UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. £0460
November 19,1993
'
SCIENCE AOVJSQfiY BOARD
EPA-SAB-EHC-LTR-94-Q03
Honorable Carol M. Browner
Administrator
U.S. Environmental Protection Agency
401 M Street, S.W.
Washington, DC 20460
Subject: Review of the Health Benefits for the proposed R!A for the RCRA
Corrective Action Plan Rule by the Environmental Health Commit-
tee
Dear Ms. Browner:
At the October, 1992 meeting of the SAB's Executive Committee (EC) the
Board was asked by the Office of Solid Waste and Emergency Response (OSWER) to
review the methodology for the draft Regulatory Impact Analysis (RIA), This
cost/benefit analysis is required prior to promulgation of the Agency's final Resource
Conservation and Recovery Act Corrective Action Rule. The EC, recognizing the
importance, complexity, and novelty of OSWER's work and its multi-disciplinary
character, established an ad hoc RCRA-RIA Steering Committee (RRSC) to assure
that certain aspects of the RIA - in both methodology and application -received
appropriate attention from the relevant SAB committees.
At a public meeting on January 29, 1993, the RRSC concluded, on the basis of
presentations by and discussions with OSWER personnel, that four SAB individual
committees should review the major segments of the RCRA-RIA. Specifically, the
RRSC agreed to review; a) the contingent valuation (Cvl methodology used in the
RCRA RIA analysis (CV-1, by the Environmental Economics Advisory Committee
(EEAC)}; b) the application of CV in the RCRA-RIA (CV-2, by the EEAC); c) the princi-
pal fate and transport model (MMSOILS). used in the RCRA-RIA (by the Environmen-
tal Engineering Committee (EEC)); d) the ecological risk assessment portion of the
RCRA-RIA (by the Ecological Processes and Effects Committee (EPEC)); and f) the
human health risk assessment portion of the RCRA-RIA (by the Environmental Health
Committee (EHC)),
-------
This letter was prepared by the SAB's Environmental Health Committee
following the circulation (by mail) of initial comments prepared by a Committee
Member, and a public teleconference held on September 24, 1993. The report
focuses on the risk assessment methodology used to generate the estimated impacts
on human health resulting from proposed corrective action at RCRA facilities. The
March 23, 1993 memorandum from Richard Guimond to Dr. Donald Barnes described
the Charge for the SAB's review expresses an interest (page 4) In the implications of
the fate and transport model assumptions on the ecological and human health n'sk
assessments. Before such an interest can be addressed, the risk assessments must
be reviewed to determine whether they are sound, and to suggest improvements
where they are not; otherwise comments on the implications of fate and transport
assumptions may not be meaningful.
Our report is organized into two sections - some overall strategic comments
which follow in the body of this (etter, and detailed technical points, keyed to specific
sections of the draft RIA, which are incorporated in an Appendix enclosed with this
letter.
Based on our understanding of QSWER's goals, the Committee views the draft
methodology as a screening analysis, as opposed to a more detailed and definitive
analysis. We have therefore reviewed the material with the following question in mind
as our Charge: "Is this the best that can be done to provide a method for conducting a
screening analysis?"
The screening analysis methodology produced by the Agency was a very
ambitious undertaking. It is also of great potential importance since implementing the
proposed RCRA Corrective Action regulation could cost many billions of dollars and, in
the future, these techniques for estimating and comparing costs and benefits may well
find application to other important cases as stated in the document.
In general, the methods used are well known and correspond to "much-used"
guidelines, methods and practices (GMP). These GMP have, for the most part, been
developed for use in setting prudently protective standards of exposure to individual
substances in specific regulatory situations. In the proposed RCRA/RIA methodology
the GMP are being applied to a large set of complex cases Involving multiple expo-
sures to make calculations of total impact not contemplated when the GMP were first
derived. Determining whether the GMP can be so applied, whether they are ade-
quately applied, whether their application to determine overall health impacts is
reasonable, whether the important assumptions and limitations have been dearly
-------
identified, and what reasonable and feasible improvements can be suggested has
been an objective of our review.
