The Health Effects of Exposure to Indoor Radon


BEIR VI - Executive Summary Released to EPA by permission of The National Academy
of Sciences (NAS) February 19, 1998

Biological Effects of Ionizing Radiation (BEIR) VI Report: "The Health
Effects of Exposure to Indoor Radon" INTRODUCTION This National Research
Council's report of the sixth Committee on Biological Effects of Ionizing Radiations
(BEIR VI) addresses the risk of lung cancer associated with exposure to radon
and its radioactive progeny. Radon, a naturally occurring gas formed from the
decay of uranium in the earth, has been conclusively shown in epidemiologic
studies of underground miners to cause lung cancer. There is supporting evidence
from experimental studies of animals that confirm radon as a cause of lung cancer
and from molecular and cellular studies that provide an understanding of the
mechanisms by which radon causes lung cancer.

In addition to being present at high concentrations in many types of underground
mines, radon is found in homes and is also present outdoors. Extensive measurements
of indoor radon concentrations in homes show that although concentrations vary
widely, radon is universally present, raising concerns that radon in homes increases
lung-cancer risk for the general population, especially those who spend a majority
of their time indoors at home. For the purpose of developing public policy to
manage the risk associated with indoor radon, there is a need to characterize
the possible risks across the range of exposures received by the population.
The higher end of that range of exposures is comparable to those exposures that
caused lung cancer in underground miners. The lower end of that range includes
exposures received from an average indoor lifetime exposure which is at least
one order of magnitude lower.

Risk models, which mathematically represent the relationship between exposure
and risk, have been developed and used to assess the lung-cancer risks associated
with indoor radon. For example, the precursor to this committee, the BEIR IV
committee, developed one such model on the basis of statistical analysis of
data from 4 epidemiologic studies of underground miners. The BEIR IV model has
been widely used to estimate the risk posed by indoor radon. Since the 1988
publication of the BEIR IV report, substantial new evidence on radon has become
available: new epidemiologic studies of miners have been completed, existing
studies have been extended, and analysis of the pooled data from 11 principal
epidemiologic studies of underground miners has been conducted involving a total
of 68,000 miners and to date, 2,700 deaths from lung cancer. Other lines of
scientific evidence relevant to assessing radon risks have also advanced, including
findings on the molecular and cellular basis of carcinogenesis by alpha particles.
Radon itself does not directly cause lung cancer but alpha particles from radon
progeny directly damage target lung cells to cause cancer. There is additional
information for calculating the dose of alpha particles received by the lung
from inhaled radon progeny, the topic of a 1991 follow-up report to the BEIR
IV report, the report of the National Research Council's Panel on Dosimetric
Assumptions. Finally, during the last decade, a number of epidemiologic case-control
studies that estimated the risk associated with indoor radon directly have also
been implemented.

The BEIR VI committee faced the task of estimating the risks associated with
indoor radon across the full range of exposures and providing an indication
of the uncertainty to be attached to risk estimates across this range. In preparing
this report, the BEIR VI committee, in response to its charge, reviewed the
entire body of data on radon and lung cancer, integrating findings from epidemiologic
studies with evidence from animal experiments and other lines of laboratory
investigation. The committee also considered the substantial evidence on smoking
and cancer and the more limited evidence on the combined effect of smoking and
radon. The report's elements include comprehensive reviews of the cellular and
molecular basis of radon carcinogenesis and of the dosimetry of radon in the
respiratory tract, of the epidemiologic studies of miners and the general population,
and of the combined effects of radon and other occupational carcinogens with
tobacco-smoking. The committee describes its preferred risk models, applies
the models to estimate the risk posed by indoor radon, and characterizes uncertainties
associated with the risk estimates.

THE MECHANISTIC BASIS OF RADON-INDUCED LUNG CANCER Information on radon carcinogenesis
comes from molecular, cellular, animal, and human (or epidemiologic) studies.
Radiation carcinogenesis, in common with any other form of cancer induction,
is likely to be a complex multi-step process that can be influenced by other
agents and genetic factors at each step. Since our current state of knowledge
precludes a systematic quantitative description of all steps from early subcellular
lesions to observed malignancy, the committee used epidemiologic data to develop
and quantify an empirical model of the exposure-risk relationship for lung cancer.
The committee did draw extensively, however, on findings from molecular, cellular,
and animal studies in developing its risk assessment for the general population.

