United States
Environmental Protection
Agency
Air And
Radiation
(ANR-464)
EPA400-R-92-011
May 1992
&EPA
Technical Support Document
for the 1992 Citizen's Guide
to Radon
Printed on flecyc/ed Paper
-------
-------
TECHNICAL SUPPORT DOCUMENT FOR THE
1992 CITIZEN'S GUIDE TO RADON
Radon Division
Office of Radiation Programs
U.S. Environmental Protection Agency
May 20, 1992
-------
ACKNOWLEDGEMENTS
This document was prepared by the U.S. Environmental Protection Agency's (EPA's)
Office of Radiation Programs (ORP) within the Office of Air and Radiation. Mr, Frank
Marcinowski in ORP's Radon Division was the Project Director.
The efforts of several additional key contributors within ORP were instrumental in the
development of this document, including Mike Walker (Project Director of the 1992 Citizen's
Guide), Jerry Puskin, Chris Nelson, Anita Schmidt, Marion Ceraso, Dennis Wagner, Mark
Dickson, Lisa Ratcliff, Stacy Greendlinger, Jed Harrison, Kirk Maconaughey, Dave Rowson, and
Steve Page.
Mr. Bruce Henschel within EPA's Office of Research and Development provided critical
input related to radon mitigation. He supplied important information on the effectiveness of
different mitigation technologies and played a key role in developing the cost model discussed in
Chapter 5 and supporting appendices.
Several individuals outside of EPA also provided key input on a variety of topics. Mr,
Harry Chmelynski with Sandy Cohen & Associates supplied statistical analyses of radon testing
options. Valuable input on radon mitigation was provided by a number of mitigators who are also
instructors at EPA's Regional Radon Training Centers and who have conducted research for
EPA's radon demonstration program. These individuals included Terry Brennan, Doug Kladder,
William Brodhead, Jim Fitzgerald, Terry Howell, John Anderson, and Jack Bartholomew.
Finally, 1CF Incorporated assisted ORP in preparing this document. The ICF project
team was directed by Sam Napolitano and Steve Wyngarden, and included Kerry Boshes, Freddi-
Jo Eisenberg, Sarah Stafford, and John Trever.
-------
TABLE OF CONTENTS
1, INTRODUCTION AND BACKGROUND 1-1
2, ESTIMATION OF RISKS 2-1
3, ANALYSIS OF EPA RADON TESTING OPTIONS 3-1
4. MITIGATION TECHNOLOGY 4-1
5. COST-EFFECTIVENESS ANALYSIS ' 5-1
6. RISK COMMUNICATION 6-1
7. RATIONALE FOR 1992 CITIZEN'S GUIDE 7-1
REFERENCES R-l
APPENDIX A - RADON CONCENTRATIONS IN AMBIENT AIR A-l
APPENDIX B - BACKGROUND INFORMATION ON RADON B-l
APPENDIX C - COMMONLY USED RADON MEASUREMENT DEVICES C-l
APPENDIX D - COVERAGE OF RADON TESTING POLICY D-l
APPENDIX E — RISK ANALYSIS METHODOLOGY E-l
APPENDIX F — RADON MITIGATION COST MODEL F-l
APPENDIX G - SENSITIVITY ANALYSIS OF THE COST-EFFECTIVENESS G-l
OF THE REVISED CITIZEN'S GUIDE
APPENDIX H - COST-EFFECTIVENESS OF RADON ACTION LEVELS H-1
GREATER THAN 4 pCi/L
-------
-------
CHAPTER 1
INTRODUCTION AND BACKGROUND
This document presents the wide range of technical analyses, radon risk communication
research, legislative directives, and other information that the U.S. Environmental Protection Agency
(EPA) used to shape the policies that are set forth in the 1992 A Citizen's Guide to Radon. The
document summarizes extensive technical analyses of the data that have been gathered over the past
six years.
Background on the1986 Citizen's Guide
In late 1984, the discovery of extremely high indoor radon in the Reading Prong area of
Pennsylvania, New Jersey, and New York thrust radon into the national spotlight as a major public
health problem. The government quickly responded to this problem by creating the Radon Program
in 1985 to help States and homeowners reduce the health risks of indoor radon. EPA's Radon
Program is non-regulatory, designed to provide public information and technical assistance to enable
citizens to make informed decisions on how they should protect themselves.
A Citizen's Guide to Radon has played a major role in the overall strategy of the Radon
Program. It serves as both EPA's core policy statement and primary public information brochure on
radon in homes, as well as the Federal government's principal guidance document on indoor radon.
Published jointly with the U.S. Department of Health and Human Services (HHS) in 1986, the
original Citizen's Guide provided basic background information on indoor radon and its associated
health risks, explained how radon levels in homes should be measured, and advised homeowners on
when ihey need to take action to reduce radon levels in their homes.
When EPA and HHS issued the 1986 Citizen !s Guide, there was less information on indoor
radon than is available now. The recommendations in 1986 were based primarily on limited
experience with homes whose indoor air had been contaminated with radon from uranium mill
tailings. Research in these homes showed that most could be mitigated consistently to an average
indoor radon level of 4 picocuries per liter (pCi/L). Based on this technology limitation, EPA
established 4 pCi/L as the action level at which people should fix their homes.
Since little information was available from many homes on the relationship between short-
term and annual average radon concentrations, and because the scientific community agreed that
long-term measurements provided a better estimate of a home's annual average radon level, it was
important to stress the need for long-term measurements before decisions on the need for mitigation
were made. The Agency also wanted to provide States and homeowners with the means to gauge
worst-case radon levels existing in their home or area, so that, the need for further action could be
determined without waiting 12'months for th'e results of the measurement. These two considerations
led to the development of a two-step strategy of measuring radon: (1) a short-term screening
measurement in the lowest livable level during the wintertime, intended to measure the highest
concentration to which occupants could be exposed; and (2) if the results of the screening
measurement were above 4 pCi/L, a confirmatory measurement in lived-in levels, designed to provide
an estimate of the average radon levels to which the occupants were exposed. The recommended
duration of the confirmatory measurement depended on the results of the screening measurement
and ranged from less than one week to one year.
Chapter 1: Introduction and Background Page 1-1
-------
Factors Influencing Revisions .to the Guide
Many factors influenced the development of the revised Citizen's Guide. These factors
included:
» The Indoor Radon Abatement Act, passed by Congress in 1988, which requires EPA
to issue a revised Citizen's Guide, and contains specific requirements for the new
Guide;
* Advancements in the understanding of the health risks associated with exposure to
residential radon;
» Improved understanding of the accuracy of measurements made under different
conditions for purposes of determining the need for mitigation;
» Advancements in the understanding of mitigation technology;
« Analyses of the costs and benefits of alternative action levels and testing strategies;
and
« Risk communication research findings.
These factors are briefly reviewed in the following sections and are covered in detail in the remaining
chapters of this document.
Indoor Radon Abatement Act (IRAA)
This amendment to the Toxic Substances Control Act endorses the non-regulatory strategy
of the Radon Program and provides EPA with funding and a legislative mandate for a variety of
additional activities designed to further reduce the risks of indoor radon exposure. Section 301 of
the Act establishes the national long-term goal that indoor air be as free of radon as ambient air
outside of buildings. This goal reflects the Congressional direction contained in Report language that
EPA be more protective of public health. Results from EPA's Ambient Radon Field Study, initiated
to comply with this requirement and confirm reported outdoor radon levels, are summarized in
Appendix A.
Congress provided specific guidance on the scope of the revised Citizen's Guide in section 303,
requiring EPA to provide a "series of action levels with a description of the associated health risks"
to ensure that the public understands that 4 pCi/L is not a safe level of exposure and that a
significant health risk exists at levels below 4 pCi/L. The Act also requires EPA to address -health
risks to special populations, i.e. children and smokers. In addition, Congress specified that EPA
present an analysis of the relationship between short- and long-term measurement results and their
relative impacts on alternative action levels.
Updated Radon Risk Information
The Biological Effects of Ionizing Radiation (BEIR IV) Committee of the National Academy
of Sciences issued a report (NAS 1988) on the effects of exposure to alpha radiation that includes
extensive information on the results of studies of underground miners exposed to radon decay
Chapter 1: Introduction and Background Page 1-2
-------
products. National radiation risk experts comprising the BEIRIV Committee prepared and reviewed
a radon risk projection model (the BEIR IV model). EPA used the BEIR IV model to estimate
radon risk, making slight adjustments to the model to account for background radon and to
incorporate the most recent information on dose per unit exposure in mines and homes (NAS 1991).
The Agency also (1) used new data on residential radon levels collected through EPA's National
Residential Radon Survey (U.S. EPA/Office of Radiation Programs 1991 a); (2) assembled an
independent panel of experts on radiation risk (the Agency's Science Advisory Board), who reviewed
the Agency's risk calculation methods and provided EPA with further suggestions for refinements;
and (3) conducted a detailed analysis of uncertainty associated with the radon risk estimates. EPA's
revised risk estimates fall within the ranges established by national and international scientific
organizations, including the National Academy of Sciences, the International Commission on
Radiological Protection, the National Council on Radiation Protection and Measurement, and the
United Nations Scientific Committee on the Effects of Atomic Radiation. Chapter 2 discusses EPA's
risk assessment methodology and risk estimates in detail. Appendix B provides basic background
information on radon and its decay products.
Analysis of Measurement Options
Since publishing the 1986 Citizen's Guide, EPA has conducted extensive research to determine
how well various radon testing options meet the basic purpose of testing — to enable homeowners
to make the right decision on whether to mitigate radon levels. The Agency drew from the wealth
of radon testing data and experience gained since 1986 to develop a number of testing options, each
defined by a given action and "trigger level," a certain testing location, appropriate ventilation
conditions, and different measurement durations. Using a statistical model, data on the magnitude
of measurement errors caused by various sources, and the distribution of annual average radon levels
developed from the National Residential Radon Survey, EPA then evaluated each testing option in
terms of how frequently it would result in homeowners making a correct mitigation decision (i.e., how
well a testing procedure provides a reasonably accurate basis for determining whether a home is
above the action level). EPA considered this information on the effectiveness of testing options,
together with real-world information on public acceptability, in order to select a revised testing
protocol for the 1992 Citizen's Guide. Chapter 3 provides the details of this analysis and Appendix
C provides background information on commonly used radon measurement devices.
Advancements in the Understanding of Mitigation Technology
There has been progress in the understanding of radon measurement and mitigation
technology since 1986, with a wide variety of methods now available to prevent radon entry into
homes and to reduce indoor radon concentrations after entry. A radon mitigation industry also has
emerged to provide homeowners throughout the country with the ability to reduce their radon levels.
Research by EPA's Office of Research and Development and a recently completed survey of radon
mitigation contractors (U.S. EPA/Radon Division 1990a) have shown that radon mitigation in the vast
majority of cases should leave homes with an annual average level of less than 4 pCi/L. It also will
be very common for homes to have post-mitigation levels of 2 pCi/L or less. Chapter 4 describes
current mitigation technology and its overall effectiveness and cost.
Analysis of Costs and Benefits
Additional data collected since the Citizen's Guide was first published have enabled the
Agency to conduct a detailed analysis of the costs and risk reductions (benefits) for the public if it
Chapter 1: Introduction and Background Page 1-3
-------
•followed the Guide's testing and mitigation advice. Based on this analysis, EPA has estimated (1) the
costs of the different testing and mitigation options considered for inclusion in the revised Guide (i.e.,
the total annualized costs of testing and mitigation); (2) the risk reductions provided by the different
options (Le., the annual lives saved by testing and mitigation); and (3) the cost-effectiveness of
alternate approaches (i.e., the cost per lung cancer death averted). EPA also has evaluated the cost-
effectiveness results by relating them to the public's willingness to pay to save a statistical life and by
comparing the estimated cost per life saved for radon reduction to the cost per life saved for other
health-related programs. The methods and results of this analysis are described in Chapter 5 and in
Appendices D, E, F, G, and H.
Risk Communication Research
Substantial information is now available on risk communication and, in particular, how the
public responds to radon risk messages. Researchers and public health officials report that
homeowners do not overreact or panic after receiving information on radon. Instead, the public
remains largely apathetic about radon, as evidenced by the small proportion of homes that have been
measured and fixed. Radon risk communication research has provided useful insight into why this
apathy exists and identified six key ways for public information materials to help overcome it. One
of the most important findings has been that the guidelines on testing and mitigation presented in
the 1986 Guide need to be streamlined in order to minimize barriers to public action. Homeowners
often drop out of the two-step testing process, especially if it means having to wait for an entire year
to get the results of the recommended long-term follow-up measurement. A 1990 telephone survey
(Johnson 1990), for example, found that only 9 percent of the participants were willing to conduct
a one-year radon test. In fact, it appears that those homeowners who do mitigate are acting on the
basis of short-term measurements in the lowest livable area of their home. Current research on radon
risk communication and a summary of how that research has been incorporated into the revised
Guide is provided in Chapter 6.
Chapter 1: Introduction and Background Page 1-4
-------
CHAPTER 2
ESTIMATION OF RISKS
Section I: Overview
Radon has been classified as a known human carcinogen based on extensive data from
epidemiologic studies of underground miners. The carcinogenicity of radon has been well
established by the scientific community, including the World Health Organization's International
Agency for Research on Cancer (IARC 1988), the Biological Effects of Ionizing Radiation (BEIR
IV) Committee of the National Academy of Sciences (NAS 1988), the International Commission on
Radiological Protection (ICRP 1987), and the National Council on Radiation Protection and
Measurement (NCRP 1984). In addition, the Centers for Disease Control, the American Lung
Association, the American Medical Association, and the American Public Health Association have
recognized radon as a significant public health problem. Radon has been characterized as "a
potentially important cause of lung cancer in the general population, which is exposed through
contamination of indoor air by radon from soil, water, and building materials" (Samet 1989).
To assess the risk of lung cancer
associated with residential exposure to radon, MVM^M^MM^^B^^MVMMHMMMMMMi
EPA currently uses the risk projection model
developed by the BEIR IV Committee (NAS THE AIMS OF THIS CHAPITER
1988) with minor modifications by EPA, EPA
adjusted the BEIR IV relative risk model to C1) To document EPA's approach for
incorporate the most recent information on estimating the lung cancer risk
dose per unit of exposure in mines and homes attributable to radon exposure.
from the report: Comparative Dosimetiy of
Radon in Mines and Homes (NAS 1991), and to (2) To Provide informatlon critjcal to
account for background radon. EPA's current the interpretation ot the risk
estimate of risk incorporates the estimate of the estimates, including information on
average national residential radon level from uncertainty and the eftectof
the recently completed EPA National smoking.
Residential Radon Survey (U.S. EPA/Office of ..^^mmm^mmmiimmm^^^^^^^^^^^
Radiation Programs 1991 a). Additionally, the
Agency has included in this radon risk assessment a detailed analysis of the uncertainty associated
with the lung cancer risk estimates, based on the approach employed in the National Institutes of
Health report on the development of radioepidemiological tables (NIH 1985).
Using the modified BEIR IV model, EPA estimates that the number of lung cancer deaths
per year in the U.S. due to residential radon exposures is approximately 14,000, with an uncertainty
range of 7,000 to 30,000. This estimate is based on the Census Bureau's estimate of 250 million
people for the residential U.S. population as of October 1, 1990 (Bureau of the Census 1992a).
1 The term radon when used in this chapter refers to radon-222 and its decay products. Of the short-lived radon decay
products (pulouium-21'8, lead-214, bismuth-214, and polonium-214), the polonium isotopes contribute most of the
radiologically significant dose to the lung. The risk from inhaled radon-222 is small compared to the risk from inhaled radon-
222 decay products.
Chapter 2: Estimation of Risks Page 2-1
-------
EXHIBIT 2-1
ESTIMATES OF LUNG CANCER RISK FROM EPIDEMIOLOGICAL
STUDIES OF UNDERGROUND MINERS EXPOSED TO RADON
Study Population
Czech Uranium
Miners
Ontario Uranium
Miners
New Mexico
Uranium Miners
Swedish Iron Miners
(Malmberget)
Colorado Plateau
Uranium Miners
Eldorado
(Beaverlodge)
Uranium Miners
Newfoundland
Fluorospar Miners
Average
Exposure
(WLM)
313
226
40-90
111.4
81.4
834
20.2
382.2
Relative
Risk
Coefficient
(%/WLM)^
1.92
1.5
0.5-1.3
1.4i2/
1.8
3.6
1.4^
.45
0.6^
3.28
2.6^
0.9
Reference
Thomas et al. 1985
Sevc et al. 1988
Muller 1984
NAS 1988
Samet et al. 1991
Radford & St. Clair Renard 1984
NAS 1988
Thomas et al. 1985
NAS 1988
' Howe et al. 1986
NAS 1988
Morrison et al. 1988
-The relative risk coefficient is the fractional increase above the baseline lung cancer incidence or mortality
rate per WLM.
-'Estimate based on reanalysis of the data by the NAS with the cooperation ot" the principal investigators.
When cumulative exposures were equal, low exposures over longer periods produced greater
lung cancer risk than high exposures over short periods.
Increased lung cancer risk with radon exposure has been observed even after controlling for,
or in the absence of, other mine exposures such as asbestos, silica, diesel fumes, arsenic,
chromium, nickel and ore dust.
Chapter 2: Estimation of Risks
Page 2-4
-------
• Increased lung cancer risk has been observed in miners at relatively low cumulative exposures
(e.g., 40-70 working level months (WLM) in the study of Ontario uranium miners).
» Non-smoking miners exposed to radon have been observed to have an increased risk of lung
cancer.
In their review of five of the major studies of underground miners, Samet and Hornung
(1990) noted that although there were significant differences in the studies, all demonstrated an
exposure-response relationship between exposure to radon and excess lung cancer, and excess relative
risk per WLM estimates that were "remarkably homogeneous." The following sections review findings
from eight major studies, including those reviewed by Samet and Hornung, This is not meant to be
an exhaustive review, but rather represents an overview of the findings from a number of the
prominent studies of miners. The studies reviewed here were conducted in Czechoslovakia, Ontario
(Canada), New Mexico (U.S.), Sweden, the Colorado Plateau (U.S.), Australia, Northern
Saskatchewan (Canada), and Newfoundland (Canada). The miners in these studies continue to be
followed over time, and updated analyses are conducted as new data become available.
The Czechoslovakia!! study included four large cohorts, with a total of 9,403 uranium miners
(Samet and Hornung 1990). Average cumulative exposures for the cohorts ranged from 3.2-3.03
WLM, and average length of follow-up varied from 6-30 years. Excess lung cancer mortality was
found largely in the cohorts with the highest cumulative exposures and the longest follow-up periods.
However, the study also detected a significant increase in the risk of lung cancer from levels of
cumulative exposure as low as 50-99 WLM. The attributable risk of lung cancer was observed to
increase with age at initial exposure, and the overall attributable risk was estimated to be 20 per
WLM per 10 person-years. When cumulative exposures were equal, low exposures over long periods
were found to produce higher risk than high exposures over short periods. The combined effect of
cigarette smoking and exposure to radon progeny was observed to be approximately additive.
Strengths of this study include the extensive exposure data and long follow-up periods (25-30 years
for two of the four cohorts). The study is limited by incomplete smoking information for the cohorts
that had the highest exposure and the longest follow-up, and by the use of analytical methods that
did not readily allow for comparison with other studies.
The Ontario, Canada study included 15,984 uranium miners, a large number of whom were
exposed at relatively low levels (Samet and Hornung 1990). The mean cumulative exposures of the
miners in this study were 40 WLM and 90 WLM, depending upon the method used to estimate
exposure, and the mean period of follow-up was 15 years. At exposure levels of 40-70 WLM, lung
cancer mortality was found to be significantly increased. The attributable risk of lung cancer was 7
or 3 per WLM per 106 person-years, again depending upon the exposure estimate. The excess
relative risk was calculated to be 1.3% per WLM. The study also showed a decrease in risk with time
since exposure. Strengths of this study are the large number of miners exposed to radon progeny at
relatively low levels, and the consideration of exposures received in other types of hard-rock mining.
The study is currently limited by the fact that the members of the cohort were still too young at the
end of the follow-up period (median age was 49 years) for observation of the complete temporal
2A working level is defined as any combination of short-lived radon decay products in 1 liter of air that will result in the
ultimate emission of 1.3 x 105 MeV of potential alpha energy. A working level month is defined as the exposure to 1 working
level for 170 hours (1 working month).
Chapter 2: Estimation of Risks Page 2-5
-------
expression of excess lung cancer risk from exposure to radon progeny. There was also a lack of
smoking information for study subjects.
The New Mexico study included 3,469 uranium miners who had worked at least one year
underground in New Mexico uranium mines prior to December 31, 1976 (Samet et al. 199lb). The
follow-up period extended through 1985, and the mean cumulative exposure level for the cohort was
111.4 WLM. The risk of lung cancer in this cohort was increased for exposure categories above 100
WLM; the excess relative risk increased by 1.8% per WLM. The study data were consistent with a
multiplicative interaction between radon and smoking. Additionally, lung cancer risk was observed
to increase more steeply for those less than 55 years of age. The authors noted that the study may
suffer from random misclassification of radon exposure, and possible systematic bias in exposure
estimates. However, the authors cite the extensive database on which exposure estimates were based
as one of the strengths of the study.
The Swedish study included 1,415 iron miners, with an average cumulative exposure of 81
WLM (Samet and Hornung 1990), This study had the longest follow-up period of the eight studies;
the average time in the study was 44 years. Of the 1,415 miners, 1,294 were observed between 1951
and 1976, during which time 50 lung cancer deaths occurred. Based on Swedish national mortality
rates, 14.6 lung cancer deaths were expected. The study found significant excess risk for cumulative
exposures above 80 WLM, and attributable risk of lung cancer was estimated to be 19 per WLM per
10 person-years. Through an analysis that required many assumptions, the combined effect of
smoking and exposure to radon progeny was estimated to be additive. Silica, diesel exhaust, arsenic,
chromium, and nickel were discounted as potential confounding exposures because they were present
in small concentrations in the mines. The strengths of this study are the length of the follow-up
periods and the thorough confirmation of cause of death. The study is limited by uncertainty in the
accuracy of exposure estimates, by the lack of smoking information for all subjects, and by the use
of different methods of exposure assignment for lung cancer cases and non-cases.
The U.S. Public Health Service Colorado Plateau study included 4,127 uranium miners
working in Colorado, New Mexico, Arizona, and Utah (Samet and Hornung 1990). Of these, 3,346
were included in the risk assessment conducted by the National Institute of Occupational Safety and
Health. The average cumulative exposure level of this cohort was 821 WLM. The average length
of follow-up was 22 years. In the range of exposure below 600 WLM, excess relative risk of lung
cancer was found to be 1.2% per WLM, and attributable risk of lung cancer was 6.3 per WLM per
106 person-years. Relative risk was observed to decrease with time since cessation of exposure, and
miners first exposed at older ages were at increased risk of lung cancer compared to those exposed
at younger ages. As in the Czech study, low exposures over long periods of time were found to be
more hazardous than high exposures for short periods of time, when cumulative exposures were
equal. The relationship between smoking and exposure to radon progeny was estimated to be slightly
submultiplicative. Strengths of this study are the large number of miners in the study, the close and
lengthy follow-up, and the availability of smoking information for most study subjects. The high levels
of exposure that characterize this study may limit its applicability to risk estimation al lower levels of
exposure. Consequently, in their analysis of these data for the development of a radon risk model,
the BEIRIV Committee excluded these very high exposure levels (above 2,000 WLM). In addition,
although the most recent complete data analysis was based on smoking information collected in 1969,
smoking histories have recently been updated and are being incorporated into a new analysis of the
radon-lung cancer relationship in this miner cohort (Roscoe 1991).
Chapter 2: Estimation of Risks Page 2-6
-------
The Colorado Plateau cohort was also the data source for a study of lung cancer risk in non-
smokers exposed to radon (Roscoe et al. 1989). For this study, a sub-cohort of 516 male uranium
miners who never smoked cigarettes, pipes, or cigars was selected from the original cohort of 4,127
miners, and was followed from 1950-1984. When the non-smoking miners were compared to non-
smoking U.S. veterans, the miners were 12.7 times more likely to have died of lung cancer than were
the veterans. When compared to male non-smokers in the general U.S. population, the miners were
9.3 times more likely to have died of lung cancer. No lung cancer deaths were found among non-
smoking miners with less than 465 WLM of cumulative radon exposure. However, the limited
number of non-smoking miners in the study who were exposed at lower levels severely restricted the
statistical power of the study to detect risks at these lower exposures relevant to residential levels.
The authors of this study concluded that: "exposure to radon daughters in the absence of cigarette
smoking is a potent carcinogen that should be strictly controlled" (Roscoe et al. 1989).
The Australian study included 2,574 uranium miners (1,433 underground workers, 1,141
surface workers) at the Radium Hill mine which operated in eastern South Australia from 1952-1961
(workers were followed through 1987). Compared to other mines of that period worldwide,
cumulative exposures of underground workers were low, with a mean of 7.0 WLM and a median of
3.0 WLM. When compared with the surface workers, lung cancer risk was significantly increased in
the underground workers with exposures above 40 WLM. The authors concluded that it was unlikely
that the increase in risk was due to differences in smoking habits or other confounders (such as
asbestos, dust levels, diesel fumes). Strengths of the study include low exposure levels, a long follow-
up period (more than 30 years for most workers) and relatively detailed historical information on
radon gas concentrations at the mine. However, the authors noted that assumptions and
approximations were necessary to estimate personal radon-daughter exposures. Additionally,
according to the authors, the study was limited by the small size of the workforce and their inability
to trace over one-third of the workers beyond the end of employment at Radium Hill.
The Eldorado (Beaverlodge) study included 8,487 workers employed at the Eldorado
Resources Limited Beaverlodge Uranium Mine in Northern Saskatchewan (Howe et al. 1986).
Follow-up was carried out from 1950-1980. The average cumulative exposure, weighted by person-
years at risk was 20.2 WLM (NAS 1988). A highly significant linear relationship between dose and
increased lung cancer risk was observed, with an estimated relative risk coefficient of 3.28% per
WLM. This study is notable for the low cumulative exposure levels and for the low silica exposures
and the absence of any diesel machinery in the mines. The authors note that the study was limited
by the lack of any smoking information on the miners, and by incomplete information on the cohort's
work experience at mines other than Beaverlodge where exposure to radon daughters may have
occurred.
Subsequent to this study, a case-control study was undertaken within this cohort to examine
the potential contribution of smoking and non-Beaverlodge work experience to the relative risk
coefficients from the original Beaverlodge study (L'Abbe et al. 1991). The study included 46 lung
cancer cases and 95 controls that were enrolled in the Beaverlodge uranium mine cohort study
between 1950 and 1980. The authors concluded that cigarette smoking and other mining experience
were unlikely to have made an important contribution to the relative risk coefficients reported in the
original study. The study did not definitively distinguish between a multiplicative or an additive
relationship between radon exposure and cigarette smoking, although the authors suggest that their
data offer limited support for the multiplicative relationship.
Chapter 2: Estimation of Risks Page 2-7
-------
The Newfoundland study included 1,772 Newfoundland underground fluorspar miners with
an average cumulative exposure of 382.8 WLM, followed from 1950-1984 (Morrison et al. 1988). The
study reported significantly elevated numbers of cancers of the lung, salivary gland and buccal cavity
and pharynx. A highly significant association was observed between radon exposure and the risk of
death from lung cancer. The relative risk coefficient was estimated at 0.9% per WLM, and those first
exposed before age 20 had the highest relative risk coefficients. Smoking miners were observed to
have relative and attributable risk coefficients comparable to those of non-smoking miners. The
strengths of this study include the collection of information on cigarette smoking status for 48% of
the cohort, and a long follow-up period (1950-1984). Study limitations include the lack of any radon
measurements prior to 1960, and an inability to trace a large number of the workers (approximately
16% of the underground cohort as originally defined was excluded because of a lack of adequate
personal identifying information).
Occupational data on exposure to radon and lung cancer risk continue to be made available
through ongoing follow-up of existing miner cohorts, and through newly reported studies. The
National Cancer Institute has acquired data from 11 cohort studies and in collaboration with the
various principal investigators is undertaking a joint analysis. These studies represent all major
cohorts of radon exposed miners currently available, and will include 2,300 lung cancer cases. The
analysis is expected to be completed in 1992 (Lubin 1991).
Residential Studies
There is uncertainty associated with the projection of lung cancer risk from occupational
radon exposures to the general population for residential exposures (see Section V for discussion of
uncertainty). Residential studies have been undertaken in an attempt to directly define and examine
the association between lung cancer incidence and exposure to radon progeny in the home.
There are a wide variety of residential radon epidemiologic studies that are already completed,
ongoing, or being planned. Residential studies conducted to date can be divided into two basic
categories, those that examine the association between radon and lung cancer based on group
characteristics (e.g., ecologic studies), and those that examine the association based on individual
characteristics (e.g., case-control studies).
Ecologic studies compare frequencies of exposure and disease between groups, rather than
between individuals, or assess changes in disease frequency and exposure frequency over time for one
or more groups. Selection of the group is often based on geographic parameters; Ecologic studies
are relatively inexpensive and convenient, and are useful for generating hypotheses; however, they
are not recommended for identifying the cause of a disease as they cannot: (1) determine the joint
distribution of the disease (e.g., lung cancer) and the study factor (e.g., radon exposure) within each
group, or (2) separate the effects of two or more variables (Kleinbaum et al. 1982). The use of
ecologic (group) data to draw conclusions about effects in individuals (e.g., development of lung
cancer) may yield causal inferences that suffer from an "ecological fallacy;" an association or the lack
of an association between a study factor and disease that is observed at the level of the group may
not hold at the individual level The study groups being compared may differ in many factors other
those being considered in the study, and any one of these may be the underlying cause of the
difference in their disease states.
For ecologic studies of radon risk, the average radon exposure and average lung cancer
incidence or mortality for each group can be identified. However, there is no way to relate the level
Chapter 2: Estimation of Risks Page 2-8
-------
of radon exposure for an individual to that individual's health status. The number of lung cancer
patients with a history of elevated radon exposure or the number of individuals with elevated radon
exposure who develop lung cancer is not determined. Because the unit of analysis is the group, this
type of study is incapable of assessing individual smoking history or mobility, both of which are very
important variables in radon risk assessment.
Completed ecologic studies have generally used characteristics of a geographical region, such
as geology or radon in water, as surrogates for radon exposure and correlated them with lung cancer
rates in the same area. Most of these studies showed associations between lung cancer incidence or
mortality and exposure to radon. However, since individual exposures were not considered, results
can only suggest that residential exposure to radon progeny increases lung cancer risk (Samet 1989).
The 1989 International Workshop on Residential Radon Epidemiology endorsed the use of
case-control studies for residential radon research, and specifically recommended against the further
use of ecologic studies because of their limitations for testing causal hypotheses (U.S. DOE/Office
of Energy Research 1989).
Case-control studies involve the comparison, with respect to exposure, of individuals who
have the disease of interest (cases) and those who do not (controls). Case-control studies are useful
for testing causal hypotheses for specific diseases. They do not require extensive follow-up, and ha%re
a greater ability to control for potentially confounding variables (e.g., smoking, diet) than ecologic
studies, since they collect information on study subjects individually.
Case-control studies have been used to more directly test the relationship between lung
cancer incidence or mortality and residential radon exposure (Samet 1989). Early case-control
studies, conducted in Sweden, the United States, and Canada, generally used housing characteristics
as surrogates for radon exposure. A number of these studies found associations between the radon
exposure surrogates and increased lung cancer incidence or mortality.
