nomenclature for SEM is inconsistent
Use ofmg Flkg - mg FIL -mglkg -mg/L" (the authors switch back
and forth throughout the document)
Use ofRfD - daily intake equivalent - RfD-based daily intake limit
- RfD-Equivalent - OW RfD-equivalent
Indicate concerns and any specific suggestions for
improving or enhancing the uncertainty discussion
>> Lack of quantification and estimation of uncertainty
>> Lack of clarity on selection of average values
>> RSC uncertainty needs to be addressed
>> Discussion on dynamic nature of fumigant market is needed
>> Uncertainty related to the use of CSFII should be discussed
>> Uncertainty related to DWI and choice of 90th percentile needs to
be further discussed
> Discussion of uncertainty specific to key data set should be
expanded
> Discussion of analytical uncertainty should be expanded
> Sensitivity analysis would be helpful (RSC values calculated using
high and low estimates of fluoride intakes)
H-3
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
Indicate deficiencies in the descriptions of the studies selected
as representative and any suggestions you have for
improvement. Recommendations for alternative studies
> Inadequate information about food studies (more information needed on
consumption rates)
> Representativeness of studies limited on a few individual or geographic areas is
questionable
> Explanation for choice of studies is needed
> Suggestion for alternative approach - USDA fluoride database coupled with
CSFII - Collaboration with ORD
> More information on quality of various exposure methodologies is needed
> Choice of mean over median for ICR data needs to be further explained
> Specific suggestions for studies to be added to specific sections are offered
> Comparison to the CDV Federal-Provincial-Territorial Committee on
Drinking Water (CDW) for Health Canada is needed
> Hospital study by loves is not likely to be representative of adult food intakes
* Singer study is likely not representative of current dietary intake patterns
Is selection of specific data elements to represent average
exposures for each of the age groups scientifically justified and
clearly and objectively described?
> Selection of age groups needs further discussion
Not dear if age groups are similar for underlying cited studies and summary table (6.1)
> The average food exposure for infants is probably too high
Insufficient data describing Ophaug study to examine relevance for youngest age class
> The average drinking water exposure for infants is probably too high
> The discussion of the contribution of sulfuryl fluoride to dietary exposure
from food is difficult to follow
There are multiple sulfuryl fluoride issues
> Estimation of total exposure need to be revisited (bias related to use of
90th percentile drinking water exposure and average food consumption)
> Text on the criteria used to identify the key data/studies is needed
> Methods used for fluoride analysis need further description
> Better description of variations in fluoride concentration in specific food
groups (for example beverages) is needed
H-4
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
Comment on the validity of basing the food intake estimate for
the 1940's on the McClure (1943) publication. Suggest
alternative approach for estimating food intake for the 1940's
Me Clure data appears to overestimate fluoride concentration
when compared to USDA 2005 data. Not clear why it is later
stated this comparison provides confidence in the McClure data
Background on the need to estimate the fluoride concentration
form the 1940's is needed
Discussion of the dental fluorosis response from Dean's studies
needs to be provided and further explanation on the dose
response analysis is needed
More detail is needed to explain the importance of McClure data
in chapter 8 (section 8.2 is particularly confusing)
Comment on the validity of basing the food intake estimate for
the 1940's on the McClure (1943) publication. Suggest
alternative approach for estimating food intake for the 1940's
Some type of quantitative uncertainty analysis needs to be
performed to address limitations of using McClure data
Differences in age groups and body weights between Dean and
McClure raise concern
The role of the McClure data in the derivation of RfD needs to be
better explained
Selection of 0.5 ppm as estimate for fluoride in solid foods needs
be better explained
Analytical techniques used in the McClure study compromise
results, they may over or underestimate (due to interferences)
H-5
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
Citations
> Provide citations (and, where possible, pdfs or hard
copies) for any references you suggest EPA
should consider adding to the document, and describe
where you suggest these references be
added
Refer to pre-meeting comments
Provide additional comments and/or further
suggestions for improving the document
> Not clear if age groups are similar for underlying cited studies and
summary table (6.1)
> Use of term intake needs further clarification
> Cross reference in tables is confusing and sometimes incorrect
> Insufficient data describing Ophaug study - there are multiple
questions related to his methods and calculations and adequacy of
use for several age groups
> Multiple questions related to Van Winkle study and its combination
with the Ophaug study
I- Bias related to use of 90th percentile drinking water exposure and
average food consumption
> Representativeness of studies limited on a few individual or
geographic areas is questionable (concerns raised over Jackson et al
study)
> Confusion between upper bound on consumption with an upper
bound on contaminant concentration needs to be addressed
H-f
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
Provide additional comments and/or further
suggestions for improving the document
> Hospital study by Taves is not likely to be representative of
adult food intakes
> Singer study is likely not representative of current dietary
intake patterns - multiple assumptions need to be revisited
> Assuming 0.5 < y year group ingests no beverages is incorrect
> Treatment of beverages for the other age classes is confusing
> There are multiple sulfuryl fluoride issues (see pre-meeting
comments)
Provide additional comments and/or further
suggestions for improving the document
> Description of 1KB status for studies described
> Explanations for choice of units and conversion are needed for
beverages and foods
> Nomenclature for SEM is inconsistent
> Table 2-50 is of limited use
> Definition of term SMCL is needed
> Recent reference for temperature dependant concentrations
for fluoride is needed
> Term margin of exposure is incorrectly used
> RfD interpretation of exposures is incorrect
H-7
-------
EPA Response to Peer Review of "Fluoride Exposure and Relative Source Contribution Analysis'
Provide additional comments and/or further
suggestions for improving the document
> Explanation of origin of 20 mglday values used to calculate MCLG
is needed
i> Need to present analytical techniques in order of
development
i> Optimum fluoride level needs definition
i> Need to mention fluoride input from pesticides will be treated
separately
> Several spelling mistakes were identified
> Several mistakes in captions for figures were identified
> Consistent style is needed (suggest past tense)
l» Some abbreviations are incorrect
l» Specific style suggestions were offered
l> Issues associated to RfD - daily intake equivalent - RfD-based
daily intake limit - RfD-Equivalent - OW RfD-equivalent
------- |