July 14, 1999
EPA-SAB-RAC-ADV-99-010
Honorable Carol M. Browner
Administrator
U.S. Environmental Protection Agency
401 M. Street, S.W.
Washington, D.C. 20460
Re: Advisory on Proposed EPA Methodology for Assessing Risks from Indoor
Radon Based on BEIR VI: White paper
Dear Ms. Browner:
The enclosed Advisory was developed by the Radiation Advisory Committee
(RAC) of the Science Advisory Board (SAB) in response to a request dated February
22, 1999 from Mr. Stephen D. Page, Director of the Office of Radiation and Indoor Air
(ORIA) to provide advice on the application in a "White Paper" of the National Academy
of Science's (MAS) Biological Effects of Ionizing Radiation Committee report (BEIR VI).
In particular, the request to the RAC seeks advice on the following charge questions:
a) Is the overall approach of using the BEIR VI age-concentration model
acceptable?
b) What advice does the RAC have on the refinements and extensions we
(ORIA staff) are considering?, and
c) Have we (ORIA staff) adequately accounted for the sources of
uncertainty?
The RAC held a public meeting on March 24, 25, and 26, 1999 at which it was
briefed by, and had technical discussions with ORIA staff, as well as a writing session
by the Committee. The advisory resulting from these meetings responds both to the
three Charge questions briefly discussed below, as well as addressing other issues
identified during the public meetings.
Overall, the Committee was impressed with the quality and focus of the ORIA
effort as presented to us in the draft "White Paper." In general, ORIA has proposed a
reasonable method for using the National Research Council (NRC)/ National Academy
of Sciences (NAS) findings from BEIR VI to form an Agency radon risk model. The
authors should be commended for a very thorough effort in considering most aspects
of this complex risk assessment. The following comments are intended to help ORIA
-------
improve a very good product, and to help ORIA sharpen its approach and
communicate its recommendations more clearly. Our response to the charge
questions and issues beyond the charge are highlighted and summarized below.
a) Question 1: Is the overall approach of using the BEIR VI age-
concentration model acceptable?
The Committee recommends use of a model that provides risk estimates
intermediate between those of the concentration and duration models. This
recommendation is supported by the predictions of other models discussed in BEIR VI,
which also provide intermediate risk estimates. However, the Committee chose not to
recommend any specific method for modeling radon risks, in the belief that ORIA is
better equipped to make that decision.
b) Question 2: What advice does the RAG have on the refinements and
extensions we (ORIA staff) are considering?
The Committee generally supports the extensions of the BEIR VI models that
ORIA proposes to make the EPA radon model most useful for Agency purposes. In
particular, the Committee supports expanded treatment of smoking prevalence by age
and urges ORIA to continue to investigate how to distinguish between the risks for
current smokers and former smokers. The Committee also supports the change of the
definition of risk from "excess" to "etiologic" through the inclusion of radon-related lung
cancers that would be incurred in persons who would have died later from lung cancer
related to other causes. ORIA should also investigate expressing risk in terms of years
of life lost, rather than simply counting cases of early mortality.
c) Question 3: Have we (ORIA staff) adequately accounted for the sources
of uncertainty?
While ORIA has discussed many of the important uncertainties in the radon risk
estimates, extending the BEIR VI discussion in some areas and providing quantitative
uncertainty estimates for some of the input variables, many of the uncertainties remain
unquantified. The White Paper, at this stage of its development, provides little feeling
for the overall uncertainties in the risk estimates stemming from all these input
uncertainties. The Committee, therefore, recommends additional effort to identify,
discuss, and quantify uncertainties to the extent possible with available resources.
Further, identifying the input uncertainties of the risk estimates contributing most to the
uncertainties would help in guiding future research.
d) Beyond the Charge: The need to use the radon risk model for situation-
specific assessments.
The Committee also provided some advice beyond the Charge in Section 4 of
-------
this advisory. The Agency should be sure that the final radon risk model can be used
for situation-specific assessments that require a user-specific mix of sex, age, and
smoking status in the studied population. Further, ORIA should provide easily
understood tools that would allow the model to be used outside ORIA - even by the
general public - to estimate radon risks for a variety of situations. However, ORIA
should be sure to provide cautions and caveats about the interpretation of risk
calculations and about the degree of uncertainty in the modeling procedures.
The RAC appreciates the opportunity to provide this advisory to you and we
hope that it will be helpful. We look forward to the response of the Assistant
Administrator for Air and Radiation to the advisory in general and to the specific
comments and recommendations in this letter in particular.
Sincerely,
Is/
Dr. Joan M. Daisey, Chair
Science Advisory Board
/s/
Dr. Stephen L. Brown, Chair
Radiation Advisory Committee
Science Advisory Board
-------
NOTICE
This report has been written as 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.
-------
ABSTRACT
On March 24-26, 1999, the Science Advisory Board's Radiation Advisory
Committee conducted an advisory for the Office of Radiation and Indoor Air (ORIA) on
a White Paper concerning proposed methodologies for assessing risks from indoor
radon, which was based on the National Academy of Sciences/National Research
Council Biological Effects of Ionizing Radiation (BEIR) VI report.
The Committee found that ORIA has proposed a reasonable method for
extending the findings from BEIR VI to form an Agency radon risk model, and made a
thorough effort in considering most aspects of this complex task. The comments
offered are intended to help ORIA improve a good product, sharpen its approach, and
communicate its recommendations more clearly.
A model that would provide risk estimates between those of the concentration
and duration models was recommended by the Committee, although an exact method
was not proposed. This recommendation is supported by other models discussed in
BEIR VI, which yield intermediate risk estimates.