Four case-control studies of residential radon exposure and lung cancer risk have been
completed recently. Unlike most of the earlier studies that used surrogates for exposure, these
studies took measurements in the homes of study subjects to determine radon exposure. The results
of these studies were mixed, with one study showing a statistically significant trend of increasing lung
cancer risk with increasing radon concentration (Schoenberg et al. 1990), one study showing a
borderline significant increase in lung cancer risk with estimated radon exposure (Pershagen in press
1992), and two studies showing no statistically significant association between radon exposure and
lung cancer risk (Ruosleenoja 1991, Blot et al. 1990). The following is a brief review of findings from
these four studies conducted in the U.S. (New Jersey), Sweden, Finland, and China.
A study of radon and lung cancer in New Jersey women with 480 lung cancer cases and 442
controls (Schoenberg et al. 1990) has reported a statistically significant trend of increasing lung cancer
risk with increasing residential radon concentration. This trend was observed even after adjusting for
smoking, age, and occupational history. Additionally, the observed increase in relative risk (3A% per
WLM), although associated with considerable uncertainty, is consistent with the increases reported
in the studies of underground miners, supporting the extrapolation of data from the miner studies to
the residential setting. However, the estimates of risk for the different categories of radon
concentration, when taken individually, were not statistically significant. Furthermore, the authors
of the study cite the small number of subjects with high radon exposures, the possibility of selection
Chapter 2: Estimation of Risks Page 2-9
-------
biases, and the incomplete cumulative exposure assessments as reasons for exercising caution in
interpreting the study results.
A study of Swedish women with 210 lung cancer cases and 400 controls has found a
borderline significant increase in lung cancer risk with estimated radon exposure in women whose
average radon level exceeded 4 picocuries per liter (pCi/L), Positive trends were observed correlating
the relative risk of lung cancer with the estimated cumulative residential radon exposure. Adjustment
for the percentage of time actually spent in the home reduced the evidence for a dose-response
relationship. The authors state that their findings suggest that radon exposure is of importance for
the risk of developing lung cancer, particularly in the younger age groups. However, they cautioned
that the small sample size limited the ability of the study to discriminate between various models in
describing the association between estimated radon exposure and lung cancer risk (Pershagen et al.
in press 1992).
A study of Finnish men with 291 lung cancer cases and 495 controls found no statistically
significant relationship between indoor radon exposure and the risk of lung cancer (Ruosteenoja
1991). The exposure-response curve initially showed an increase but then deviated downward at the
higher exposure levels (above 7.4 pCi/L). The authors postulated that this may have been due to
random variation or systematic bias. The authors also noted that the final sample size was too small
to detect the risk that was expected based on findings from the miner studies.
A study of Chinese women, which included 308 lung cancer cases and 356 controls, showed
no statistically significant association between radon exposure and lung cancer risk, regardless of
cigarette smoking status (Blot et al. 1990). A non-significant trend of increased lung cancer risk with
increased radon exposure was observed among residents that were exposed to high levels of radon.
The authors of the study advised cautious interpretation of these results, however, given the
potentially important confounding effect of high indoor air pollution in the study area. Another issue,
in light of the ethnic and cultural differences between China and the U.S., is the applicability of these
results to the U.S. population.
A number of residential radon case-control studies that are presently in progress were
presented at the International Workshop on Residential Radon Epidemiology in July of 1989 in
Alexandria, VA (Samet et al. 1991 a). The principal invesLigators of these studies as well as other
scientists discussed a variety of issues at the workshop, including the issue of sample size. Workshop
attendees concluded that the majority of the residential radon studies, when taken individually, have
samples too small to characterize the exposure-response relationship between radon and lung cancer.
As indicated in an analysis conducted by Lubin et al. (1990), the number of lung cancer cases needed
to characterize the exposure-response relationship between indoor exposure to radon and lung cancer
risk may be in the thousands. For this reason, the study investigators have decided to undertake a
pooling of results from the international case-control residential radon studies as they are completed.
By pooling the data and thus increasing sample size, the investigators hope to characterize the
exposure-response relationship. The Second International Workshop on Residential Radon
Epidemiology, held in Alexandria, VA in July of 1991, determined that an initial pooling effort would
not be possible until 1994. Exhibit 2-2 summarizes 11 ongoing case-control studies of residential
radon exposure and lung cancer risk.
Chapter 2: Estimation of Risks Page 2-10
-------
9
to
EXHIBIT 2-2
ONGOING CASE-CONTROL STUDIES OF RESIDENTIAL RADON EXPOSURE AND LUNG CANCER RISK
So
I
Study Site
U.S./CN
U.S./FL
U.S./MO
U.S./NJ
U.S./UT & So. ID
Belgium
Canada
France
Sweden
U.K.
West Germany
Principal
Investigator
Jan Stolwijk
Heather Stockwell
Michael Alavnnja
WlIHnm Nicholson
Joseph Ly<»n
Andre l*offyn
E. G. Letourneau
Margot Tirmarcbe
Goran Pershagen
Surah Darby
H. Erich Wichmann
Funding Agency
National Institute of Environmental Health
Sciences
National Cancer Institute
National Cancer Institute
National Institute of Environmental Health
Sciences
National Institute of Environmental Health
Sciences
National Fund tor Science Research
Canadian Health and Welfare
University of Brest/National Institute of Health and
Medicnl Research
National Institute of Rndiation Protection and the
Swedish Cancer Society
Imperial Cancer Research Fund/National
Radiological Protection Board
Federal Ministry of Environment
Estimated
Completion
1993
1992
1993
1994
1994
1994-96
1993
1994-96
1992
1994
1996
# Lung
Cancer
Cases
1000
300
600
500
750
KM
750
600
150(1
600
3200
Duration of
Radon
Measurement
1 year
I year
I year
I year/4 days
I year/6 months
6 months
6 months
3 months
3 moil II is
6 months
1 year
Source; Second International Workshop on Residential Radon Kjiiriviniology, Alexandria, VA, 1991.
A coordinated case-control study involving France, Germany, Luxembourg, the United Kingdom and Belgium is being sponsored by the Commission of European Communities,
The coordinator is Dr. A. 1'offyn nt the Slate University of Ghent in B< iyiiim.
I
-------
Animal Studies
Further information on the deleterious health effects associated with exposure to radon has
been provided by experimental studies of animals. According to the BEIR IV Committee, animal
experiments conducted in both the United States and in France have confirmed the carcinogenicity
of radon, and have provided insight into the nature of the exposure-response relationship, as well as
the modifying effects of exposure rate (NAS 1988), The following is a summary of relevant findings
to date from animal studies:
« Health effects observed in animals exposed to radon and radon decay products
include lung carcinomas, pulmonary fibrosis, emphysema, and a shortening of life-
span (U.S. DOE/Office of Energy Research 1988a).
• The incidence of respiratory tract tumors increased with an increase in cumulative
exposure and with a decrease in rate of exposure (NAS 1988).
» Increased incidence of respiratory tract tumors was observed in rats at cumulative
exposures as low as 20 WLM (NAS 1988).
* Exposure to ore dust or diesel fumes simultaneously with radon did not increase the
incidence of lung tumors above that produced by radon progeny exposures alone
(DOE/Office of Energy Research 1988a).
• Lifetime lung-tumor risk coefficients that have been observed in animals (between
1 and 5 x 10 /WLM) are similar to the lifetime lung-cancer risk coefficients observed
in human studies (DOE/Office of Energy Research 1988a).
• In a study of rats exposed to radon progeny and uranium ore dust simultaneously,
it was observed that the risk of lung cancer was elevated at exposure levels similar
to those found in homes. The risk decreased in proportion to the decrease in radon-
progeny exposure (Cross et al. 1991).
Section HI: Exposure Assessment
EPA estimates annual U.S. lung cancer deaths due to radon by extrapolating risks observed
in miners to the general public based on estimated cumulative radon exposures, expressed in terms
of WLM. Exposure is often estimated indirectly, using assumptions or mathematical models. For
radon, EPA has directly estimated exposures to the public based on the results of the National
Residential Radon Survey (NRRS) (U.S. EPA/Office of Radiation Programs 1991a). The NRRS
included approximately 6,000 randomly selected homes that were statistically representative of all
residential structures across the fifty United States. Alpha track detectors were placed in each home
for a period of one year, and a questionnaire was used to gather information on occupant living
patterns, smoking habits, house construction, weatherization, and heating, ventilation, and air
conditioning characteristics. The survey was implemented in 1989 and 1990 and completed in 1991.
An estimate of average exposure in the general population must include information on both
the magnitude and duration of exposure. Based on results of EPA's NRRS, the current estimate of
the average national residential radon concentration (i.e., magnitude o£ exposure) is 1,25 pCi/L (based
Chapter 2; Estimation of Risks faSe 2-12
-------
on the average of all lived-in levels) (U.S. EPA/Office of Radiation Programs 199la). Using this
concentration, EPA has calculated the average annual U.S. cumulative exposure to radon to be 0.242
WLM. This estimate assumes that, on average, 75% of a person's day is spent in the home (i.e.,
duration of exposure), and the equilibrium factor is 0.5 (see section V.2.b for definition of the
equilibrium factor).
The primary source of residential radon concentrations is soil gas which enters the house
through cracks or openings in the foundation. However, radon can also enter the home through
ground water sources, and be released into indoor air during household water use such as showering
and washing clothes. It is estimated that this source accounts for approximately 5% of the total
indoor air concentration in houses served by ground-water sources. On average every 10,000 pCi/L
of radon in water contributes about 1 pCi/L of radon to indoor air. Generally, the contribution to
lung cancer risk from radon that enters the home through water is small compared to the risk from
radon that enters the home through soil gas. In addition, certain building materials and, to a lesser
degree, natural gas may occasionally make a small contribution to indoor radon concentrations. This
document focuses on the lung cancer risk associated with inhalation exposures to radon. Although
the ingestion of drinking water contaminated with radon may add some minor risk of developing
stomach or other internal organ cancer, this risk is small compared to the risks from breathing indoor
air containing radon (U.S. EPA/Radon Division 1991).
There is some uncertainty associated with the estimate of the U.S. population's average
annual cumulative radon exposure. The average residential radon level, the occupancy factor, and
the equilibrium factor all contribute to the uncertainty in the estimation of average residential
exposure. This uncertainty has been factored into EPA's overall estimate of risk as discussed in
Section V.
Section IV: Characterization of Exposure and Response
The purpose of exposure-response assessment is to describe the relationship between
exposure to the substance of interest and the occurrence of an effect or response (e.g., cancer).
Subsequently, an estimate of the potential response at various levels of exposure can be made based
on this relationship. Usually this means extrapolating from the effects observed at high levels of a
substance administered to experimental animals, or at occupational exposures noted in epidemiologic
studies, down to the generally lower levels of exposure to the general population.
Data on the relationship of exposure and response from human epidemiologic studies are
clearly preferable to animal data when predicting risks to human populations. Uncertainties
associated with applying a pattern of exposure and response observed in animals to humans are due
to obvious differences, including differences in genetic makeup, size, life span, metabolism,
absorption, and excretion of a substance.
EPA's estimates of lung cancer risks to the general population due to radon are based on
human exposure-response data from epidemiologic studies of underground miners. Unlike many
other environmental pollutants, therefore, special procedures for extrapolating from animals to
humans (with their associated uncertainties) are not needed for radon. Further, although it is
necessary to extrapolate from the risk observed in miners at occupational levels of exposure to
estimate the risk from residential exposure, the extrapolation is not large. Average cumulative
lifetime radon exposures to the general public are only slightly below those for which excess risk can
Chapter 2: Estimation of Risks Page 2-13
-------
be demonstrated in miners (about 40-70 WLM). For example, exposure to the annual average U.S.
cumulative radon exposure of 0.242 WLM over a lifetime (74 years) is equivalent to a cumulative
exposure of approximately 18 WLM, and exposure to a concentration of 4 pCi/L over a lifetime is
equivalent to a cumulative exposure of approximately 57 WLM, assuming that the equilibrium factor
is 0.5 and the fraction of time spent indoors is 75% (see Exhibit 2-13 for more comparisons of WLM
and pCi/L based on these assumptions).
Assumption of Linear Exposure-Response Relationship
In assessing residential radon risk, EPA assumes that the exposure-response relationship is
linear at low exposures and exposure rates. This assumption is consistent with the evidence for
linearity at a wide range of cumulative exposures in the radon epidemiologic studies of underground
miners. There is no evidence of a threshold for lung cancer response from radon exposure, that is,
a level of radon exposure below which no increased risk of lung cancer would exist. It is generally
recognized that even low doses of alpha radiation produce DNA damage that cannot always be
repaired. Research further indicates that at low doses of alpha radiation the dose-response
relationship for cell transformation and tumorigenesis is linear and independent of dose rate (NCRP
1980; NAS 1988).
The BEIR IV Committee has stated that "linear models are adequate for extrapolation to low
doses of high linear energy transfer (LET) radiations" (NAS 1988). The Committee noted that the
only significant evidence against linearity in their analysis was the observation of a decrease in risk
per WLM at very high exposure levels (above 2,000 WLM), and these levels were excluded from their
primary analysis (NAS 1988). Significant increases in lung cancer mortality have been observed, in
miners at a wide range of cumulative exposures, including low levels comparable to a lifetime
residential exposure at 4 pCi/L. In developing their radon risk projection models, both the ICRP 50
and BEIR IV Committees assumed that the relationship between cumulative radon progeny exposure
and lung cancer risk is linear at low exposures. The EPA's Science Advisory Board (SAB)
recommended that the Agency continue to assume a linear exposure-response relationship when
extrapolating risks from miners to estimate the risks from residential exposures (U.S. EPA 1988).
Risk Projection Models
1. History
Several models have been developed to characterize the lung cancer risk associated with
exposure to radon daughters in the residential environment, based on studies of miners exposed to
radon. EPA has relied primarily on relative risk projection models to estimate radon risks to the
public. This section reviews the models that EPA used in the past to calculate radon risk estimates.
Section 2 that follows describes EPA's current approach to the estimation of radon risks.
In general terms, the relative risk models assume that the rate of excess lung cancer due to
radon increases with the baseline rate of lung cancer in the population due to all other causes. This
means that the estimated number of deaths due to radon will vary depending on the baseline lung
cancer mortality rate, which varies with age, sex, and other factors such as the prevalence of smoking
in the population. Since smoking accounts for most of the baseline lung cancer risk, this approach
is consistent with the assumption that radon and smoking act multiplicatively in causing lung cancer
(Puskin and Yang 1988). A multiplicative relationship exists when the relative risk associated with
Chapter 2: Estimation of Risks Page 2-14
-------
combined exposure to both factors is approximately equal to the product of the relative risks
associated with exposure to either of the factors alone.
Prior to 1988, EPA used a constant relative risk model to project radon-induced lung cancer
risk in the general population. The model was of the form:
r(a)=r0(a) [1 + BW*] (1)
where:
r(a) represents the lung cancer mortality rate at age a from all causes including radon
exposure,
r0(a) is the age-specific baseline lung cancer mortality rate,
fi is the relative risk coefficient, and
W' is the cumulative "exposure equivalent" up to age a-10.
EPA's SAB recommended assuming a range of 1 to 4%/WLM for 8. The "exposure
equivalent" was an adjustment to the cumulative exposure estimate (WLM) to account for age-
specific differences in airway dimensions and surface area, respiratory frequency, and tidal volume.
EPA assumed that the dose per WLM was 40% lower for residential exposures than for exposures
received in underground mines.
The constant relative risk model was used in conjunction with a lifetable analysis governed
by 1980 U.S. vital statistics, to project a risk factor of 460 fatal cancers/106 person-WLM with an
uncertainty range of 230-920 fatal -cancers/106 person-WLM. EPA's estimate of annual lung cancer
deaths was 5,000-20,000 lung cancer deaths per year (U.S. EPA/Office of Radiation Programs 1987).
In May 1988, EPA reviewed its risk assessment methodology in response to the publication
of the BEIR IV report (MAS 1988) and the issuance of a draft risk assessment for the National
Emission Standards for Hazardous Air Pollutants (NESHAPS) rulemaking on radionuclides under
the Clean Air Act (U.S. EPA/Office of Radiation Programs 1989). The SAB recommended that EPA
use both the BEIR IV and the ICRP 50 reports to determine a range o£ radon risk estimates, and
suggested using the average of the central values calculated with these two risk projection models to
derive EPA s risk estimate (U.S. EPA 1988). The SAB did not recommend use of the NCRP 78
model, which is an absolute risk model. Absolute risk models have been described as less appropriate
for the estimation of lifetime radon risk, as they do not assume a temporal correlation with the
baseline lung cancer rate (ICRP 1987). Moreover, the NCRP model presumes that the effects of
radon and cigarette smoke are additive, contrary to evidence from the Colorado Plateau Study. The
SAB also recommended that EPA assume that indoor and mining exposures give the same dose per
WLM of cumulative radon exposure. This was also the assumption used by the BEIR IV Committee.
For assessing the risks due to childhood exposures, the SAB recommended that EPA rely on the
ICRP 50 model. This model assumes a three-fold higher risk coefficient for exposures to people
under 20 years of age. EPA used the BEIR IV and the ICRP 50 models in conjunction with a
lifetable analysis and 1980 mortality data to estimate risk factors (lung cancer deaths/106 person-
WLM) and annual deaths from lung cancer.
Chapter 2; Estimation of Risks Page 2-15
-------
In deriving these rates with both models, EPA adjusted the age-specific baseline rates of lung
cancer mortality, r0(a), by eliminating deaths due to average background radon exposure. Relative
risk is normally used to determine the excess cases of a disease in an exposed population compared
to an unexposed population. But in the case of radon, exposure is universal. The 1980 age-specific
baseline rates of lung cancer include a radon contribution to lung cancer deaths. Therefore, EPA
derived "radon free" background lung cancer rates, that is, reduced to reflect the removal of the age-
specific radon contribution to these rates. EPA assumed that the average background radon exposure
was 0.25 WLM/year. The net effect of this adjustment was to reduce the lifetime risk estimates given
by the BEIR IV model by approximately 15% (U.S. EPA/Office of Radiation Programs 1989). In
this way, the relative risk models do not project higher rates of excess radon-induced lung cancer
based on lung cancer rates that already include lung cancers due to background radon. In their
calculation of lifetime population risks from indoor radon exposure, the ICRP 50 Committee
corrected the baseline lung cancer rate assuming an average exposure rate of 0.2 WLM/year (ICRP
1987). The BEIR IV Committee did not incorporate this correction in their risk estimates (NAS
1988). EPA adjusted both models assuming an average exposure rate of 0.25 WLM/year.
EPA's calculations using the adjusted BEIR IV model yielded a risk factor of 305 lung cancer
deaths per 10 person-WLM, and an estimated 18,300 annual lung cancer deaths. The calculations
using the adjusted ICRP 50 model yielded a risk factor of 420 lung cancer deaths per 10 person-
WLM, and an estimated 25,200 annual lung cancer deaths. EPA then averaged the values calculated
using each of these models separately. This gave a risk factor of 360 lung cancer deaths per 106
person-WLM and an estimated 21,600 annual lung cancer deaths.
The following sections discuss the ICRP 50 and the BEIR IV models and the adjustments
made by EPA.
a. ICRP 50 Model
The ICRP 50 Committee developed their constant relative risk model based on the lung
cancer incidence observed in three of the miner cohorts (U.S., Ontario, and Czechoslovakian uranium
miners) and on information from the studies of the atomic bomb survivors (ICRP 1987). The model
assumes a constant rate of excess lung cancer risk due to radon over time and age; that is, the
increase in relative risk for a given exposure is constant over time after exposures end. The ICRP
50 model is written:
r(a) = r0(a) + r0(a)pW(a.10) (2)
where:
r(a) represents the rate of lung cancer mortality from all causes (including radon) at a specific
attained age a,
r0(a) is the age-specific baseline lung cancer mortality rate (corrected for background radon
exposure),
p is the relative risk coefficient for radon-induced lung cancer,
Chapter 2: Estimation of Risks Page 2-16
-------
"Wfa-10) is the cumulative radon progeny exposure up to age a except for the previous 10
years. The 10 years represent the assumed minimum latency period between exposure and
cancer development, and
r0(a)pW/a JQN represents the age-specific rate of excess lung cancer risk due to radon.
The ICRP 50 model does not assume that the excess risk of lung cancer in miners due to
radon can be directly transferred (extrapolated) to the general population. To adapt a relative risk
model derived from studies of miners for use in estimating risks to the general population, the ICRP
made certain adjustments. The radon relative risk coefficient from the miner data (p) was modified
by the ICRP Committee to reflect presumed differences between mine and indoor radon exposures.
First, because of potential co-carcinogenic influences that may be present in the mines but
not indoors (e.g., exposures to vapors, dust, and other forms of radiation), the ICRP assumed that
the risk coefficient for indoor exposures would be only 80% of that from mine exposures.
Second, due to potential differences between indoor and mine exposure conditions, the ICRP
Committee assumed that the absorbed dose of alpha radiation per unit of cumulative radon exposure
(WLM) for the general population is only 80% of that for miners.
Third, based on findings from studies of the atomic bomb survivors, the ICRP Committee
assumed a three-fold higher risk coefficient for exposures to people under 20 years of age. This had
the effect of increasing overall lifetime risk by approximately 40% (Puskin and Nelson 1989).
When these correction factors are applied to the miner relative risk coefficient, the ICRP
Committee's risk coefficients for indoor exposures to radon are:
p = 1.9% per WLM for exposures at ages <, 20 years
P = 0.64% per WLM for adult exposures > 20 years
When EPA used the ICRP 50 model to estimate lung cancer risks to the general population,
a slight modification was made to these ICRP Committee risk coefficients (U.S. EPA/Office of
Radiation Programs 1989). EPA eliminated the assumption that the absorbed dose of alpha radiation
per unit of cumulative radon exposure (WLM) for the general population is only 80% of that for
miners, and assumed instead that there was a 1:1 relationship between dose per unit exposure in
mines and in homes. This was based upon a recommendation from the SAB and evidence considered
by the BEIR IV Committee. The recalculated risk coefficients are then:
p = 2.4% per WLM for exposures at ages <_ 20 years
JJ = 0.8% per WLM for adult exposures > 20 years.
b. BEIR IV Model
The BEIR IV Committee developed their relative risk projection model based on detailed
analyses of four major cohort studies of underground miners (NAS 1988). These were the U.S.,
Ontario, and Saskatchewan (Beaverlodge) uranium miners and the Swedish (Malmberget) iron
miners. The Committee reviewed both the original study data as well as unpublished material, and
conducted a reassessment of the data using new statistical methods.
Chapter 2: Estimation of Risks Page 2-17
-------
The Committee found a pattern of increased risk of lung cancer that was very consistent
among the four groups. In the U.S. uranium miners, a decline in relative risk with age and with time
since exposure was observed. There was no consistent effect on relative risk from age at first
exposure or from rate of exposure. As additional data are collected from ongoing follow-up of the
miner cohorts, the lifetime pattern of risk from radon exposure may be further clarified.
The BEIR IV modified relative risk model departs from a constant relative risk model (such
as the earlier EPA model and the ICRP 50 model) by including additional assumptions about the
effects of time since exposure (TSE) and of age. The BEIR IV preferred TSE model can be written
as:
r(a) - r0(a)[l + Py(a) (Wt + 0.5W2)] (3)
where:
r(a) is the lung cancer mortality at attained age a due to all causes including radon exposure,
r0(a) is the age-specific baseline lung cancer mortality rate,
y(a) is the age-specific adjustment to the relative risk coefficient for radon (P = 0.025), with
y(a) =1.2 when a < 55 years
= 1.0 when a is 55-64 years
= 0.4 when a >_ 65 years.
The y(a) adjustment decreases the radon-induced lung cancer risk with age. This
incorporates the Committee's finding that excess relative risk in the miners decreased with
age at risk.
(Wj + 0.5W2) represents cumulative lifetime exposure up to age a modified as follows:
W^ = cumulative exposure occurring from 5-15 years before age a, and
W2 = cumulative exposure up to age a-15 years.
Since W2 is reduced by 50%, the model gives less weight to exposures more distant in time
since exposure. This reflects the Committee's conclusion that risk decreases with time since exposure
as modeled for the four cohort studies of miners. Hence, the relative risk coefficient (ji = 0.025)
effectively varies from 0.5%/WLM to 3.0%/WLM, depending upon age at risk and time since
exposure (Puskin and Nelson 1989). According to the model, 0.025y(a)r0(a)(Wj + 0.5W2) represents
the rate of excess lung cancer due to radon.
2. Current Risk Projection Approach - Adjusted BEIR IV Model
In 1991, the Office of Radiation Programs requested that the SAB review proposed revisions
to EPA's radon risk assessment methodology. The SAB recommended that the Agency use only the
BEIR IV model for assessment of risk from residential exposure to radon. The recommendation to
use only the BEIR IV model and discontinue use of the ICRP 50 model was based on several new
pieces of information. The first was evidence from epidemiologic studies of a decrease in lung cancer
Chapter 2: Estimation of Risks Page 2-18
-------
risk with time since exposure, which had been incorporated into the BEIR IV model, but not the
ICRP 50 model. The second was the publication of the BEIR V report (NAS 1990) and a study of
Chinese miners exposed to radon gas (Lubin et al. 1990). These publications found no evidence of
dependence on age at exposure for lung cancer. This was not consistent with the increased risk to
children assumed in the ICRP 50 model (for further discussion of age dependence of risk see Section
V).
EPA has made two adjustments to the BEIR IV model in estimating radon risks. The first
modification was an adjustment of the age-specific baseline lung cancer mortality rate by eliminating
deaths due to an average background exposure of 0.242 WLM/year, reducing the lifetime risk
estimates by about 10%. The second modification was based on findings in the NAS' report,
completed under a grant from EPA, Comparative Dosimetty of Radon in Mines and Homes (NAS
1991). The purpose of this report was to compare the dose to the bronchial epithelium per unit of
radon progeny exposure in mines and homes, based upon a number of physical and biological factors
which are expected to differ in the two environments. Among the factors considered in the report
are the following: age, sex, aerosol size distribution, the unattached fraction of radon progeny, the
breathing rate and route (oral vs. nasal), the pattern and efficiency of deposition of radon progeny,
the solubility of radon progeny in mucus, and the growth of aerosols in the respiratory tract (for
further discussion of this report see Section V.l.d.). This comparison of exposure-dose relations in
the mining and home environments indicated that the dose per unit of exposure to radon progeny
is approximately 30% lower in the home environment. This finding was expressed in terms of a K-
factor (K = 0.7).
Risk is presumed to be proportional to the dose received by target cells in the bronchial
epithelium. Therefore, in calculating the risks for residential exposures, it is appropriate to multiply
the risk coefficient in the BEIR IV model by a factor of K.
Taking K = 0.7 for both sexes and all ages, the modified BEIR IV model becomes:
r(a) = r0(a)[l + 0.0175 Y(a) (Wt + O.SW2)j (4)
where the parameters are as described in equation 3, with the difference that p = 0.0175 as a result
of the adjustment for the factor K = 0.7.
The most recent estimates of the deaths/person-WLM and annual U.S. lung cancer deaths
calculated using the BEIR IV model are discussed in Section V of this chapter.
3. Lifetable Derivation: Rate of Radon-Induced Excess Lung Cancer Death
Because data are not available on the actual incidence rate of lung cancer death due to radon
for the general population, the risk projection models are used to calculate a rate. (Although death
certificates may identify cause of death, it is not possible to differentiate between the lung cancer clue
to radon and that due to smoking, or any other cause.)
To calculate a lifetime, age-averaged rate for radon-induced lung cancer in the general
population, EPA has used a standard lifetable analysis in conjunction with the BEIR IV model as
adjusted by EPA (U.S. EPA/Office of Radiation Programs 1989). The lifetable approach was used
to account for the effect of competing causes of death. For each year of the lifetable calculation, the
rate of age -specific death due to all causes competes with the rate of age-specific radon-induced lung
Chapter 2: Estimation of Risks Page 2-19
-------
cancer death. This Is important because of the inherent delay between radon exposure and the
occurrence of lung cancer. During and after the period of radon exposure and before cancer appears,
a potential lung cancer victim is exposed to other causes of death (e.g., auto accidents, cardiovascular
disease). When people die from these other causes, they are no longer in the "pool" of people at risk
of dying from lung cancer due to radon.
EPA uses a lifetable for a general population that reflects the 1980 U.S. age-specific mortality
rates and vital statistics. The lifetable starts out with the "birth" of 100,000 hypothetical persons who
are followed from birth to death, with no member living longer than 110 years. Therefore, in each
year of the lifetable, all members of the "birth cohort" (minus those that have died from various
causes in the previous year, including radon-induced lung cancer) are the same age.
Using the BE1R IV model, the lung cancer deaths due to a given cumulative radon exposure
(person-WLM) can be calculated for each year of the lifetable over a full life span, from birth to 110
years. The deaths for each year are calculated based on the cumulative radon exposures for that year
and for the preceding years as determined by the model. The age-specific rate of lung cancer death
due to radon derived with the BEIR IV model is G.0175y('a)r0(aXWj + -5W2) (see previous
discussion of the BEIR IV model).
In deriving the risk per unit exposure, EPA assumes each member of the lifetable receives
a constant low level of radon exposure (0.001 WLM/year) from birth to death. (A constant low level
of exposure over 110 years is used because the lifetime risk estimate would no longer increase linearly
if there were high cumulative radon exposures.) Thus, from the beginning of the lifetable, 100,000
persons are subjected year after year to the age-specific mortality rates of the 1980 U.S. population
and to an increased age-specific lung cancer mortality rate due to low level radon exposure
experienced up to that year.
As each age-specific radon-induced lung cancer mortality rate is calculated (e.g., for age 10-
11), it is applied to the surviving lifetable population for that year. This generates the excess deaths
that year for the cumulative radon exposure experienced by the lifetable population up to that year
(i.e., Y excess lung cancer deaths per Z person-WLM). These deaths are then reflected in the
overall mortality rate from all causes for the next year (e.g., for age 11-12). Therefore, due to deaths
from all causes, each year there are fewer and fewer of the original 100,000 members of the lifetable
population still alive and at risk from radon exposure. The lung cancer mortality for cumulative
radon exposure is recalculated each year of the lifetable based on the revised population and
cumulative radon exposure for that next year.
After this has been done for each of the 110 years in the lifetable, the excess radon-induced
lung cancer deaths for each year can be summed to give a total lifetime number of deaths for the
population. The collective exposures to the cohort from each year (i.e., over all ages) are also
summed to give the lifetime population exposure.
4. Applying the Rate of Radon-Induced Lung Cancer to the U.S. Population
EPA's estimation of annual deaths due to radon (see Section V) is based on a steady-state
population, i.e., one that does not change from year to year. The steady-state population is
constructed to have people of all the same age groups, and in the same proportion, as those in each
year of the lifetable cohort (i.e., covering newborns to 110 year-olds). For each year of exposure, the
population will have the same proportion of exposure (person-WLM) by age as a birth cohort
Chapter 2: Estimation of Risks Page 2-20
-------
receiving a constant lifetime exposure. This is essentially the population that would result from a
constant birth rate of 100,000 per year (birth cohort) with the age-specific mortality rates of each
cohort defined by the 1980 mortality rates and vital statistics. Therefore, the number of radon-
induced lung cancer deaths per person-WLM for constant lifetime exposure to the single birth cohort
is the same as that for an annual exposure to all members (at various ages) of the stationary
population (U.S. EPA/Office of Radiation Programs 1989).