The committee's review of the cellular and molecular evidence was central to
the specification of the risk model. This review led to the selection of a linear
non-threshold relation between lung-cancer risk and radon exposure. However,
the committee acknowledged that other relationships, including threshold and
curvilinear relationships, cannot be excluded with complete confidence, particularly
at the lowest levels of exposure. At low radon exposures, typical of those in
homes, a lung epithelial cell would rarely be traversed by more than one alpha
particle per human lifespan. As exposure decreases, the insult to cell nuclei
that are traversed by alpha-particles remains the same as at higher exposures,
but the number of traversed nuclei decreases proportionally. There is good evidence
that a single alpha particle can cause major genomic changes in a cell, including
mutation and transformation. Even allowing for a substantial degree of repair,
the passage of a single alpha particle has the potential to cause irreparable
damage in cells that are not killed. In addition, there is convincing evidence
that most cancers are of monoclonal origin, that is, they originate from damage
to a single cell. These observations provide a mechanistic basis for a linear
relationship between alpha-particle dose and cancer risk at exposure levels
at which the probability of the traversal of a cell by more than one alpha particle
is very small, that is, at exposure levels at which most cells are never traversed
by even one alpha particle. On the basis of these mechanistic considerations,
and in the absence of credible evidence to the contrary, the committee adopted
a linear-non-threshold model for the relationship between radon exposure and
lung-cancer risk. However, the committee recognized that it could not exclude
the possibility of a threshold relationship between exposure and lung cancer
risk at very low levels of radon exposure.

Extrapolation from higher to lower radon exposures is also influenced by the
inverse dose-rate effect, an increasing effect of a given total exposure as
the rate of exposure is decreased, as demonstrated by experiments in vivo and
in vitro for high-LET radiation, including alpha particles, and in miner data.
This dose-rate effect, whatever its underlying mechanism, is likely to occur
at exposure levels at which multiple particle traversals per cell nucleus occur.
Mechanistic, experimental, and epidemiologic considerations support the disappearance
of the effect at low exposure corresponding to an average of much less than
one traversal per cell location, as in most indoor exposures. Extrapolating
radon risk from the full range of miner exposures to low indoor exposures involves
extrapolating from a situation in which multiple alpha-particle traversals of
target nuclei occur to one in which they are rare; such an extrapolation would
be from circumstances in which the inverse dose-rate effect might be important
to one in which it is likely to be nonexistent. These considerations indicated
a need to assess risks of radon in homes on the basis of miner data corresponding
to as low an exposure as possible, or to use a risk model that accounts for
the diminution of an inverse exposure-rate effect with decreasing exposure.

The committee also reviewed other evidence relevant to the biologic basis of
its risk assessment approach. For the combined effect of smoking and radon,
animal studies provided conflicting evidence on synergism, and there is uncertainty
as to the relevance of the animal experiments to the patterns of smoking by
people. Early attempts to identify a molecular "signature" of prior
alpha-particle damage through the identification of unusual point mutations
in specific genes have not yet proven useful, although approaches based on specific
chromosomal aberrations show some promise, and all the principal histologic
types of lung cancer can be associated with radon exposure. Available evidence,
albeit limited, supports the likelihood that a typical human population would
have a broad spectrum of susceptibility to alpha-particle-induced carcinogenesis.

THE BEIR VI RISK MODELS For estimating the risk of indoor radon, the committee
chose an empirical approach based on analysis of data from radon-exposed miners.
Other approaches that the committee considered but did not use included a "dosimetric"
approach, and use of "biologically-motivated" risk models. A dosimetric
approach, in which radon risks are estimated by applying risk estimates from
A-bomb survivor studies to estimates of radiation doses delivered to the lung,
was not pursued because of the major differences in the type of radiation and
exposure patterns compared with radon-progeny exposure. A biological-based approach
to modeling with a description of the various processes leading to radon-induced
cancer was not followed primarily because of the present incomplete state of
knowledge of many of these processes.

The committee turned to the empirical analysis of epidemiologic data as the
basis for developing its risk model. Two sources of information were available:
data from the epidemiologic studies of underground miners and data from the
case-control studies of indoor radon and lung cancer in the general population.
Both groups include ever-smokers and never-smokers. Although the case-control
studies provide direct estimates of indoor radon risk, the estimates obtained
from these studies are very imprecise, particularly if estimated for never-smokers
or ever-smokers separately, because the excess lung cancer risk is likely to
be small. Other weaknesses of the case-control studies are errors in estimating
exposure and the limited potential for studying modifying factors, particularly
cigarette smoking. Nonetheless, the committee considered the findings of a meta-analysis
of the 8 completed studies.