The Committee generally supports modifications of the BEIR VI models intended
to improve the usefulness of the EPA radon model, including expanded treatment of
smoking prevalence by age and continued investigation on distinguishing the risks of
current and former smokers. While ORIA identified and quantified numerous important
uncertainties in the radon risk estimates, further identification, discussion, and
quantification is desirable.
The final radon risk model should be made usable for assessments that require
specific mixes of sex, age, and smoking status. Further, easily used tools should be
provided so that the model can be used outside of ORIA to estimate radon risks for a
variety of situations.
KEY WORDS: cancer risks, indoor radon exposures, radon models, radon risk
-------
U.S. ENVIRONMENTAL PROTECTION AGENCY
SCIENCE ADVISORY BOARD
RADIATION ADVISORY COMMITTEE
March 24-26, 1999
CHAIR
Dr. Stephen L. Brown, R2C2 Risks of Radiation and Chemical Compounds, Oakland, CA
MEMBERS AND CONSULTANTS
Dr. William Bair, Retired, Richland, WA
Dr. Vicki M. Bier, University of Wisconsin, Madison, Wl
Dr. Thomas F. Gesell, Idaho State University, Pocatello, ID
Dr. David G. Hoel1, Medical University of SC, Charleston, SC
Dr. Richard W. Hornung, University of Cincinnati, Cincinnati, OH
Dr. Janet Johnson, Shepherd Miller, Inc., Ft. Collins, CO
Dr. Donald Langmuir, Hydrochem Systems Corp., Golden, CO
Dr. Jill Lipoti, New Jersey Department of Environmental Protection, Trenton, NJ
Dr. Ellen Mangione, Colorado Department of Health, Denver, CO
Dr. John W. Poston, Sr.1, Texas A&M University, College Station, TX
Dr. Genevieve S. Roessler, Radiation Consultant, Elysian, MN
SAB Staff
Dr. Jack Kooyoomjian, Designated Federal Officer, US EPA, Science Advisory Board,
Washington, DC
Ms. Diana Pozun, Management Assistant, US EPA, Science Advisory Board, Washington, DC
Drs. Poston and Hoel did not attend the RAC' s March review meeting. Dr. Poston provided written
comments, and Dr. Hoel reviewed the draft advisory after the March meeting.
iii
-------
TABLE OF CONTENTS
1. EXECUTIVE SUMMARY 1
1.1 Question 1. Is the overall approach of using the BEIR VI age-
concentration model acceptable? 2
1.2 Question 2. What advice does the RAC have on the refinements and
extensions we (ORIA staff) are considering? 2
1.3 Question 3. Have we (ORIA staff) adequately accounted for
the sources of uncertainty? 2
1.4 Beyond the Charge 3
2. INTRODUCTION 4
2.1 Background 4
2.2 Charge to the SAB 5
3. RESPONSE TO THE CHARGE 6
3.1 Question 1: Is the overall approach of using the BEIR VI age-
concentration model acceptable? 6
3.2 Question 2: What advice does the RAC have on the refinements and
extensions we (ORIA staff) are considering? 8
3.2.1 Overview 8
3.2.2 Baseline Adjustment 8
3.2.3 Smoking Patterns 8
3.2.4 Etiologic Definition 9
3.3 Question 3: Have we (ORIA staff) adequately accounted for the
sources of uncertainty? 9
4. COMMENTS BEYOND THE CHARGE 12
4.1 Uses of the EPA Radon Model 12
4.2 Documentation of the Model 12
APPENDIX A- DETAILED COMMENTS A-1
APPENDIX B- DETAILED EDITORIAL COMMENTS B-1
APPENDIX C- GLOSSARY OF TERMS AND ACRONYMS C-1
REFERENCES R- 1
IV
-------
1. EXECUTIVE SUMMARY
Radon is well established as a cause of lung cancer in miners (particularly
uranium miners) through the inhalation of its radioactive decay products. Radon decay
products in indoor air are also widely assumed to cause lung cancer. However, the
available epidemiology for residential exposures to radon is ambiguous and does not
provide a solid quantitative basis for evaluating the magnitude of the lung cancer risk.
As a result, the quantitative analysis of the miner data has been extrapolated from
radon decay product exposures in mines to lower residential exposures.
The methods used to perform this extrapolation from mines to indoor air have
been debated, however. The National Research Council (NRC) of the National
Academy of Sciences (NAS) in March, 1999 released the report entitled "Health Effects
of Exposure to Radon" ("BEIR VI") on the risk of indoor radon2. The Report discusses
several mathematical models capable of making the extrapolation, with an equal
preference for two models that produce somewhat different estimates of the population
lung cancer burden attributable to indoor radon: the "concentration" model and the
"duration" model. Both models also consider the effects of age, sex, and smoking
status-never smoker (NS) vs. ever smoker (ES). For the relatively low levels of radon
exposure encountered in homes, the concentration model predicts about 40% greater
risk per unit of cumulative exposure. Using data on the average levels of radon in
homes and the characteristics of the 1995 U.S. population with respect to age, sex,
and smoking status, the NRC estimated the cumulative cancer burden (lung cancer
deaths per year) from radon in homes with the two models.
EPAs Office of Radiation and Indoor Air (ORIA) is in the process of deciding
how to use the NRC findings in its own risk assessments for radon decay products.