Because these estimates of annual lung cancer deaths due to radon were derived using a
lifetable defined by 1980 statistics, they are specific to the 1980 population. Since a relative risk
approach is used, the estimated number of excess lung cancer deaths due to radon exposure will
change as these background rates of lung cancer mortality change. As new mortality rates and vital
statistics become available, EPA will revise its estimates of radon risk.
Section V: Risk Characterization
EPA's characterization of the risk associated with residential exposure to radon has two
primary components: calculation of the numerical estimates of risk, and presentation of the
information needed to judge the overall significance of the risk estimates.
In addition to deriving estimates of risk from low-level residential radon exposures and the
number of lung cancers induced by radon each year using the BEIR IV model, EPA has identified
and quantified, to the extent possible, the major sources of uncertainty in these estimates and has
made a quantitative estimate of overall uncertainty. This includes a discussion of the uncertainty
associated with both the risk factor and the estimate of average residential radon exposure. Since
the uncertainty analysis is critical to the interpretation of the risk estimates, this analysis is presented
first, followed by EPA's current estimates of risk.
Uncertainty Issues
In the past, EPA has quantified uncertainty using the ICRP uncertainty range which
represents a range of risk coefficients from the miner studies (ICRP 1987). The Agency's current
uncertainty analysis is based on the methodology used in the National Institutes of Health report on
the development of radioepiderniological tables (NIH 1985). According to this approach, each
component of uncertainty is treated as independently affecting risk by a multiplicative factor, and the
upper and lower bounds for each component of uncertainly are treated as the limits of a 90%
"confidence interval." These are referred to as "credibility intervals" by NIH to emphasize the
element of subjectivity involved in their estimation. In estimating the overall uncertainty, each
component is further assumed to be distributed lognormally on each side of the nominal geometric
mean (G) of the upper and lower bounds. The lower and upper bounds are then defined by
multiplicative factors L and U, such that the limits of a 90% confidence interval are given by LxG
and UxG, respectively. The selection of a lognormal distribution is somewhat arbitrary, and much
of the justification is one of practicability and mathematical tractability.
Sources of uncertainty that do not readily lend themselves to quantification were not included
in the uncertainty analysis, but were addressed qualitatively.
Chapter 2: Estimation of Risks Page 2-21
-------
1. Risk Factor and Associated Uncertainty
The following are sources of uncertainty associated with the estimate of lifetime risk that were
addressed in the uncertainty analysis: (a) statistical variability in the miner data, (b) projection of risk
beyond period of epidemiologic follow-up (projection of risk over time), (c) age dependence of risk,
(d) extrapolation from mines to homes, (e) influence of mine exposures other than radon, (f)
exposure-rate effect, (g) extrapolation to females, and (h) relationship between radon risk and
smoking.
Of the eight sources of uncertainty that were addressed, four were included in the quantitative
analysis. These were: variability in the miner data, time projection, age sensitivity, and extrapolation
from mines to homes based on bronchial dose.
a. Statistical Variability in the Miner Data
As discussed in the BEIR IV Report, the statistical variability in the miner data is most readily
addressed in terms of a constant relative risk model. It is recognized that this is not the model
preferred by the BEIR IV Committee for describing the temporal and age dependence of the risk;
however, the uncertainty in the time and age projections can be treated separately.
The BEIR IV Committee analyzed four cohorts of miners. Exhibit 2-3 shows the risk
coefficient derived from each study, assuming a constant relative risk model, and, in parentheses, the
respective multiplicative standard error (MSB).
EXHIBIT 2-3
BEIR IV ESTIMATES OF RISK COEFFICIENTS FOR
A CONSTANT RELATIVE RISK MODELS
Cohort % Increase in Relative Risk/WLM
p (MSB)
Eldorado (Beaverlodge) 2.6 (1.5)
Ontario 1.4(1.6)
Malmberget 1.4 (2.6)
Colorado Plateau 0.6(1.5)
- Source: NAS 1988.
The values in Exhibit 2-3 are based on an internal analysis in which the miner cohort
effectively served as its own control (NAS 1988). Results from an external analysis of each cohort
Chapter 2: Estimation of Risks Page 2-22
-------
differ only slightly from these values. Treating the errors as lognormally distributed, the combined
estimate of the risk coefficient and standard error from the four studies is calculated to be:
In P = ; (5)
Lj(
l/(ln2 S) =
where:
the PJ values are the respective risk coefficients estimated for the individual studies, and
the Sj values are the corresponding multiplicative standard errors.
Substituting from Exhibit 2-3 above, the combined estimate for p is 130%/WLM (S = 1.27).
This compares with the BEIR IV combined estimate of 134%/WLM and an MSE of about 1.3. The
small difference between the estimates given in BEIR IV and those given here (calculated by EPA)
may reflect rounding errors or a difference in method for arriving at the combined estimate (the
BEIR IV report does not describe its calculation).
As noted in BEIR IV, agreement among the studies is quite good and, on statistical grounds,
there is no reason to reject any of the four studies as an outlier. Further support for the reliability
of the estimate is provided by other studies not included in the BEIR IV analysis. A noteworthy
example is a recent analysis of the Czechoslovakian uranium miner data that derived the estimate of
1.5%/WLM (Sevc et al. 1988). In addition, based on an analysis of 8 miner cohorts, Lubin (1988)
obtained a combined estimate for p of 1.5%/WLM with an MSE of 1.2. It was noted that none of
the four studies considered by BEIR IV can be regarded as a statistical outlier. It is nonetheless
instructive to examine the sensitivity of the combined estimate for p to possible bias in a single study.
If, on the one hand, the cohort yielding the highest estimate (Eldorado) is removed from the analysis,
the estimate of p is reduced from 1.30 to 0.90%/WLM. On the other hand, removing the Colorado
cohort, which yields the lowest estimate, increases the combined estimate to 1.93%/WLM. Thus, the
risk factor is fairly robust with respect to potential systematic error in any one study.
Upper and lower confidence limits on p associated with the statistical variability in the miner
data are readily obtained from the estimated MSE of 1.3: U = L"1 -(1.3)1-64 = 1.5. That is,
considering only statistical variability in the miner data, the 90% confidence interval for P is (0.013/1.5
- 0.013 x 1.5), or about 0.0087 to 0.0195.
It is important to emphasize that this estimate of uncertainty in the miner data does not
include uncertainty associated with systematic errors in the assignment of exposure to individual
miners, and the possible influence of mine exposures other than radon.
b. Projection of Risk Over Time
When data from all four of the BEIR IV miner cohorts were combined, the resulting risk
coefficient was 1.3%/WLM. This value represented a best estimate for the limited period of follow-
up in the cohorts, assuming a constant relative risk model and a 5-yr minimum latency. Further
statistical analysis, however, revealed an apparent decrease in the relative risk with (1) age at risk and
(2) time since exposure. Consequently, the BEIR IV Committee preferred model incorporates age-
at-risk and time-since-exposure dependent parameters. The net effect of these model changes was
Chapter 2: Estimation of Risks Page 2-23
-------
to reduce the estimated lifetime risk due to lifetime exposure by about one-third. In the preferred
model, the risk coefficient effectively varies from 0.5 to 3%/WLM, and the lifetime risk estimate is
about the same as would be projected using a constant relative risk model with p = 0,9%/WLM.
Although there is statistically significant evidence of a temporal fall-off in risk in the data for
four cohorts, much uncertainty remains as to the rate and eventual degree of the fall-off. Some of
the observed decline may be due, at least in part, to changes in miner smoking habits over time or
other unknown confounding factors. However, the BEIR IV Committee concluded that it was
unlikely that the decrease could be accounted for simply by changes in the smoking habits of the
miner cohorts. Additional follow-up on temporal changes in the smoking habits of the miners may
help settle this issue.
To obtain a reasonable bound on the possible underestimate in risk due to uncertainty over
the time projection, a constant relative risk model was assumed. As indicated above, this assumption
increased the risk estimate by about a factor of U = (1,3%/WLM)/(0.9%/WLM) - 1.5.
An estimate of the corresponding lower bound was obtained using an approach suggested in
the BEIR IV Report (NAS 1988). Although the preferred model assumes an excess risk projection
over a full lifetime, useful follow-up on the miners extends only for 30-40 y after their first exposure.
To obtain a reasonable lower bound estimate, therefore, it was assumed that the excess risk ends
completely about 40 y subsequent to exposure. As pointed out in BEIR IV, this is equivalent to
stipulating that W2, in the preferred model, includes only exposure accrued 15-40 y before the age
at risk. Quantitatively, the effect of such a postulated cut-off would be to reduce the general
population risk estimate by about 33%, i.e., by a factor of L = 1/1.5 (see Exhibit 2-4).
As shown in Exhibit 2-4, most of the effect of such a cut-off is associated with childhood
exposures. The estimated risk due to exposures at age 0 and 10 is reduced by a factor of 100 and
10, respectively. For all exposures between ages 0 and 20, the average risk is decreased by a factor
of 7.5; by comparison, for all exposures above age 20, the average risk is decreased by only 10%.
EXHIBIT 2-4
AVERAGE LUNG CANCER MORTALITY RISK PER 106 WLM BY EXPOSURE
WITH AND WITHOUT 40- YR CUTOFF IN EXPRESSION OF RISK^.
Projection
Limit
None
40 Yr
0
322
3.4
Exposure Group by Age
10 0-20
328 328
32 44
(y)
+20
294
264
GROUP,
All
304
205
-' Source: Calculated using the BEIR IV model (uncorrected for difference, between dosimetry in homes and
mines). The baseline lung cancer rate was corrected for the contribution of background radon exposure, which was
assumed to be 0.23 WLM/y for this calculation.
Chapter 2: Estimation of Risks Page 2-24
-------
c. Age Dependence of Risk
The BEIR IV Committee "did not find an effect of age at first exposure, after controlling for
other correlates of age" (NAS 1988). However, their study population was comprised entirely of
adults. Direct evidence on the sensitivity of children to radon is very limited. The oniy such
information currently available is from a study of Chinese tin miners that included some individuals
who started mining before age 13 (Lubin et al. 1990). No enhancement in the risk per unit exposure
was seen in these individuals as compared to the rest of the cohort. However, given the sparseness
of the data and the lack of control for smoking, no definite conclusions can be drawn from this study.
Atomic bomb survivors irradiated before age 20 show a highly statistically significant greater
relative risk for radiogenic induction of cancer than those irradiated after age 20. Also, for lung
cancer, a higher relative risk is observed for those irradiated before age 20, but the increase in risk
is not statistically significant (Shimizu el al. 1988). Guided by the results obtained on the atomic
bomb survivors (through 1978), the authors of the ICRP 50 report on radon risk assigned a 3 times
higher relative risk coefficient to exposures received before age 20 (ICRP 1987).
The data on atomic bomb survivors are suggestive of a higher risk of radiogenic lung cancer
in children, but epidemiologic follow-up is still too short for those exposed as children to express
many lung cancers; consequently, no firm conclusions can be drawn. Additional follow-up should
settle the question, but the applicability of the findings to the case of chronic radon daughter
exposures would remain an issue.
As an upper limit on the possible effect of age at exposure on the BEIR IV risk estimate, the
lifetime risk for a constant exposure rate was calculated using the preferred BEIR IV model revised
to incorporate a 3 times higher risk coefficient for all exposures before age 20. The net effect of the
modification was to increase the general population risk by about 60%. Accordingly, for this source
of uncertainty, U is taken to be 1.6.
While there is no basis for postulating that the lungs of children are inherently less sensitive
to the carcinogenic effects of radon daughters, a temporal fall-off in risk could effectively make the
lifetime risk from childhood exposures very small This possibility has already been addressed in the
previous section and incorporated into the uncertainty relating to the temporal projection. Thus, with
respect to age sensitivity, L is taken to be 1.0.
d. Extrapolation from Mines to Homes
As discussed previously, the risk coefficients used in the BEIR IV risk model were adjusted
by a factor K = 0.7 to correct for an estimated lower bronchial dose per WLM in homes as compared
to mines, based on information contained in the recent report, Comparative Dosimelry of Radon in
Mines and Homes (NAS 1991). The NAS' report did not attempt to quantify the overall uncertainty
in K, but it did indicate how much change would occur under some alternative sets of assumptions.
That information was used here to develop an estimate of the uncertainty in K.
It is important, in this connection, to distinguish uncertainty from variability. The value of
K appropriate to an individual will vary widely across the population, depending on personal
characteristics (e.g., age) and on conditions in the home. For example, Table 3-4 of the NAS' report
shows large standard deviations on K (typically, about 30%). In large part, these reflect calculated
Chapter 2: Estimation of Risks Page 2-25
-------
changes in K due to variability in the aerosol conditions that people are exposed to, not uncertainties
in the value of K most characteristic of average exposure conditions.
Numerous assumptions have to be made in calculating the dose to target cells in the lung
from a given radon exposure, and each of these assumptions is associated with some uncertainty. For
example, the NAS' Committee considered various alternative classes of target cells; primarily,
secretory vs. basal cells, and the whole bronchial tree vs. the upper (lobar/segmental) bronchi only.
No preference was expressed regarding the type of cell; estimates were generally given for secretory
and basal types separately or as the mean of the two. Assuming that secretory or basal cells alone
are the target cells would increase or decrease the estimate of K by 6%, respectively, relative to the
mean. The preferred estimates were based on target cells being distributed over the whole bronchial
tree; if only the upper bronchi were to be considered, K would have been decreased by 5%.
Much of the uncertainty in K pertains to the dose contributed by unattached radon decay
products in the home. According to estimates endorsed by the NAS* Committee, it was considered
that about 2/3 of the total dose in homes comes from the unattached fraction. However, subsequent
to the publication of new dosimetric information, the Committee presented revised estimates based
on a modified set of assumptions. The modifications included a higher nasal deposition of unattached
daughters and a smaller particle size in some mine locations. With the higher nasal deposition, the
contribution of the unattached fraction in homes fell to roughly 50% (James 1991), and the estimate
of K for each age group was reduced by about 20%.
The fraction (f_) of radon daughter alpha-energy which is "unattached" (i.e., present as an
ultrafine fraction) varies considerably between houses and, over time, even within a single house. The
NAS' report adopted a value of 8% as typical for the indoor environment, and cited a range of 7%
to 10% for indoor measured values in the absence of smoking. However, given differences in heating
and cooking methods, air cleaning equipment, ventilation, etc., this range would seem to be too
narrow. In this connection, the report noted that t" may exceed 16% in houses with low aerosols.
Tobacco smoke provides additional attachment sites for radon decay products; consequently,
f is expected to be lower in houses with a smoker present. The NAS' Committee assumed that, on
average, f is reduced from 8% to 3% during waking hours in houses with a smoker, and to 1%
during actual smoking. As a consequence, for waking hours, K is reduced by about 40% in houses
with a smoker present. Taking into account sleeping hours, the reduction in average K is roughly
20%. The reduction would be less under the assumptions discussed above (higher nasal deposition
of unattached daughters and smaller particle size in mines), since the contribution to dose from the
unattached fraction would be less. Another aspect that should be considered in this connection is
the higher equilibrium factor (defined in Section V.2.b.) expected in homes with tobacco smoke
(Vanmarcke et al. 1988). Thus, for a given radon concentration, smoking will increase radon
daughter concentration but decrease the unattached fraction. These changes will have opposite
effects on dose and it is unclear whether the average dose, at a given radon concentration, would
actually be higher or lower in homes with a smoker.
A critical assumption made by the NAS' Committee, which substantially reduced their estimate
of K, pertains to the hygroscopic growth (by taking up and retaining moisture) of aerosols inside the
respiratory tract. Except for certain nonhygroscopic aerosols produced by cooking and vacuuming,
the Committee assumed a doubling of the aerosol particle size in the respiratory tract. Due to
differences in the initial size distribution of the aerosols, assuming hygroscopic growth led to a
reduction in calculated dose for homes but an increase for mines, especially during heavy exercise.
Chapter 2; Estimation of Risks Page 2-26
-------
With no hygroscopic growth, K would be increased by about a factor of 1.6 (1.5, with modified
assumptions on nasal deposition and particle size).
Summarized in Exhibit 2-5 are the average adult values calculated for K by the NAS' panel,
with and without the modifications in nasal deposition and particle size, with and without growth
(doubling) of aerosol size in the respiratory tract (NAS 1991). For children, the values of K would
typically be increased by about 10%.
EXHIBIT 2-5
AVERAGE ADULT VALUES FOR K ADJUSTED FOR
NASAL DEPOSITION AND HYGROSCOPIC GROWTHS7
Assumptions on Nasal
Deposition and Particle Growth No Growth
Size
Unmodified 0.70 1.1
Modified 0.58 0.88
*' Source: NAS 1991.
Another source of uncertainty in K is associated with uncertainty over the activity-weighted
size distributions of radon progeny, both in mines and in homes. However, this uncertainty is not
quantified in the NAS' Radon Dosimetry report,
In view of the recent data indicating higher nasal deposition of the unattached fraction and
the expectation of at least some hygroscopic growth of particles in the respiratory tract, it would
appear that the dose/WLM is, on average, higher in mines than in homes (i.e., K < 1). Taking into
account possible enhanced nasal deposition of unattached daughters in conjunction with an assumed
size doubling of hygroscopic particles, as well as some allowance for additional sources of uncertainty
inherent in the NAS' analysis, K = 0.5 is assumed to be a reasonable lower bound. These
conclusions must be regarded as tentative and subject to change in light of better information on
aerosol conditions in mines and homes. Accordingly, a symmetric multiplicative uncertainty range U
= L~l = 1.4 is taken around the central value of K = 0.70.
e. Other Mine Exposures
Uncertainty associated with the presence of agents such as ore dust, arsenic, fluorides, silica
and diesel fumes in the mining environment has been considered but not quantified in this analysis.
It has been suggested that these other mine exposures may have enhanced the risk of lung
cancer in the miners by increasing cell turnover (NAS 1991). It is important to note, however, that
the potential effects of other mine exposures have been taken into consideration in many of the
miner studies. Sevc et al. (1984) reported that in the Czech uranium mines, a comparison of areas
with similar radon levels but different arsenic levels showed no variation, in lung cancer incidence.
Chapter 2: Estimation of Risks Page 2-27
-------
Archer et al. (1985) conducted an analysis of five mining or milling groups and found that lung cancer
rates correlated with radon exposure regardless of silica dust levels. Animal studies have shown that
exposure to ore dust or diesel fumes simultaneously with radon does not increase the incidence of
tumors produced by radon daughter exposures (Cross et al. 1991). The miner studies were conducted
in different types of mines where environmental pollutants other than radon were present to varying
degrees, and radon was the common exposure. The similarity in the estimates of lung cancer risk per
WLM of radon exposure provides evidence of the role of radon as the primary carcinogen.
Further analysis of the potential effect of other exposures on lung cancer risk in the
underground miners will be undertaken in the previously described joint analysis of 11 miner cohorts
(Lubin 1991), which proposes to critically evaluate the variation of the exposure-response relationship
with age, time after exposure, cigarette smoking, and other mine exposures.
f. Exposure-Rate Effect
Exposure levels in homes are generally lower than those the miner cohorts experienced.
Laboratory studies indicate that the risk per unit dose from alpha irradiation, in general, and from
radon daughters in the lung, specifically, is maximal at low dose rates. The same trend has been
found in human studies (NAS 1988, Sevc et al. 1988, Howe et al. 1986). If risk decreases with
increasing exposure rate, then the estimate of the risk coefficient derived from miner studies may in
some cases be biased low relative to what is appropriate for use at the lower exposure rates in
residences. In particular, it often has been suggested that the high exposure rates prevailing in the
Colorado mines may be responsible for the comparatively low risk coefficient derived from the
Colorado Plateau cohort study. A quantitative estimate of possible bias in the BEIR IV analysis
associated with exposure rate might then be obtained by excluding the Colorado Plateau cohort from
the analysis. As discussed in the previous section, this exclusion would increase the combined
estimate of p by about a factor of 1.5, i.e., from 130%/WLM to 1.93%/WLM.
g. Extrapolation to Females
The BEIR IV Committee assumed that the same age-specific relative risks found for men
could be applied to women. However, U.S. male lung cancer rates are substantially higher than
female rates. As a result, the BEIR IV model projects a much higher risk of lung cancer attributable
to radon for men than for women. There is, however, no evidence that males are inherently more
susceptible to the disease; rather, most of the difference in baseline rates seems to be explained by
differences in smoking habits. It follows that a great deal of the uncertainly regarding dependence
of radon risk on gender revolves around the question of the radon-smoking interaction.
h. Relationship Between Radon Risk and Smoking
Issues surrounding the possible form of the radon-smoking interaction and its possible effect
on radon risk are complex and rather intractable given the limited information available. No attempt
is made here to quantify the uncertainty due to this source, but rather to outline what appear to be
the major problems.
Information on the interaction between smoking and ionizing radiation in causing lung cancer
is somewhat conflicting. Data on the atomic bomb survivors suggest additivity but are also consistent
with a supra-additive or even a multiplicative interaction (Prentice et al. 1983). According to BEIR
Chapter 2: Estimation of Risks Page 2-28
-------
V, despite lower smoking exposures, women have about the same excess risk of lung cancer
attributable to radiation as men; thus, women appear to have higher relative risks.
The most important data currently available on the interaction between radon and smoking
come from the Colorado Plateau miners, for which there is individual smoking information on all
miners (changes in smoking habits in the cohort subsequent to 1968 have not yet been reported for
the entire cohort however). Statistical analyses of this cohort yield results that are consistent with
a multiplicative but not an additive model; however, a wide range of supra-additive and even supra-
multiplicative models also give acceptable fits (NAS 1988). Other human data are rather limited,
including support for everything from a multiplicative to a protective effect of smoking on radon risk
(Lubin 1988, L'Abbe et al. 1991, Samet et al. 1991b, Sevc et al. 1988, Radford and St. Clair Renard,
1984). Animal studies, too, are conflicting: one study showing a synergism with tobacco smoke
(Chameaud et al. 1980); another, a protective effect (Cross et al. 1982).
The EPA model of radon risk is based on the BEIR IV assumption that radon and smoking
act multiplicatively in causing lung cancer. According to this model, the radon risk is proportional
to the baseline lung cancer rate, so that as the baseline rate varies over time (due largely to changes
in smoking habits), the projected rate of radon-induced lung cancers varies in parallel. Therefore,
neglecting possible changes in average exposure rates, it is, according to the model, the proportion
of all lung cancers attributable to radon that remains constant over time, not the absolute rate of
radon-induced lung cancers. Thus, for past years when lung cancer rates were much lower than they
have been recently, the model also projects much lower rates of radon-induced lung cancers.
An important element of uncertainty concerns the temporal dependence of the smoking-radon
interaction. Young children are not generally smokers, but some of them will smoke when they get
older. By employing a relative risk model in which the risk of lung cancer due to childhood exposure
increases with the baseline rate at the age at risk, EPA effectively assumes that the risks from
childhood radon exposure and adult smoking multiply. This may not be unreasonable since animal
data suggest that tobacco smoke exposure acts synergistically with prior radon exposure in causing
lung cancer (Chameaud et al. 1980), Moreover, such an assumption is consistent with the widely held
view that radiation is primarily an initiator of carcinogenesis while tobacco smoke acts both as an
early and late stage carcinogen (Doll 1978). Another issue concerns risks to former smokers, either
due to radon exposures received prior to cessation of smoking or afterward. Additional follow-up
on the U.S. miners may shed light on this question.
Finally, it should be noted that the risks calculated here are based on 1980 lung cancer
mortality rates. These rates are continually changing in response to changes in smoking patterns. This
limits the ability of the current risk estimates to predict future radcwvinduced lung cancers.
i. Overall Uncertainty in the Risk Factor
Summarized in Exhibit 2-6 are the upper and lower multiplicative uncertainties due to the
sources of error that could be estimated reasonably. With regard to the epidemiologic studies of
miner cohorts, there is additional uncertainty not addressed in this analysis associated with systematic
errors in the assignment of exposures to individual miners based on area radon and radon decay
product monitoring and, in some cases, highly questionable extrapolations over space and time. Also
not included is the possible effect of other mine exposures. In extrapolating the miner data to the
situation of radon in homes, additional uncertainty exists regarding the dependence of radon risk on
exposure rate, gender, and smoking. These also are not addressed quantitatively here.
Chapter 2: Estimation of Risks Page 2-29
-------
EXHIBIT 2-6
QUANTITATIVE ESTIMATES OF UNCERTAINTIES
Source of Uncertainty
Statistical Variability in Miner Data
Projection of Risk Over Time
Age Dependence of Risk
Extrapolation from Mines to Homes
Uj
1.5
1.5
1.6
1.4
IN RADON RISK^
(Li)"1
1.5
1.5
1.0
1.4
a/
Source: Calculated in text.
Following, in outline, the calculations presented in the National Institutes of Health report
on the development of radioepidemiological tables (NTH 1985), each source of uncertainty (i) is
characterized by a lognormal probability distribution where the geometric mean Gj = (UjxLj)1yi and
where the geometric standard deviation (GSD) Sj is given by Uj/Lj = Sj3'29. The latter equation is
equivalent to the assumption that the 90% "confidence interval" of the respective distribution is
(GjxLj, GjxUj). The combined uncertainty distribution due to all four sources is then lognormal with
a geometric mean G given by
In G = In Gj + In G2 + In G3 + In G4 (6)
and a GSD given by
In2 S = In2 S, + In2 S2 + In2 S3 + In2 S4 (7)
Substituting from Exhibit 2-6 then, G = 1.26(5). The 90% "confidence interval" encompassing all
four sources of uncertainty is then (GxL, GxU) where U2 = 1/L2 = S3'29. Solving, U = 1/L = 2.02.
The tour components of uncertainly were defined relative to a central estimate of risk based
on the BEIR IV model, adjusted to correct for an assumed 30% lower dose per WLM in homes
compared to mines. Upper and lower bound estimates that reflect the combined uncertainty are then
obtained by multiplying the central estimate by GxU = 2.56 and G/U = 0.625, respectively. Thus,
according to this analysis, the actual risk could be about 2.6 times higher than estimated by the
adjusted BEIR IV model or about 1.6 times lower.
The adjusted BEIR IV model, used in a lifetable calculation in conjunction with U.S. 1980
vital statistics, yields a risk factor of 224 lung cancer deaths per 10 person-WLM, for constant
lifetime exposure. Multiplying this value by G, the geometric mean risk estimate is 283 lung cancer
deaths per 10 person-WLM. The estimated uncertainty range is then found to be 140 to 570 lung
cancer deaths per 10 person-WLM.
Chapter 2: Estimation of Risks Page 2-30
-------
As discussed previously, this range does not include several potentially important sources of
uncertainty, such as errors in miner dosimetry and the effects of exposure rate, gender, and smoking,
which cannot be quantified at present. In general, these uncertainties could produce underestimates
or overestimates of the risk. Also, it should be emphasized that the uncertainties in risk to specific
subpopulations — including children, never smokers, and former smokers — are considerably higher.
2. Average Residential Radon Exposure and Associated Uncertainty
The average residential exposure is estimated to be:
E - C [F x 0.01WL/(pCi/L)][ Q x 51.6 WLM/WL-y] (8)
where C is the average radon concentration in homes, F is the average equilibrium factor, and ft is
the average occupancy factor. Based on a review of available information, EPA has determined that
the previously employed values of F = 0.5 and Q = 0.75 are still appropriate for estimating exposure
and risk, but the estimate of C is revised slightly downward from 1.29 to 1.25 pCi/L (a discussion of
the estimates for these parameters and their uncertainties is provided below). Substituting into the
equation above, the estimated average exposure is 0.242 WLM/y,
a. Estimate of Average Radon Concentration
EPA's National Residential Radon Survey (NRRS) of U.S. residences estimates that the
annual average radon concentration to which people are exposed in their homes is about 1.25 pCi/L.
This value reflects an average over all frequently occupied areas for each home, averaged over all
homes. The standard error in the measurement is 0.07, implying a 90% statistical confidence interval
of about 1.14 to 1.36 pCi/L.
The radon level in' each house, as determined by the NRRS, represents year-long
measurements taken at one or two locations on each floor defined as a frequently occupied area.
Frequently occupied areas are defined as any level of the home, including basements, which contain
a family room, living room, playroom, den, or bedroom. Also included as frequently occupied areas
are basements which may not contain one of the previously mentioned room types, but are occupied
by a resident for more than four hours per day, such as a frequently utilized workshop. The annual
average radon concentration estimate of 1.25 pCi/L does not reflect the time spent on floors not
classified as frequently occupied ares. Occupancy data, which reflect the total time each individual
spent in the house and on each floor of that house, were collected in the NRRS, but they primarily
reflect summertime activity patterns, which are likely to be atypical; consequently, they were not used
to adjust the estimate of the average indoor radon concentration. Had floor occupancy data been
used to weight the floor measurements, the estimate of the annual average radon concentration
would have been diminished only slightly — by about 6%.
In order to allow for these additional uncertainties discussed in the preceding paragraph, an
estimate is adopted of the overall uncertainty in the average radon concentration to which people
are exposed in their homes, a range of 1.1 to 1.4 pCi/L, which is slightly wider than the statistical 90%
confidence interval from the NRRS cited above.
Chapter 2: Estimation of Risks Page 2-31
-------
b.
Estimate of Average Equilibrium Factor (F)
The equilibrium factor (F) is defined as the ratio of the potential alpha energy concentration
in the existing mixture to that which would exist if all short-lived daughters were in equilibrium with
the radon present (NCRP 1988), Measured values of F are highly variable between houses (or even
within a single house), depending on ventilation rates, aerosol concentrations, and other factors. The
equilibrium factor rises with increasing aerosol concentration because the radon decay products attach
more rapidly, reducing plate-out onto walls and other surfaces. Consequently, F will be increased in
the presence of a smoker. From a risk standpoint, the situation is less clear, however, since
concomitant with the increase in F will be a decrease in the unattached fraction, fD; the changes in
F and L are predicted to have opposite effects on lung dose.
P'
Based primarily on measurements in 21 houses reported by George and Breslin (1980), EPA
has used a value of 0.5 for F. The NAS' radon dosimetry report suggests that the average value is
more likely to lie between 03 and 0.4, but that conclusion seems to rest on measurements carried out
in a relatively small number of houses — mostly European. Recently, two large surveys of houses in
the U.S. have been completed. The first was a survey of 200 houses conducted by the State of New
Jersey that gave a mean equilibrium factor of 0.45 (NJ Department of Environmental Protection
1989). The second, a multi-state survey of 113 homes conducted by Radonics, yielded an average
value of 0.51 (Radonics 1991). The data from the latter were found to be approximately lognormally
distributed; the geometric and arithmetic means of the distribution were 0,54 and 0.56, respectively.