* The methods developed in Chapter 7 and Appendix E are obviously the result
of much careful and thoughtful work. Care has been taken to lay out and define the
risk assessment process used and its components. The basic assumptions used in
each of the steps are dearly stated, (including the way in which numerical values were
chosen to bring in as much realism as possible at each step through the use of site
specificity where possible}. Alternatives considered and different ways used to
calculate the end results are set forth so as to give an idea of sensitivities or possible"
alternative results (e.g., the decision to analyze a "less than 100 percent effective-
ness" corrective action scenario). Lastly, possible biases and uncertainties inherent in
the whole procedure are identified.
The net effect of the effort was to produce a construct formed of many carefully
selected parts with Which a formal estimate of cancer population risk can be made,
both before and after corrective action, and with which a formal estimate can be made
of the numbers of persons exposed to contaminant levels which have some probability
(extent unknown) of producing adverse, non-carcinogenic effects , also before and
after corrective action. Thus, at least formally, the effect of the regulation can be
measured in two different ways, one for cancer and one for non-cancer effects.
Although those responsible for developing this construct can be congratulated
on their achievement, another group of workers might have made different specific
choices at various junctures (for example, some other length of time than the 9 year
exposure period, or the 128 year timeframe). However, it is doubtful if, given the state
of science, a more "rigorous" method with greater certainty of giving "right" answers is
possible at the present time.
As a construct, given both the state of science and the impossibility of validat-
ing the calculations through any realistically achievable, actual measurements, the
results reported must be regarded as coming from an enormously complex, logically
consistent, but mainly hypothetical calculation - as must the results of most low
exposure-level risk assessments. We do not use the term "hypothetical" In a negative
sense; rather, as carefully devised as it is, the construct cannot give, at best, more
than a rough estimate of the actual situation. Using these results in any kind of
cost-benefit balancing must be done with this ambiguity firmly in mind. We suggest
that in the final R1A, this point be made abundantly clear to avoid even the possibility
that the results will be treated as definitive by anyone.
-------
There are some areas where we suggest possible improvements, i It would be
very useful to include a way to estimate, even crudely, the fraction of the population
presumably exposed to significant levels of contaminants (HI > 1) who actually
manifest adverse, non-cancer effects. Without some attempt at providing such
estimates for the most important cases, the cost/benefit calculation remains seriously
incomplete. By including such estimates, the monetization of both cancer and non-
cancer effects avoidance benefits can be done in a formal sense, and that portion of
the cost/benefit calculation would at least be present in the overall screening analysis.
The Committee suggests that there is a methodology available to estimate the
size of the population affected at a specified level of effects. We suggest (and this is
probably the most significant revision we propose) a change in the basic approach
used for relating toxicant exposure and effects. The proposed methodology can be
expanded to include use of the benchmark dose concept for determining adverse
effects, including reproductive and developmental toxicity, as described by Crump
(1984) and Kimmel and Gaylor (1988), and offered as a suggestion in the EHC's
report (EPA, 1990) reviewing proposed revisions to the EPA Guidelines for Develop-
mental Toxicity, From this benchmark dose, a straight line can then be drawn to the
dose level that represents a 100-fold margin-of-safety» plus any relevant uncertainty
factors, i.e., the reference dose for developmental toxicity. The reference dose is
considered to be an exposure level unlikely to cause human developmental effects.
At the beginning of the low-dose extrapolation below the benchmark dose, the actual
probabilities of adverse effects may be close approximations of the linear extrapola-
tion. As the extrapolation progresses toward the reference dose, the calculated
probability of adverse effects may exceed reality to some unknown extent due to the
presumed threshold phenomena operative for most developmental toxicants.
Use of this model would allow both pre- and post-remediation quantification of
reduction of risk in a manner markedly consistent with that for low-dose extrapolations
for carcinogenesis. This calculation could be used to monetize remediation efficacy.