ORIA must decide whether to adopt one of the two NRC models, to create a hybrid
model, to adopt one of the other models discussed by the NRC but not preferred, or to
create its own model. If ORIA selects an existing model, it must decide what
modifications, if any, are necessary to adapt the model for Agency use. ORIA has
produced a "White Paper" (EPA. 1999) discussing the issues of model selection and
adaptation and presenting its preliminary conclusions about each issue. It has
tentatively chosen to use the concentration model with three important modifications
regarding smoking prevalence by age, accounting for radon-related cancers in the
national cancer statistics, and identification of premature death in persons who would
have later died of lung cancer.
ORIA requested that the Radiation Advisory Committee ( RAC, or "the
Committee") of the Science Advisory Board (SAB) provide advice on its radon modeling
Whenever the Committee refers to the health efects of radon, the role of the radon decay
products is implicit.
1
-------
efforts by answering three charge questions. The Committee's responses to the
questions appear in Section 3 of this Advisory and are summarized below. In general,
ORIA has proposed a reasonable method for using the NRC findings to form an
Agency radon risk model. The authors should be commended for a very thorough
effort in considering most aspects of this complex risk assessment.
1.1 Question 1. Is the overall approach of using the BEIR VI age-concentration
model acceptable?
The Committee believes that the differences in the two primary BEIR-VI models
would have vanished if the NRC had used a continuous rather than a discretized
representation of radon exposure rates, durations of exposure, and cumulative
exposures. Therefore, the ORIA rationale for choosing the concentration model on the
basis of biological plausibility and simplicity of use is not compelling. If ORIA could
base its risk assessment on a continuous model, a choice between models would not
be necessary. Because that option may not be feasible in the near term, the
Committee recommends that ORIA adopt a model that provides risk estimates
intermediate between those of the concentration and duration models. This
recommendation is supported by the predictions of other models discussed in BEIR VI,
which also provide intermediate risk estimates. The Committee chose not to
recommend any specific method for modeling radon risks, in the belief that ORIA is
better equipped to make that decision.At a minimum, ORIA should examine the
strengths and limitations of the categorizations within each model and develop a
rationale and consensus on which categories and ranges should be chosen.
1.2 Question 2. What advice does the RAC have on the refinements and
extensions we (ORIA staff) are considering?
The Committee generally supports the modifications to the BEIR VI models that
ORIA proposes to make the EPA radon model most useful for Agency purposes. In
particular, we support the expanded treatment of smoking prevalence by age and urge
ORIA to continue investigating how to distinguish between the risks for current smokers
and former smokers. We also support the change of the definition of risk from "excess"
to "etiologic" through inclusion of radon-related lung cancers that would be incurred in
persons who would have died later from lung cancer related to other causes. ORIA
should also investigate expressing risk in terms of years of life lost, rather than simply
counting cases of early mortality.
1.3 Question 3. Have we (ORIA staff) adequately accounted for the sources of
uncertainty?
ORIA has discussed many of the important uncertainties in the radon risk
estimates, extending the BEIR VI discussion in some areas and providing quantitative
uncertainty estimates for some of the input variables. However, many of the
-------
uncertainties remain unquantified, and the White Paper at this stage of development
provides little feeling for the overall uncertainties in the risk estimates stemming from all
these input uncertainties. The Committee, therefore, recommends additional effort to
identify and quantify uncertainties to the extent possible with available resources.
Although we recognize the difficulties of such an effort, some guidance on the overall
reliability of the estimates would be welcome. Identifying the input uncertainties
contributing most to the uncertainties in the radon risk estimates through some kind of
sensitivity analysis would help in guiding future research. Most important, it would
provide the public with an indication of the uncertainties associated with prediction of
radon-caused lung cancer deaths.
1.4 Beyond the Charge
The Committee also provided some advice beyond the Charge as presented in
Section 4. In brief, the Agency should be sure that the final radon risk model can be
used for situation-specific assessments that allow for a user-specified mix of sex, age,
and smoking status in the population of concern. Further, it should provide easily
understood tools that would allow the model to be used outside ORIA-even by the
general public-to estimate radon risks for a variety of situations. However, ORIA
should be sure to provide cautions and caveats about the interpretation of risk
calculations and about the degree of uncertainty inherent in the modeling procedures
and results.
The White Paper is not intended for general distribution and assumes a high
level of expertise and familiarity with BEIR VI that would not be found in most
audiences. When ORIA produces its official risk assessment methodology report, more
detailed explanations should be provided, especially on those issues where EPAs
approaches are substantial extensions of those of BEIR VI.
-------
2. INTRODUCTION
2.1 Background
Radon is well established as a cause of lung cancer in miners (particularly
uranium miners) through the inhalation of its radioactive decay products ("radon
progeny"). The epidemiologic evidence from studies of miners is supported by results
from animal studies and radiobiological data. Radon decay products in indoor air are
also widely assumed to cause lung cancer. The available epidemiology for residential
exposures to radon is ambiguous. Although a meta-analysis of a group of residential
studies does provide some support for the assumption that radon in indoor air
contributes to the risk of lung cancer (Lubin and Boice. 1997), these studies do not
provide a solid quantitative basis for evaluating the magnitude of the lung cancer risk.
As a result, the quantitative analysis of the miner data has been extrapolated from mine
exposures to lower residential exposures.