Exhibit 2-7 summarizes various measurements of F. For the most part, they represent "grab
sample" measurements and/or measurements taken in a small number of houses. Only the Radonics
study (1991) seems to reflect measurements taken over several days in a large number of houses. In
MEASURED
Location
Austria
Finland
Norway
Sweden
nonsmoker
smoker
United States
cellars
living areas
West Germany
no smoke
smoke
VALUES OF
F
0.60
0,47
0.50
0.44
0,46
0.5 i
0,52
0.63
0.33
0.45
0.51
0.37
0.34
0.30
0.50
EXHIBIT 2-7
THE EQUILIBRIUM FRACTION, F, IN HOMES
Reference
Steinha usler ei ai. 1980
MakeJainen 1980
Stranden et a!. 1979
Swccljemark 1983
Jonassen and Jensen 1989
Jonasscn and Jensen 1989
George and Breslin 1980
George and Breslin 1980
Israeli 1985
NJ Dept. of Environ. Protection 1989
Radonics, Inc. 1991
Wieke and Porstendorfer 1982
Keller and Folkerfs 1984
Porstendorfer 1987
Porstendorfer 1987
Chapter 2; Estimation of Risks
Page 2-32
-------
light of the available information, F = 0.5 will continue to be employed as a nominal point estimate;
an uncertainty range of 0.35 to 0.55 will be adopted.
c. Estimate of Average Occupancy Factor (O)
The occupancy factor, which is the percent of time spent in the home, naturally varies with
lifestyle. For example, a study sponsored by the Electric Power Research Institute and conducted
by GEOMET found values of 61.6%, 91.5%, and 86.2% for fully employed persons, housewives, and
elderly individuals, respectively (GEOMET 1981). Exhibit, 2-8 summarizes estimates of occupancy
from the literature. For the most part, these estimates reflect analyses of detailed reports by-
individuals as to their activities over a fixed time period. However, it should be cautioned that several
of the U.S. estimates are based on identical or overlapping data sets.
Reference
Oakley
Moelter & Underbill
GEOMET
New Jersey
U.S. EPA/OHEA
U.S. EPA/ORP
Brown
Francis
Roy & Courtay
EXHIBIT 2-8
PERCENTAGE OF TIME SPENT
Date
1972
1976
1981
1989
1989
1991b
1983
1987
1990
AT HOME
Country
US
us
us
us
us
us
UK
UK
France
Avg. Occupancy
79.5
74.4
75.3
71
64-73
61.5
75
77
76.7
The estimates for 0 vary from about 60% to 80%. The lowest value is the National
Residential Radon Survey finding of 61.5%; this estimate is highly questionable, however, given the
lack of detail obtained in the data acquisition and the concentration on summertime activity patterns.
Based on the data in Exhibit 2-8, EPA will continue to employ an occupancy factor of 75%, as a
nominal point estimate for calculating exposure and risk, with an uncertainty range of 65% to 80%.
d. Overall Uncertainty in the Estimate of Exposure
The estimated annual radon progeny exposure is proportional to the radon concentration (C),
the equilibrium factor (F), and the occupancy factor (O). As discussed above, the average values for
these parameters and the associated uncertainty ranges (in parentheses) are estimated to be: C =
Chapter 2: Estimation of Risks Page 2-33
-------
1.25 pCi/L (1,1-1.4 pCi/L), F = 0.5 (0.35-0.55), and Q = 0.75 (0.65-0.8). The geometric means of the
uncertainty ranges are CG = 1.24, FG = 0.439, and QG = 0.721. From Equation 8 it follows that the
geometric mean of the uncertainty range for exposure is:
GE - (0.242 WLM/y) (1.24/1.25)(0.439/0.5)(0.721/0.75) (9)
Hence, GE = 0.203 WLM/y. Treating the uncertainty distribution for each factor above as lognorrnal
and the respective ranges as 90% confidence intervals, the corresponding 90% confidence interval
for the uncertainty in exposure will be (GxL, GxU), where
In2 U/L = In2(1.4/l.l) + In2(0.55/0.35) + In2(0.8/0.65) (10)
Solving this equation, U/L = 1.76, By symmetry of the lognorrnal distribution, U = 1/L = 1.32.
Thus, the annual exposure is estimated to be between 0.203/1.32 and 0.203 x 1.32 WLM/y; i.e., from
0.154 to 0.268 WLM/y.
Numerical Estimates of Lung Cancers Induced by Residential Radon Exposures
Based on an average annual exposure of 0.242 WLM, the estimated number of radon-induced
lung cancer deaths (led), per year, according to the modified BEIR IV model is:
N = (224 led per 106 person-WLM) (0.242 WLM)(25Gxl06 persons)
- 13,600
An uncertainty range for N can be estimated from the calculated uncertainty distributions for
the risk factor and for the annual exposure. The former was approximated by a lognorrnal
distribution with geometric mean GR = 283 led per 10 person-WLM and upper and lower 90%
confidence bounds equal to 2.02xGR and GR/2.02 (140 and 570). Similarly, for the latter, GE was
0.203 WLM/y and the corresponding uncertainty bounds were 1.32xGF and GE/1.32 (0.154 and 0.268
WLM/y). The upper and lower uncertainty bounds on N are then given by NL = 250 xlO6
(GRxGExL) and ND = 250xl06 (GRxGExU), where:
In2 U = In2 1/L = In2 2.02 + In2 1.32, or (11)
U = 1/L = 2.13. (12)
The uncertainty distribution for N is then lognorrnal with geometric mean GN = 25()xl06,
(GRxGE) = .14,400 Icd/y, and "90% confidence limits," GN/2.13 and 2.13xGN. Thus, it is projected
that for the general population, 6,740 to 30,600 Icd/y are attributable to residential radon exposures.
EPA has continued to update its estimates of radon risk as exposure estimates have been
refined and as new information has become available on the radon risk projection models and
comparative dosimetry in mines and homes. Exhibit 2-9 shows the evolution of radon risk estimates
over time, including the current estimate of approximately 7,000 to 30,000 lung cancer deaths per year
due to residential radon exposure.
Chapter 2: Estimation of Risks Page 2-34
-------
EXHIBIT 2-9
EPA'S RADON RISK ESTIMATES
Model/Approach Date of Range of Estimated Annual
Estimate Lung Cancer Deaths
EPA Model 1986 5,000-20,000
Average of ICRP 50 and BE1R IV 1988 8,000 - 43,000
BEIR IV Model (as adjusted by EPA) 1992 7,000 - 30,000
Section VI: The Effect of Smoking Status on Radon Risk
The revised Citizen's Guide includes charts on the risk of radon exposure for smokers and
never smokers. The charts present risks in terms of lung cancer deaths/1,000 persons and also in
comparison to the risk of death from different types of accidents. This section presents the approach
used to estimate the lifetime lung cancer risk/person from radon for smokers and never smokers.
EPA estimates of annual lung cancer deaths due to radon presented earlier in this chapter
are calculated for the general population, consisting of smokers and non-smokers. Since the risk of
lung cancer from radon exposure appears to be enhanced by cigarette smoking, it is of interest to
estimate the variation in radon risk by smoking category (current, former, and never smoker).
The assumption underlying the estimation of radon risk by smoking category is that radon risk
varies in proportion to smoking risk. Therefore, for example, if lung cancer risk is more than twice
as great in a current smoker compared to a person in the general population, then the radon risk for
a current smoker also would be more than twice as great as that of a person in the general
population. This assumption is consislenl with the BEIR IV Committee conclusion thai radon and
smoking act multiplicatively in causing lung cancer (NAS 1988). For further discussion of the
relationship between radon and smoking in causing lung cancer, see Section V of this chapter.
I. Smoking Risk Data
The data source for the prevalence and relative risks of smoking was the Surgeon General's
Report, Reducing the Health Consequences of Smoking (DHHS 1989), which included the Cigarette
Smoking Supplement to the 1985 National Health Interview Survey (NHIS 1985), and the American
Cancer Society's Cancer Prevention Study II (American Cancer Society 1988). The National Health
Interview Survey provided information on the prevalence of smoking in the U.S., and the American
Cancer Society's CPSII study provided the relative risks for current and former smokers (see Exhibit
2-10).
To date, only the female data for CPS-II have been analyzed in detail; hence, estimates of
risk for specific smoking levels (packs per day) or time since cessation cannot be made for males or
the general population. For this reason, this analysis considers only the broad categories of current
smokers and former smokers, with no further breakdown of smoking by subgroup.
Chapter 2: Estimation of Risks Page 2-35
-------
It should be cautioned that the relative risks by smoking category derived in the CPS-II study
may not accurately reflect the effect of smoking on lung cancer risk in the U.S. population. Subjects
in CPS-II are more representative of middle-class white Americans than of the U.S. population as
a whole. As a result, exposures to occupational and environmental carcinogens in the CPS-II group
are likely to be lower than average. So, for example, non-smokers in the CPS-II group may have a
lower incidence of lung cancer than non-smokers in general. In the case of current or former
smokers, the study group may differ from the general population with regard to their smoking
patterns, as well as other relevant factors.
2. Average Relative Risk
Exhibit 2-10 summarizes data from the Surgeon General's Report on smoking prevalence
(proportion of the population in each smoking category) and relative risks of lung cancer by sex and
by smoking category (current, former, and never smokers). These data were used to calculate a
separate average relative risk of lung cancer for all males and for all females (see "All" under each
gender in Exhibit 2-10). The relative risk of smoking is usually calculated as the ratio of the lung
cancer mortality rate of a smoker group to that of a never smoker group. The average relative risk
calculated here differs in that it is given by the ratio of the lung cancer mortality rate in the total
PREVALENCE AND
Category
Males
Current smoker
Former smoker
Never smoked regularly
All (average)^
Females
Current smoker
Former smoker
Never smoked regularly
All (average)-^
EXHIBIT 2-10
LUNG CANCER RISKS
Prevalence-^
0327
0.291
0.382
1.0
0.275
0.171
0.554
1.0
OF CIGARETTE SMOKING
Relative
Risfcfe/
22.36
9.36
1.0
10.42S/
11.92
4.69
1.0
4.64S/
Lung Cancer
Deaths/105 persons^
14,118
5,909
631.4
6,579
5,593
2,200
469.2
2,177
T'
-' Proportion of population in each smoking category; data from Surgeon General's Report (DHHS 1989).
^ Defined as the ratio of the age-adjusted lung cancer mortality rate in the current and former smoker category to
(divided by) the age-adjusted lung cancer mortality rate in the never smoker category; data from Surgeon General's
Report (DHHS 1989).'
- Calculated by EPA, as described in text.
- Includes current, former, and never smokers.
Chapter 2: Estimation of Risks Page 2-36
-------
male or female population (including never smokers) to the lung cancer mortality rate in the never
smokers for each respective gender. This approach to the calculation of relative risk is based on the
methodology employed in the National Institutes of Health report on the development of
radioepidemiological tables (NIH 1985).
Thus the average relative risk for all males is:
RR = Ii Pi RRj (13)
where p( and RRj are the prevalences and relative risks for all smoking categories including those
persons who never smoked regularly. Thus, the average relative risk for all males is
RR = 0.327x22.36 + 0.291x9.36 + 0.382x1.0 = 10.42
relative to males who never smoked regularly. Similarly, for all females, the average relative risk is
RR = 0.275x11.94 + 0.171x4.69 + 0.554x1.0 = 4.64
relative to females who never smoked regularly.
3, Lung Cancer Mortality Calculations
The average relative risks calculated in Section 2 were used in conjunction with 1980 vital
statistics and decennial life tables (NCHS 1983, NCHS 1985) to calculate lung cancer deaths in male
and female birth cohorts (all smoking categories combined) of 100,000. The number of lung cancer
deaths yielded by these calculations was 6,579 and 2,177 per 100,000 males and per 100,000 females,
respectively (see Exhibit 2-10).
The number of lung cancer deaths in cohorts of 100,000 of each smoking category of males
and of females was then calculated. The lung cancer death rate for each category was assumed to
be proportional to the relative risk for that smoking category of males or females. For example, the
lung cancer death rate in male current smokers (M,CS) was given by:
lcd(M,CS) = (22.36/10.42) (6579 lcd/100,000)
= 14,118 lcd/100,000.
The estimated lung cancer deaths/100,000 for all smoking categories and both sexes are listed in
Exhibit 2-10.
For the general population (defined as males and females combined), the number of lung
cancer deaths was subsequently estimated by adjusting the number of lung cancer deaths calculated
for males and for females according to the sex distribution of the birth cohort (1.051 male: 1 female),
Thus, the number of lung cancer deaths for the general population would be (1.051x6579 +
2177)/2.051 = 4433 lung cancer deaths (see Exhibit 2-11). To calculate the number of lung cancer
deaths in each smoking category of the general population, the male and female lung cancer rates
Chapter 2: Estimation of Risks Page 2-3 7
-------
were combined. As an example, consider the case of current smokers. The fraction of the current
smokers birth cohort that is male was estimated to be;
f(M,CS) = 1.051 p(M,CS)
' ' (14)
1.051 p(M,CS) + p(F,CS)
where p(M,CS) and p(F,CS) are the prevalences of current smokers among males and females,
respectively. From Exhibit 2-10, p(M,CS) = 0.327 and p(F,CS) = 0.275. Solving, f(CS,M) = 0.5555.
The fraction of current smokers that is female is then given by:
f(CS,F) = l-f(CS,M) = 0.4445 (15)
In a birth cohort of 100,000 (male and female) current smokers, therefore, the number of lung cancer
deaths would be:
lcd(CS) = (0.5555)(14,118)-K0.4445)(5593) = 10,329
Lung cancer deaths in birth cohorts of 100,000 (male and female) former smokers and 100,000 (male
and female) never smokers were calculated following the same approach, and are presented in Exhibit
2-11.
EXHIBIT 2-11
LUNG CANCER RISK OF CIGARETTE SMOKING
IN THE GENERAL POPULATION^
Category Lung Cancer Deaths/10" persons
Current Smoker 10,329
Former Smoker 4,579
Never Smoked Regularly 537.4
All (Average) 4,433
i
^ Combined male and female data from Exhibit 2-10. Sex ratio at birth, M:F = 1.051:1.
The lung cancer death rates in each smoking category in the general population were
compared to the average lung cancer death rate for the general population to obtain the radon risk
multipliers presented in Exhibit 2-12. For example, the number of lung cancer deaths/100,000 current
smokers in the general population is 10,329. This is a factor of 2.33 times the 4,433 lung cancer
deaths expected in the general population, averaging over all smoking categories. It follows from the
presumed multiplicative interaction between radon and smoking that the radon risk among current
Chapter 2: Estimation of Risks Page 2-38
-------
smokers also would be about 2.33 times the radon risk for the general population. Risk multipliers
were derived for former and never smokers as well, using this approach. As shown in Exhihit 2-12,
the lung cancer risk (and thus the radon risk) for never and former smokers in the general population
are estimated, respectively, to be 0.121 and 1.03 times that of the general population.
EXHIBIT 2-12
RISK MULTIPLIERS BY SMOKING CATEGORY
IN THE GENERAL POPULATION^
Category Risk Multiplier
Current Smoker : 2.33
Former Smoker 1.03
Never Smoked Regularly 0.121
All (Average) 1.00
§/
Calculated in text.
4. Lifetime Radon Risk by Smoking Category
The above risk multipliers were used in conjunction with a standard lifetable analysis, 1980
decennial vital statistics and the EPA-adjusted BEIR IV relative risk model (K = 0.7) to estimate the
lung cancer risk for never smokers and for current smokers at several radon levels. As before, in
using the BEIR IV model, the lung cancer baseline risk was adjusted for an annual background radon
exposure of 0.242 WLM.
For a more rigorous calculation of the excess risk due to radon, the baseline lung cancer risk
for current smokers and for never smokers would be multiplied by the risk multipliers 2.33 and 0.121,
respectively. This would mean constructing new litetables for both groups. However, there are a
number of difficulties with this approach. For example, age-specific death rates are generally not
available for smokers and never smokers separately. A simpler approach is to multiply the excess
lung cancer risks (not the baseline risks) due to radon for never smokers and for current smokers, by
their respective risk multipliers. For example, in the case of current smokers, the excess risk R(a)
is estimated as:
R(a) = 2.33 x [r(a) - r0(a)J (16)
where r(a) is the age dependent total lung cancer risk and r0(a) is the age dependent baseline lung
cancer risk. Simplifying the calculation gives risk estimates that are within 10% of the values
calculated using the same approach described in the BEIR IV report (NAS 1988), without requiring
changes to the lifetables.
Chapter 2: Estimation of Risks Page 2-39
-------
Exhibit 2-13 shows the risks tor never smokers and current smokers along with the risks for
the general population, for selected exposure levels. Levels are expressed in both pCi/L and WLM/y.
The relationship between the two is as follows:
WLM/y = [F x 0.01 WL/(pCi/L)] x [Q x 51.6 WLM/WL-y] pCi/L (17)
where:
F, the equilibrium factor, is assumed to be 0.5, and
Q, the fraction of time spent indoors, is assumed to be 0.75.
For further discussion of the average equilibrium factor and average occupancy factor, see Section
V.
EXHIBIT 2-13
LIFETIME LUNG CANCER RISK FOR NEVER SMOKERS,
CURRENT SMOKERS, AND THE GENERAL POPULATION
Radon
pCi/L
20
10
8
4
2
1.25
0.40
Level
WLM/y
3.87
1.94
1.55
0.774
0387
0.242
0.077
Lifetime
Never Smokers
7.72xlO'3
3.87xlO-3
s.ioxitr3
1.55xlO'3
7.76X104
4.85x1 0"4
l.SSxlO'4
Lung Cancer Risk
Current Smokers
1.35xl04
7.11xlO'2
5.74x1 0'2
2.93x10-2
L48xlO-2
9.29xlO-3
2.98xlO"3
(per person)
General Population
6.15xlQ-2
3.14xlQ-2
2.52xlO'2
1.27xlO"2
6.39xlO'3
4.00xlO~3
1.28xlO'3
5, Discussion
The approach used here for calculating the risks for various smoking categories is
approximate. It does not account for life-shortening due to smoking, which reduces the relative
contribution of smoking-attributable lung cancer deaths. An analysis of the effect of smoking
presented in BEIR IV only partly corrected for life-shortening because it neglected all smoking-
related causes of death other than lung cancer. EPA calculations show that use of the BEIR IV
methodology would produce a difference of less than 10% in the risk multipliers presented here. In
light of the larger potential sources of error associated with the calculation, these differences would
appear to be insignificant. The chief sources of error and uncertainty associated with the analysis of
the effects of smoking on radon risk are the uncertainty over Ihe actual form of the interaction (e.g.,
multiplicative or submultiplicative); the variations in the relative risk of smoking by age and gender;
Chapter 2: Estimation of Risks Page 2-40
-------
the changes in age-specific relative risks for smokers as smoking habits and types of cigarettes change
over time; and the influence of environmental and passive cigarette smoke on radon risk.
Section VII: Conclusions
For most pollutants, the assessment of human risk is based on data from animals exposed to
very high levels of a given substance. The large and obvious uncertainty associated with this
extrapolation is usually compounded by the fact that the substance may be administered to laboratory
animals in a way that is different from human exposure (e.g., oral versus inhalation). In addition, the
animals are often exposed to levels of a substance several orders of magnitude greater than typical
human exposures.
Radon is a known human carcinogen. Estimates of lung cancer risk for the general
population from inhalation of radon and radon progeny are derived using human data on the
occurrence of lung cancer in underground miners due to inhalation exposures. Furthermore, the
average lifetime cumulative radon exposures to the general population are only slightly below those
for. which increased risk can be demonstrated in underground miners. Hence, low-exposure
extrapolation is less of an issue for radon risk assessment than it is for the assessment of risk for most
other environmental carcinogens.
There is extensive evidence supporting the classification of radon as a known human
carcinogen. There is some uncertainty associated with the projection of lung cancer risk from
occupational radon exposures to the general population for residential exposures. Although the
epidemiologic data on increased risk from indoor exposures are limited, some of the studies suggest
that the relative risk coefficients for the general population and for underground miners are
comparable.
Uncertainty is always present to some extent in risk assessment for environmental pollutants.
However, as more data become available for radon, the uncertainties associated with estimation of
lung cancer risk may diminish. Follow-up of the underground miners will continue to clarify the
relationship of radon and smoking and the nature of the exposure-response relationship for radon
exposure and lung cancer risk over a lifetime. The results of epidemiologic studies of indoor radon
exposure may help to further define the relationship at low exposure levels. Advances in dosimetric
modeling and the findings of the NAS' report on comparative radon dosimetry (NAS 1991) have
improved the extrapolation of risks from miners to the public. As new information becomes available,
EPA will continue to update its estimates of luog cancer risk from residential radon exposure.
Chapter 2: Estimation of Risks Page 2-41
-------
-------
CHAPTER 3
ANALYSIS OF EPA RADON TESTING OPTIONS
EPA recommends that individuals living in the following types of housing test their homes for
radon: single family homes, apartment units within multi-unit structures (if they are below the third
floor), mobile homes with permanent foundations, and units in-group quarters below the third floor
(e.g., college dormitories). Among other changes, the proposed revision of the Citizen's Guide
recommends a new procedure for testing home radon levels. EPA narrowed the options down to six
and chose a final testing procedure from among
those six approaches. This chapter describes •••••••••••••^••^^•"••^•"•^••••^
the six options and explains the process the TttE AIMS OF THIS CHAPTER
Agency used to analyze all of them before '
recommending one for the public to use. ^ To describe the options for radon
testing procedures EPA considered
This chapter is organized into two major recommending to the public.
parts. The first provides background
information, reviews the major considerations £) To expia^ how EPA analyzed the
important to developing testing procedures, and effectiveness of these options in
summarizes the results of an analysis of the correctly categorizing homes above
effectiveness of the testing options (Sections I, or below the threshold requiring
II, and III). The second part of the chapter radon mitigation.
presents the details of the analysis and closes
with general conclusions drawn from the ••••••••••^^••••••••••••••••••••••B
analytic results (Sections IV, V, and VI). This
second part describes the analytic process, but does not replicate the analysis. Therefore, EPA has
cited rather than included technical background information. Finally, Appendix C supports this
chapter by providing basic background information on commonly used measurement devices.
Section I: Background
The original Citizen's Guide recommended that individuals perform (1) a short-term
(screening) measurement in the lowest livable level of the home under "closed-house" conditions
during the winter months, and (2) confirmatory long-term tests in the Hved-in levels of the home if
the result of the short-term test was above 4 picocuries per liter (pCi/L). The recommended duration
of the confirmatory tests depended on the short-term test results and varied from less than one week
to a full year. The short-term test was designed to obtain the most reproducible estimate of the
highest (worst-case) radon concentration likely to be found in that particular house at any time. The
long-term measurement was intended to be an estimator of the annual average concentration in Hved-
in levels of the home, being made over an entire year under normal ventilation conditions. An
exception to this process was when the short-term result was so elevated as to necessitate a more
rapid confirmation. Descriptions of short-term and long-term tests are given below.
originally considered more than six options. However, for reasons discussed in Section II (Appropriate Testing
Location), EPA focused this analysis on the six options that recommend testing on the lowest Hved-in level.
Chapter 3: Analysis of EPA Radon Testing Options Page 3-1
-------
Short-term measurements, which last from 2 to 90 days, aim to give a quick assessment of
the level of radon in a home. Short-term screening measurements, which EPA recommended be
taken in the winter months in the "lowest livable" level of homes (the bottom-most inhabitable
space) under "closed-house" conditions, tend to produce conservative (i.e., high) estimates of annual
average radon levels. Short-term tests also can be conducted in the lowest lived-in level of homes
(a room that is used regularly) to produce estimates that more closely represent the annual average
radon concentration.
Long-term measurements are usually taken over the course of an entire year under normal
ventilation conditions. Because long-term measurements take into account seasonal fluctuations in
radon levels, and because they are performed under the ventilation conditions normally encountered
by homeowners, they tend to be better indicators of the actual radon concentrations to which
individuals in a home are exposed.
While long-term measurements tend to be better indicators of the average radon
concentration in a house over the course of a year, short-term measurements are simpler to make
and provide faster results. In cases where radon concentrations are extremely high, fast results are
desirable in order for mitigation to begin quickly.
An individual's risk of getting lung cancer from radon is directly proportional to his or her
integrated exposure, and the annual average concentration in lived-in levels of the home is the most
readily measurable quantity that approximates that exposure. However, it is not necessary to know
the actual risk in order to act to reduce it; a homeowner only needs to know that the radon
concentration is above some action level. Since the public seemed willing to take action based only
on short-term tests (see Chapter 6), EPA needed to adopt a recommended testing protocol for the
revised Guide that would allow homeowners to make the best possible mitigation decision based on
short-term test results.
In revising the Citizen's Guide, EPA has focused on how well options meet the basic purpose
of testing - to enable homeowners to make the right decision on whether to mitigate radon levels.
When analyzing those options utilizing short-term measurements, the primary concern was not on
how well short-term measurements could estimate long-term averages. Rather, the central issue was
how well a testing option provides a reasonably accurate basis tor determining whether there is a
need to reduce radon levels in a home. Errors could result when (1) a test indicated that mitigation
should occur when actual radon levels were below the action level (false positive), or (2) a test
indicated that mitigation should not occur when actual radon levels were above the action level (false
negative). Combining the percentages of both types of errors provides an estimate of the rate to
which each option might misclassify homes for mitigation. As discussed in the next section, the
misclassification rate for various testing schemes was one of five major factors that was important to
EPA in evaluating testing options.
Section II: Major Considerations in Developing Testing Procedures
EPA's goal is to recommend a radon testing procedure in homes that will lead the public to
decisions about radon reduction that are timely, simple, cost-effective, and protective of human
health. Before developing possible testing procedures, EPA identified five major factors that were
important to consider in developing and evaluating options. These considerations are listed in
summary form in Exhibit 3-1. The first factor is the action level used to determine whether a home
Chapter 3: Analysis of EPA Radon Testing Options Page 3-2
-------
EXHIBIT 3-1
CONSIDERATIONS IN DEVELOPING RADON TESTING OPTIONS
• Appropriate Action Level
• Appropriate Testing Location
« Appropriate Ventilation Conditions
• Acceptability to Potential Users
• Effectiveness of Short-Term and Long-Term Measurements
requires mitigation. The second concern is the location in the home where testing occurs. The third
factor is the ventilation condition (i.e., "open-house condition" versus "closed-house condition")
under which tests are performed. The fourth concern is the acceptability of the procedure to
potential users, and the final factor is the effectiveness of relying on short-term measurements. Each
of these factors is explained in more detail in the text that follows.
Appropriate Action Level — EPA is recommending an action level of 4 pCi/L for several
reasons. First, the Office of Research and Development's (ORD's) research on mitigation
effectiveness and the Office of Radiation Programs' mitigator survey suggest that elevated levels of
radon can be reduced to 4 pCi/L more than 95 percent of the time. Results from the mitigator
survey indicate that 2 pCi/L can be achieved about 70 percent of the time, while the ORD research
suggests this estimate may be even higher (U.S. EPA/ORD 1989; U.S. EPA/Radon Division 1990a).
Reducing the action level to 2 pCi/L, therefore, could result in perhaps as many as 30 percent of
homes with elevated levels being unable to achieve the action level. Second, lower action levels
introduce more uncertainty in the measurement results. Measurement device error increases to
approximately 50 percent at 2 pCi/L. This device error in conjunction with the larger fraction of
homes (of total homes testing) that have radon levels around 2 pCi/L would result in a threefold
increase in false negatives and a twofold increase in false positives over those expected-at a 4 pCi/L
action level.
Appropriate Testing Location — In the 1986 Citizen's Guide, EPA recommended that, for
screening purposes, a short-term measurement be taken on the "lowest livable" level, i.e., the lowest
area of a home that is used or has the potential to be used as a living space during the winter
months. This definition would include an unfinished basement that is in a condition such that it could
be converted to a bedroom, playroom, den, etc. For a confirmatory test (if the short-term test result
was above 4 pCi/L), the 1986 Guide recommended a follow-up test in the lived-in levels of the home.
Lived-in level is defined as a level of a house that is used regularly, such as a living room, playroom,
den, or bedroom, but not a kitchen or bathroom, where high humidity conditions or the operation
Chapter 3: Analysis of EPA Radon Testing Options Page 3-3
-------
of an exhaust fan could affect the validity of the test. EPA considered both "lowest livable" and
"lowest lived-in" testing locations in formulating the testing protocol in the revised Guide.
The choice between "lowest livable" and "lowest lived-in" level affects where homes with
basements will place radon detectors. Roughly 50 percent of homes nationwide have basements; only
half of these homes, or almost 25 percent of the national total, have basements that are used as a
lived-in level (U.S. EPA/Office of Radiation Programs 1991a). Accordingly, a recommendation that
devices be placed in the "lowest livable" level would result in 50 percent of the devices being placed
in the basement. A recommendation that devices be placed in the "lowest lived-in" level would
result in only 25 percent of devices being placed in the basement, and the remaining 75 percent being
placed on the first floor. A difference in the testing location therefore would affect 50% of homes
with basements that are not presently lived-in; this is 25% of all U.S. homes.
Keeping the short-term test in the lowest livable area as compared to moving it to the lowest
lived-in area would reduce false negative results by a factor of 1.5; however, the number of false
positive results would increase by a factor of 2. Measurements taken in the lowest lived-in area strike
a more equitable balance between false positives (public money spent with more limited benefits in
many cases) and false negatives (public health protection) than do livable area measurements.
Additionally, recent research (Harley et al. 1991) suggests that basement short-term measurements
overstate personal exposure by a factor of 3 to 5, while first floor short-term measurements are, on
average, only 30 percent greater than occupant exposure. Since the goal of radon measurement on
which a mitigation decision relics is to assess occupant exposure (U.S. EPA 1991a), lived-in level
measurements are better predictors than livable level measurements. Based on this finding and the
recommendation of EPA's Science Advisory Board (SAB) (U.S. EPA 1992b), short- or long-term
measurements that are to be used for the purpose of making mitigation decisions should be made on
the lowest lived-in level since it more closely approximates that concentration to which inhabitants
are exposed. Therefore, only lived-in level testing is considered in the various testing options
discussed below.
Appropriate Ventilation Conditions — EPA examined whether tests should be made under
"open-house" or "closed-house" conditions. In open-house conditions, tests are made with windows
and other ventilating passageways either closed or open as they would be when a test is not being
conducted. In closed-house conditions, all windows and other ventilated passageways are closed, to
the extent possible. Overall, EPA stresses the value of closed-house conditions in the final revised
Guide. Specifically, the final Guide recommends keeping windows and outside doors closed "as much
as possible" during short-term testing. For short-term tests lasting just 2 or 3 days, the final Guide
recommends closing windows and outside doors at least 12 hours before beginning the test.
Acceptability to Potential Users — EPA realized that, in real terms, the testing procedure
recommended in the original Citizen's Guide was technically sound, but was not viable because many
individuals dropped out before obtaining long-term measurements of the radon level in their homes.
Current Radon Program experience indicates that long-term tests are a deterrent to action and that
few people take long-term follow-up tests (see Chapter 6). In fact, the majority of people who
mitigate their homes are currently doing so based on a single short-term measurement.
Effectiveness of Short-Term and Long-Term Tests — Long-term tests are more indicative of
annual exposure than are short-term tests. Although a few of the testing procedures incorporate the
use of long-term tests, experience indicates that people are unwilling to take such tests. (Please refer
to Chapter 6 for more details on the public's unwillingness to use long-term tests.) Because short-
Chapter 3: Analysis of EPA Radon Testing Options Page 3-4
-------
term tests are completed in a short period of time, they are much more appealing to the public. A
procedure that does not incorporate long-term testing may be more effective in actual risk reduction.
Therefore, EPA assessed the feasibility of testing procedures utilizing short-term tests. Before
recommending a testing procedure not based on long-term testing, EPA sought to examine the
implications of relying OE short-term tests as indicators of whether or not mitigation is necessary.