The example given here has been for developmental toxicity, but it is suggested that
the report also attempt similar type calculations for reproductive toxicity and other
non-carcinogenic toxicologic endpoint assays. (It should be noted that this method
does not lend itself to quantification of effect severity, e.g., deft lip verses auricular
tags, but it does permit quantification of classes of adverse effects.)
Addressing other areas, we note that the term "population risk" (and related
terms in connection with both cancer and non-cancer adverse effects) is employed
correctly vis-a-vis cancer, but not with non-cancer effects. As discussed in more detail
L _..
-------
in the Appendix, the population risk for cancer is correctly given as an estimate of the
number of people affected by cancer within the exposed populations. In contrast, for
non-cancer effects, estimates of the number of people who are exposed at levels
exceeding the Hazard Index (HI > 1) do not necessarily coincide with the numbers
presumed affected; indeed, the fraction of those actually expected to be affected by
non-cancer effects among the numbers cited is likely to be very small, possibly zero.
Also, within the cancer/non-cancer dichotomy, consideration might be given in some
way to those autoimmune diseases such as lupus, and certain genetic diseases
whose causes may be related to environmental factors. The Agency should be more
explicit in distinguishing cancer and other disease conditions with respect to risk and
"population at risk," The SAB's Environmental Engineering Committee, in its review of
the RCRA MMSoils model (EPA-SAB-EEC-94-002) had similar concerns about the
EPA practice of using different approaches to cancer and non-cancer risk assessment
(as well as on some misidentification of critical endpoints for some common landfill
constituents).
The benefits of abating disease are not monetized in the RIA document At
some point, the decrease in cancer cases and the decrease in numbers exposed to
possible risks of non-cancer adverse effects may need to be balanced, along with
other benefits (either monetized or not) against the dollar costs of corrective action.
This is a difficult if not impossible aim to achieve in any objective way. Alternatively,
using existing methodologies, the results of the calculations could be used to estimate
direct monetary benefits of cancer avoidance (medical costs avoided, lost productivity,
etc...) as well as indirect costs (pain and suffering, damage to family relations,
damage to quality of life, etc...). However, a similar calculation cannot be made for
non-cancer effects based on the results of the proposed screening method. (The
SAB's Environmental Economics Advisory Committee, as part of its review of the
RCRA/RIA, is considering the monetization of health benefits.) Another consideration
which might be taken into account in estimating impact or monetizing effects is age of
onset of a fatal and/or disabling disease; i.e., cancer in an 80-year old will have
different societal and personal impacts than cancer in a 24-year old person.
It would be useful to estimate for cancer, the number of individuals "at risk"
(already done for non-cancer effects) so as to have comparable numbers of people at
risk for cancer and non-cancer effects. An estimate of the population exposed at
levels of exposure of concern for cancer (i.e., levels leading to a lifetime individual risk
of 10"6 or greater) would yield such estimates recognizing the fact that what is of
concern is not identically defined in the two cases.
-------
Lastly, we urge increased emphasis on the collection and management of good
exposure data as a foundation for this, and other efforts by the Agency. The impor-
tance of good exposure data can not be underestimated. From what we now under-
stand, concentrations predicted via the MMSQILS model (as reviewed by the Environ-
mental Engineering Committee) are subject to large uncertainties, affecting exposure
estimates significantly, and thereby affecting the results of the risk assessment. More-
over, the screening methodology will produce results which are completely useless
and inaccurate if chemicals released from the subject facilities are not included in the
assessment. We note that several chemicals which have caused problems around
municipal landfills are not included in Table E-1 - particularly methane and hydrogen "
sulficle. These chemicals may well also be generated at solid waste management
units at RCRA facilities. Although not generally considered a toxic problem, methane
has accumulated In houses and caused explosions; this is indeed a public health
problem. Hydrogen sulfide is produced when sulfur containing compounds such as
gypsum are buried and become wet. Hydrogen sulfide releases from several landfills
have produced documented health effects in people living in nearby communities.
Although the importance of exposure data is discussed in the Appendix, we believe
that it is such a basic and important consideration that it needs to be highlighted in our
comments especially since some problematic chemicals have not been Included in the
assessment.