The methods used to perform this extrapolation from mine air to indoor air have
been debated, however. The National Research Council (NRC) of the National
Academy of Sciences in March, 1999 released the report "Health Effects of Exposure to
Radon" ("BEIR VI") on the risk of indoor radon3. The report discusses several
mathematical models capable of making the extrapolation, with an equal preference for
two models that produce somewhat different estimates of the population lung cancer
burden attributable to indoor radon. One model, the "concentration" model, uses the
average level of exposure (in working levels - WL) along with cumulative exposure (in
working level-months - WLM) to predict risk, while the other, the "duration" model, uses
duration of exposure (in years) along with WLM. Both models also consider the effects
of age, sex, and smoking status-never smoker (NS) vs. ever smoker (ES). For the
relatively low levels of radon exposure encountered in homes (generally much less
than 0.5 WL), the concentration model predicts about 40% greater risk per unit of
cumulative exposure. Using data on the average levels of radon in homes and the
characteristics of the 1995 U.S. population with respect to age, sex, and smoking
status, the NRC also estimated the cumulative cancer burden (lung cancer deaths per
year) from radon in homes with the two models.
EPAs Office of Radiation and Indoor Air (ORIA) is currently in the process of
deciding how to use the NRC findings in its own risk assessments of radon. ORIA
must decide whether to adopt one of the two NRC models, to create a hybrid model, to
adopt one of the other models discussed by the NRC but not preferred, or to create its
own model. If ORIA selects an existing model, it must decide what modifications, if
any, are necessary to adapt the model for Agency use. ORIA has produced a "White
Whenever the Committee refers to the health effects of radon, the role of the radon decay
products is implicit.
-------
Paper" discussing the issues of model selection and adaptation and presenting its
preliminary conclusions about each issue (EPA. 1999). It has tentatively chosen to use
the concentration model, modified to account for the following:
1. age-specific smoking prevalence data;
2. early deaths in people who would later have died from lung cancer not
associated with radon; and
3. an adjustment to remove radon-attributable cancer deaths from the
baseline lung cancer risk derived from national cancer mortality statistics
so that the application of the relative risk concentration model will be
consistent.
ORIA is also extending the uncertainty analyses provided in the BEIR VI report.
After considering the comments in this Advisory and perhaps from other
observers, ORIA will produce one or more reports similar to the "Blue Book" (EPA.
1994) that will describe its radon risk methodology and its application to various
problems.
On February 22, 1999, the Director of ORIA, Mr. Stephen D. Page, requested an
advisory of Dr. Donald G. Barnes, Director of the SAB, on the white paper. The RAC
engaged in the advisory at its March 24-26, 1999 public meeting. The Committee
prepared the initial draft advisory during the March public meeting. In the course of the
review, the Committee also identified other issues beyond the charge. These issues
deal principally with uses and documentation of the EPA radon model.
2.2 Charge to the SAB
ORIA requested that the Radiation Advisory Committee ("the Committee") of the
Science Advisory Board (SAB) provide advice on its radon modeling efforts by
answering the following three charge questions:
Question 1. Is the overall approach of using the BEIR VI age-concentration model
acceptable?
Question 2. What advice does the RAC have on the refinements and extensions we
(ORIA staff) are considering?
Question 3. Have we (ORIA staff) adequately accounted for the sources of
uncertainty?
-------
3. RESPONSE TO THE CHARGE
In general, ORIA has proposed a reasonable method for using the NRC findings
to form an Agency radon risk model. The authors should be commended for a very
thorough effort in considering most aspects of this complex risk assessment. The
following comments are intended to help ORIA sharpen its approach and communicate
its recommendations more clearly.
3.1 Question 1: Is the overall approach of using the BEIR VI age-concentration
model acceptable?
The Committee is generally supportive of ORIAs proposal to selecta single
radon risk model rather than maintaining two separate models for future analyses. Use
of both models could well cause confusion among those outside the scientific
community. Based on information presented in BEIR VI and the White Paper, the
estimates of the U.S. lung cancer burden due to residential radon are somewhat
different when using the concentration model vs. the duration model-both of which are
methods for taking into account the inverse exposure rate effect. The concentration
model appears to be more conservative in the sense that it estimates approximately
40% more annual deaths nationwide, a difference that might be politically significant
even if not statistically significant. ORIA needs to consider the importance of such a
perceived difference.
While ORIA has presented two arguments to support its selection of the
concentration model over the duration model for a variety of risk assessment purposes,
neither argument is unassailable. The argument based on biological plausibility
assumes that radon acts principally as an initiator of cancer. The possibility that radon
has cancer promoting activity (e.g., through cell killing and subsequent cell
proliferation) should not be dismissed. The argument based on simplicity falters if
exposure rate is interpreted as average radon decay product concentrations (working
levels) over an extended duration of exposure, as it appears the NRC used for each
miner in its epidemiologic evaluation.
The estimates from both risk models seem to depend on the cut points selected
for exposure and exposure rate in the analysis of the miner data. If the cut points had
been selected differently, both the concentration and duration models could have
yielded different results. A model such as the Cox proportional hazards model (Cox.
1972) can be constructed incorporating both cumulative exposure and either exposure
rate or duration as continuous variables representing different aspects of protracted
exposure. The choice between the latter variables should yield identical likelihoods,
due to the fact that cumulative exposure in WLM is simply the product of intensity of
exposure (measured in WL) and duration of exposure in months. Therefore, the use of
multiplicative relative risk models, such as those used in BEIR VI, means that inclusion
-------
of either duration or concentration makes the other unnecessary. The primary reason
that the BEIR VI concentration and duration models differ in their estimates of the
overall risks appears to be that each of these variables was categorized independently
and their product is not necessarily equal to cumulative exposure. If ORIA could base
its risk assessment on a continuous model, a choice between models would not be
necessary.
The rationale in the White Paper for choosing the concentration model,
therefore, becomes arbitrary. We have already questioned the biological argument and
the simplicity argument. Orally, ORIA also offered the argument that the concentration
model would be more easily understood by the general public. Although the
argument has some merit, it is in itself not a compelling reason to choose the
concentration model to represent the inverse dose-rate effect.