In developing its risk models, the committee started with the recently reported
analyses by Lubin and colleagues of data from 11 studies of underground miners--uranium
miners in Colorado, New Mexico, France, Australia, the Czech Republic, and Canada;
metal miners in Sweden; tin miners in China; and fluorspar miners in Canada.
The data for 4 studies were updated with new information. These 11 studies offered
a substantially greater data resource than had been available to the BEIR IV
committee. The 11 epidemiologic studies covered a range of mining environments,
times, and countries, and their methods of data collection differed in some
respects.

The committee analyzed the data with a relative-risk model in which radon exposure
has a multiplicative effect on the background rate of lung cancer. In particular,
the committee modeled the excess relative risk (ERR), which represents the multiplicative
increment to the excess disease risk beyond background resulting from exposure.
The model represents the ERR as a linear function of past exposure to radon.
This model allows the effect of exposure to vary flexibly with the length of
time that has passed since the exposure, with the exposure rate, and with the
attained age. The mathematical form of the model for ERR is:

ERR = (w5-14 + 15-24 w15-24 + 25+w25+)age z

The parameter represents the slope of the exposure-risk relationship for the
assumed reference categories of the modifying factors. Exposure at any particular
age has 4 components: exposure in the last 5 years--excluded as not biologically
relevant to cancer risk--and exposures in 3 windows of past time, namely 5-14,
15-24, and 25 or more years previously. Those exposures are labeled w5-14, w15-24,
and w25+, respectively, and each is allowed to have its own relative level of
effect, 5-14(set equal to unity), 15-24, and 25+, respectively. With this weighting
system, total exposure can be calculated as w* = w5-14 + 15-24w15-24 + 25+w25+.
The rate of exposure also affects risk through the parameter z; thus, the effect
of a particular level of exposure increases with decreasing exposure rate, as
indexed either by the duration of exposure or the average concentration at which
exposure was received. The ERR also declines with increasing age, as described
by the parameter age.

Based on this analysis, the committee developed two preferred risk models referred
to as the exposure-age-concentration model and the exposure-age-duration model.
These two models differ only with respect to the parameter z, which represents
either duration of exposure or the average concentration over the time of the
exposure. The models were equally preferred by the committee. The new models
are similar in form to the BEIR IV model, but have an additional term for exposure
rate and more-detailed categories for the time-since-exposure windows and for
attained age.

RISK ASSESSMENT The committee's risk models can be used to project the lung-cancer
risk associated with radon exposure, both for individuals and for the entire
US population. To extend the models that were developed from miner data to the
general population, the committee needed to make a set of assumptions on the
following key issues.

Lung Dosimetry of Radon Progeny Physical and biologic differences between the
circumstances of exposures of male miners working underground and of men, women,
and children in their homes could lead to differing doses at the same exposures.
The committee estimated the value of a dimensionless parameter, termed the "K
factor" in prior reports, that characterizes the comparative doses to lung
cells in homes and mines for the same exposure. Using a model to estimate the
dose to the cells in the lung, and incorporating new information on the input
parameters of the model, the committee found that the doses per unit exposure
in mines and homes were essentially the same. Thus, K is calculated to be about
1 for men, women and children (age 10 years), and slightly above K=1 for infants
(age 1). Consequently, a value of 1 was used in making the risk projections.

Extrapolation of Risks at Higher Exposures to Lower Exposures Average exposures
received by the miners in the epidemiologic studies are about one order of magnitude
higher than average indoor exposures, although the lowest exposures of some
miners overlap with some of the highest indoor exposures. To estimate risks
of indoor radon exposures, it is thus necessary to make an assumption about
the shape of the exposure-risk relationship across the lower range of the distribution
of radon exposures.

The committee selected a linear-non-threshold relationship relating exposure
to risk for the relatively low exposures at issue for indoor radon. This assumption
has significant implications for risk projections. Support for this assumption
came primarily from the committee's review of the mechanistic information on
alpha-particle-induced carcinogenesis. Corroborating information included evidence
for linearity in the miner studies at the lower range of exposures, and the
linearity and magnitude of risk observed in the meta-analysis of the case-control
studies, which was fully consistent with extrapolation of the miner data. Although
a linear-non-threshold model was selected, the committee recognized that a threshold-that
is, a level of exposure with no added risk-could exist and not be identifiable
from the available epidemiologic data.