From knowledge of the relationship between short-term measurements and annual averages
gained from EPA/State Residential Radon Surveys (SRRS), the Agency believed that although short-
term measurements are imperfect indicators of annual averages, they could potentially serve
effectively in the intended use EPA was considering. EPA reasoned that people measuring radon
levels in their homes do not require an accurate measurement of the annual average; instead,
homeowners require only a reasonably accurate basis for determining whether they need to mitigate.
Testing options need to have low misclassification rates for mitigation decisions to be acceptable.
EPA recognized that the distribution of radon levels in homes that is believed to exist would work
in the Agency's favor in using short-term tests. Exhibit 3-2 shows the distribution of annual average
radon concentrations suggested by EPA's National Residential Radon Survey (NRRS) for homes that
should be tested according to EPA's recommendations. Over 82 percent of the homes have radon
levels below 2 pCi/L, Another 7 percent of the homes have levels between 2 and 3 pCi/L. (These
estimates are based on lived-in level averages.) Therefore, errors in the use of short-term testing for
nearly 90 percent of the homes would have to be very substantial for these homes to be misclassified
as above the action level. The largest chances for problems exist in the range of 3 to 5 pCi/L, where
there are roughly 6 percent of all houses. The skewed distribution of radon levels towards the lower
end of the range could serve as a strong moderating force to any potential misclassification errors.
EPA was able to assess how short-term measurements could be used to make radon mitigation
decisions, as it had carefully tested large numbers of homes and compiled data to enable such an
evaluation. Beginning in 1987, EPA and States had cooperatively conducted 40 state-wide surveys
of radon levels in homes. For these surveys, which were designed to eliminate bias using standard
statistical techniques, EPA collected short-term radon measurements from nearly 60,000 homes and
paired short-term/long-term measurements from over 1,000 homes. EPA used testing protocols and
maintained a strong quality assurance and control program to ensure that the radon testing data
collected could be relied on in EPA evaluations of the significance of the radon problem and in
determining recommendations for public action.
As explained in more detail later in this chapter, EPA used both the SRRS and NRRS results
in a variety of ways in the analysis of radon testing options. For example, the Agency used the SRRS
results to analyze the relationship between short- and long-term measurements. The Agency used
the NRRS results to determine the distribution of annual average radon concentrations in homes
across the nation.
EPA initially considered the implications of a single, short-term test that people would
conduct in the lowest lived-in level of their homes. This simple approach did not offer enough
confidence that homes would be correctly classified in relation to the action level. Therefore, EPA
concentrated on analyzing options that relied on multiple tests. One of the options was very similar
to the short-term/long-term testing approach in the original Citizen's Guide, except the test would
occur in the lowest lived-in level (as opposed to the lowest livable level). This option is a reliable
approach that can be compared to options using short-term confirmatory tests. EPA evaluated five
other options that relied on multiple tests in a shorter time interval than the first option. All six
options were analyzed to estimate their rate of misclassification of homes for mitigation.
Chapter 3: Analysis of EPA Radon Testing Options Page 3-5
-------
EXHIBIT 3-2
DISTRIBUTION OF RADON LEVELS IN U.S. HOMES FOR WHICH EPA RECOMMENDS RADON TESTING
70%
60%
I 30%
s,
10%
0%
62.0%
20.7%
"1.1W
1.9
3.7%
0.7%
0-1 1-2 2-3 3-4 4-5
Annual Average Radon Concentration (pCi/L)
— Average over All Living Levels -
Source: U.S. EPA/Office of Radiation Programs I991a.
5-10
Section III: Summary of the MisclassiOcation Analysis Results for EPA's Testing Options
This section describes the sk testing options that the Agency considered in developing a
testing protocol for the revised Citizen's Guide. It then describes the criteria used to judge the
effectiveness of each option, and presents the results of the Agency's analysis in summary form. The
options considered are diagrammed in Exhibit 3-3. For this document, EPA labeled the sk options
in the order that makes their presentation most comprehensible. The Agency did not necessarily
create or evaluate options in the order in which they are described in this chapter.
When describing testing alternatives, the Agency made a distinction between action level and
trigger level. The action level was the level above which EPA recommended that a home be
mitigated. It also could be the same point where additional testing should occur. In contrast, the
trigger level was the level above which the Agency recommended in Option D that an individual
conduct further long-term or short-term testing to determine if a home was above the action level.
The action level was fixed at 4 pCi/L. The trigger levels in Option D were fixed at 2 pCi/L
(additional long-term testing required) and 10 pCi/L (additional short-term testing required).
Chapter 3: Analysis of EPA Radon Testing Options
Page 3-6
-------
EXHIBIT 3-3
ILLUSTRATION OF TESTING OPTIONS CONSIDERED
Initial Short-Term Test
f
Below 4pCi/L:
No Further Action
Required
LT ©©ouffBrm
LT Is
Initial Short-Term Test
f
Below 2pCi/L:
No Further Action
Required
No Action If
Below 4 pCi/L
Iyr Be
Initial Short-Term Test
I
f
Below 4pCi/L:
No Further Action
Required
f
$
f
STs I®
Initial Short-Term Tests
On All Living Levels in Home
f
Average of All STs
Below 4pCi/L:
No Further Action
Required
initial Short-Term Test
I
f
Below 4pCi/L:
No Further Action LT
Required
LTsi
LTh
IT;
initial Short-Term Test
*
Follow-Up Short-Term Test
I
f I
Either ST
Below 4pCi/L:
No Further Action
Required
ST = Short-Term Test LT = Long-Term Test
-------
• Option A — Individuals would • make an initial short-term measurement in the
lowest lived-in level. If the measurement exceeded the action level set
by EPA, the individual would be advised to make a long-term (i.e.,
year-long) measurement. If this measurement exceeded the action
level set by EPA, the Agency would recommend mitigation. [Note:
The original Citizen's Guide recommended a testing procedure similar
to Option A, except that short-term tests were conducted on the
lowest livable level, and the recommended duration of the
confirmatory test was not necessarily one full year.]
• Option B — Individuals would make an initial short-term measurement in the
lowest lived-in level. If the measurement exceeded the action level,
the individual would be advised to make a confirmatory short-term
measurement. If the average of the two short-term measurements
exceeded the action level set by EPA, the Agency would recommend
mitigation. [Note: EPA proposed this testing procedure in the
revised draft of the Citizen's Guide, published in 1990.]
• Option C — Individuals would make an initial short-term measurement in the
lowest lived-in level. If the result exceeded the action level, the
individual would make either a confirmatory long-term or short-term
measurement. This essentially gives individuals the choice of Options
A and B. If a confirmatory long-term measurement were made, the
results would be compared to the EPA action level, as in Option A.
If a confirmatory short-term measurement were made, the average of
the two short-term tests would be compared to the EPA action level,
as in Option B. If either method resulted in a value which exceeded
the action level, the Agency would recommend mitigation.
• Option D — Individuals would make an initial short-term measurement in the
lowest lived-in level of their home. If the measurement exceeded a
trigger level of 10 pCi/L, the individual would be advised to make a
short-term test to confirm that the home required mitigation. If the
measurement was below 10 pCi/L but above a second trigger level of
2 pCi/L, the individual would be asked to make a confirmatory long-
term measurement. If the average of the two short-term tests (for
homes above 10 pCi/L) or the result of the long-term measurement
exceeded the action level, the Agency would recommend mitigation.
• Option E — Individuals would make simultaneous short-term measurements on all
lived-in levels of their home, one measurement device per level. For
single level homes, the Agency would recommend that individuals
perform two tests in different rooms. If the average of all test results
indicated that the radon level in the home exceeded the action level,
the Agency would recommend mitigation.
« Option F — Like Option B, this option requires that individuals make an initial
short-term measurement in the lowest lived-in level. If the
measurement exceeded the action level, the individual would be
Chapter 3: Analysis of EPA Radon Testing Options P®ge 3-8
-------
advised to make a follow-up short-term measurement. If the
confirmatory measurement also exceeded the action level, the Agency
would recommend mitigation.
EPA used the following six quantitative measures to evaluate the simplified testing
procedures. Each measure was expressed as a percentage of the total homes tested, except the fifth
and sixth measures, which are explained further below.
Correct classification percentage — the percentage of homes tested for which the testing procedure
came to the correct conclusion (i.e., that homes with high radon levels should be mitigated
and all others left alone).
Error rate — homes for which the testing procedure came to the incorrect conclusion (i.e., either
false negative or false positive). This is the misclassification rate.
False negatives — homes tested in which the procedure incorrectly concluded that mitigation was not
required. False negatives can be presented as a percent of all homes or as a percent of actual
positives (i.e., homes that actually have radon levels greater than 4 pCi/L regardless of test
results).
False positives — homes tested in which the procedure incorrectly concluded that mitigation was
required. False positives can be presented as a percent of all homes or as a percent of all
positive tests.
False positives with quantifiable benefits — homes with false positive test results with actual annual
radon levels between 2 and 4 pCi/L (i.e., homes that have low starting radon concentrations,
but will still receive significant risk reduction through mitigation). False positives with
quantifiable benefits are presented as a percent of all positive tests.
False positives with non-quantifiable benefits — homes with false positive test results with actual
annual radon levels below 2 pCi/L. These homes also will also receive risk reduction through
mitigation, though the risk reduction will be less than that in homes with actual annual radon
levels between 2 and 4 pCi/L. Because the cost-effectiveness analysis presented in Chapter
5 assumes that mitigation only lowers homes to an average of 2 pCi/L, the actual benefits of
fixing homes with initial levels below 2 pCi/L is left unquantified. False positives with these
"non-quantifiable" benefits are presented as a percent of all positive tests.
Exhibit 3-4 schematically shows how misclassification rates depend on the annual average
lived-in area radon concentration, the result of the testing procedure, and the action level. Among
other items, the exhibit shows that the inaccuracies in testing procedures matter only when they
produce results on the opposite side of the action level from the correct annual average (i.e.,
misclassifications). The exhibit also shows that the correct classification percentage and the error rate
are complementary, and that the false negative percentage and the false positive percentage add up
to the error rate. In the remainder of this chapter, EPA refers to the error rate as the
misclassification rate.
To determine the effectiveness of testing procedure options, EPA performed statistical
analyses of the misclassification rates associated with each option. EPA was interested in how
frequently a given test procedure correctly indicated whether a home should be mitigated. Exhibit
Chapter 3: Analysis of EPA Radon Testing Options Page 3-9
-------
EXHIBIT 3-4
THE ACCURACY OF MITIGATION DECISIONS
BASED ON SHORT-TERM MEASUREMENTS
Annual Average < Action Level Annual Average > Action Level
Short-Term
Measurement
< Action Level
Short-Term
Measurement
> Action Level
CORRECT CLASSIFICATION
Although the short-term
measurement may vary from the
annual average, both are below
the action level. The short-term
test correctly indicates that
mitigation is unnecessary.
CORRECT CLASSIFICATION
Although the short-term
measurement may vary from the
annual average, both are at or
above the action level. The
short-term test correctly indicates
that mitigation is necessary.
Correct Classification Percentage _
(corresponds to total unshaded cells) ~~
Error Rate _
(corresponds to total shaded cells) ~~
Number of times testing led to a correct conclusion
total homes tested
Number of times testing led to an incorrect conclusion
total homes tested
Note: Often home mitigations will reduce radon levels to 2 pCi/L or lower. Consequently, mitigations undertaken due to false
positive short-term testing results where the annual average radon concentration is above 2 pCi/L (but less than 4 pCi/L)
raav result in significant risk reduction.
3-5 summarizes the results of EPA's statistical analysis of the six testing procedures it considered.
As recommended by the SAB (U.S. EPA 1992b), all results are based on the true radon levels being
equal to the annual average radon levels in homes, across all lived-in levels, as determined by EPA's
National Residential Radon Survey (U.S. EPA/Office of Radiation Programs 1991a). Results are
based on an action level of 4 pCi/L (the level used in both the original and revised Citizen's Guide).
Chapter 3: Analysis of EPA Radon Testing Options
Page 3-10
-------
EXHIBIT 3-5
TESTING PROCEDURES EPA CONSIDERED
AND THEIR MISCLASSIF1CATION RATES
(Action Level = 4 pCi/L)
Test Accuracy
Correct
Classification
Error Rate
• False Negatives
as a Percent of
All Tests
False Negatives
as a Percent of
Actual Positives
False Positives
as a Percent of
All Tests
False Positives
as a Percent of
Ail Positive
Tests
False Positives
with
Quantifiable
Benefit
False Positives
with Non-
Quantifiable
Benefit
Option A
98.1%
1.9%
1.9%
30.2%
0.0%
0.0%
0.0%
0.0%
Option B
93.7%
6.3%
23%
36.5%
4.0%
49.9%
28.9%
21.0%
Option C
94.1%
5.9%
2.3%
36.5%
3.6%
47.4%
27.4%
20.0%
Option D
98.5%
1.5%
0.5%
7.9%
1.0%
15.2%
8.2%
7.0%
Option E
94.7%
5.3%
1.5%
23.8%
3.8%
44.3%
30.9%
13.4%
Option F
94.8%
5.2%
2.8%
44.4%
2.4%
40.4%
26.5%
13.9%
The accuracy of the testing procedure options also can be expressed in terms of the annual
lung cancer deaths that would be averted (i.e., the annual lives that would be saved) and the annual
deaths from lung cancer that would not be prevented (i.e., the number of lung cancer deaths in
homes that received a false negative test result). For the purposes of these calculations, EPA
assumed that 100 percent of the population followed the testing procedures and mitigated their
homes when results were above the action level (see Appendix E for a more detailed explanation).
Exhibit 3-6 presents the annual lives saved due to mitigations (where risk reductions were
quantifiable) under each of the testing procedure options. The exhibit also shows the lives not saved
due to false negative testing results.
Chapter 3; Analysis of EPA Radon Testing Options
Page 3-11
-------
EXHIBIT 3-6
ANNUAL LIVES SAVED AND NOT SAVED
Testing
Option
A
B
C
D
E
F
UNDER
Annual Lives
Saved in Homes
with True Positive
Results
2,160
2,040
2,050
2,630
2,330
1,820
TESTING PROCEDURE OPTIONS27
Annual Lives
Saved in Homes
with False Positive
Results
0
200
180
50
240
140
Total Lives
Saved
Annually
2,160
2,240
2,230
2,680
2,570
1,960
Annual Lives Not
Saved in Homes
with False- Negative
Results
620
750
740
160
450
970
Estimates assume 100% testing and mitigation.
Section IV: Approach for Analyzing Misclassification Rates of Each Testing Procedure Option
To help determine which testing procedure option Lo recommend to the public, EPA focused
on how well testing options provided a reasonably accurate basis for determining whether there was
a need to mitigate radon levels in a home. EPA wanted to estimate the misclassiflcation rates
associated with each option. Misclassifications result from the inaccuracies of test measurements in
estimating the annual average with respect to the action level
Because even Option A, which relies only on long-term tests for confirmatory purposes, relies
on a short-term test for initial screening, all of the options EPA considered have misclassifications
resulting from short-term testing. To analyze the misclassificalion rates that result from the short-
term testing components of each of the options, the Agency used a statistical approach that relied
to some extent on empirical data. The six steps involved in this statistical approach are summarized
in Exhibit 3-7. These steps are explained in more detail in the text that follows according to the step
numbering in the exhibit.
The development of EPA's statistical approach is further documented in two reports. The first report describes the
Agency's empirical analysis of existing radon measurement data (Marcinowski 1990). The second report details EPA's
development of a quantitative approach to predict short-term misclassification rates (Chmelynski 1992).
Chapter 3: Analysis of EPA Radon Testing Options
Page 3-12
-------
EXHIBITS-?
STEPS INVOLVED IN EPA'S
STATISTICAL ANALYSIS OF TESTING OPTIONS
(1) EPA developed a statistical model that explained a short-term test's inaccuracy
in estimating the annual average.
(2) For each option, EPA determined what test procedure outcomes were possible,
depending on the results of the initial and confirmatory tests.
(3) EPA considered the misclassifications resulting from all possible outcomes for
a given option separately.
(4) To determine how many homes would not require a confirmatory test and how
many false negatives there would be among homes not retesting, EPA modeled
the outcome of the initial test by developing a joint probability distribution of
short-term measurements and annual average radon concentrations.
(5) Given the joint probability distribution developed above, EPA modeled the
outcome of subsequent tests by developing conditional probability distributions
based on the annual averages in homes that would be required to retest.
(6) EPA determined the overall misclassification rates for each option by summing
the probabilities of misclassification under all possible outcomes for the option.
Step 1 — EPA's first step was to develop a statistical model that could explain how short-term
measurements could deviate from annual averages. EPA's model is presented in Equation 1. After
starting with the annual average as an expected value for a short-term measurement, EPA's model
includes components that could contribute to deviations from annual averages. The model includes
components for three error types — measurement error, temporal error, and spatial error — as well
as the bias of radon concentration on the testing door level, compared to the overall house annual
average. EPA developed upper and lower bound estimates for all three error types used in its model.
Lowest Living Level (fmml x Lawe^ Livt^ Levd]
nr.._ ^*._. = {Average Floor Bias )
Short-Term
Measurement
measurement temporal
spatiai)
Chapter 3: Analysis of EPA Radon Testing Options Page 3-13
-------
An annual average is the mean radon level over time for a particular house. It reflects an
average of the radon concentrations in all lived-in levels of a home. To calculate the
misclassifications associated with each option, EPA needed to know the distribution of annual
averages, since these determine what short-term measurements will he observed. For its analysis,
EPA assumed that nationwide annual average radon levels had the distribution suggested by the
Agency's National Residential Radon Survey (U.S. EPA/Office of Radiation Programs 1991a) for
homes that EPA recommends test.
Measurement error is the error associated with either a testing device itself or in the reading
of an exposed device by a laboratory technician. EPA's analysis covered only two types of
measurement devices, charcoal canisters and alpha track detectors. Other commonly used
measurement devices, such as electrets and continuous monitors, would perform similarly, if not
better. If the latter were the case, then the misclassifieation rates listed in Exhibit 3-5 would be
reduced slightly. A description of commonly used measurement devices is included in Appendix C.
For its analysis, EPA assumed that there was an equal likelihood of using charcoal canisters
and alpha track detectors. Initially, EPA intended to model the measurement device error associated
with charcoal canisters and alpha track detectors separately. For this purpose, EPA obtained data
on charcoal canisters from EPA/State surveys (U.S. EPA/Radon Division 1990c) and data on alpha
track detectors from a New York State survey commissioned by the State government (Perritt 1990).
The EPA/State surveys provided paired charcoal canister and alpha track detector data on over 1,000
homes. The New York State survey had information on over 2,000 homes. After comparing these
two sources of data, EPA concluded that no significant difference in measurement error existed
between these two types of devices. EPA data show that typical measurement device error for both
charcoal canisters and alpha track detectors is approximately 15 percent in the lab and 30 percent in
the field, in the vicinity of 4 pCi/L (U.S. EPA/National Air and Radiation Laboratory 1991). The
percentage measurement error is higher at lower radon levels, and lower at higher radon levels.
Temporal error is the error associated with the time of year a short-term measurement was
taken. EPA assumes that the annual average is constant, but that within any day, month, or year
there will be fluctuations in the radon level. At any time in the year, therefore, the exact relationship
of a short-term measurement to the annual average will not be known exactly. This inexactness is
the temporal error.
Since there are seasonal differences in the radon level of a home, the false positive rate could
be greater in the winter than in the summer, or the false negative rate could be greater in the
summer than in the winter (Marcinowski 1990; Condon et al. 1990), However, Radon Program
Experience indicates that measurements are currently being conducted during all seasons of the year.
This indicated the need for a testing protocol that could be used throughout the entire year, so the
analysis focused on testing occurring equally during all seasons.3
Research indicates that temporal errors are the biggest contributor to the overall error
associated with short-term measurements. EPA simulated charcoal canister and alpha track device
measurements separately in its analysis because of the difference in temporal errors associated with
3It is unreasonable to assume that closed-house conditions would be maintained for a 90-day test. Because there is a
lack of data on compliance with closed-house conditions, EPA assumed both open-house and closed-house conditions would
occur with equal frequency for testing periods of this duration.
Chapter 3: Analysis of EPA Radon Testing Options Page 3-14
-------
these devices due to their different recommended durations of exposure. Because of the significance
of seasonal variation in radon levels, EPA estimates that a measurement is subject to a temporal error
ranging from as low as 40 percent, for certain tests made under closed-house conditions using alpha
track detectors, to as high as 80 percent, for certain tests made under open-house conditions using
charcoal canisters. These estimates have been based on an EPA study comparing short-term
measurements to annual averages (U.S. EPA/Radon Division 1990d). Because data in this study are
from a cold climate area with a long heating season and since the seasonal variability is greater in
areas with long heating seasons, the estimate of temporal error is conservative (i.e., high).
Chmelynski (1992) provides more detail on the methods used to estimate temporal error.
Spatial error is the error associated with the relationship between the radon concentration
on the lowest lived-in floor, where short-term tests are made, and the overall radon concentration
in a house. Spatial error does not account for the entire difference between the radon level on a
particular floor and the overall home average, but only the unpredictable component of this
difference. EPA estimates that short-term measurements are subject to spatial error that is about
40 percent of a measurement. This estimate has been based on data gathered for a study comparing
basement radon levels to first floor radon levels (Ronca-Battista 1989).
Since annual averages in most cases take into account the radon levels on more than one
floor within a house, short-term measurements may be inaccurate if the floor on which they are taken
does not tend to have the same radon level as the home as a whole. EPA identified the predictable
difference between the radon level on a given floor and the overall home average as the floor bias.
When radon measurements (short-term or long-term) on all lived-in levels in a home are considered
together, the resulting aggregate has no floor bias, since it would be expected to approach the house
overall average. Houses with only one lived-in level would not have any floor bias, since the lived-in
level average would be the same as the overall house average.
Research indicates that for 75 percent of homes, the lowest lived-in level is the first floor
(Sterling et al, f 985). Research also indicates that for most homes, the first floor average is roughly
reflective of the overall home annual average (Peake 1990). Accordingly, the Agency assumed for
its modeling that there would be no floor bias associated with lowest lived-in level measurements.
The components of Equation 1 are multiplicative (reflecting how lowest lived-in floor bias and
error components affect the annual average) rather than additive. To make this equation more
conducive to statistical analysis, EPA applied a logarithmic transformation to both sides of the
equation. Equation 2 shows that the log of a single short-term measurement can be modeled using
five additive components.
f Lowest Living Level"
log Short-Term
I Annual \ , (Lowest Living Level\
(Average) + l°% ( Floor Bias }
(2)
( Measurement , ,, ,
+ log e + log € ,+ log e
["-•& ^measurement ° temporal o ^
Chapter 3: Analysis of EPA Radon Testing Options F*a§e
-------
Steps 2 and 3 — To make its calculation of the overall misclassification rates for an option, EPA
needed to calculate separately the miselassification rates associated with each possible outcome of the
option (i.e., what type of confirmatory test, if any, would be required based on the measurement
observed in the initial test). In Exhibit 3-8, the different outcomes possible under each option are
graphically separated into different boxes. The types of error associated with each possible outcome
are also indicated. In this exhibit, unshaded areas for each option indicate the range of initial short-
term measurements for which no confirmatory tests are required. Shading indicates areas where
errors are dependent on the combination of initial short-term and confirmatory tests. Option E is
completely unshaded, unlike other options, because for this option no confirmatory tests are
performed.
Step 4 — For each option, EPA began its analysis of raisclassification rates by focusing on the
possible outcomes of the initial test. Based on the statistical model shown in Equation 2, EPA
developed a joint probability distribution of annual average radon concentrations and expected short-
term measurements. A joint probability distribution indicates the probability with which certain short-
term measurements can be expected to be observed at a given annual average. EPA needed to
develop a joint probability distribution of annual averages and short-term measurements to determine
the proportion of homes in which a confirmatory test would be required under each option, assuming
the national distribution of annual average radon levels.
Because a joint probability distribution indicates the distribution of short-term measurements
that will occur for a given distribution of annual averages, it can be used to calculate the
misclassification rates attributed to testing outcomes not requiring a confirmatory test. Options A,
B, C, and F all require one initial short-term test with an action level of 4 pCi/L. Exhibit 3-8
indicates that the initial short-term test components in these options will produce false negatives in
the same way, since all four options have one initial measurement and the same action level. The
overall misclassification rates of these four options vary due to differences in the confirmatory test
required when the initial measurement is above the action level. Since Option D requires further
testing if the initial measurement is above 2 pCi/L, this option has fewer false negatives associated
with the initial testing. Option E, unlike other options, bases mitigation decisions on the average of
several initial short-term tests. This option requires no further testing regardless of the radon level
observed using short-term testing. Consequently, Steps 5 and 6 are not required for this option.
Step 5 — Using the joint probability distribution of annual averages and short-term testing that it
developed for Options A, B, C, D, and F, EPA developed conditional probability distributions to
model the outcome of the possible subsequent tests required under each option. Conditional
probability distributions indicate the expected frequency of results considering only the subset of a
population that has already met a given condition.
In EPA's analysis, the subset of the population being considered was only those homes that
had short-term measurements above the action level (Options A, B, C, and F) or the trigger level
(Option D). For each possible outcome in an option, EPA developed a separate conditional
probability distribution. In all cases, the conditional distributions took into account the subpopulation
required to retest and whether a long-term or short-term test was required. As part of its analysis
of conditional distributions, EPA focused on the probabilities of false positives occurring at homes
with annual average radon levels between 2 pCi/L and 4 pCi/L. If these homes are mitigated, it is
likely that the radon levels in them will decrease and the occupants will receive some benefits from
lowering their exposure to radon.
Chapter 3: Analysis of EPA Radon Testing Options Page 3-16
-------
EXHIBIT 3-8
COMPARISON OF FALSE NEGATIVES AND FALSE POSITIVES UNDER
RADON TESTING OPTIONS THAT EPA CONSIDERED
Option A
I 4 pd/L Action Level For Options
| A-C. Initial Test Above This Level
i Leads to Confirmatory Testing,
If ST test is below action level, no retest
is required. False negatives may result
Option B
If ST test is below action level, no retest
is required. False negatives may result
Option C
If ST test is below action level, no retest
is required. False negatives may result
Option D
2pCi/L
Trigger Level
WpCVL
Trigger Level
If ST is below 2 pCi/L,
no retest is required.
False negatives may result
Option E
If average of ST measurements is below
action level, false negatives may result
I If average of ST measurements
J is above action level,
• false positives may result
Option F
If ST test is below action level, no retest
is required. False negatives may result
4 pd/L Action Level
ST = Short-Term Test
LT= Long-Term Test
Note: Shaded a/was indicate results dependent on both initial and confirmatory tests.
-------
Step 6 — To determine the overall misclassification rates for each option, EPA summed probabilities
in the joint probability distribution of annual averages and initial short-term tests and in the
conditional probability distributions of houses required to retest. To calculate the total of false
negatives, EPA summed probabilities that a testing outcome would classify a home as being below
the action level when the annual average was actually above the action level. To calculate the total
of false positives, EPA summed probabilities that a testing outcome would classify a home as being
above the action level when the annual average was actually below the action level. EPA aggregated
false negatives and false positives for possible outcomes under each option to generate the results
presented in Exhibit 3-5. Because EPA's analysis in essence is a comparison between the accuracy
of the option relying on a confirmatory long-term test and the accuracies of options relying to varying
degrees on confirmatory short-term tests, the Agency simplified its analysis by assuming that long-term
tests did not result in any misclassifications. This is because temporal variation is the largest
contributor to the misclassification rate and for year-long tests, this error term is negligible. The
misclassification for those options utilizing a long-term confirmatory test is insignificant and is
represented as resulting in no misclassifications in the results.
Section V; Analysis of Options
Options A, B, C, and F begin with a single short-term test, followed by a confirmatory long-
term or short-term test if the result of the initial test is above the action level. Because these four
options begin with the same procedure, they share the same rate of false negatives among homes not
requiring a subsequent test. The difference in misclassification rates among these four options is due
strictly to differences in the procedures they use for follow-up testing. To determine the
misclassification rates associated with follow-up tests under each option, EPA developed conditional
probability distributions for the subset of homes expected to need confirmatory testing.
Like Options A, B, C, and F, Option D also begins with a single short-term test. Option D,
however, has a trigger level of 2 pCi/L, and an additional trigger level at 10 pCi/L that is discussed
further below in the documentation of EPA's analysis of Option D. Option D's lower trigger level
of 2 pCi/L for initiating confirmatory testing leads to a lower rate of false negatives than Options A
through C since homes with measurements between 2 to 4 pCi/L are subject to long-term tests. Only
false positives happen in measurements above 10 pCi/L, since in the interval of 2 to 10 pCi/L long-
term testing is done.
Option E, unlike other testing alternatives, consists of multiple initial tests and does not
require confirmatory testing. Accordingly, for this option EPA analyzed only the misclassification
rates due to initial testing. EPA's approaches for determining the misclassification rates for the six
options are discussed in more detail below.
OPTION A — This alternative recommends a confirmatory long-term test whenever the
results of the short-term measurement are above the action level. As discussed above, the
misclassifications for long-term measurements are insignificant and, therefore, all long-term tests were
represented as resulting in correct classifications of homes concerning the need for mitigation.
Accordingly, to compute the misclassification rates associated with this option, EPA needed only to
consider the probabilities that false negatives would occur with the initial short-term test. These
probabilities were obtained from a joint probability distribution of annual averages and expected
results of the initial short-term test. This joint probability distribution was based on Equation 2,
Chapter 3: Analysis of EPA Radon Testing Options Page 3-18
-------
which states the assumed relationship between short-term measurements and long-term annual
averages. See Steps 1 through 3 of Exhibit 3-9 for the derivation of the joint distribution.
OPTION B — This option is similar to Option A, except that it requires a confirmatory
short-term test when the initial short-term measurement exceeds 4 pCi/L. Some false negatives result
from this testing procedure when the initial short-term measurement is below the action level. Both
false negatives and false positives are associated with the option when the initial short-term
measurement is above 4 pCi/L, depending on the outcome of the second test. Exhibit 3-9 illustrates
the key steps in determining false positives for Option B.
To analyze the misclassification rates associated with Option B, EPA separately considered
the range of initial short-term measurements below the action level and above the action level. For
its analysis of Option A, it already had analyzed the probability of obtaining false negatives associated
with short-term measurements below 4 pCi/L.
To calculate the misclassification rates associated with the average of two short-term
measurements with a 30-day waiting period between them, EPA modified its statistical model. When
EPA was analyzing the misclassifications associated with only one short-term test, as in Option A, it
took into account the temporal error associated with a radon measurement made at one given time
of the year. If a second short-term measurement were made at the same time as the first (as in the
case of Option E, described below), the Agency would still be dealing with the same amount of
temporal error. In statistics, the strength of the relationship between sequential samples is called
serial correlation. In the absence of serial correlation, the averaging of two measurements would lead
to a 1//2 reduction of the temporal error associated with short-term measurements, and hence to a
large reduction in the misclassification rate. To determine the degree to which a second sample
would reduce temporal error, EPA needed to take serial correlation into account in its statistical
model. In considering Option B, the Agency did this at a point in its analysis corresponding to Step
5 in Exhibit 3-7.