If using a continuous radon risk model is not feasible in the near term, the
Committee recommends that ORIA adopt a model that produces estimates
intermediate between the concentration and duration models. For example, the
constant relative risk (CRR) model fitted to miners with relatively low exposures or the
meta-analysis model of the residential studies, both estimated in BEIR VI, produce risk
estimates greater than the duration model but less than the concentration model. Both
of these models are also appealing since they represent exposures more relevant to
the indoor radon problem. However, the Committee recognizes that the latter models
do not include coefficients that modify risk for attained age and time since exposure,
which reduces their appeal for estimating risk using the EPA lifetable approach.
Rather than recommend a specific model for use by EPA, the Committee prefers to
defer the final model choice to EPA, with the recommendation that the weight of
evidence seems to indicate that the most accurate estimates are between those
produced by the duration and concentration models.
The Committee notes that cancer risk theory predicts differing risks for the same
cumulative exposure depending on the timing and pattern of exposure, with the
direction of the differences depending on the mechanism of carcinogenesis (e.g.,
initiation vs. promotion) (Crump and Howe, 1984). As proposed, the concentration
model is probably better at dealing with time-varying exposures than are the other
proposed models, but all are deficient in some respect and further research in this
arena would be useful. The use of average exposure rates over periods of time in
WLM/month may have caused some bias in the radon risk estimates. If the variation in
the estimated average exposure rates was randomly distributed about the true
exposure rates in each mine, the bias would be toward lower risks4. If the variation was
systematic (e.g., exposure rates declining with time for most miners), the bias might be
BEIR VI actually did use time-varying exposures by calculating each miner's cumulative
exposure at specific age intervals. Bias toward the null would occur if WLM estimates
were unbiased but subject to random misclassification.
-------
in the other direction.
3.2 Question 2: What advice does the RAC have on the refinements and
extensions we (ORIA staff) are considering?
3.2.1 Overview
The proposed methodology for calculating radon risk described in the White
Paper is very sophisticated (EPA. 1999). It builds upon the BEIR VI report conclusions
(NRC. 1998). The White Paper takes the information about radon risk that was
developed by the NRC and puts it into a form that can be used by the EPA for a variety
of purposes including, but not limited to, updating its public information aimed at
reducing residential radon exposures. For instance, ORIA improved the usefulness of
the radon risk estimate by the inclusion of smoking patterns from the data supplied by
the Department of Health and Human Services.
The refinements EPA has used in deriving the risk numbers appear to be
reasonable, although in light of the level of uncertainty in the basic model at low radon
concentrations, some of these refinements may have little effect on the effective range
of the risk estimates. However, the Committee found it difficult to determine how much
of the increase in estimated cases of lung cancer in the US between the 1992 EPA
estimate and the various examples given in the White Paper is due simply to the
increase in size and change in composition of the population on which the risk is
projected, and how much of the increase is due to increased risk per person.
3.2.2 Baseline Adjustment
The Committee had difficulty understanding ORIA's adjustment to the baseline
risks to account for the contribution of radon to the current lung cancer statistics. The
Committee agrees that ORIA is correct to avoid double counting of risks in the relative
risk approach. However, when the adjustment is applied, it effectively makes the
baseline cancer rates applicable to a hypothetical population exposed to a radon level
of zero in the home.5 It does not correct for the ambient (outdoor) radon level.
Because no population exists with a zero exposure to radon, the results are difficult to
understand, interpret, and communicate. It would be easier to understand a baseline
adjustment consistent with the state of the art in radon mitigation, which would bring
the radon level in residences to a credible but non-zero level.
3.2.3 Smoking Patterns
5Although BEIR-VI was directed specifically at residential radon, its methods can be
used, with small modifications, to predict risks from workplace radon.
-------
The Committee strongly supports an ORIA effort to incorporate a category for
former smokers. The general U.S. population has a higher prevalence of former
smokers than in the 11 cohorts of miners used in the BEIR VI report. Also, the relative
risk for lung cancer due to smoking declines with time from cessation of smoking. This
decrease should be reflected in lower numbers of excess lung cancers if these
estimates are used in ORIA's life table approach. This approach has the added benefit
of demonstrating to the public that the best strategy includes reducing indoor radon
levels as well as eliminating or reducing cigarette smoking.
EPA proposes to add a discussion on how changes in smoking patterns might
impact estimates of risk. It is even more important to discuss how changes in radon
exposure through the installation of a mitigation system could affect risk.
3.2.4 Etiologic Definition
The NRC used an "excess" definition of risk in which only the excess of lung
cancer cases over those which would have occurred through other causes are counted.
ORIA is proposing to shift to an "etiologic" definition in which cases of radon-induced
lung cancer are counted even if they occur in people who would have died of lung
cancer (from other causes) later on. Because the etiologic definition takes into account
the burden of earlier occurrence of cancer, the Committee supports this change.
It is interesting that for the example of the female NS (page 15, Fig.3), virtually
all of the lung cancer mortality for ages up to 50 to 55 years is attributable to radon.
There is a school of thought that years of life lost, or even quality-adjusted life years
(QALY) are more appropriate parameters with which to gauge detriment than is simply
the number of premature deaths (Holmes, 1995). With this in mind, the apparent fact
that lung cancer deaths attributable to radon begin 10-15 years earlier than those
from remaining causes should perhaps be highlighted. Certainly, the Committee
supports the presentation of the radon risk as the probability of dying prematurely from
a radon-induced lung cancer.
3.3 Question 3: Have we (ORIA staff) adequately accounted for the sources of
uncertainty?