Exposure Rate At higher exposures, the committee found evidence in the miner
data of an inverse exposure-rate effect. Theoretical considerations suggested
that the inverse exposure-rate effect found in the miner data should not modify
risks for typical indoor exposures. Consequently, the exposure-rate effect in
the lowest range of miner exposure rates was applied for relevant indoor exposures
without further adjustment.

Combined Effect of Smoking and Radon Apart from the results of very limited
in-vitro and animal experiments, the only source of evidence on the combined
effect of the 2 carcinogens (cigarette smoke and radon) was the data from 6
of the miner studies. Analysis of those data indicated a synergistic effect
of the two exposures acting together, which was characterized as submultiplicative,
i.e., less than the anticipated effect if the joint effect were the product
of the risks from the two agents individually, but more than if the joint effect
were the sum of the individual risks. The committee applied a full multiplicative
relation of the joint effect of smoking and exposure to radon, as done by the
BEIR IV committee, and also a submultiplicative relationship. Although the committee
could not precisely characterize the joint effect of smoking and radon exposure,
the submultiplicative relation was preferred by the committee because it was
found to be more consistent with the available data.

Risks for Women The risk model is based on epidemiologic studies of male miners.
The effect of radon exposure on lung cancer risk in women might be different
from that in men because of differing lung dosimetry or other factors related
to gender. The K factor was calculated separately for women and men, but did
not differ by gender. The committee also could not identify strong evidence
indicative of differing susceptibility to lung carcinogens by sex. Consequently,
the model was extended directly to women, with the assumption that the excess
risk imposed by radon progeny estimated from the male miners multiplies the
background lung cancer rates for women, which are presently substantially lower
than for men.

Risks Associated with Exposures in Childhood Evidence was available from only
one study of miners on whether risk was different for exposures received during
childhood, during adolescence, and during adulthood. There was not a clear indication
of the effect of age at exposure. The committee made no specific adjustment
for exposures received at earlier ages. The K factor for children aged 10 was
calculated as 1 and the value for infants was only slightly higher (about 1.08).

Characterization of Radon Risks In making its calculations, the committee used
the latest data on lung cancer mortality for 1985-1989 and for smoking prevalence
for the U.S. in 1993. To characterize the lung cancer risk posed to the population
by indoor radon, the two models for the exposure-risk relationship were applied
to the distribution of exposures received by the population to estimate the
burden of lung cancer sustained by the population as a result of indoor radon
exposure. To characterize risks to the population, we have used the population
attributable risk (AR), which indicates how much of the lung cancer burden could,
in theory, be prevented if all exposures to radon were reduced to the background
level of radon in outdoor air. The AR estimates include cases in ever-smokers
and never-smokers. To characterize the risk to specific individuals, the committee
calculated the lifetime relative risk (LRR), which describes the relative increment
in lung-cancer risk resulting from exposure to indoor radon beyond that from
exposure to outdoor-background concentrations of radon.

Radon-Attributable Risks LRRs were computed using the committee's risk models.
Estimates were computed for exposure scenarios which reflect concentrations
of indoor radon of interest. Table ES-1 shows the estimated LRRs for lifetime
exposures at various constant radon concentrations. The LRR values are quite
similar for the preferred 2 models: exposure-age-concentration and exposure-age-duration.
The LRR values estimated by the BEIR VI models and the BEIR IV model are also
similar, in spite of the addition of exposure rate to the new models. As anticipated,
LRR values increase with exposure. Women have a somewhat steeper increment in
LRR with increasing exposure because of differing mortality patterns.

Attributable risks for lung cancer from indoor radon in the US population were
computed with the committee's 2 preferred models and compared with the BEIR
IV results. Based on the National Residential Radon Survey, the committee assumed
a log-normal distribution for residential radon concentration, with a median
of 24.3 Bqm-3 (0.67 pCi/L-1) and a geometric standard deviation of 3.1 (Marcinowski
1994). The AR was calculated for the entire US population and for males and
females and ever-smokers and never-smokers under the preferred submultiplicative
model (Table ES-2). For the entire population, the ARs calculated with the new
models ranged from about 10% to 14% and were higher than estimates based on
the BEIR IV model. Under the submultiplicative assumption which was described
on page ES-9, the attributable risk estimates for ever-smokers tended to be
lower than estimates for never-smokers, although the numbers of cases are far
greater in ever-smokers than in never-smokers.