To determine how much serial correlation there is between two short-term measurements,
EPA studied the variations in radon levels in homes over time in the course of its Butte, Montana
study (U.S. EPA/Radon Division, 1990d). The reduction in uncertainty obtained using a second,
serially correlated short-term measurement was determined by computing the serial correlation of
various short-term averages separated by a set number of days using data on daily radon levels in
these homes. When the short-term measurements are made under closed-house conditions, a serial
correlation of approximately 0.4 was obtained for 2- to 7-day tests separated by 30 days. For 90-day
tests separated by 90 days, the serial correlation was lower, approximately 0.1. Variance reduction
factors were then computed, and these factors were then applied to the temporal error term in
Equation 2. The reduction in temporal error due to seasonal variation resulting from a second,
uncorrelated measurement would be approximately 30 percent (1 - [1//2]). The estimated reduction
in temporal error for 2- to 7-day tests separated by 30 days was 16 percent, compared to 26 percent
for back-to-back 90-day tests.
After estimating the amount of reduction in uncertainty that a confirmatory short-term
measurement would bring, EPA developed a conditional probability distribution of the average of two
sequential short-term tests among houses with an initial short-term test above 4 pCi/L. Evaluation
of the required conditional distribution was based on the joint (trivariate) distribution of two serially
correlated short-term tests and the long-term annual average, as described in Step 5 of Exhibit 3-9.
Chapter 3: Analysis of EPA Radon Testing Options Page 3-19
-------
EXHIBIT 3-9
HOW FALSE POSITIVE MISCIwVSSIFICATION KATES WERE CALCULATED FOR OPTIONS
Step (1) —EPA Developed a Model for Expressing Short-Term Measurements
in Terms of the Long-Term Annual Average and Several Components
Contributing to Deviations from this Average (See Equation 2 in Text)
l«f
{Lowest Living Levet\
Short-Term
{ Measurement }
log ff nnml}
\Average)
. I Lowest Living Level]}
+ log 1 Floor flL Jj
Step (2) — EPA Assumed Values or Ranges of Values for All Additive
Components Based on EPA Studies
Lowest Living Floor Bias — none
Measurement Error — 30% at 4 pCi/L
Temporal Error — 40% to 80%
Spatial Error — 40%
Step (3) — EPA Developed a Joint Probability Distribution to Predict the
Homes Requiring Confirmatory Testing. The model in Step (1) estimates the
conditional distribution of Sl\ given the annual average, t'(S'f t,AA),
Probability (ST. > 4pCilL, AA < 4pCilL] = f" f* /(ST., AA) dST. dAA
1 J$r,=4 JAA=O J *
where STj = Initial Short-Term Measurement, AA = Annual Average
Step (4) — EPA Analyzed the Percentages of Initial Short-Term Measurements
Above 4 pCi/L for Certain Ranges of Annual Averages (percentages are
expressed as proportion of homes testing)
True Annual Average Range
0 - 1 pCi/L
1 - 2 pCi/L
2 - 3 pCi/L
3 - 4 pCi/L
Short-Term Tests Above 4 pCi/L
1.9%
2,4%
1,9%
1,6%
Step (5) — EPA Developed a Conditional Probability Distribution for the
Average of Two Short-Term Tests, Taking into Account Serial Correlation,
Given the Distribution of Annual Averages and the Initial Measurement
| sr,
h**) JsTi-t Is
'Srt«ii~STs
iu^i$T^/(AA'STt)'
9 Shert-Term Mtmummtnts> AA = Annual Average
Step (6) — EPA Estimated the Percentage of Short-Term Measurements above
4 pCi/L and Summed Them to Get Total False Positives
True Annual Average Range
0 - 1 pCi/L
1 - 2 pCi/L
2 - 3 pCi/L
3 - 4 pCJ/L
0 - 4 pCi/L
Option B False Positives
0.4%
0,9%
1,1%
1.2%.
3.6%
-------
To obtain the overall misclassification rate for Option B, EPA aggregated the probabilities of
obtaining false negatives when follow-up testing was not required with the probabilities of obtaining
false negatives and false positives when a confirmatory test was required. (In addition to leading to
a false positive by confirming an initial short-term test, the confirmatory test can lead to a false
negative if it does not agree with an initial test when the radon level is actually high.) These results
are presented in Exhibit 3-5.
OPTION C — To analyze the misclassification rates associated with Option C, EPA viewed
this testing procedure as a combination of Options A and B. Those options, respectively, require
initial short-term measurements to be followed by a confirmatory long-term or short-term
measurement.
Since Option C allows individuals to use either a confirmatory long-term or short-term test,
EPA's first task in analyzing the misclassification rates associated with this option was to determine
the frequencies with which each type of test would be used. EPA relied on a study it had done of
the public's willingness to use long-term tests. This study indicated that only 9 percent of the public
was willing to use long-term tests (Johnson 1990). Accordingly, EPA assumed for its analysis of
Option C that for initial short-term measurements that were above the action level, 9 percent would
be followed by confirmatory long-term measurements and that the remaining 91 percent would be
followed by confirmatory short-term tests.
Once EPA had assumed how the public would respond to the flexibility in the testing
procedure offered under Option C, it could determine the misclassification rate of the option by
combining its analytical approaches for Options A and B. The Agency weighted the results of these
two options with factors of 0.1 and 0.9, respectively, to simulate the choice of long-term and short-
term testing procedures by the public. The results are presented in Exhibit 3-5.
OPTION D — Like Option C, this option can be viewed for analytical purposes as a
combination of Options A and B. Option D, however, does not give the public the choice between
long-term and short-term tests, but rather specifies which type of confirmatory test will be used, based
on the concentration of the initial short-term measurement. For measurements between 2 pCi/L and
10 pCi/L, a confirmatory long-term test is required. For measurements above 10 pCi/L, only a
confirmatory short-term test is required. Of course, for measurements below 2 pCi/L, no
confirmatory test is required at all.
To determine the total amount of misclassification associated with Option D, EPA considered
the three possible outcomes of the initial short-term test separately. Exhibit 3-8 illustrates what types
of misclassification errors are associated with these three outcomes. Because initial short-term
measurements below the lower trigger level of 2 pCi/L do not include any scenarios leading to
mitigation, some false negatives occur. Roughly 93 percent of all homes in the population considered
did not require a confirmatory measurement. Initial short-term measurements above the lower trigger
level (2 pCi/L), but below the higher trigger level (10 pCi/L), were represented as not resulting in
any misclassifications at all, since the misclassifications for long-term measurements were insignificant.
Roughly 4 percent of all homes that should test had this outcome. Results above the higher trigger
level always indicated that mitigation was necessary, since the average of the two short-term tests was
always at least 5 pCi/L. In practice, however, if there is a large disparity between the results of the
first and second test, additional evaluation may be necessary to determine the need for mitigation.
Chapter 3: Analysis of EPA Radon Testing Options Page 3-21
-------
Less than 3 percent of all homes in the population had initial short-term tests with measurements
above 10 pCi/L.
Based on its analysis of Options A through C, EPA had already obtained a joint probability
distribution of annual averages and initial short-term measurements. From this joint probability
distribution, EPA determined the probability of false negatives due to initial short-term tests
incorrectly providing measurements o£ less than 2 pCi/L.
Since EPA assumed that no misclassifications would be associated with long-term
measurements, it determined that all initial short-term measurements at least equal to 2 pCi/L but
below 10 pCi/L would result in correct final classifications, regardless of the accuracy of the initial
short-term test.
EPA knew that all initial short-term measurements above 10 pCi/L would require mitigation.
To determine the number of false positives associated with this outcome, EPA aggregated the
probabilities of having a false positive above 10 pCi/L from the joint probability distribution. (EPA
did not have to take into account the serial correlation between the two short-term tests under this
option, since it assumed that the outcome would always result.in an assessment that mitigation was
necessary.)
To determine the overall misclassification rate associated with Option D, EPA combined the
probabilities that false negatives would occur due to initial short-term measurements below 2 pCi/L
with the probabilities that false positives would occur due to initial short-term measurements above
10 pCi/L. The results are presented in Exhibit 3-5.
OPTION E — Unlike Options A through D, Option E requires that short-term tests be
conducted on all levels of a home that are lived-in levels (i.e., floors on which individuals spend more
than 4 hours of their time). Because this testing procedure would result in a number of different
testing patterns based on home type and lifestyle, EPA repeated its analysis of this option for three
different housing situations that could occur. The summarized results presented in Exhibit 3-5 reflect
a weighted average of the misclassifications tor the different housing situations. The first situation
EPA considered was an individual living on one level. In this case, the individual would make two
tests on the same level, in different locations. The second situation EPA considered was an individual
living on two levels, either basement and first floor or first floor and second floor. The third situation
EPA considered was an individual living on three levels.
After the Agency had determined the reduction in error components in Equation 1 under all
three scenarios, EPA derived three separate joint probability distributions for this option. All of
these distributions differed slightly from the single short-term measurement/annual average joint
probability distribution developed for Options A through D. Because each of the three scenarios had
a different joint probability distribution, each also had a different misclassification rate.
OPTION F — Like Option B, this option required a confirmatory short-term test when the
initial short-term measurement exceeded 4 pCi/L. As Exhibit 3-8 illustrates, some false negatives
resulted from this testing procedure when the initial short-term measurement was below the action
level. Both false negatives and false positives were associated with the option when the initial short-
term measurement was above 4 pCi/L, depending on the outcome of the second test. Unlike Option
B, however, this option does not average test results.
Chapter 3: Analysis of EPA Radon Testing Options Page 3-22
-------
To analyze the misclassification rates associated with Option F, EPA modified the approach
that it had taken for the analysis of Option B. After estimating the amount of reduction in
uncertainty that a confirmatory short-term measurement would bring given the serial correlation
between two sequential measurements, EPA developed a conditional probability distribution of the
probability of outcomes for a second short-term among houses with an initial short-terrn test above
4 pCi/L. To obtain the overall misclassification rate for Option F, EPA aggregated the probabilities
of obtaining false negatives when follow-up testing was not required with the probabilities of
obtaining false negatives and false positives when a confirmatory test was required. (In addition to
leading to a false positive by confirming an initial short-term test, the confirmatory test could lead to
false negatives if it does not agree with an initial test when the radon level is actually high.)
Section VI: General Conclusions
Analysis of options that rely on short-term tests to decide whether to mitigate existing radon
levels shows that these tests can be used effectively. For the action level of 4 pCi/L, the Agency
found that short-term testing procedures can lead to a reasonably low level of misclassification of
homes with respect to the action level. EPA found that the skewed distribution of radon levels
towards the lower end of the distribution served as a strong countervailing factor to any
misclassifications created by short-term testing errors.. The results are based on examination of the
most commonly used radon measurement devices and consideration of testing throughout the year.
Option A, which requires a single short-term measurement followed by a confirmatory long-
term test, parallels the testing procedure recommended in the original Citizen's Guide, except that
testing would be conducted on the lowest lived-in level. Under Option A, 98 percent of homes were
correctly classified with regard to the need for mitigation. To determine how testing procedures
relying on a short-term confirmatory' test would compare to this option, EPA developed and analyzed
five other options (Options B through F). EPA found that all of these options produced results
yielding at least 94 percent correct classifications. Similarly, each option has the potential to save a
significant number of lives, although there is some variation in the potential number of lives saved
and the number of lives not saved under each testing procedure.
Given these results, the Agency determined that on the basis of misclassification, none of the
options should be rejected. EPA decided that other factors, such as procedure simplicity, should be
weighed with the trade-off of increased levels of proper classification in making the decision on the
testing option to recommend in the revised Citizen's Guide. As stated previously, current experience
indicates that few people are actually taking long-term follow-up tests, and that most people who
mitigate do so based on a single short-term measurement. This is supported by risk communication
research that indicates that only 9 percent of the population is willing to conduct annual tests (see
Chapter 6 for a more detailed discussion). This suggests that a testing protocol that relies exclusively
on long-term confirmatory tests is unlikely to be followed by most of the public. Therefore, although
Options A and D have more desirable error rates than other options, it is likely that compliance with
these options would be low. Selection of either of these two options is apt to result in a situation
in which decisions are most often made based on a single short-term measurement when homeowners
do not complete the testing process. By explicitly providing homeowners a process that calls for more
than one short-term test, all of the other options considered would be more likely to result in people
having better information to use in reaching a mitigation decision.
Chapter 3: Analysis of EPA Radon Testing Options Page 3-23
-------
On the other hand, since long-term measurements are more desirable than short-term
measurements and since there are some individuals who are willing to conduct long-term tests, the
use of long-term tests should not be precluded from the testing protocol. As a result, Options B, E,
and F, which do not include long-term tests, are also undesirable.
Option C offers an effective compromise between these different approaches. It promotes
long-term testing by people who are willing to conduct long-term tests and recommends an effective
short-term test as an alternative for people who are not willing to conduct a long-term test.
Therefore, consistent with the advice from EPA's SAB, the revised Guide recommends Option C
because it should maximize the total risk reduction the public would gain through future testing and
mitigation.
Chapter 3: Analysis of EPA Radon Testing Options Page 3-24
-------
CHAPTER 4
MITIGATION TECHNOLOGY
This chapter explains different types of
available and emerging radon mitigation THE AIMS OF THIS CHAPTCR
techniques, their ettectiveness, and costs. 1 he
chapter is based primarily on research that ,.,, ~ , ., ,.,,£. . ,
r-n!T> rw f'n I i r^ i . (I) To describe different radon
EPAs Office of Research and Development ' ... ,. . , ..
/r-M-,r-,% i j j j i c mitigation techniques.
(ORD) has conducted and results from a survey . '
of commercial radon miticators that EPA's ,«N ^ . c .. .,
„„,. /-r, .... r, ,nn-n\ i t j (2) To present information on the
umce ot Radiation Programs (ORr) conducted ~. ... , , . r ...v.
. . „„„ & v / etfectiveness and cost of different
in 1989. • , , .
techniques.
Section I: Mitigation Methods
Preventing radon entry and reducing radon concentrations after entry are two common
mitigation strategies. Preventing entry is often the best strategy since it has a high probability of
success, even in locations with very high radon levels. Techniques that reduce radon after entry are
most appropriate for buildings with relatively low radon levels, where radon entry cannot be
prevented, or in which increased ventilation could provide valuable benefits in addition to radon
reduction.
Regardless of the strategy considered, any mitigation plan must take into account a number
of considerations. In addition to controlling radon, the mitigation system should be unobtrusive,
quiet, and capable of indicating system failure. It should be economical and easy to maintain and
operate. Mitigation systems also must be a permanent part of the building rather than portable or
window-mounted devices that can be removed when the building is sold.1
Methods that Prevent Radon Entry
Gaseous elements contained beneath the earth's surface can migrate through openings in soil
or rock and be released from the ground, depending on a number of factors. Soil gas is drawn
indoors by the differential between relatively low air pressure in a house and the higher air pressure
in the soil. Depressurization of the surrounding soil and foundation can prevent radon-bearing soil
gas from entering a house. Experience also shows that depressurization may be made more effective
by sealing cracks and other openings in floors and walls. Sub-slab depressurization is the preferred
approach because of its effectiveness, applicability, and case of installation. Whenever any active
system is installed, the mitigator should test the house for combustion backdrafting.
Sub-slab depressurization (also called sub-slab suction) is a popular mitigation technology.
Once soil gas accumulates in the soil and aggregate underlying the concrete slab in a basement or
slab-on-grade house, it can migrate indoors through any openings in the slab. Active sub-slab
depressurization systems use pipes running down through the slab and up to a fan that vents the gas
According to the Interim Mitigation Standards for the EPA Radon Contractor Proficiency Program.
Chapter 4: Mitigation Technology Page 4-1
-------
outdoors. The fan reduces air pressure in the soil, thereby reversing the pressure gradient and
causing air to flow out of the house and into the soil through any openings in the slab. Passive sub-
slab depressurization systems also can be used, but are less effective than active sub-slab
depressurization. Soil depressurization under a plastic or rubber membrane, or sub-membrane
depressurization, is used in buildings with earth-floored crawlspaces or basements (Hubbard et al.
1987). Its function is essentially the same as sub-slab depressurization, except that the mitigator
constructs a barrier as a collection cover rather than using the existing slab as a barrier.
If a perforated drain tile is present, drain-tile depressurization can be used to create a
negative pressure field surrounding the foundation walls to draw soil gas away from possible entry
routes. Block-wall depressurization consists of using a fan and duct work to draw suction on the
hollow interior cavities of a concrete block wall By maintaining a lower pressure in the void network
within the block wall than in the basement, this technique forces the flow of soil gas to be outward
rather than into the basement. In a baseboard depressurization system, a "baseboard" is installed
around the entire perimeter of the basement and fans are used to depressurize the block wall cavities.
This system evenly distributes the pressure field and may simultaneously handle wet foundation
problems (Fowler et al. 1988; Henschel 1987).
Sub-slab and block-wall pressurization have been tested as alternatives to depressurization
in a number of government research projects. Essentially, these techniques reverse the fan in a soil
depressurization system, creating a positive shield of diluted air around a building. These methods
have had mixed results (Henschel and Scott 1987; Hubbard et al. 1987; Pyle et al. 1988; Turk et al.
1986). In certain conditions, they perform better than soil depressurization. (The selection of these
techniques requires caution when pesticides have been applied nearby.)
Basement pressurization prevents soil gas from entering the building and reduces radon
concentrations by dilution (Nitschke et al. 1985; Turk et al. 1986). It can be accomplished by using
a fan to pull indoor air from the upstairs lived-in area and blow it into the basement. This process
creates a higher air pressure in the basement than outside, causing indoor air to flow through cracks
and holes in the basement out into the soil. Basement pressurization should be considered an
alternative for houses that are not compatible with soil depressurization techniques. Adequate long-
term data are not available to determine the effect of basement pressurization on energy costs or
whether condensation will occur as basement air is forced out through cracks above grade.
Sealing opening!* to the soil is helpful as a supplementary activity to other methods as it
closes radon entry pathways and can improve the extension of depressurization on the soil. However,
the results of sealing are unpredictable when it is used alone as the primary mitigation method.
Success seems to be based on the number of cracks and holes in the slab or foundation, as well as
their connection to radon transport mechanisms. If the house is situated in a permeable soil with
elevated radon concentrations in the soil gas, it is very difficult to make a sealing-only technique
work. If there is only one primary transport mechanism connected to a single entry point, sealing
could make a difference, assuming the entry route could be sealed.
Isolation is another form of sealing that can be used to prevent radon entry. This technique
closes the openings between a large area of the substructure and the adjacent lived-in space, thereby
isolating exposed soil and rock under, around, or within a house that can be both a major source and
entry route for radon (Nitschke et al. 1985; Turk et al. 1986). When used in combination with
ventilation, isolation can be a very effective mitigation approach. Unfortunately, many spaces cannot
be easily isolated from the lived-in space and ventilated without causing winter freezing problems or
Chapter 4: Mitigation Technology Page 4-2
-------
summer moisture problems. Crawlspaces lend themselves to isolation and ventilation, unless they
contain air distribution duct work that makes isolation difficult. Lived-in space radon concentrations
can be reduced by passively ventilating a crawlspace, depressurizing the crawlspace, pressurizing the
erawlspace, or creating a balanced flow. The two primary goals are to (1) decouple the lived-in area
from the soil, and (2) reduce concentrations in the crawlspace by dilution.
Radon entry also can be deterred by reducing negative pressure in the lowest level of the
house, since this pressure is the main driving force for infiltrating soil air. Negative pressure can be
reduced by sealing thermal bypasses at the ceiling level or supplying makeup (combustion or exhaust)
air to appliances such as furnaces, boilers, and bath fans that depressurize a house by using and
exhausting air to the outside. Currently, reducing negative pressure is not a primary mitigation
technique, but is used to enhance the performance of other mitigation efforts or to prevent a radon
problem in new construction (very little research has been done to evaluate it as a primary mitigation
approach). This technique also may reduce a home's heating needs,
Methods that Remove Radon from a Building
Once radon has entered a building, a variety of mitigation techniques can be used to reduce
the radon concentrations. Any radon problem can be reduced to some extent by blowing air through
the house. However, the use of increased ventilation is usually limited to homes in mild climates with
less than 20 pCi/L of radon because of the energy costs and comfort problems that would be
associated with the addition of large amounts of ventilation air. Best results are obtained when
basement areas are ventilated. Ventilation also can provide additional benefits such as the dilution
of other indoor air contaminants.
Mechanically powered ventilation is used either to blow air into the house or to supply and
exhaust air simultaneously. The indoor air pressure is increased (i.e., negative pressure is reduced)
if air is blown into the house, possibly leading to higher radon reductions than would be expected
from dilution alone. However, in climates where heating is necessary, blowing unconditioned air into
the house also would tend to increase energy usage and may increase the risk of moisture problems
in the building shell. If one fan is blowing in the same amount of air that another fan is removing,
the indoor pressure level will not be affected. Therefore, if other factors such as wind and
temperature remain constant, the influx of radon will remain the same and the radon concentration
will go down in proportion to the increase in the ventilation rate. If mechanically powered ventilation
is used only to exhaust air from the house, radon may be increased due to the negative pressure
induced inside (Hubbard et al. 1987; Lencheck et al. 1987).
Ventilation systems can be designed with or without a heat recovery system. A typical system
without heat recovery can employ a duct that draws outdoor air into the cold-air return of a forced-
air furnace when the furnace is operating (Pyle et al. 1988). When the furnace fan is not operating,
fresh air enters only in response to pressure differentials between outside and inside air, thereby
reducing pressure differentials. The furnace heats this air and distributes it to the house. This small
amount of air will probably not move the house from negative pressure to positive pressure, but it
will reduce the negative pressure in the basement.
Heat recovery ventilators (HRVs or air-to-air heat exchangers) are designed to bring in the
same amount of air that the system is exhausting. The incoming air stream is preheated by the
outgoing airstream (Fowler et al, 1988; Henschel and Scott 1987; Lenchck et al. 1987; Nitschke et
al, 1988). This process reduces the energy costs and associated discomfort. Because the balanced
Chapter 4; Mitigation Technology Page 4-3
-------
flow should have no impact on basement pressure, radon gas reduction will occur only through
dilution. The radon concentration will be reduced by the inverse of the ventilation increase. If a
basement is large, leaky, or has greatly elevated radon levels, it will be very difficult to make a
significant difference with ventilation.
Another radon removal device circulates interior air through alternating charcoal filtration
beds. As interior air circulates through one bed that accumulates radon and its decay products,
exterior air is circulated through the second bed and exhausted outside. Thus, the beds collect radon
and its decay products from the indoor air which are then transported outside by the flow of outside
air. The airflows alternate between beds to maintain radon removal efficiency.
Passive ventilation is achieved by opening windows while ensuring that the house is not
depressurized. Passive ventilation works by diluting the indoor radon concentration and by
neutralizing the pressure differentials between the inside and the outside. In one study, by opening
all the windows the leakage area of the house was increased by a factor of 10 to 20, and radon
concentrations in the house were reduced by a factor of 4 to 10 (Hubbard et al. 1987). Because this
method can be used only when the outside weather permits, it is not an adequate permanent solution.
Other Methods under Development
There are several other methods that have not yet been shown to reduce radon health risks,
but which are still the subject of research and may prove to be significant after further improvement.
These alternate methods include air cleaners and space ionization. EPA does not recommend the
use of these techniques.because they can be easily removed from the structure.
Air cleaners that are commonly used to remove airborne particulates and condition indoor
air undoubtedly remove some radon decay products. However, many questions remain concerning
the relative health effects of the decay products that are not removed (the unattached fraction).
Space ionization removes radon decay products from indoor air by ionizing and circulating
the air. Many units may be needed to treat an entire house effectively. Also, no easy way currently
exists to monitor the performance of units that remove decay products without affecting radon gas
concentrations. Use of these devices should be considered only where preventative techniques are
not feasible, or as a temporary technique until other controls are implemented.
Section II: Effectiveness and Cost
This section summarizes the effectiveness and cost of various radon mitigation techniques.
Effectiveness is measured by the reduction in initial radon levels, and cost includes all the associated
costs of employing each technique. There are two primary sources of information on mitigation
effectiveness and cost: EPA research and demonstration projects and the Private Sector Radon
Mitigator Survey (U.S. EPA/Radon Division 1990a). Each one of the sources is summarized
individually. (The Office of Radiation Programs also has constructed a mitigation cost model that
uses the information summarized here along with additional information supplied by private mitigators
to estimate costs. That model is explained in Chapter 5.)
Chapter 4: Mitigation Technology Page 4-4
-------
Effectiveness
Several research projects on the reduction in radon levels achieved by various mitigation
techniques have been conducted by EPA's ORD. Pre- and post-mitigation radon concentrations were
measured in the basement for houses with basements, and in the lived-in area for houses with slab-
on-grade and crawlspaces. Various measurement methods were used in the different projects (e.g.,
alpha track detectors, charcoal canisters, continuous monitors). ORD's program was designed to
reduce radon levels in homes with very high initial radon levels and often was conducted in houses
that had been identified as difficult to fix.
Through this research, ORD has found that radon levels in almost all homes can be reduced
to less than 4 pCi/L (U.S. EPA/ORD 1989). In fact, radon levels very often were reduced to less
than 2 pCi/L, even though the mitigation efforts were only trying to reach an action level of 4 pCi/L.
EPA believes that radon reduction to 2 pCi/L or lower would have been even more common if it
were not for the fact that many of the homes examined had very high initial radon levels. Based on
this experience, the data suggest that available mitigation technologies are able to reduce radon levels
in homes above 4 pCi/L down to 2 pCi/L, on average. Such reductions can be expected in houses
with different foundation types, including basement, slab-on-grade, and crawlspace homes.
Similar evidence of the ability of mitigation techniques to reduce radon levels was compiled
in ORP's Private Sector Radon Mitigator Survey, a nationwide survey of about 340 private sector
radon mitigators in the fall of 1989. Exhibit 4-1 summarizes the Gnal results from this survey for
detached houses.2
EXHIBIT 4-1
PERCENT OF PRE- AND POST-MITIGATION RADON LEVELS
CITED IN PRIVATE SECTOR RADON MITIGATOR SURVEY
(U.S. EPA/Radon Division I990a)
PRE-MITIGATION POST-MITIGATION
pCi/L Percent Homes pCi/L Percent Homes
< 4 1% < 1 41%
4 - 10 23% 1 - 2 28%
10 - 20 37% 2 - 3 18%
20-50 29% 3-4 10%
> 50 10% > 4 3%
The question asked of private mitigators was "What were the radon concentrations in the last building you mitigated
of each of the following types: detached house, school, workplace?" The answers for detached houses were used to
develop Exhibit 4-1.
Chapter 4: Mitigation Technology Page 4-5
-------
The exhibit shows mitigators reporting that they have reduced radon levels to below 4 pCi/L
in 97 percent of the cases reported. Mitigators have also reduced radon levels to 2 pCi/L, or less in
69 percent of the cases reported. Examination of the pre-mitigation radon levels in these homes
suggests that the private mitigator survey is reporting results from a set of homes that have a higher
proportion of units with elevated radon levels (above 10 pCi/L) than will exist nationally in the homes
that remain above the action level (i.e., 4 pCi/L) and therefore need to be mitigated. Also, although
mitigators were trying to get radon concentrations below EPA's action level, they were not necessarily
trying to reduce radon levels to 2 pCi/L or less. Therefore, EPA expects even better results for the
owners of the remaining homes that follow EPA's advice and have mitigators try to reduce radon
levels as much as possible. EPA believes that it is reasonable to assume that mitigators should be
able to get all homes above 4 pCi/L down to an overall average of 2 pCi/L (with some homes still
above 4 pCi/L and many homes attaining levels below 2 pCi/L).3
Based on the findings of ORD's own research and the mitigator survey, the Agency concludes
that radon mitigation in the vast majority of cases (more than 95% of the time) should lead to radon
reductions that leave homes with an annual average level of less than 4 pCi/L. It also will be very
common for homes to have post-mitigation levels of 2 pCi/L or less. Results from the mitigator
survey indicate that 2 pCi/L can be achieved approximately 70 percent of the time, while the ORD
research suggests this estimate may be even higher. ORD's research shows that, on average, homes
above 4 pCi/L can be mitigated to 2 pCi/L. In the fraction of homes where this lower level can be
reached, reductions to 2 pCi/L or less should occur without inordinate effort, as the private sector
survey results are based on mitigation experience when radon mitigators were trying to bring homes
down to an action level of 4 pCi/L. Therefore, homeowners should expect that mitigators can provide
radon reductions at, or well below EPA's action level.
Lowering high radon levels often requires considerable technical knowledge and special skills.
As a result, the revised Citizen's Guide recommends that homeowners use mitigation contractors who
have passed the EPA Radon Contractor Proficiency (RCP) Program tests. Research indicates that
remedial actions taken by homeowners themselves are generally less effective than mitigation by
contractors, and often are not followed by retesting to verify their effectiveness (Doyle et al. 1990).
Nevertheless, EPA recognizes that some homeowners may want to mitigate their homes themselves
and has produced several publications to help guide "do-it-yourself" efforts. The revised Citizen's
Guide also recommends that homeowners contact their State radon office for further information on
do-it-yourself mitigation.
Cost
The cost of a mitigation technique depends primarily on two factors: the cost of installing
the system and the cost of operating and maintaining the system. Installation costs vary depending
on foundation type, size and structure of the home, and initial radon level. Operating costs vary
depending primarily on whether there is an energy penalty resulting from the loss of conditioned air,
the price of electricity to operate active systems, and various maintenance requirements. Therefore,
operating costs are usually presented as a range.
'Additionally, when EPA considered setting the action levels as low as 2 pCi/L, the Agency recognized that over 60
percent of the homes that should be fixed when the action level was 2 pCi/L would be between 2 and 4 pCi/L. These
homes would only need to achieve a 50 percent reduction, or less, to meet that action level. Often they would probably
have reductions well below 2 pCi/L.
Chapter 4: Mitigation Technology Page 4-6
-------
From EPA's Private Sector Radon Mitigator Survey, the mean of the average reported
mitigation cost in 1989 was approximately $1,200 and the median was $1,100.4 Cited average costs
ranged from $50 to $5,000, but about 85 percent of the average costs reported were between $500
and $2,000. More recent EPA research (U.S. EPA 1991 b) suggests that it may be more appropriate
to consider $2,500 as the upper end of the range.
Exhibit 4-2 presents estimated ranges of installation and operating costs for various dominant
mitigation techniques. This exhibit is a result of the collaborative effort of EPA's ORP and ORD,
and is based largely on the results of past research. The costs shown are typical ranges for most
homes, although the actual cost of installing and operating a mitigation system in a home may be
more or less than the range indicated in the exhibit. For active sub-slab depressurization, the exhibit's
figures are further backed by a detailed study of the installation and operating exists of soil
depressurization techniques (U.S. EPA 1991). More details on different aspects of radon mitigation
are provided in Appendix F, which supports Chapter 5.
As shown in Exhibits 4-2, typical installation costs for radon mitigation techniques range from
$100 to $3,000. However, depending on the type of house, installation costs can be above the upper
end of this range. In addition to installation costs, most mitigation systems (except sealing) will have
annual operating costs ranging from around $70 to $700. The effectiveness of each system will
depend on the characteristics of the house to be mitigated.
''The question asked in the survey was "What are the average charges for the mitigation techniques you use in
detached houses?" EPA then calculated the mean and median of the responses.