We look forward to receiving your response to our comments.
Sincerely,
Dr, Raf mond C. Loehr, Chair
Science Advisory Board
C
Dr. Arthur C. Upton, Chair
Environmental Health Committee
ENCLOSURE
-------
APPENDIX
SPECIFIC COMMENTS: CHAPTER 7
- Page 7-7. last para.: The SAB/EEC, in Its review of the MMS01LS transport model
as part of the RCRA/RIA review, apparently has reservations about the use of such a
genera! model to predict concentrations for a wide variety of different, specific cases,
along with other reservations, about it. It would be desirable to contact the SAB/EEC
(and the liaison members) to determine what bias or uncertainty this might introduce
into the risk assessment, what might be done instead, and under what criteria
MMSOILS might not be adequate, with suitable precautionary considerations, for a
screening analysis,
- Page 7-1S. references 17 and 18: It would be useful to review tie SAB's reviews
of these documents, which took place subsequent to the dates on them, to ensure that
the comments therein are adequately taken into account in this proposed methodolo-
gy; perhaps the SAB's review should be added as a reference.
- Page 7-22. top Of the pane: The selection of the nine-year period needs more
justification or explanation,
- Pace 7-23. last sentence of 1st para.: This is an inadequate risk descriptor (See
comments re Paoe 7^41. 2n4j>ara. and further related comments as to why),
- pane 7-29.1st paragraph: the word "may" appears twice; delete the second occur-
rence,
- Page 7-41. 2nd para.: The statement,"... and 25,000,000 persons experiencing
non-cancer health effects over the 128 year modeling period,.." is an example of an
inaccurate interpretation of the extent of the population whose exposure exceeds an
HI of 1.0, (The next sentence is another example as is the wording in Exhibit 7-17).
The 25,000,000 people correspond to the population exposed at levels yielding an HI
of 1.0 or higher, not those "...experiencing non-cancer health effects...". Exposure at
or above effect levels does not equal effect, only the probability of effect (Suppose a
news story were to appear asserting that EPA estimates that 25,000,000 people suffer
from eta.,!)' The number experiencing non-cancer health effects will actually only be a
fraction of the 25,000,000 - 25, 250 or maybe as many as 2,500. The number of
people exposed at an HI of 1.0 or greater is therefore not a measure (or a descriptor)
of population risk: the number of such people who do actually experience non-cancer
health effects is (as in the case of cancer.popuiation risk is the number of people
adversely affected). The numbers of people so exposed and the numbers actually
experiencing non-cancer health effects can, rn fact, easily be in reverse order for
different cases so that the figure of 25,000,000 is, again, not a measure of risk. This
wording should be changed (as should similar wording elsewhere in the text) to reflect
A-l
-------
this fact and to emphasize it so that the user/reader will not be misled. As mentioned
above, one of the matters the SAB's Environmental Economics Advisory Committee is
addressing in their portion of the RCRA/RIA review is the question of monetizing the
health risks so as to calculate the benefits obtained directly from reducing the risks
through regulation. Use of the 25,000,000 figure in such monetization under the
impression that it measures the number of people experiencing non-cancer health
effects would lead to grossly high dollar values.
- Page 7-42. 1st Dam.: This paragraph is a most important observation. It could
imply that most of the cost and most of the benefit of Corrective Action could be
attributed to this one site. It would be highly desirable to quote the site population
figures in this paragraph, compared to the relevant totals, and to underline or highlight
the paragraph to help ensure that its importance for the rest of the analysis is not
missed by the reader.