These AR estimates for the general population are further broken down with
respect to the distribution of indoor concentrations in Table ES-3. This analysis
provides a picture of the potential consequences of alternative mitigation strategies
that might be used for risk-management purposes. The findings were the same
for the committee's 2 models. The radon concentration distribution is highly
skewed, with homes with higher radon concentrations contributing disproportionately
to AR. Only 13% of the calculated AR is estimated to be contributed by the 50%
of homes below the median concentration of about 25 Bqm-3 (0.7 pCi/L-1) and
about 30% by homes below the mean of about 46 Bqm-3 (1.25 pCiL-1). Homes above
148 Bqm-3 (4 pCi/L-1), the current action level established by the Environmental
Protection Agency, contribute about 30% percent of the AR. This contribution
to the total AR is indicative of the potential magnitude of avoidable deaths
with a risk management program based on the current action guideline. While
10-15 percent of all lung cancers are estimated to be attributable to indoor
radon, eliminating exposures in excess of 148 Bqm-3 (4 pCi/L-1) would prevent
about 3 to 4 percent of all lung cancers, or, about one-third of the radon-attributable
lung cancers.

The ARs were re-estimated with assumption of thresholds, levels below which
cancer risk is not increased, at 37, 74, or 148 Bqm-3 (1, 2, or 4 pCi/L-1).
Even though the committee assumed that risk was most likely linear with exposure
at lower levels, this analysis was conducted to illustrate the impact of assuming
a threshold on risk-management decisions. Assuming an action level of 148 Bqm-3
(4 pCi/L-1) for mitigation, postulating a threshold reduces the total number
of lung cancer deaths that are attributable to indoor radon and also the number
of lung-cancer deaths that can be prevented by reducing levels in homes to zero.
For assumed thresholds below 148 Bqm-3 (4 pCi/L-1), there is little impact on
the estimated numbers of preventable lung cancers by mitigation of homes with
radon concentrations above 148 Bqm-3 (4 pCi/L-1).

Table ES-1: Estimated lifetime relative risk (LRR) of lung cancer for lifetime
indoor exposure to radona Exposure-age-concentration model Exposure-age-duration
model Exposureb Male Female Male Female WLM/y Jhm-3/y Bqm-3 pCiL-1 WL Ever-
smoker Never- Smoker Ever- Smoker Never- Smoker Ever- Smoker Never- Smoker Ever-
Smoker Never- Smoker 0.10 0.00035 25 0.7 0.003 1.081 1.194 1.089 1.206 1.054
1.130 1.059 1.137 0.19 0.00067 50 1.4 0.005 1.161 1.388 1.177 1.411 1.108 1.259
1.118 1.274 0.39 0.00137 100 2.7 0.011 1.318 1.775 1.352 1.821 1.214 1.518 1.235
1.547 0.58 0.00203 150 4.1 0.016 1.471 2.159 1.525 2.229 1.318 1.776 1.352 1.819
0.78 0.00273 200 5.4 0.022 1.619 2.542 1.694 2.637 1.420 2.033 1.466 2.091 1.56
0.00546 400 10.8 0.043 2.174 4.057 2.349 4.255 1.809 3.053 1.915 3.174 3.12
0.01092 800 21.6 0.086 3.120 7.008 3.549 7.440 2.507 5.058 2.760 5.317 aBased
on a submultiplicative relationship between tobacco and radon.

bExposures are represented by concentrations in bequerels per cubic meter (Bqm-3),
picocuries per liter (pCiL-1), or Working Levels (WL), assumed to be constant
for home occupancy at the 70% level and 40% equilibrium between radon and its
progeny, and also by joules-hours per cubic meter per year (Jhm-3/y) and Working
Level Months per year (WLM/y). For definitions of these terms, see the Glossary
at the end of this report.

Table ES-2: Estimated attributable risk (ARa) for lung cancer death from domestic
exposure to radon using 1985-89 U.S. population mortality rates based on selected
risk models Model Population Ever- smokersb Never- smokersb Males Committee's
preferred models Exposure-age-concentration 0.141 0.125 0.258 Exposure-age-duration
0.099 0.087 0.189 Other Models CRRc (<50 WLM) 0.109 0.096 0.209 BEIR IV 0.082
0.071 0.158 Females Committee's preferred models Exposure-age-concentration
0.153 0.137 0.269 Exposure-age-duration 0.108 0.096 0.197 Other Models CRRc
(<50 WLM) 0.114 0.101 0.209 BEIR IV 0.087 0.077 0.163 aAR = the risk of lung
cancer death attributed to radon in populations exposed to radon divided by
the total risk of lung cancer death in a population. bBased on a submultiplicative
relationship between tobacco and radon. cCRR = constant relative risk.