Although the discussion of uncertainties is a good start and many of the sources
of uncertainty are adequately treated, some sources of uncertainty and their
importance have not been adequately discussed. The White Paper does not provide a
good sense of the overall uncertainty of the risk estimates in quantitative terms or
which of the component uncertainties is most important. In some cases, even the
qualitative identification and discussion of uncertainties could be improved.
For example, the results of the Cohen study are dismissed using the arguments
that ecologic studies are not valid for establishing causation (due to possible
-------
confounding), and also that the results are biologically implausible and inconsistent
with residential case-control studies. Cohen ( Cohen. 1990) has attempted to account
for confounding by smoking and other factors to the extent possible using population
data. The RAC recognizes that ecologic studies have a limited ability to address
confounding, because data on individuals are not available; however, the results of
such studies should be acknowledged in a discussion of uncertainties. At low
exposure levels, the case-control studies, in general, seem to show very little effect.
Finally, the assertion of biological implausibility ignores the possibility that at low levels,
radiation may stimulate the immune system (hormesis) (Sakamoto, Miyamoto and
Watabe. 1987). Although the Committee is not advocating the hermetic hypothesis or
other threshold or nonlinear exposure-response relationships, from the evidence now
available, a threshold exposure (i.e., a level of exposure below which radon has no
effect) cannot be excluded. Therefore, we believe the White Paper should
acknowledge in its uncertainty assessment the possibility that radon may not increase
lung cancer risks at very low levels of exposure.
More discussion would also be useful on the uncertainties attributable to
potential misestimation of miner's exposures, and on how changes in smoking patterns
or cigarette composition would affect calculated risks. With respect to miner's
exposures, not only are there errors in average exposure estimates (which may cause
bias, as noted in the White Paper), but individuals with the same average exposure
may have different timing of exposure, and hence different lung cancer risks.
Moreover, the relevance of the Chinese cohort data seems questionable, for a variety of
reasons. Finally, susceptibility due to either genetic or environmental factors might be
different in miners than in the general population, and this possibility contributes to
uncertainties in the residential risk estimates.
With respect to smoking patterns, there is great variability between occasional
and heavy smokers. Although the Committee realizes that it may not be possible to
factor this variability into the model, perhaps data are available to at least characterize
the extent of variability in smoking patterns. There is also uncertainty about the relative
risk of radon for former smokers, and about radon risks to passive smokers. Finally,
differences in lung dose per WLM exposure could vary by age, sex, and smoking
status.
As other examples, reductions in radon exposure not only will change the
distribution of exposures used in the cancer burden estimates, but may also require the
adjustment in the risk coefficients that account for the contribution of the baseline
radon exposure to observed lung cancer rates. Moreover, if the equilibrium fraction for
radon progeny correlates with the radon level in homes, residential risk estimates
would be affected.
It is extremely important that the sources of uncertainty be discussed in some
detail so that the reader can understand the assumptions made in the model, the
10
-------
range of possible values for each parameter, and-at least qualitatively-the impact of
these uncertainties on the resulting estimates. Although the Committee realizes that
quantification will often be difficult if not impossible, ORIA should attempt a sensitivity
analysis to identify which of the input uncertainties contribute most to the overall
uncertainties in the risk estimates.
Finally, although uncertainty is discussed in Section VI, statements of
uncertainty are not provided in the rest of the document. While it is clear that the
numbers in the White Paper are for illustration only and do not represent EPA policy at
this time, qualitative expressions of uncertainties should be associated with numerical
estimates in any future descriptions or applications of this methodology that are
intended for broader distribution.
11
-------
4. COMMENTS BEYOND THE CHARGE
4.1 Uses of the EPA Radon Model
Although the White Paper provides several examples of uses to which the radon
risk model could be put, its only numeric example is an estimate of the lung cancer
burden (deaths per year) attributable to residential radon. Much of the discussion of
model choice and modifications seems to focus on this use. However, the models will
undoubtedly be used for situation-specific assessments as well, with possibly even
more important policy implications. It therefore seems desirable to provide the users
with the ability to input their own mix of sex, age, and smoking status, rather than to
have them be forced to use the national mix at some specific point in time.
However the risk model is finally formulated, it would be desirable for ORIA to
provide tools that would allow it to be used outside ORIA-even by the general public-
to estimate radon risks for a variety of situations. The tools should be easy to
understand and use, but they should be accompanied by cautions about the
interpretation of risk calculations and about the degree of uncertainty inherent in the
modeling procedures and their results to minimize the potential for misuse of the risk
estimates. Guidance should be provided on uses of the models that would be
inappropriate or misleading.
4.2 Documentation of the Model
The White Paper is generally quite readable for an audience familiar with the
radon risk literature. However, it assumes a high level of expertise and familiarity with
BEIR VI that would not be found in most audiences. When ORIA produces its official
risk assessment methodology report, more detailed explanations should be provided,
especially on those issues where EPAs approaches are substantial extensions of those
of BEIR VI. Further, explanations need to be clearer and better justified, especially
concerning why EPA chose the "concentration" versus the "duration" model, as well as
the methodology for adjusting baseline lung cancer death rates for the existing
distribution of residential radon levels. The latter concept seems to have been made
more difficult than necessary, and several of the Committee members were confused
by the discussion. Although it does not appear that the White Paper itself will be
distributed widely, it will be important to keep these points in mind to the extent that
this document serves as the basis for later documents and as a reference for health
professionals in communicating with the public. Moreover, derivation of some of the
equations used would be helpful to the critical reader. For example, equation (4)~
determination of lung cancer rates for non-smokers from population lung cancer rates-
-would be clearer if a short derivation were included either in the text or as an
appendix.