Chapter 4: Mitigation Technology Page 4-7
-------
EXHIBIT 4-2
ESTIMATED INSTALLATION AND OPERATING COSTS FOR
VARIOUS RADON MITIGATION TECHNIQUES (1991 $f
Technique
Sub-slab Suction
(Sub-slab
Depressurization)
Drain-tile
Suction
Block-wall
Suction
Sump Hole
Suction
Sub-membrane
Depressurization
in Crawlspace
Natural
Ventilation in a
Crawlspace
Sealing of Radon
Entry Routes
House
.(Basement)
Pressurization
Natural
Ventilation
Heat Recovery
Ventilation
Typical
Radon
Reductions
80 - 99%
90 - 99%
50 - 99%
90 - 99%
80 - 99%
0 - 50%
0 - 50%
50 - 99%'
Variable
25 - 50% if
used for
full house;
25 - 75% if
used for
basement
Typical Range of
Installation
Costs
(Contractor)
$800 - 2,500
$800 - 1,700
$1,500 - 3,000
S800 - 2,500
51,000 - 2,500
$200 - 500 if
additional vents
are installed; $0
if no additional
vents
$100 - 2,000
$500 - 1,500
$200 - 500 if
additional vents
installed; $0 if no
additional vents
$1,200 - 2,500
Typical Operating Cost
Range for Fan
Electricity and
Heated/Cooled Air
Loss (Annual)
$75 - 175
$75 - 175
$150-300
$100 - 225
$50 - 175
May be some energy
penalties
None
$150 - 500
$100 - 700
$75 - 500 for
continuous operation
Comments
Works best if air can move easily in
the material under the floor slab.
Works best if drain-tiles form
complete loop around the house.
Only in houses with hollow block-
walls; requires sealing job of major
openings.
Works best if air can move easily to
sump under slab, or if drain-tiles form
complete loop.
Less heat loss than natural ventilation
in cold winter climates.
Costs are variable
Normally used in combination with
other techniques. Requires proper
materials and careful installation.
Works best with tight basement that
can be isolated from outdoors and
upper floors.
Significant heat and conditioned air
loss; operating cost dependent upon
utility rales and amount of ventilation.
Limited use; works best in a tight
house and when used for basement;
less conditioned air loss than natural
ventilation.
costs provided in this exhibit represent the range of typical costs for reducing radon levels in homes above 4 pCi/L down
to radon levels below 4 pCi/L. In most cases homes are reduced to an average of about 2 pCi/L.
Chapter 4: Mitigation Technology
Page 4-8
-------
CHAPTER 5
COST-EFFECTIVENESS ANALYSIS
This chapter presents an analysis of the
risk reductions and costs for the public if it ••^•••••••••••••••••••MBHHMHMMMM
followed the radon testing and mitigation advice 'Tim AIMS OF THIS CHAPTER
that EPA otters in the revised Citizens Guide to
Radon. The first section of the chapter TQ t m anafefe of
provides an initial overview ot the analysis. This reductions and costs of the testing
overview is followed by a section that describes , ... , - r^-n A
, ,J , . , , and mitigation advice EPA
the testing and mitigation options that the ,"" . ., ;,.
e . ,. to . . y ., , recommends to the public.'
Agency considered m revising the Guide as well
as the portion of housing stock and residential } Tf) ^^ an evaluation of the cosl,
population that would be targeted by each. The effectiveness of the options
next two sections analyze, in turn, the risk • , ,
,. , , ,. , , considered.
reductions and costs tor the public that result
from the options that EPA considered. The mmm^mmmm^mmm^mam^^**mm*m^m^*m*m
final section examines the cost-effectiveness of
each alternative.
Section I: Overview of the Analysis
This analysis estimates the annual risk reductions that would occur if the public followed the
program prescribed in the revised Citizen's Guide to Radon, the annualized costs the public would
incur, and the overall cost-effectiveness of the prescribed program as reflected by the cost per life
saved. The analysis considers three options, with each consisting of the basic testing approach
recommended in the revised Guide (Option C as defined in Chapter 3) coupled with different action
levels for triggering the mitigation of a home. EPA recognized that the other testing options
discussed in Chapter 3 also could be cost-effective in providing large risk reductions if the public
followed the advice it had provided, but believed that these options had other weaknesses as outlined
earlier that ruled out their selection. Following EPA's Science Advisory Board's concurrence on
using Option C, the Agency focused its cost-effectiveness analysis on the action levels it should chose
rather than on testing approaches.
The first option includes an action level of 4 pCi/L, which is the action level recommended
in the revised Citizen's Guide. However, as stated in several places in the revised Guide, EPA
recognizes that mitigation down to lower levels may be appropriate because radon levels less than
4 pCi/L still pose a health risk. Also, most homes today can be reduced to 2 pCi/L or below, and
Congress has set a long-terra goal that indoor radon levels be no more than outdoor levels, which are
typically below 2 pCi/L. As a result, the cost-effectiveness analysis also considers two other options
with lower action levels, 2 pCi/L and 3 pCi/L.
1The Agency also examined action levels of 8 pCi/L and 20 pCi/L. These action levels are discussed in Appendix H.
Chapter 5; Cost-Effectiveness Analysis Page 5-1
-------
The results for each option are compared to the figures that EPA's Guidelines for Performing
Regiilatoiy Impact Analysis (1983) suggest that the public is willing to pay to save "statistical lives."
The figures presented in these guidelines represent the value the public places on reducing risks of
death from all types of causes, given empirical evidence of the public's willingness to either pay to
reduce small risks or receive payments for accepting those risks. The results also are compared to
the cost-effectiveness of EPA's prior testing and mitigation advice in the original Guide and to the
cost-effectiveness of other health and safety programs.
The analysis examines seven major components for each option: (1) the population of homes
and people covered by the advice; (2) radon distributions in the affected homes; (3) the number of
homes that should mitigate given the effectiveness of radon testing; (4) the risk reduction associated
with lower radon exposure due to mitigation; (5) the unit costs associated with testing and mitigating
homes; (6) the total costs of testing and mitigation for the group covered; and (7) the costs per life
saved. The major steps in the analysis are summarized in Exhibit 5-1.
EXHIBIT 5-1
MAJOR STEPS IN THE COST-EFFECTIVENESS ANALYSIS
Step 1; Determine the Universe Covered by the Testing and Mitigation Advice Option
Step 2: Estimate the Radon Distribution in Affected Housing Units2'
Step 3: Calculate the Number of Housing Units that Should Mitigate Given the
Estimated Accuracy of Radon Testing
Step 4: Analyze the Annual Risk Reductions Resulting from Mitigation
Step 5: Determine the Unit Costs of Testing and Mitigation
Step fy. Develop the Total Annualized Costs of Testing and Mitigation
Step 7: Calculate the Cost per Life Saved
-See Appendix D for a description of the affected housing units.
Throughout the analysis, a 100 percent public compliance rate with EPA's advice on testing
and mitigation is assumed. There is no consideration of actions the public already has taken in
response to the 1986 Citizen's Guide and independent State programs.
Section II; EPA's Approach to Reducing Residential Radon
This section provides a brief description of the residential radon problem, which, as opposed
to radon problems in schools and other buildings, is the focus of the revised Citizen's Guide. It also
describes EPA's advice to the public in the revised Citizen's Guide on how to reduce the risks from
Chapter 5: Cost-Effectiveness Analysis Page 5-2
-------
radon exposure and defines the residential population (and housing stock) that the Guide is trying
to convince to take action to reduce radon exposures,
Residential Radon Problem
Radon can pose a significant health problem in any building where it can accumulate. Its
greatest risk to the public appears to result from its accumulation in homes, where the public is
estimated to spend about 75 percent of its time. Radon can emanate from certain building materials
or be released at harmful levels in the home from water taken from ground-water sources, but it
usually reaches elevated levels in homes due to soil gas entry through the home foundation. EPA's
revised Citizen's Guide focuses on testing for radon in homes and taking actions to reduee radon entry
or mitigate its effects. The Agency estimates that radon levels in the United States housing stock
lead to 7,000 to 30,000 lung cancer deaths in the residential population annually. EPA's central
estimate is that about 14,000 lung cancer deaths result each year from residential radon levels. The
vast majority of the exposure occurs in single-family homes, mobile homes with permanent
foundations, and the lower levels of apartment buildings and group quarters, such as college
dormitories. Chapter 2 describes EPA's risk estimates for residential radon in more detail and
Appendix B provides basic background information on radon.
Revised Citizen's Guide's Approach
In the revised Citizen's Guide, EPA advises homeowners initially to conduct a short-term test
in the lowest lived-in level of their homes. If the results of the test are above the established action
level of 4 pCi/L, homeowners are advised to obtain a follow-up measurement using either a short-
term or long-term test. The higher the initial short-term result, the greater the certainty that the
long-term average is also above the action level. With this in rnind, and to keep homeowners from
being unduly exposed to extremely elevated levels, EPA recommends that a short-term follow-up test
be used if the results of the first short-term test are above 10 pCi/L. If homeowners conduct a long-
term follow-up test, they are advised to mitigate their homes if the results of the long-term test are
at or above the action level. If homeowners conduct a short-term follow-up test, they are advised
to consider mitigating their homes if the average of the results of the two short-term tests is 4 pCi/L
or more. The revised Guide also notes that homeowners can further reduce their lung cancer risk
by mitigating homes that are below 4 pCi/L.
Homeowners are advised to mitigate their homes as quickly as possible. The mitigation
technology that a person chooses to install will depend on the type of housing unit, the number of
floors in the unit, the foundation type, and its interior space use. Most people will probably use one
of the major technologies such as active subslab depressurization, sealing/plugging, natural ventilation,
and heat recovery ventilation. Chapter 4 contains a more detailed discussion of these technologies.
In the revised Citizen's Guide to Radon, EPA establishes an action level of 4 pCi/L. The
action level, i.e., the radon level at which homeowners are advised to mitigate, effectively determines
the number of homes that should mitigate. Lowering the action level increases the number of homes
that should mitigate and thus increases the scope of the radon program as well as the national risk
reductions and total costs associated with the program. As described in the overview, this analysis
also examines two alternative action levels, 2 pCi/L and 3 pCi/L.
Chapter 5: Cost-Effectiveness Analysis Page 5-3
-------
Coverage of Revised Citizen's Guide
Exhibit 5-2 presents the total number of housing units and the total population that would
be affected by EPA's advice to test and fix homes for radon. In the revised Citizen's Guide to Radon,
all single-family homes, apartment units that are below the third floor, mobile homes with permanent
foundations, and units in group quarters (e.g., college dormitories, military barracks) that are below
the third floor should test for radon. EPA took the numbers of single-family homes, apartments in
multi-unit structures, and mobile homes that were used in the analysis of the radon testing population
from initial releases of 1990 Census data provided by the Bureau of the Census,2 EPA's radon
testing policy would lead to the testing of about 83 million housing units that are occupied by about
215 million people. This represents almost 83 percent of the housing stock that is regularly used and
about 87 percent of the residential population. Appendix D provides greater explanation of how the
estimates in Exhibit 5-2 were derived.
Section HI: Risk Reduction
The risk analysis calculated the risk reductions resulting from the mitigation of homes with
elevated radon. EPA estimated the annual risk reduction that occurred from each option examined
in terms of annual lung cancer deaths averted, i.e., annual lives saved. The Agency also calculated
annual deaths from lung cancer that are not prevented because of false negative test results (i.e.,
negative test results for a home that actually has a radon level above 4 pCi/L). The number of homes
mitigated, and thus the risk reductions that occur, depends on (1) the action level established by EPA
in each option, (2) the distribution of radon in housing units that test, and (3) the effectiveness of
the radon tests. It also critically depends on the effectiveness of mitigation techniques that reduce
radon levels.
This section initially describes the analyses conducted to determine how effective EPA's
testing approach would be in classifying homes correctly using testing Option C (defined in Chapter
3) at various action levels. This description is followed by a summary of the assumptions made about
the effectiveness of mitigation actions in lowering radon levels. The section closes with a explanation
of how EPA calculated risk reductions based on this information and provides the results.
Results of Radon Testing
According to EPA's testing advice, an initial short-term screening test should be conducted
in the lowest lived-in level of a home. If the results of the initial test are above the action level
established by EPA, either a short-term or long-term follow-up test should be conducted. To
determine the frequencies with which each type of test would actually be used by the public to make
the follow-up measurement, EPA conducted a study of the public's willingness to use long-term tests.
The study concluded that only 9 percent of the public were willing to use long-term tests (Johnson
1990). Accordingly, this analysis assumes that 9 percent of all follow-up tests will be long-term tests
and 91 percent will be short-term tests.
"EPA used 1990 Census data released in June 1991 (from Summary Tape File 1A) that provided official Census
disaggregated estimates of population characteristics and types of structures as of April 1990. The Census Bureau's
residential population estimate at that time was 248.7 million people, which by October 1990 was estimated to have grown
to 250 million people (the basis of Chapter 2's risk calculation). As of May 1992, the residential population is 255 million
people (Bureau of the Census, 1992b).
Chapter 5: Cost-Effectiveness. Analysis . Page 5-4
-------
9
'
I
i
i
t
EXHIBIT 5-2
1990 HOUSING UNITS AND RESIDENTIAL POPULATION
COVERED BY EPA'S RADON TESTING POLICY
(in 1000s)
HOUSING UNITS
Single-Family
Multi-Units
Mobile Homes
Group Quarters
TOTAL
Percent of All Units
Percent of Units Intended
for Regular Use/
Percent Population
Total Units*1
65,762
27,981
8,521
3,383
105,647
100%
Illll^illllliicSssI
Total Units
Intended for
Regular Use*'
62,461
26,569
8,067
3,383
100,480
95%
100%
Units that Should
Test for Radon-'
62,461
17,801
403
2,267
82,932
78%
83%
Coverage Criteria from
Radon Testing Policy
All single family homes
All units below 3rd floor
All units on permanent
foundation
AH units below 3rd floor
|l|t !lilli;l;lllllllilK?
SSSKS mrn:iKy?::mf:;:;•:->:•:-:-; :••-: ;•:-
100%
Radon Testing
Population
174,891
35,239
726
4,487
215,344
:;;:::::;::.::::::::::::gv:-:^:;^::;:^:::-::::::;:;^;:::;y;X::;:
87%
I
1
Note: Totals may not sum due to rounding.
* All estimates, except group quarters are from 1990 Census data. Number of group quarters estimate based on assumed occupancy of about two persons per unit and
1990 Census estimate of persons living in group quarters. Mobile homes also include relatively small numbers of trailers and "other" units that could not easily be removed
from the estimate.
^ All estimates derived from available 1990 Census data on numbers of total units and excluding units with vacancies that were due to seasonal, occasional, and "other"
(undefined) uses. All these cases were assumed to be situations in which owners would not test their units since they were not intended to have regular usage. Vacant
units for rent or sale for year-round use were covered and used in estimating the units that should test in the next column.
- Estimates of units that should test based on EPA testing policy (i.e., that anyone living in detached houses, including mobile homes with permanent foundations should
test for radon (U.S. EPA/Office of Public Affairs 1988)). Applicable criteria appear in the next column.
* Occupancy rates were derived from published and unpublished Census data and assumptions about similarities in unit usage between structures that did and did not
have data available for direct computations of occupancy rates. These rates were used to estimate the population residing in each type of unit.
-------
EPA generated distribution tables for each action level based on testing devices used, number
of home floors, and testing conditions,3 (For a more detailed explanation of radon tests, please refer
to Chapter 3.) Based on 1990 Census data and data from the Characteristics of New Housing: 1990
Current Construction Reports (Bureau of the Census 1991), the weights for single floor vs. multi-floor
homes were determined to be 46 and 54 percent, respectively. The same approach that was used to
derive the mitigation classification rates in Chapter 3 was also used here, except the focus was on
options that had different action levels combined with the same Option C testing approach.
Exhibit 5-3 presents the weighted testing distribution for the action level of 4 pCi/L. The
distribution is based on the actual annual average radon level in the home and the radon level
EXHIBIT 5-3
JOINT PROBABILITY DISTRIBUTION OF RADON TESTING RESULTS
FOR AN ACTION LEVEL OF 4 pCi/L
Intervals of
Actual Annual
Average Radon Level
(pCi/L)
0-1
1-2
2-3
3-4
4-6
6-8
8-10
10-20
>20
TOTAL^
Radon Level According to Test Results (pCi/L)
0-1
1-2
2-3
3-4
SKSHSwSS
iimi
llllil
KwXwXv.x-x-;
miiii
FALSE NEGATIVES
0.0007
0.0001
0.0000
0.0000
0.0000
0,5493
0.0036
0.0005
0.0001
0.0000
0.0000
0.1899
0.0061
0.0011
0.0003
O.OOOi
0.0000
0.1114
Illlltff
llllil
• 2.2%
0.0075
0.0016
0.0005
0.0002
0.0000
0.0726
4-6
6-8
8-10
10-20
>20
FALSE POSITIVES - 3.6%
0.0035
0.0060
0.0058
0.0052
0.0012
0.0021
0.0025
0.0029
0.0005
0.0008
0.0010
0.0013
0.0004
0.0006
0.0009
0.0013
0.0000
0.0000
0.0001
0.0001
Mill:
ilHIll
mil
iiilii
iiiii
0.0318
iilili
illiil
llllfl;!
Iliiiil
0.0181
111111
ilfflli
iliiHi
0.0101
lfi!li
llllll
il!l!
iiilill
0.0139
litii
isiiii
»;->:w»:vX-;-'v
f|||lti
ilil!
0.0030
TOTALS'
0.6198
0.2074
0.0717
0.0382
0.0373
0.0121
0.0067
0.0062
0.0006
1.0000
-Totals may not sum due to rounding. Very small values rounded to zero in the exhibit.
3The Agency generated distribution tables using information in a report entitled 'An Evaluation of the Performance of
Alternative Short-Term Radon Testing Procedures in Homes with Pending Real Estate Transactions'. Chmelynski, H. 1992.
Chapter 5: Cost-Effectiveness Analysis
Page 5-6
-------
indicated by the test results. From this exhibit, it is apparent that close to 8 percent of the housing
stock should mitigate their homes based on their test results (i.e., 3.6 percent with false positive
results plus 4.0 percent with true positive results). About 53 percent of these houses have actual true
positive results; 47 percent of the homes actually have false positive results. Very often homes with
false positive results also will receive risk reductions when mitigated and, therefore, substantial
benefits are still derived from fixing homes below the action level. (This is discussed further in the
next section on mitigation effectiveness.) The Agency also prepared unique distributions for the
other action levels. From these distributions, EPA calculated the fractions of the population that
obtain false positive, true positive, false negative, and true negative test results.
Exhibit 5-4 presents estimates of the homes testing, retesting, and mitigating at various action
levels, as well as the population in each set of these homes. Of the 83 million homes that should
initially test, about 10 million homes (12 percent) would be expected to retest using EPA's selected
action level of 4 pCi/L. About 6 million of these homes will have results above the action level and
should mitigate (8 percent of homes initially testing). Close to 17 million people live in the homes
that should mitigate. This is about 8 percent of the occupants in homes that should test and close
to 7 percent of the residential population of the U.S.
EXHIBIT 5-4
ESTIMATES OF HOMES TESTING AND MITIGATING
AT SELECTED ACTION LEVELS AND RESIDENTS IN THOSE HOMES (in 1000s)
HOUSING UNITS
Initial Testing
Retesting
Mitigating
Action Level (pCi/L)
2
3
4
RESIDENTS
82,900
82,900
82,900
Initial Testing
21,900
14,400
10,000
Retesting
16,000
9,700
6,400
Mitigating
Action Level (pCi/L)
2
3
4
215,300
215,300
215,300
57,000
37,300
26,000
41,600
25,200
16,500
This distribution reflects the results expected to be obtained using the testing advice in the 1992 Citizen's Guide.
Therefore, it includes homes that have tested once and received a result below the action level and homes that have tested
twice. Homes that retest with a short-term test average the results of the two tests. Homes that retest with a long-term test
use only the results of the long-term test since long-term tests are assumed to always result in correct classification of homes.
Chapter 5: Cost-Effectiveness Analysis
Page 5-7
-------
Effectiveness of Mitigation
For the purposes of estimating the benefits (risk reduction) of various policy decisions, this
analysis assumes that homes with annual average radon levels above 2 pCi/L that mitigate will reduce
radon levels to a post-mitigation annual lived-in area level of 2 pCi/L.5 This is a level of radon
mitigation that EPA believes most homes can achieve on average, as discussed in Chapter 4.
Homes with average radon levels below 2 pCi/L that mitigate (due to false positive test results) are
likely to receive some reduction in radon levels, but the amount cannot be quantified. Therefore,
the Agency conservatively assumed no reduction in risks. The Agency estimated radon exposure for
the testing population using results from EPA's National Residential Radon Survey (U.S. EPA/Office
of Radiation Programs 1991 a), EPA also derived the mean radon exposure for each interval in the
testing population distribution using data from the Survey. The exposure reduction achieved in each
interval from mitigation was then calculated using Equation 1.
Radon Exposure ,, „ , „ .
n , f . Mean Radon Post rii
Reduction m = r . r . , - , ,.„. _. T , W
„.,, , Exposure in Interval Mitigation Level
Calculation of Risk Reductions
This analysis assumes that the reduction in risk associated with a reduction in radon exposure
.of 1 pCi/L was 43.2 lives saved annually per million people receiving the reduction. This risk factor
was derived from EPA's risk assessment methodology that is explained in Chapter 2. The central
estimate of the radon risk factor is used in this analysis, rather than the upper bound risk estimate.
The upper bound estimate would have increased all the risk estimates by about 2.5 times and
improved later cost-effectiveness estimates by about 60 percent. Using the central risk factor, the
number of radon deaths averted per million people was calculated for each radon concentration
interval as shown in Equation 2.
Radon Deaths Averted Radon Exposure ', -,
. , . , D j .- ft-IT w * i R^k Reduction o\
in Interval = Reduction in pCifL Interval x „ W
Per Million People (From (1))
'Mitigation is assumed to lower homes to an average of 2 pCi/L, regardless of the action level.
"This assumption is based on review of the results from post-mitigation testing, which is often conducted in the lowest
lived-in level in a home. Therefore, it could potentially lead to a conservative estimate of the average reduction of radon
throughout the home (i.e., it could lead to an underestimate of the risk reduction). In Appendix G,.the sensitivity of this
assumption is examined.
7EPA converted its central risk factor (2.24 x 104 lifetime risk) into an annual risk reduction factor for homes based on
a conversion of 0.193 working level months per pCi/L and assuming an average life expectancy of 74, years.
Chapter 5: Cost-Effectiveness Analysis . Page 5-8
-------
Annual lives saved (due to both true positive and false positive test results) and annual lives
lost (due to false negative test results) were then calculated for each radon concentration interval
using the weighted testing distribution and the testing population as shown in Equation 3.
Annual Lives
Saved (Lost)
in pd/L Interval
Testing
Population
(in Millions)
Fraction
of Population
in pdjL Interval
Radon Deaths Averted
in Interval
Per Million People
(From (2))
(3)
Exhibit 5-5 presents the results of the risk analysis for all three action levels and provides an
estimation of the percentage of risk that is above each action level.9 The estimated 2,200 lives saved
annually with an action level of 4 pCi/L is roughly 17 percent of the lives that could potentially be
lost due to residential radon exposure.10 The action levels of 2 pCi/L and 3 pCi/L would reduce
the estimated number of lung cancer deaths by 23 percent and 19 percent, respectively. More detail
on how the risk assessment was conducted is provided in Appendix E.
EXHIBIT 5-5
ANNUAL LIVES SAVED AND NOT SAVED
UNDER ALTERNATIVE ACTION LEVELS FOR MITIGATION
Action Level
2 pCi/L
3 pCi/L
4 pCi/L
Percentage of Pre-
Mitigation Risk
Above Lhe Action
Level
55%
42%
32%
Annual Lives Saved
(Due to True and False
Positive Test Results)
3,100
2,600
2,200
Annual Lives Not Saved
(Due to False Negatives)
460
750
740
^Equation 3 is used to calculate both annual lives saved and annual lives lost. If the lesl results are true or false positives,
lives are saved due to the reduction in radon exposure. If the results are false negatives, lives are lost due to no reduction
in radon exposure.
'The results in Exhibit 5-5 are based on a central estimate of 13,430 radon-induced lung cancer deaths per year. This
estimate was developed using the Census Bureau's residential population figure of 248.7 million people as of April 1990,
which is lower than the figure of 250 million people as of October 1990 used in Chapter 2.
10The reduction is lower than the percentage of risk that exists in homes above 4 pCi/L because mitigation will, on
average, lower homes to 2 pCi/L, and because false negative results prevent mitigation from fixing all homes above 4 pCi/L.
(Additionally, fixing many homes with false positive results does lead to significant levels of risk reduction.)
Chapter 5: Cost-Effectiveness Analysis
Page 5-9
-------
Section IV; Costs
EPA estimated the costs of testing and the costs of mitigating of homes based on the most
recent information available, and employed basic assumptions that are consistent with those used in
the risk analysis. The Agency calculated all costs as total present value costs. Annualieed costs were
then derived to show the total costs of each alternative in annual constant dollars over the time
period of the analysis. All testing and installation of mitigation systems were assumed to occur
immediately. The time period used for the analysis to estimate operating and maintenance (O&M)
expenses was 74 years. This is actually the average life expectancy of the U.S. population, today and
a time period representative of at least the average life of a home,11 The Agency chose this time
period as an analytic convention to make the analysis comparable to other EPA analyses that estimate
cancer cases averted based on an assessment of the lifetime risks and then convert those estimates
into annual rates of cancer incidence avoided. A three percent discount rate was used to calculate
both the present value of the total costs and the annualized costs.
Testing Costs
The analysis assumes that the average initial homeowner testing cost is $28. This is based on
the assumption that half of the homeowners would use charcoal canister tests (approximately $20 per
test) and half would use alpha track detectors (approximately $35 per test).1"3 For homes that
retest, the cost of a second short-term test was again assumed to be $28. Long-term tests were
estimated to cost homeowners $100 per test. Based on a study conducted by EPA on the public's
willingness to use long-term tests (Johnson 1.990), this analysis assumes that short-term tests are used
in follow-up testing 91 percent of the time and long-term tests are used 9 percent of the,time.15
The number of homes that test initially is assumed to be all homes covered by the revised Citizen's
Guide advice, i.e., approximately 83 million homes. The number of homes that retest was calculated
from the distribution of radon in the affected population using Equation 4.
>r , ,. „ Fraction of Number of
Number of Homes TT ~, . , ... „ ™ _^ , ~ „ , (&\
Th t K t t = Homes That Initially Test x Affected W
Above Action Level Homes
"Examination of housing data on demolition rates suggests that the average home life is longer than 74 years.
12The three percent discount rale is based on the opportunity cost of money to the homeowner recognizing that most
of the expenses will result from yearly operating and maintenance expenses.
15Some homeowners will use more expensive devices, especially during real estate transactions. The assumption here is
that the vast majority of homeowners will use charcoal canisters or alpha track detectors, since CRMs and electrets aren't
"do-il-yourself devices.
MThis assumption is based on homeowners using an alpha track detector over a long period of time in a single location
or multiple locations; using charcoal canisters over several seasons; or having tests conducted by a contractor. This estimate
is meant to be an overall average cost of follow-up by the 9 percent of homes retesting with a long-term test, which is a very
modest percentage of total costs. The testing costs rover actions of homeowners and do not include additional tests mitigators
may perform on site while they fix a home.
15The sensitivity analysis presented in Appendix G looks at the annual lives saved and cost incurred if 100% of the public
used long-term confirmatory, tests.
Chapter 5': Cost-Effectiveness Analysis Page 5-10
-------
Total initial testing and retesting costs were calculated as shown in Equation 5 and Equation 6.
T * r i •+ • i Number of ~ , ,.
Total Imttal m Homgs ^ Cos of (5)
Testing Cost Mtially Test Initial Test
TI ^ i Number of ~ . /. ,-, _ /.
Tbtoi! „ J , Cost of Q1<_ Cost of Q<_.
Retesting Cost = ' * Short-term Test x /e Long-term Test x ^
Mitigation Costs
The costs of radon mitigation were calculated using a model developed by EPA. The model
includes both upfront costs of system installation and operating and maintenance costs. The model
provides a weighted average cost of mitigation based on characteristics of the housing stock such as
foundation type, radon levels in the homes to be mitigated, and mitigation methods employed. The
model produces different weighted average costs for each action level. For an action level of 4 pCi/L,
the model produces a lifetime (74 years) weighted average cost of about $6,550 per home. About
$1,500 of this expense (close to 25 percent), is for upfront costs of correctly installing systems. Close
to $1,250 of this upfront cost is direct installation of the system and trie remainder is for price
estimation and post-mitigation activities conducted to ensure proper system operation. The largest
portion of mitigation costs results from O&M costs and is roughly $5,050, or about 75 percent of the
total costs. The majority of this cost comes from active systems that use electricity and require
homeowners to pay for increased ventilation in the home or additional space conditioning because
of exhalation of air beneath the foundation due to depressurization efforts in subslab systems.
A similar upfront and O&M cost relationship exists in the costs for the lower action levels.
The total weighted average costs in those cases are $6,150 and $6,350 for 2 pCi/L and 3 pCi/L,
respectively. The lower costs are due to an increasingly larger percentage of homes at lower radon
levels that enter into the calculation of the weighted average costs. In the majority of situations
(especially in homes using ASD), the costs of reducing radon levels in homes above all three action
levels down to 2 pCi/L is about the same. However, mitigations in the smaller set of homes that use
sealing and ventilation technologies cost less at the lower action levels, because the mitigation systems
need to provide lower percentage reductions in radon levels.
An explanation of how the model works is provided in Appendix F. The model was designed
to estimate costs for mitigation contractors for single-family detached homes (nearly 69 percent of
the homes covered). This analysis assumes that mitigation costs for other types of housing units are
the same as for single-family detached homes. This may overestimate the costs for these other units,
since it fails to recognize economies of scale and other factors in mitigating single-family attached
homes, multi-units, group quarters, and mobile homes. Of the housing stock that should be tested
for radon, these units represent 7 percent, 21 percent, 3 percent, and 0.5 percent, respectively. The
analysis assumes that all radon mitigation system installations occur immediately and that the
mitigation systems operate over the 74-year period of the analysis. In addition to routine annual
operating costs, biennial testing and repair and replacement costs are considered in developing
lifetime home costs (U.S. EPA/Radon Division, 1991), The cost model does not make a distinction
Chapter 5: Cost-Effectiveness Analysis Page 5-11
-------
between homes that are mitigated due to true positive test results and homes that arc mitigated due
to false positive results. The number of homes that are mitigated was calculated using Equation 7.
,7 , ,. „ Fraction of • Number of
Number of Homes „ ru • £ * ~* AJ* + j
•n * »>•*• * = Homes That Retest x Affected
That Mitigate A, , .. , , „
0 Above Action Level Homes
(7)
Total mitigation costs were calculated using Equation 8.
Total
Mitigation Costs
Number of Weighted Average
= Homes x Cost of
Mitigated Mitigation
(8)
Total Costs
Exhibit 5-6 presents the results of the cost analysis for all three action levels for testing and
mitigation. It shows that if the public fully followed EPA's advice with an action level of 4 pCi/L, the
annualized cost is about $1.5 billion. About 6 percent of the cost would be for testing and about 94
percent cost would be for mitigation. About 87 percent of the testing costs is incurred in the initial
round of testing.