Table ES-3: Distribution of Attributable Risks for U.S. Males from indoor residential
radon exposure under BEIR VI models Exposure-age-concentration model Exposure-age-duration
model Exposure Range (Bqm-3) % of Homes in Range Contribution to AR Contribution
to AR Actual % Cumulative % Actual % Cumulative % 0 - 25 49.9 0.018 12.8 12.8
0.013 12.8 12.8 26 - 50 23.4 0.026 18.5 31.3 0.018 18.4 31.2 51 - 75 10.4 0.020
14.2 45.5 0.014 14.2 45.4 76 - 100 5.4 0.015 10.5 56.0 0.010 10.5 55.9 101-150
5.2 0.020 13.9 69.9 0.004 13.9 69.8 151 - 200 2.4 0.013 9.2 79.1 0.009 9.2 79.0
201 - 300 1.8 0.014 9.6 88.7 0.010 9.7 88.7 301 - 400 0.7 0.007 5.2 93.9 0.005
5.3 94.0 401 - 600 0.4 0.006 4.5 98.4 0.005 4.6 98.6 601 + 0.4 0.002 1.5 99.9
0.001 1.6 100.2 Total 100.0 0.141 100.0 0.099 100.0

These AR estimates can be translated into numbers of lung-cancer deaths (Table
ES-4). In 1995, there were approximately 157,400 lung-cancer deaths-95,400 in
men and 62,000 in women-in the United States. Most occurred in smokers and it
is estimated that 95% of cases occurred in men and 90% in women. Table ES-4
shows the estimated lung-cancer deaths in the United States attributable to
indoor radon progeny exposure under the BEIR VI models. A review of the data
presented in table ES-4 reveals some differences in the calculated radon-attributable
lung-cancer deaths using the exposure-age-concentration model and the exposure-age-duration
model. Further variability is evident for both models depending on the approach
used to estimate the influence of cigarette-smoking on lung-cancer risk. The
use of the two models with two approaches to dealing with smoking yields an
array of estimates of lung-cancer risk attributable to radon exposure, and provides
an indication of the influence of the model and of incorporating the effects
of tobacco-smoking on the projections of population risk. The range of calculated
values, however, is not a complete reflection of the uncertainty in estimating
the lung-cancer risks of radon exposures and especially for never-smokers at
low levels of radon exposure.

Uncertainty Considerations Quantitative estimates of the lung cancer risk imposed
by radon are subject to uncertainties--uncertainties that need to be understood
in using the risk projections as a basis for making risk-management decisions
(see table ES-5). Broad categories of uncertainties can be identified, including
uncertainties arising from the miner data used to derive the lung-cancer risk
models and the models themselves, from the representation of the relationship
between exposure and dose, from the exposure-distribution data, from the demographic
and lung-cancer mortality data, and from the assumptions made in extending the
committee's models from the exposures received by the miners to those received
by the general population. The committee addressed those sources of uncertainty
qualitatively and, to a certain extent, quantitatively.

The committee's models of lung-cancer risk were based on analyses of data from
epidemiologic studies of miners. There are undoubtedly errors in the estimates
of exposures to radon progeny for the miners, and information was limited on
other key exposures including cigarette smoking and arsenic. The committee could
not identify any overall systematic bias in the exposure estimates for radon
progeny, but random errors might have led to an underestimation of the slope
of the exposure-risk relationship. Although 6 of 11 study cohorts had some smoking
information, sparse information on smoking limited the committee's characterization
of the combined effects of smoking and radon-progeny exposure and precluded
precise estimation of the risk of radon-progeny exposure in never-smokers.

The committee's models may not correctly specify the true relationship between
radon exposure and lung cancer risk. The models assume a linear-multiplicative
relationship without threshold between radon exposure and risk. While the miner
data provide evidence of linearity across the range of exposures received in
the mines, the assumption of linearity down to the lowest exposures was based
on mechanistic considerations that could not be validated against observational
data. Alternative exposure-risk relations, including relations with a threshold,
may be operative at the lowest exposures. However, the committee's analysis
showed that assumption of a threshold up to exposures at 148 Bqm-3 (4 pCiL-1)
had little impact on the numbers of lung-cancer deaths theoretically preventable
by mitigation of exposures above that level.