The quantity "exposure rate" and its units are not used consistently in the
12
-------
document. In Table 2, and on page 14 (Section 3b), the units for exposure rate are
WL; however, on page 13, in an example used to bolster arguments against the
duration model, exposure rate is expressed in WLM/y. Furthermore, ORIA needs to
clarify that the actual quantity used in BEIR VI is WLIW month, not the instantaneous
concentration as expressed by WL. A related question regards the duration model. If
exposure rate is expressed in WL, are comparisons between exposures in mines and
dwellings with this model comparable, given that occupancy times for mines (~24%)
and dwellings (assumed to be 70%) differ by nearly a factor of three? Equal radon
decay product concentrations in mines and dwellings will lead to quite different
average exposure rates when the averages are taken over a day or more.
The definitions of smoking status (ever smoker, never smoker, and former
smoker) need to be sharpened to avoid ambiguity. Number of cigarettes smoked daily
or in total, duration of smoking, age at first smoking, and time since cessation of
smoking could be used to define the categories. Moreover, the uncertainties
introduced by these category definitions need to be discussed. Using only two or three
categories for a continuum of behavior introduces the same kinds of difficulties as
defining the increments of age or exposure rates used in the risk equations. How
should "passive" smokers (e.g., children or spouses of heavy smokers) be classified?
Although the title and the Introduction clearly state that the White Paper is
intended to address methodology, the document ends abruptly. Perhaps the fact that
this is just methodology should be recapped at the end, and there should be a
paragraph or two that says what EPA will do next. The three uses of the methodology
as outlined in the Introduction should be repeated and expanded upon there.
After releasing its methodology report, a next step would be to prepare a paper
for a peer-reviewed journal such as Health Physics. This Committee has recommended
previously that EPA be more active in their activities within the radiation protection
societies. EPA can gain much credibility by making the radiation safety professionals
more aware of these solid scientific contributions.
13
-------
APPENDIX A- DETAILED COMMENTS
A-1 Comments on Units , Terminology, Tables, and Figures
a) The Uncertainty Analysis section of the White Paper refers to "errors in
dosimetry." These are not really errors in dosimetry, but errors in
exposure estimates.
b) On page 3: There needs to be more of an explanation of Working Level.
How does it equate to pCi/L?
c) Be sure all details such as units, etc. are given in Tables and Figures to
make these self-explanatory.
d) Tables and figures need to be labeled more clearly...for example, where
table titles say "number", the symbol # or the word "number" should also
appear immediately adjacent to the appropriate column or row. Every
attempt should be made to facilitate a reader's correct and timely
interpretation of the tables.
e) The risk estimates in Table 7 should be multiplied by 10"4.
A- 2 Recommendations for Clarifying the Report
a) In Section D, it is stated that the MAS preferred models are applicable to
any population. Is this really the case? What about populations with
compositions different from the US average? Populations within the U.S.
may differ substantially in age and sex distribution, smoking prevalence,
race or ethnicity, and other factors potentially influencing risk.
b) On page 6, more explanation for the value of 0.9, by which p is adjusted
downward, would be useful.
c) On page 26, the basis for assuming a constant rate of change is not
explained. Given only two data points from which to extrapolate, a linear
extrapolation might just as well be postulated. Having decided to do a
constant rate of increase projection, a formula for the more general case
should be given, rather than just a numerical example. It could take the
form:
Moreover, exactly how the prevalence rates were adjusted downward "to
A-l
-------
be in agreement with estimates from the OSH" is not clear.
d) On page 32 (Section C2), the White Paper states that the rate of lung
cancer in females is approaching that in males. However, female lung
cancer mortality rates are still less than one half those of males. (EPA.
1994).
e) On page 4, more explanation of why the "inverse dose rate" effect makes
biological sense is needed.
f) On page 3 (paragraph 2), the statement that "...the right hand side...is
multiplied by a factor K..." requires further explanation/justification.
g) On page 3 (paragraph 3), some explanation of "...life table techniques..."
would be helpful: how they are used, why it makes sense to use them
here. This could be done in a footnote.
h) On page 3 (paragraph 3), the White Paper states that the methodology
"...subtracted off the estimated radon-induced lung cancer deaths..."
Further explanation is needed.
i) On page 5 (paragraph 1), the White Paper states that "...the dose to
target cells...was typically about 30% lower for a residential exposure as
compared to an equal WLM exposure in mines." Some additional (short)
explanation of how MAS reached this conclusion would be helpful (e.g.,
footnote, appendix, etc.).
j) On page 6 (paragraph 2), the White Paper states that attributable risk "...
is only weakly dependent on lung cancer rates." Why?
k) On page 9 (paragraph 1), the White Paper states that "...BEIR VI did not
provide numerical estimates of the risk per WLM." This is an important
point. Why did BEIR VI decide not to do this? Was it beyond their
charge? Did they feel that the information was too weak to support such a
move? A word of explanation as to why EPA needs to do this would be
important.
I) On page 13 (paragraphs 3-4) the White Paper states that "...we prefer
the concentration model..." The White Paper needs to provide more than
the two-paragraph explanation currently devoted to this-perhaps some
discussion of the likely convergence of the 2 models would be helpful.