-- Page 7-42. 2nd para., last two sentences: The last of the two sentences corrects
the statement of the first one; I suggest eliminating the first one and using the sense
of the second one, instead, to give a correct impression in the first place,
- Page 7-42. section entitled " Number of Facilities..."Exhibit 7-18 and all other
sections of the material under review where this comment applies: The cancer and
non-cancer population "risk" figures should not be combined in any way and indicated
to be somehow of the same type since they are intrinsically differently defined; they
should preferably be presented separately. The cancer figures represent an estimate
(however uncertain) of the number of cancer cases whereas the non-cancer figures
represent the number of Individuals who merely might become non-cancer health
effects cases. Any kind of "sum" or statement of "jointness" is meaningless unless
very carefully labelled. Stating the number or percentage of sites where the number
of cancer cases is expected to be insignificant (less than one in a million, say) and in
which the HI is less than one {and therefore the number of people in which non-
cancer health effects might not occur in significant numbers if at all) is a
not-misleading statistic that might be helpful to the reader/user. The text and Exhibits
need to be modified to not mix the two types of estimates in a misleading way. For
example, in Exhibit 7-18, the wording No Risk might be changed to 'No' Risk and the
wording % Risk might be changed to % Risk ind Possibly At Risk. Other
examples abound; for example, in Exhibit 7-24 (a very useful Exhibit), even though the
non-cancer ordinate has a parenthetical statement defining what is really meant, the
title of the chart should bt changed to CUMULATIVE NON-CANCER POPULATION
POSSIBLY AT RISK and the title of the ordinate should be changed accordingly.
Throughout the entire text the distinction between the two types of population
estimates should be carefully maintained; population risk, in the case of cancer, and
population at risk.in the case of non-cancer effects. There are many such instances
and I suggest the text be carefully edited to find and change each and every one of
A-2
-------
them (for example, where the term 'risk/effects'1 is used one might use, instead "risk/at
risk," or "risk/concern" or "effects/possible effects," etc.
- Sections 7.4.1. 7.4.2 and 7.4.3: These two sections are well done and very wej-
come.
SPECIFIC COMMENTS: APPENDIX E
- page 5. first paragraph: Too flat and sharp a distinction is made between carcino-
gens and non-carcinogens on the basis of the existence or nonexistence of thresh-
olds. I would suggest the following words: "it has been the custom to suppose that for
nofecardnogenic or systemic effects protective physiological mechanisms exist that
must be overcome before the adverse effect is manifested. This may not, in fact, be
universally true and for the non-carcinogenic effects of lead, for example, it appears
not to be true. Similarly thresholds are thought to be absent in the case of cancer,
Le., any level of exposure, however small, could result in cancer although mere are a
very few instances now known in which this assumption may not hold. Nonetheless,
in this method, the existence of thresholds will be assumed as usual in the case of
non-carcinogens (except for lead, as discussed below) and the lack of thresholds will
be assumed for carcinogens, also as usual**
These words describe just what is being done, and the real assumptions being
made, for the user of the method.
- page S. second paragraph: The use of "benchmark" to describe RfDs could be
confused with the term "benchmark dose," which quite different from and RfD. We
suggest changing the subject sentence to read "For many chemicals the RfD ap-
proach has been used as a basis for regulatory decisions in relation to potential
impacts on human health."
- Page 27. first paragraph: This is an entirely sensible way to handle the aggregation
of risks from exposures to multiple carcinogens, it should be mentioned here, howev-
er, that synergism (and/or antagonism) is possible (referring to section 7.4.3 of
Chapter 7 where it is already mentioned),
— Pane 27, second paragraph: The second sentence, beginning "Ratios of contami-
nant level..." should be added to as follows: " ... non-carcinogenic health affect for
exposure to a particular contaminant The hazard quotients for different contaminants,
even If they have equal RfDs, cfo not necessarily Indicate which substance poses the
greater risk." (the material in italics is the added material) - Also, regarding the HI, the
SAB/EHC commented in considerable depth somewhat over a year ago on the
limitations of the use of the Hi in its review of the "Risk Assessment Guidelines for
Superftind Sites". We suggest that the Agency needs to refer to the SAB/EHC review,
especially its Appendix, and to include mention of the principal limitations of the use of
A-3
-------
the HI, here, in Appendix E, since the HI is incorrectly used in identifying populations
as being significantly exposed to contaminants having non-cancer adverse health
effects.
- page 27.last paragraph (E.3.3): Here, again, the comments made on Chapter 7
need to be taken into account on the subject of the difference between population risk
and population at risk. The first sentence of this paragraph, as well as further
statements within it, need modification.