Table ES-4: Estimated number of lung cancer deaths for the U.S. for 1993 attributable
to indoor residential radon progeny exposure Number of Lung Cancer Deaths Attributable
to Radon Progeny Exposure Population Number of Lung Cancer Deaths Exposure-age-
concentration Model Exposure-age-duration Model Malesa Total 95,400 12,500b
8,800b Ever-smokers 90,600 11,300 7,900 Never-smokers 4,800 1,200 900 Femalesa
Total 62,000 9,300 6,600 Ever-smokers 55,800 7,600 5,400 Never-smokers 6,200
1,700 1,200 Males and Females Total 157,400 21,800 15,400 Ever-smokers 146,400
18,900 13,300 Never-smokers 11,000 2,900 2,100 aAssuming 95% of all lung cancers
among males occurs among ever-smokers; 90% of lung cancers among females occurs
among ever-smokers. bEstimates based on applying a smoking adjustment to the
risk models, multiplying the baseline estimated attributable risk per exposure
by 0.9 for ever-smokers and by 2.0 for never-smokers, implying a submultiplicative
relationship between radon-progeny exposure and smoking.

Additional sources of uncertainty in the risk projections reflect the approach
used to evaluate possibly differing lung dosimetry for miners and for the general
population, the limited information on cigarette smoking, and the lack of data
on risks of exposures of children and women.

The committee applied new quantitative methods for uncertainty analysis to
evaluate the impact of variability and uncertainty in the model parameters on
the attributable risk. Since not all sources of uncertainty could be characterized,
this analysis was intended to be illustrative and not to replace the committee's
more comprehensive qualitative analysis.

The quantitative analysis conducted by the committee provided limits within
which the AR was considered to lie with 95% certainty. For the exposure-age-concentration
model, the uncertainty interval around the central estimate of AR (14%) ranged
from about 10 to 26%. This range reflects a substantial degree of uncertainty
in the AR estimate, although the shape of the uncertainty distributions indicated
that values near the central estimates were much more likely than values near
the upper and lower limits. For the exposure-age-duration model, the uncertainty
interval ranged from 8 to 19% and was centered at about 10%. The committee also
computed uncertainty limits for the simple constant-relative-risk model fitted
to the miner data below 0.175 Jhm-3 (50 WLM), which is based on observations
at exposures closest to residential exposure levels. The latter analysis, which
minimizes the degree of extrapolation outside the range of the miner data, led
to uncertainty limits of 2-21%, with a central estimate of about 12%.

Table ES-5: Sources of uncertainty in estimates of lifetime risk of lung cancer
mortality resulting from exposure to radon in homes.

I. Sources of uncertainty arising from the model relating lung cancer risk
to exposure. A. Uncertainties in parameter estimates derived from miner data

Sampling variation in the underground miner data; Errors and limitations in
the underground miner data; a). Errors in health effects data including vital
status and information on cause of death; b). Errors in data on exposure to
radon and radon progeny including estimated cumulative exposures, exposure rates
and durations; c). Limitations in data on other exposures including data on
smoking and on other exposures such as arsenic. B. Uncertainties in application
of the lung cancer exposure-response model and in its application to residential
exposure to the general U.S. population

Shape of the exposure/exposure rate response function for estimates at varying
exposures and exposure rates; Temporal expression of risks; Dependence of risks
on sex; Dependence of risks on age at exposure; Dependence risks on smoking
status.

II. Sources of uncertainty arising from differences in radon progeny dosimetry
in mines and in homes

III. Sources of uncertainty arising from estimating the exposure distribution
for the U.S. population exposure distribution model

Estimate of the average radon concentration; Estimate of the average equilibrium
fraction; Estimate of the average occupancy factor. IV. Sources of uncertainty
in the demographic data used to calculate lifetime risk

Effects of Radon Exposure Other Than Lung Cancer Health effects of exposure
to radon progeny other than lung cancer have been of concern, including other
malignancies and non-malignant respiratory diseases in miners. The findings
of several ecologic studies in the general population have indicated a possible
effect of radon exposure in increasing risk for several types of non-lung cancers
and leukemias. A pooled analysis of 11 miner studies, differing in one study
from the data used by the committee, showed no evidence of excess risk for cancers
other than the lung. The committee concluded that the findings in the miners
could be reasonably extended to the general population and that there is no
basis for considering that effects would be observed in the range of typical
exposures of the general population that would not be observed in the underground
miners exposed at generally much higher levels.