How different are the results of these 2 models likely to be?
m) On page 19 (paragraph 1), the White Paper contrasts "AR" and "... the
A-2
-------
risk per WLM ..." There are several epidemiologic approaches used in the
document, which makes the document confusing. Perhaps a glossary
with a brief discussion of how each is used and which is the most
pertinent/revealing for this discussion would be helpful (or just making
the text clearer regarding the utility of each measure).
n) On pages 30-31, the discussion of uncertainties in miner data is nicely
done. However, the Committee members were not all familiar with
"submultiplicative" (page 31, paragraph 4) as a scientific term.
o) On page 9, please explain item 2.
p) On page 13 (paragraph 3), this discussion could be strengthened with
reference to observations of an inverse dose rate effect in experiments
with alpha emitters in laboratory animals. However, observations at high
exposure levels do not necessarily imply similar behavior in the residential
radon exposure range. In the proposed model, risk per WLM does not
increase further below 0.5 WL. The possibility that the increase could
either continue or reverse should be discussed in the uncertainties
section. The relationship between the inverse dose rate effect and dose
or dose rate could be discussed.
q) On page 32 (Section 2), could there be a difference in dosimetry between
males and females?
r) The assumption of few new smokers after age 21 should be documented.
s) On page 33 (paragraph 2), what about the relatively short life-spans of
the Chinese miners?
t) On page 34 (last paragraph), the Committee recollects that in some of the
animal experiments cited, radon and smoking exposures generally
occurred on the same day in both groups (smoking followed by radon or
vice versa). If this is the case, then they would not be very relevant for
assessing the relative susceptibility of children to radon if they became
smokers many years later. ORIA should check this point.
A-3
-------
APPENDIX B- DETAILED EDITORIAL COMMENTS
a) On page 2, the citation (MAS, 1990) is not in the references. Should it be
MAS (1988)?
b) On page 14, "0.69 for a < 75 y" should be "0.69 for a > 75 y".
c) On page 29 (paragraph 2), the White Paper states that "...the
committee's preferred uncertainty estimates were obtained from the CRR
model." This model should have been introduced earlier in the document
as another model with which EPA staff does not agree.
d) On page 10 (next to last line), should this be hPop(a)7
e) On page 29 (line 2), omit one "these."
g) On page 32 (line 2), the statement should say "There are very few data".
B- 1
-------
APPENDIX C- GLOSSARY OF TERMS AND ACRONYMS
ADV Advisory
BEIR Biological Effects of ionizing Radiation
Ci Cune
CRR Constant Relative Risk
EPA U.S. Environmental Protection Agency (U.S. EPA, EPA, or "the Agency")
ES Ever Smoker
L Liter
MAS National Academy of Sciences
NRC National Research Council
NS Never Smoker
ORIA Office of Radiation and indoor Air (U.S. EPA)
p Pico (one trillionth, e.g., 10"12 Ci is a picocurie)
QALY Quality-Adjusted Life Years
RAC Radiation Advisory Committee (of the U.S. EPA/SAB/RAC)
SAB Science Advisory Board (of the U.S. EPA/SAB)
U.S. United States
vs Versus
WL Working Levels
WLM Working Level-Month
y Year
C-l
-------
REFERENCES
1. Cohen, B. L.. 1990. "A Test of the Linear-No Threshold Theory of Radiation
Carcinogenesis," J. Environ. Res. 53:193-220
2. Cox, D.R.. 1972. "Regression Models and Life Tables," J. Royal Stat. Soc.,
Series B, 34:187
3. Crump, K. S. and R. B. Howe. 1984. Cancer model with time-dependent
dosing. Risk Analysis 4:163-176.
4. Environmental Protection Agency. 1999. Proposed EPA Methodology for
Assessing Risks from Indoor Radon Based on BEIR VI. Office of
Radiation and Indoor Air, February Draft
5. Environmental Protection Agency. 1994. Revised Methodology for Estimating
Radiogenic Cancer Risks, EPA 402-R-93-076, June (EPA Blue Book)
6. Holmes, A. M. 1995. A QALY-based societal health statistic for Canada, 1985.
Social Science and Medicine 41(10): 1417-27
7. Landis, S. H., T. Murray, S. Bolden, T.A. Wingo. 1998. Cancer Statistics, 1998.
CA (A Cancer Journal for Clinicians), 48(1): 6-29
8. Lubin, J. H., and J. D. Boice. 1997. Lung cancer risk from residential radon:
meta-analysis of eight epidemiologic studies. J. Natl. Cancer Inst. 89:49-
57
9. National Research Council. 1998. Health Effects of Exposure to Radon. National
Academy Press, BEIR VI, Washington, DC. (Prepublication Copy),
Thursday, February
10. Sakamoto, K., M. Miyamoto, and N. Watabe. 1987. Effect of low-dose total body
irradiation on tumor control. Can to Kagaku Ryoho (Japan) 14:5. 1545-
1549. May
-------
DISTRIBUTION LIST
Deputy Administrator
Assistant Administrators
EPA Regional Administrators
EPA laboratory Directors
Deputy Assistant Administrator for Office of Policy
Director, Office of Strategic Planning and Environmental Data
Director, Office of Policy Analysis
Director, Office of Regulatory Management and Evaluation
Deputy Assistant Administrator for Air and Radiation
Director, Office of Radiation and Indoor Air
Director, Office of Radiation Programs
EPA Headquarters Libraries
EPA Regional Libraries
-------
United States Science Advisory EPA-SAB-RAC-ADV-99-010
Environmental Board July 1999
Protection Agency Washington DC
ADV|SORY:
ASSESSING RISKS FROM
INDOOR RADON
AN ADVISORY BY THE RADIATION
ADVISORY COMMITTEE ON
PROPOSED EPA METHODOLOGY
FOR ASSESSING RISKS FROM
INDOOR RADON
R-
------- |