The problem is that the estimates of population risks associated with carcino-
gens, despite their well known weaknesses, at least purport to make some kind of
estimate, using a model which may or may not apply in a given case, of the number of
people affected (true population risk) as a result of exposure whereas counting the
number of people with HI > 1 for non-carcinogens estimates the number of people at
exposure levels such that they are potentially affected but not necessarily affected
(this is not population risk). Whereas the estimates for carcinogens at least attempt to
get at the number of people affected, the method for non-carcinogens does not
attempt to do so since it includes those exposed and affected, plus those exposed but
not affected in its count. Unless some effort is made to reconcile the two methods,
the result of the non-carcinogen procedure is inconsistent with that for carcinogens
and the two results are not comparable. Moreover, as mentioned above, the
"population-at-risk" result for non-carcinogens does not measure risk and should not
be said or inferred to do so here or elsewhere in the report: it estimates only one
factor in characterizing risk, namely, an estimate of whether an exposed population is
exposed to possibly meaningful levels of the agents involved, but it takes no account
of the probability of such a population actually exhibiting adverse effects. This point
needs to be made and maintained clearly in the text to avoid any misunderstanding by
users if the method.
A-4
-------
REFERENCES
Crump, K.S. 1i84. A new method for determining allowable daily intakes. Fund.
Appi. Tox/co/. (4):854-871.
EPA. 1990. Review of proposed developmental guidelines. EPA-SAB-EHC-iO-013.
Kimmel, CA and Gaylor, D. W. 1988. Issues in qualitative and quantitative risk
analysis for developmental toxicology. Risk Analysis (8);15-20.
R-l
-------
U.S. ENVIRONMENTAL PROTECTION AGENCY
SCIENCE ADVISORY BOARD
ENVIRONMENTAL HEALTH COMMITTEE
RCRA RIA HEALTH BENEFITS REVIEW
CHAIRMAN
Dr. Arthur Upton, Santa Fe, NM
MEMBERS AND CONSULTANTS
Dr. William B, Bunn, Mobile Administrative Services, Company, Inc., Princeton, NJ
Dr. Bavid Gaylor, Department of Health & Human Services, National Center for Toxicological
Research, Jefferson, AR
Dr. Rolf Hartung, School of Public Health, University of Michigan, Ann Arbor, Ml
Dr, Rogene F, Henderson, Inhalation Toxicology Research Institute, Albuquerque, NM
Dr. Marshall Johnson, Jefferson Medical College, Philadelphia, PA
Dr. Nancy K. Kim, New York Department of Health, Albany, NY
Dr. David Howe Wegrnan, University of Lowell, Lowell, MA
Dr. Bernard Weiss, University of Rochester School of Medicine, Rochester, NY
DESIGNATED FEDERAL OFFICER
Mr. Samuel Rondberg, Environmental Health Committee, Science Advisory Board (1400F),
U.S. Environmental Protection Agency, Washington, D.C, 20460
STAFF SECRETARY
Ms. Mary L. Winston, Environmental Protection Agency, Science Advisory Board (140QF),
Washington, D.C. 20460
-------
NOTICE
This report has been written as a part of the activities of the Science Advisory
Board, a public advisory group providing extramural scientific Information and advice
to the Administrator and other officials of the Environmental Protection Agency. The
Board is structured to provide balanced, expert assessment of scientific matters
related to problems facing the Agency. This report has not been reviewed for
approval by the Agency and, hence, the contents of this report do not necessarily
represent the views and policies of the Environmental Protection Agency, nor of other
agencies in the Executive Branch of the Federal government, nor does mention of
trade names or commercial products constitute a recommendation for use.
-------
Distribution List
Administrator
Deputy Administrator
Assistant Administrators
Deputy Assistant Administrator for Research and Development
Deputy Assistant Administrator for Water
EPA Regional Administrators
EPA Laboratory Directors
EPA Headquarters Library
EPA-Regional Libraries
EPA Laboratory Libraries
------- |