The committee reviewed new studies of non-malignant respiratory disease in
uranium miners. A case series of uranium miners with pulmonary fibrosis supported
the possibility that exposures to radon progeny may cause fibrosis of the pulmonary
interstitium, but the case series is insufficient to establish a causal link
to radon progeny specifically.

CONCLUSIONS Radon is one of the most extensively investigated human carcinogens.
The carcinogenicity of radon is convincingly documented through epidemiologic
studies of underground miners, all showing a markedly increased risk of lung
cancer. The exposure-response relationship has been well characterized by analyses
of the epidemiologic data from the miner studies, and a number of modifiers
of the exposure-response relationship have been identified, including exposure
rate, age, and smoking. For residences in the United States, a large national
survey provides information on typical exposures and on the range of exposures.

On the basis of the epidemiologic evidence from miners and understanding of
the genomic damage caused by alpha particles, the committee concluded that exposure
to radon in homes is expected to be a cause of lung cancer in the general population.
According to the committee's two preferred risk models, the number of lung-cancer
cases due to residential radon exposure in the United States was projected to
be 15,400 (exposure-age-duration model) or 21,800 (exposure-age-concentration
model). Although these represent the best estimates that can be made at this
time, the committee's uncertainty analyses using the constant relative risk
model suggested that the number of cases could range from about 3,000 to 32,000.
(The 95% upper confidence limit for the exposure-age-concentration model was
approximately 38,000, but such an upper limit was highly unlikely given the
uncertainty distributions.) Nonetheless, this indicates a public-health problem
and makes indoor radon the second leading cause of lung cancer after cigarette-smoking.

The full number of attributed deaths can be prevented through radon mitigation
only by eliminating radon in homes, a theoretical scenario that cannot be reasonably
achieved. Nonetheless, the burden of lung-cancer deaths attributed to the upper
end of the exposure distribution is expected to be reduced by lowering radon
concentrations. Perhaps one-third of the radon-attributed cases (about 4% of
the total lung-cancer deaths) would be avoided if all homes had concentrations
below the Environmental Protection Agency's action guideline of 148 Bqm-3 (4
pCiL-1); of these, about 87% would be in ever-smokers. It can be noted that
the deaths from radon-attributable lung cancer in smokers could most efficiently
be reduced through tobacco-control measures, in that most of the radon-related
deaths among smokers would not have occurred if the victims had not smoked.

The committee's model and general approach to assessing lung-cancer risks posed
by indoor radon and cigarette-smoking are subject to considerable uncertainty
because of gaps in our scientific knowledge of effects at low levels of exposure.
This uncertainty should be reduced as an improved understanding develops of
molecular and cellular events in the induction of lung cancer at low levels
of exposure to radon and other toxicants and of the role of various factors
influencing susceptibility to lung cancer. The long-term follow up of miner
populations is strongly encouraged, as is completion of the case-control studies
of residential exposures now in progress. The committee encourages further meta-analysis
and pooling of case-control data. However, the committee recommends that new
case-control studies not be initiated until those in progress are completed,
data are analyzed and synthesized, and judgments rendered as to the likely value
of further residential studies.

Despite the limitations of existing data, the committee found key observational
and experimental data that, along with theoretical considerations in radiobiology
and carcinogenesis, provided a basis for the models developed and used to estimate
radon-attributable lung-cancer risks. The major shortcomings in the existing
data relate to estimating lung cancer risks near 148 Bqm-3 (4 pCiL-1) and down
to the average indoor level of 46 Bqm-3 (1.24 pCiL-1), especially the risks
to never-smokers. The qualitative and quantitative uncertainty analyses indicated
the actual number of radon-attributable lung-cancer deaths could be either greater
or lower than the committee's central estimates. This uncertainty did not change
the committee's view that indoor radon should be considered as a cause of lung
cancer in the general population that is amenable to reduction. However, the
attributable risk for smoking, the leading cause of lung cancer, is far greater
than for radon, the second leading cause. Lung cancer in the general population
and in miners is related to both risk factors and is amenable to prevention.