EXHIBIT 5-6
NATIONAL COSTS OF ALTERNATIVE ACTION LEVELS FOR MITIGATION
(1000s of 1991 $)
Action Level
Initial Testing Cost
Retesting Cost
Mitigation Cost
Total Cost
Total Present Value of Costs
2 pCi/L
3 pCi/L
4 pCi/L
2,322,000
2,322,000
2,322,000
757,000
495,000
346,000
98,171,000
61,726,000
41,839,000
101,250,000
64,542,000
44,507,000
Annualized Costs
2 pCi/L
3 pCi/L
4 pCi/L
78,000
78,000
78,000
26,000
17,000
12,000
3,317,000
2,086,000
1,414,000
3,421,000
2,181,000
1,504,000
Chapter 5: Cost-Effectiveness Analysis
Page 5-12
-------
Section V: Cost-Effectiveness
EPA has used cost per life saved (cost per death averted) as a measure of effectiveness of
the revised Citizen's Guide in reducing risks to the public. This section discusses how EPA calculated
cost-effectiveness and provides the results for the alternative action levels considered in revising the
Citizen's Guide. The section then evaluates these results in three different ways, by (1) relating the
results to the public's willingness to pay to save a statistical life, (2) comparing the cost per life saved
of radon reduction to the cost per life saved of other safety and health-related programs, and (3)
comparing the cost-effectiveness of the approach taken in the revised Citizen's Guide to that of the
1986 (original) Guide. The section also analyzes the sensitivity of the results to major assumptions
and parameters used in the analysis.
Cost per Life Saved of Selected Approaches to the Revised Guide
The average cost per life saved for each alternative was calculated using Equation 9.
Average Cost per _ Total Annual Lives
Life Saved AnnuaUzed Cost Saved
Exhibit 5-7 presents the average cost per life saved for all three action levels. The exhibit also
presents the incremental cost per life saved.16 The incremental cost per life saved is the added cost
per life saved of each increment of risk reduction that can be gained when a lower action level is
selected.17 The results of the cost per life saved calculations for each action level are best
evaluated in relationship to the value the public places on risk reductions. It is also helpful to look
at the cost-effectiveness of other health and safety programs. Simply comparing one action level
against another in the analysis can lead to misleading conclusions. For example, although an action
level of 2 pCi/L is the least cost-effective option analyzed here, an EPA radon program based on that
level could be more cost-effective in protecting public health than other governmental actions to
reduce other forms of risk. It could also be providing a risk reduction at a cost the public is generally
willing to pay for safety and health measures. Therefore, it is important to evaluate the results in
Exhibit 5-7 in the broader context of what government can do overall to reduce risks.
In addition, the incremental cost per life saved is a very valuable tool in evaluating the cost-
effectiveness of an option because it clearly assesses the value of the addiliona! cost incurred by
expanding the program to gain additional risk reduction. In evaluating the cost-effectiveness of the
options analyzed here, the incremental cost per life saved for each option is best compared to the
value the public places on saving a statistical life.
lfr]1ie cost per life saved of additional action levels is provided in Appendix H.
"For example, the incremental cost per life saved of an action level of 3 pCi/L is the increased annual cost of moving
from an action level of 4 pCi/L to 3 pCi/L ($677 million) divided by the increased annual risk reduction (about 400 lives
saved) -- an incremental cost per life saved of $1.7 million. Alternatively, the average cost per life saved is the direct division
of the annuitized cost of a 3 pCi/L action level ($2.18 billion) by the annual lives saved (2,600 lives saved) - an average cost
per life saved of $0.8 million.
Chapter 5: Cost-Effectiveness Analysis Page 5-13 -
-------
EXHIBIT 5-7
COST PER LIFE SAVED
UNDER ALTERNATIVE TESTING AND MITIGATION PROGRAMS
Action Level
2pCi/L
3 pCi/L
4pCi/L
Number of Lives
Saved Annually
3,100
2,600
2,200
Annualized Cost
(1000s of 1991$)
$3,421,000
2,181,000
1,504,000
Average Cost
per Life Saved
(1000s of 1991$)
$1,100
800
700
Incremental Cost
per Life Saved
(1000s of 1991$)
$2,400
1,700
700^
-Based on assumption that "no action" was the alternative EPA had to this action level.
Note: The central estimate of the radon risk factor is used in this analysis, rather than the upper bound risk estimate. The
upper bound estimate would have increased all the risk estimates by about 2.5 times and reduced the cost-effectiveness
estimates by about 60 percent.
Public Willingness to Pay to Reduce Risks
The cost per life saved indicates how cost-effective an alternative is in providing health
benefits to the public. It should be compared to the public's willingness to pay to save a "statistical
life," Le., buy risk reductions. In the past, EPA's 1983 Regulatory Impact Analysis (RIA) Guidelines
(U.S. EPA 1983) have indicated that the public appears to value a risk reduction that saves a life (in
statistical terms) for between $600,000 to $9,900,000 in 1991 dollars. A more recent study (Fisher
et al. 1989) suggests that the public places the value of a statistical life between $2,000,000 and
$10,500,000 in 1991 dollars.18
All three alternatives analyzed here have average cost per life saved values well below the
upper end for the value of a statistical life. In fact, when compared to 'the most recent study's results
of the public's willingness to pay, all three options provide an average cost per life saved that is well
below the lower end of the range.
Cost per Life Saved of Other Programs
The cost per life saved calculations for the alternatives examined in this analysis can be
compared to the cost per life saved of other health-related programs in order to evaluate how cost-
effective the radon program-is in protecting public health relative to other governmental actions
taken to reduce risk. Its value as an aid to decisionmaking is based on the premise that given the
limited availability of resources, public agencies should try to use societal resources to purchase the
greatest risk reductions possible for the public. The lower the cost per life saved of actions taken,
the more risk reduction the government could provide for the resources society can spend for health
and safety protection.
ibis study was conducted by the same researchers who provided EPA with the range in the 1983 RIA Guidelines.
Chapter 5; Cost-Effectiveness Analysis
Page 5-14
-------
Exhibit 5-8 shows the cost-effectiveness of other health and safety programs that has been
reported in various sources. The effectiveness measures are cost per life saved. Another similar
measure that is often used to evaluate the cost-effectiveness of government programs is the cost per
cancer case avoided, when there is less certainty about the percentage of the time the cancer cases
involved will lead to fatalities.19 EPA has calculated this latter measure for many of its regulatory
decisions because Agency risk estimates routinely report cases avoided, rather than fatalities. An
" EPA-sponsored review of EPA decisions over the -last decade shows that the Agency has often
decided on regulatory controls that cost several million dollars per cancer case avoided (Travis et al,
1987).
EXHIBIT 5-8
COST-EFFECTIVENESS ESTIMATES FOR SELECTED
HEALTH AND SAFETY PROGRAMS2''
Federal Program Area
Medical Screening/Care
Highway Safety
Air Transportation Safety
Occupational Safety
Range of Cost Per Life
Saved/Case Avoided
(1000s of 1991$)
63 - 510 per life saved
100 - 3,300 per life saved
100 - 1,600 per life saved
100 - 74,000 per life saved
Source
Cohen 1987
OMB 1991
OMB 1991
OMB 1991
-These values represent the ranges for regulatory decisions and health programs that have been reported in the literature
from the most authoritative sources found. They are not al! inclusive and other sources may provide different ranges because
of differences in analytic approaches and year in which studies were completed.
The studies that have provided these estimates could have significant differences in their
methodologies for calculating risk reductions and costs. Therefore, care must be taken in comparing
these results to each other and to the results for the radon program. For instance, most assessments
of the benefits from EPA rules are based on estimates of risk reductions that use conservative risk
factors analogous to the upper bound estimate for radon. If the upper bound estimate for the risk
factor had been used in this analysis, the average cost per life saved would have been reduced by
about 60 percent.
However, simple direct comparisons of the range of results in Exhibit 5-8 and the review of
EPA's own past regulatory experience to the results for the revised Citizen's Guide provide a general
sense of how cost-effective the radon program is. For example, the average cost per life saved for
Often in occupational health, food safety, and environmental protection analyses, only toe incidence of cancer cases
rather than death is calculated, because the target site and specific cancer mortality rate are unknown. In general, one in
every two cases of cancer results in a fatality.
Chapter 5: Cost-Effectiveness Analysis
Page 5-15
-------
an action level of 4 pCi/L is approximately $700,000. This falls well within the range of the cost per
life saved of highway safety, air transportation safety, and occupational safety programs and
regulation. It is slightly higher than the upper bound cost per life saved of medical screening and
health care. The $700,000 cost per life saved also compares favorably to the expenditures that EPA
has been willing to see incurred for risk reductions in the past.20
Comparisons to the 1986 Citizen's Guide
EPA also examined the risk reductions, costs, and cost-effectiveness of the 1986 Citizen's
Guide and compared its findings to what could be expected under the revised Citizen's Guide. The
original Guide called for a short-term test in the lowest "livable space" and general follow up with a
long-term test. If the long-term lest results were greater than the action level of 4 pCi/L, mitigation
was recommended.
EPA believes that about six percent of homes that should be tested had actually been tested
as of July 1991. EPA estimates that this has led to the saving of about 40 lives per year, with an
annualized cost of close to $40 million. The cost per life saved, therefore, has been close to $900,000.
The cost per life saved under the revised Guide is about 25 percent less. The main reason for this
reduction is that following the new EPA testing advice reduces the chances of false positive results.
Homes with false positive results often will install mitigation systems that could provide only relatively
small, if any, risk reductions.
Sensitivity Analysis and Examination of Higher .ActionLevels
EPA examined the sensitivity of the results of its cost-effectiveness analysis to the assumptions
it used for major parameters of the analysis. This analysis covered the action level that EPA selected
for the revised Citizen's Guide (action level of 4 pCi/L,) The results of changing major parameters
within the reasonable range of potential values that they could have still suggest that the radon
program should be cost-effective. The results of the analysis are provided in Appendix G.
In keeping with the Congressional mandate in IRAA to consider lowering the action level for
radon mitigation, the Agency has focused its attention on action levels of 4 pCi/L and lower. This
also is in keeping with EPA's mission as a public health agency to reasonably minimize risks to the
public. EPA believes it is reasonable to focus on these lower action levels since higher action levels
would remove a relatively smaller portion of the public's residential radon risk. Furthermore, having
an action level of at least 4 pCi/L was determined to be incrementally cost-effective early in the
Agency's deliberations on the Citizen's Guide. Appendix H provides a summary of the analytic results
on which EPA has based its conclusions.
'"'The cost-effectiveness of environmental controls is in terms of cancer cases avoided and not in terms of lives saved.
Any comparison of cost-effectiveness between the radon program and other environmental programs should take this
difference into consideration.
Chapter 5: Cost-Effectiveness Analysis Page 5-16
-------
CHAPTER 6
RISK COMMUNICATION
Section I: Introduction
Two conditions must be met for any successful voluntary public health program: (1) the
public must believe that the behavior or substance in question poses a significant personal threat, and
(2) the solution must be simple enough for the public to be willing to adopt. The goal of the
Citizen's Guide is to stimulate public action by convincing people that radon poses a serious health
threat that can be mitigated relatively easily. Public action entails testing for the presence of radon
and mitigating if necessary. Because it is
difficult to persuade people to take voluntary "^"^•""""""•"•"""•""^^^""""""""•"""^•i
action, it is critical to find and use the best THE AIMS OF 'fins CHAPTER
possible risk communication strategies. Risk
communication research has helped EPA (J) To summarize major results of radon
develop and refine the communication strategies risk communication research.
used in the Citizen's Guide to ensure its success
in educating the public and stimulating testing. ^ To exp]ain how these resu}ts? along
with public comments on risk
Risks associated with high radon levels communication issues, have been
in homes began receiving national publicity integrated into the final revised
about six years ago. Since then, a number of Citizen 'f Guide
researchers have studied risk communication
issues associated with radon (e.g., the reactions HHBM>HHMH|^B^HIHHHMHHMHBHBMI|HH|^B1^aBHBMHIMi
of homeowners to information about radon
risks). Most of this growing body of research has been sponsored by EPA's Office of Policy, Planning
and Evaluation (OPPE) in cooperation with universities, States, and others. These studies include
risk communication testing of the Fall 1990 draft Guide.
When the original (1986) Citizen's Guide was prepared, a major concern was avoiding public
panic. Accordingly, the tone of the GuMe was factual and informative. This approach, however,
resulted in public apathy, not panic. EPA estimates that about 5.4 million homes have been tested
for radon as of July 1991. This figure represents approximately six percent of the occupied homes
in the United States that should test (including single family homes, mobile homes with permanent
foundations, apartment units below the third floor, and units in group quarters such as college
dormitories that are below the third floor). Furthermore, of the small percentage of those who test
for radon, many do not mitigate. EPA estimates that between 140,000 and 170,000 homes in the U.S.
have been mitigated by contractors as of July 1991.- These estimates represent less than three
percent of the occupied homes believed to have radon levels above 4 pCi/L.
1This estimate is based on an update of the National Residential Radon Survey (NRRS) June 1989 results. To update
the NRRS estimate, EPA conducted a telephone survey of large primary measurement firms to obtain information on the
number of tests performed since June 1989.
2This estimate is based on an update of the Radon Division's Private Sector Radon Mitigator Survey finding that almost
100,000 homes had been mitigated by contractors as of February 1990. The estimate as of July 1991 is based on information
collected by EPA staff in informal discussions with mitigation firms, the Regional Radon Training Centers, and States.
Chapter 6: Risk Communication Page 6-1
-------
Radon risk communication has provided useful insight into why public apathy exists and
suggests directions for overcoming it. Six key findings, listed in Exhibit 6-1, have emerged from the
available research that can be applied to the revision of the Citizen's Guide.
EXHIBIT 6-1
KEY RISK COMMUNICATION FINDINGS
1. Be prescriptive as well as informative.
2. Streamline guidelines on testing and mitigation to minimize barriers to public action.
3. Overcome public denial through the use of persuasive appeals such as concern for the
family.
4. Provide an appropriate level of radon information, since too much or too little
information may result in an undesired effect.
5. Personalize the radon threat with tangible, relevant comparisons to familiar risk.
6. Stress that radon problems can be corrected but do not overstate the ease of fixing
them.
The remainder of this chapter discusses the evolution of the current Citizen's Guide from the
standpoint of risk communication research and issues. The first section (Section II) reviews the major
research results as they relate to the six key findings above. The next section (Section III) describes
how this research was incorporated into the Fall 1990 draft Guide. The following section (Section
IV) discusses the major public reactions to the Fall 1990 draft Guide and what changes were made
to incorporate public comments into the final revised Guide. The concluding section (Section V)
reiterates the main themes of the chapter.
Section II: Research Findings and Implications
This section provides a brief summary of the research that supports each of the key findings
listed in Exhibit 6-1.
1. Be Prescriptive
The original Citizen's Guide was informative, emphasizing what was known about radon risks
in order to help the citizen make an educated decision to reduce risks by testing and mitigating if
necessary. Research has shown, however, that people are more responsive to information brochures
Chapter 6: Risk Communication Page 6-2
-------
that tell them exactly what they should do. A study of different radon information formats (Smith
et al. 1987), for example, found that, of several approaches tested, the "command/quantitative
approach" that combined clear directions for action with precise information seemed to perform best
overall, "Messages" that tell people clearly what to do are the most satisfactory, and when required
actions are not emphasized, the resulting uncertainty can delay or even prevent action. Smith et al.
(1987) also found that "people dislike uncertainty and may use it as an excuse for disregarding a
radon risk message that indicates a lack of complete understanding of a risk or disagreement among
experts." Similarly, Fisher and Johnson (1990) concluded that brochures that told people what to do
rather than asked them to make their own decisions resulted in greater satisfaction with the message
received.
2. Streamline Guidelines
As outlined in Chapter 3, the original Citizen's Guide recommended that individuals perform
(1) a short-term measurement in the lowest livable level of the home under "closed-house" conditions,
and (2) confirmatory tests in the lived-in levels if the result of the short-term test was above 4 pCi/L.
The recommended duration of the confirmatory test depended on the result of the short-term
measurement and varied from less than one week to a full year. The original Guide then provided
guidance on whether and how quickly citizens should take action to fix their homes based on the
confirmatory test results. This "test-test-fix" strategy is comprised of several steps that provide
multiple opportunities for homeowners to "drop out" of the process. Compared to actions required
to protect against other household risks, such as fire and theft, radon-related protective action
recommended in the original Guide was a "multi-faceted, technical, multi-person, multi-skilled,
.recursive process" (Doyle et al. 1990) that deters action to reduce exposure.
Research has shown that it is very difficult to convince people to take even the first step to
test their home. Weinstein, Sandman, and Klotz (1987), for example, investigated a random sample
of homeowners living in or near the Reading Prong area in New Jersey and their responses to the
radon threat. The researchers found that few people in the sample had tested or planned to test
their homes for radon, even though they lived in a known high risk area. In accordance with the
recommendations in the original Guide, this hurdle essentially had to be overcome twice — once for
the initial test and again for the confirmatory test.
Research also indicates that people are especially unwilling to take a second measurement
if it means having to wait for long-term test results. For example, the Wirthlin Group's February
1990 telephone survey found that only 9 percent of the participants were willing to conduct a one-
year radon test and that 27 percent were willing to wait no more than two days for test results
(Johnson 1990). Fifty-five percent of respondents preferred a two-day test to longer-term tests. In
a similar study, researchers found that 60 percent of all respondents preferred a two-day test to
longer-term tests, 37 percent of all respondents stated that two days is the longest time that they
would be willing to wait for test results, and only 20 percent were willing- to wait for one year
(Bruskin Associates 1991).
Other data confirm that long-term follow-up testing protocols are ignored by the public.
Radon industry representatives maintain that despite EPA's recommendations, people arc not
conducting long-term tests. According to a senior representative from the American Association of
Radon Scientists and Technologists, less than 5 percent of all testing devices sold are used for long-
term testing (Walker 1991). People who have the initiative to begin a long-term test often misplace
the measuring device, lose the mailer, or forget about the test altogether over the course of the year.
Chapter 6: Risk Communication Page 6-3
-------
The experiences of State radon programs also demonstrate the public's lack of interest in long-term
testing. For example, the Pennsylvania Department of Environmental Resources actively advertises
free long-term follow-up tests for all citizens with screening test results of 20 pCi/L or higher, yet
virtually no one requests the free tests (Walker 1991).
In addition, informal telephone surveys of measurement companies have found that most long-
term testing devices are purchased for government studies, not by private citizens. Due to the lack
of demand for long-term testing devices, only a small percentage of testing companies offer them for
sale. As of July 9, 1991, only 120 testing firms out of 711 applying to the latest round of the Radon
Measurement Proficiency Program market devices that can be used for long-term testing, and many
of the devices sold by these firms may also be used for short-term testing (Walker 1991).
Finally, there is conflicting evidence about whether people who have tested for radon are
inclined to fix their home. Some studies indicate that some people may drop out of the process
before taking this final step. Doyle et al. (1989), for example, analyzed the effectiveness of the mass
media radon information and testing campaign conducted in Washington, D.C. in 1988. A grocery
store chain that teamed up with a Washington, D.C. TV station in its "Radon Watch" campaign could
not keep monitors in stock at a bargain price of $4.75. However, only about half of the buyers
actually performed the test and only a fraction with elevated levels did any credible mitigation.
Researchers estimated that, of those people in the target population that tested their homes and
needed to mitigate, only about 8 percent actually mitigated.
Other studies, however, provide evidence that once people have tested their home, they are
likely to mitigate if necessary. Weinstein et al. (1988), for example, surveyed 123 New Jersey
homeowners who discovered at least two years in advance of the study that they had elevated radon
levels in their home. Of those whose initial first-floor readings were 20 pCi/L or more, 93 percent
made home modifications. Seventy-one percent of those with readings between 8 and 20 pCi/L and
62 percent of those with readings between 4 and 8 pCi/L also modified their homes. Additionally,
a June 1990 survey of 300 people who had tested their homes found that, of the 135 who had radon
levels above 4 pCi/L, 24 percent had acted to reduce radon levels and the remaining 76 percent
indicated that they planned to act, either by mitigating or by doing a follow-up test (Wagner and
Dickson 1991).
3. Use Persuasive Appeals to Overcome Denial
Like many other public health and safety issues (e.g., seat belts) the radon issue suffers from
people's belief that "it can't happen to me." There is a significant amount of research supporting the
principle that people tend to deny that radon could be a problem in their home or their community
(e.g., Desvousges and Kollander 1986, Smith et al. 1987, Weinstein et al. 1987, Weinstein et al. 1989,
Leferman Associates, Inc. 1989, and TBWA 1990). Smith et al. (1987) evaluated how homeowners
in New York State responded to different informational brochures. The researchers found that
participants believed that their personal radon risks were lower than general population-risks of
radon. Weinstein et al. (1989) evaluated the responses to two public opinion surveys conducted in
1988 and 1989. Respondents to these surveys claimed that radon was less likely to be a problem in
their own homes than in a home selected at random in their community. Even in high risk areas,
residents tended to see a radon problem in their homes as fairly unlikely.
Research also shows that people use information about radon to rationalize why they do not
have a problem and to create excuses for inaction. For example, when told that radon is caused by
Chapter 6: Risk Communication Page 6-4
-------
uranium in soil, people hypothesize that their soil does not contain uranium. Told that a certain
percentage of homes are likely to have a radon problem, they assume their home does not. If people
learn that radon problems are related to housing characteristics, such as age or construction type, they
assume that the characteristics of theirs is such that it is not likely to be affected. Finally, if they
learn that radon problems may be more prevalent in certain geographical areas, they hypothesize that
their area is not likely to have a problem.
Research has found that this tendency to deny that high radon levels can affect "me" can best
be overcome by using a "persuasive appeal" that affects people on an emotional level and provides
a powerful risk message. TBWA focus groups found that focusing on concern for family members
was a highly effective persuasive approach.
In communications testing of the 1990 draft Citizen's Guide, Leferman Associates, Inc. (1990)
assessed consumer response to two different appeals used in radon information brochures: "protect
your family" and "dangerous intruder." Both of these themes were developed from TBWA and other
research suggesting the need to emphasize family protection, perhaps through an unsettling approach,
in order to stimulate action quickly before barriers are built in people's minds. In this
communications testing, both approaches were successful in conveying the risks of radon as the
brochures were found to be clear, easy to understand, and believable. Overall, more than half of the
participants claimed that they would be likely to test their home after studying either brochure.
Positive responses were higher, however, among those exposed to the "dangerous intruder" brochure
than among those evaluating the "protect your family" brochure. The "dangerous intruder" brochure
also evoked a higher level of immediate action. The researchers believed that the "danger" expressed
in that theme evoked a stronger need to do something than the, perhaps, more passive suggestion
to "protect."
4. Provide an Appropriate Level of Radon Information
Several risk communication studies have evaluated the success of providing various levels of
information in radon brochures. The most successful level of information is a difficult parameter to
evaluate because other variables (e.g., the way in which the information is presented) complicate the
issue. Also, "success" can be measured in terms of "success in promoting testing" or "success in
communicating accurate information." While no study has identified the "perfect" level of
information, the available risk communication studies suggest that too little information can cause
confusion and unnecessary concern and too much information can be overwhelming or create excuses
for inaction.
Smith et al. (1987), for example, examined alternative radon information formats and found
that homeowners who received a brief fact sheet rather than an informational brochure showed lower
levels of learning and were sometimes unnecessarily concerned over low radon readings. They were
also more likely to ask for more information and more willing to demand the services of a radon
diagnostician (Fisher and Johnson 1990). In a later study by Smith et al. (1990), homeowners
received one of six types of radon information material along with the result of their radon test.
Although the study did not identify any one format as being more effective than the others, it again
found that the brief fact sheet is ineffective. The fact sheet created concern about risk from radon
when it was not warranted.
Other research, in contrast, has found that readers can use additional information to create
excuses for inaction. Results of focus group testing conducted by TBWA (1989), for example, suggest
Chapter 6: Risk Communication Page 6-5
-------
3. Use Persuasive Appeals to Overcome Denial
As summarized in Section lls two "persuasive appeals" designed to affect people on an
emotional level, "protect your family" and "dangerous intruder," were found to be successful in
conveying the risks of radon. EPA used both of these themes in the draft revised Guide. For
example the draft overview to the Guide included a number of references to "keeping your family
sale" EPA focused heavily on the "dangerous intruder" theme throughout the Fall 1990 draft Guide
because it was believed, based on the research summarized above, that a strong message was needed
to overcome the public apathy that was met by the original Citizen's Guide.
4. Provide an Appropriate Level of Radon Information
The Fall 1990 draft Guide sought to strike an appropriate balance between too little
information, which can result in confusion and unwarranted concern, and too much information,
which can result in greater denial of radon risks and a greater tendency toward inaction.
5. Personalize the Radon Threat
To personalize radon risks, the Fall 1990 draft Guide included charts comparing radon risks
to an annual number of chest x-rays. In addition, EPA attempted to tailor the Guide to people in
different smoking categories by separating information on smokers and never smokers. The risks to
former smokers were addressed in a footnote to the charts. The Fall 1990 draft Guide also provided
quantitative information on risks and the number of lung cancer deaths attributable to radon to help
people fully understand and appreciate the threat.
6. Stress that Radon Problems Can Be Corrected without Overstating the Ease of
Fixing
EPA recognizes that while emphasizing the ease of corrective action may stimulate more
people to test and/or mitigate, overstating the ease of such action could cost the program some
credibility. Therefore, the 1990 draft Guide emphasized that radon problems in homes can be
remedied without overstating the ease, or understating the cost of corrective action. For example,
the Guide stated "the cost of repairing a radon problem in the home will vary from home to home.
Most homes can be fixed for about the same cost as other common home repairs. A lot depends on
the way the home is built." In a later section the Guide stated "the costs of making repairs to reduce
radon will vary depending on how your home was built. And the extent of the radon problem. But
in most cases, homes can be fixed for between $500 and $2,000."
Section IV: Reactions to the Fall 1990 Draft Guide and Changes Made in Response to
Comments
Recognizing the significance of the proposed changes in the Fall 1990 draft Guide, EPA took
direct steps to solicit input from the public, States, and other government agencies, and to carefully
weigh all comments in preparing the final version. EPA announced the availability of the draft Guide
in the Federal Register on September 24, 1990 and distributed 700-800 copies of the Guide and its
TSD for external review and comment. Although the comment period officially closed at the end of
October 1990, EPA continued to receive and accept comments well beyond that time and ultimately
Chapter 6: Risk Communication Page 6-8
-------
received 143 individual comment letters totalling over 700 pages. Virtually all sectors were
represented by the comments, including 33 States, Congress, 14 Federal agencies, numerous medical
and building associations, universities, the real estate industry, the radon testing and mitigation
industry, and private citizens. A large fraction of the public comments centered on risk
communication issues.
The final revised Citizen's Guide attempts to strike a balance between what the risk
communication research indicates is necessary and additional information provided by public
commenters. The final Guide addresses the concerns of the commenters, without sacrificing all of
its persuasive strength. This section summarizes the major concerns expressed by commenters as they
relate to each of the six key findings and describes how these concerns were incorporated into the
final version of the Guide,'
1, Be Prescriptive
Several commenters found the testing recommendations in the draft Guide to be confusing,
and many misinterpreted the guidelines to be advocating a "test-fix" strategy. In addition, there was
significant confusion among the commenters caused by the draft Guide's suggestion to "consider fixing
between 2 and 4 pCi/L." The commenters thought that this suggestion makes the final action level
unclear, and that in effect, it sets a de facto action level of 2 pCi/L.
The recommendations on the inside cover of the draft Guide were designed to encourage
public action by being brief, clear and easy to follow. The misinterpretation of the recommendations
by some of the commenters indicated to the Agency that the format or brevity of the
recommendations may have caused some confusion. Therefore, EPA revised its testing
recommendations to be much more explicit. The recommendation to fix, printed in bold at the very
beginning of the final Guide, is essentially the same as that in the draft version; only the wording has
been revised slightly to make the Guide more clear. The testing recommendations in a later section
of the Guide have been expanded and clarified, and a new section, "What your test results mean," has
been added. EPA believes that the recommendations are now very clear, and that there is no room
for misinterpretation.
With regard to the recommended action level, the 1992 Guide maintains 4 pCi/L as the level
for triggering home mitigation (see Chapters 3 and 7 for EPA's specific reasons for choosing this
action level). For example, the front inside cover the Guide states in bold: "Fix your home if your
radon level is 4 picocuries per liter (pCi/L) or higher." The final Guide also explains that there is
some health risk at levels below 4 pCi/L and that most homes today can be reduced to 2 pCi/L or
below. However, based on its risk communication research, EPA believes that this explanation simply
provides homeowners with added factual information to help them make informed decisions and will
not result in significant confusion.
2. Streamline Guidelines
Several commenters objected to the draft Guide's testing strategy, which allowed homeowners
to reach a mitigation decision based on short-term tests. The commenters believed that short-term
• EPA has prepared a separate response-to-comments document (Response to Public Comments on EPA's Draft "A
Citizen's Guide to Radon") that summarizes these and other issues in more detail and provides direct Agency responses.
Chapter 6: Risk Communication Page 6-9
-------
test results are not reliable and do not adequately approximate average annual radon levels, especially
at lower radon concentrations. Several commenters stated that short-term measurements may differ
from annual average results by a very large amount and could, therefore, lead the homeowner to
make the wrong mitigation decision.
EPA recognizes that short-term tests are not as representative as long-term tests of average
annual radon levels. However, as summarized above, research has shown that people are much more
likely to conduct short-term tests than long-term tests. Also, it is usually infeasible to conduct long-
term tests during a real estate transaction. Based on these practical limitations, the 1992 Guide
discusses the relative accuracy of short-term vs. long-term tests, emphasizing the benefits of long-term
testing, but allowing people to choose to conduct either a long- or a short-term follow-up test. This
approach represents the best balance between obtaining radon measurements that closely
approximate average year-round radon levels and ensuring that people do not drop out of the testing
process. In this sense, EPA believes that the approach recommended in the final Guide will result
in the greatest amount of radon risk reduction.
3. Use Persuasive Appeals to Overcome Denial
Many commenters addressed the degree to which EPA should attempt to stimulate the public
to test and fix their homes. A few commenters believed that the persuasive approach was effective
and/or appropriate, with at least one risk communication expert commenting that the tone should be
even more persuasive. A majority of commenters, however, believed that the approach was too
alarmist. Although the severity of this concern among the commenters varied, most believed that the
approach could confuse or alarm the public. Commenters focused their critique on the use of the
"intruder" strategy, stating that it was patronizing and misleading.
As discussed above, it was EPA's intent that the revised Citizen's Guide function more as a
persuasive brochure than as an informational document, prompting people to test their homes rather
than educating them fully on the technical aspects of radon. After weighing these comments, EPA
revised the Guide to be more informational and less "startling," while still striving to stimulate public
action. EPA also abandoned the "intruder" concept in favor of the "protect your family" theme. This
theme is displayed on the cover of the final Guide as well as being referred to throughout,
4, Provide an Appropriate Level of Radon Information
The many commenters who addressed the level of technical information in the Guide agreed
almost unanimously that more general and/or technical information should be added. The
commenters differed, however, in their suggestions of what additional information should be included.
Several commenters, for example, suggested that the Guide should provide more information on
testing and mitigation systems and techniques. Other commenters recommended that the Guide list
references for more information and indicate how these references can be obtained. Finallv, several
•j -
commenters advised that the assumptions, estimates, and uncertainties regarding risk should be better
covered in the Guide. Many of the commenters addressing this issue represented State radon offices
that were concerned that the release of the draft 1990 Guide would cause them to be inundated with
telephone calls requesting more radon information.
In response to these concerns, EPA added a significant amount of information to the Guide.
Because the Agency recognized that the inclusion of additional technical information risked sacrificing
Chapter 6: Risk Communication Page 6-10
-------
the effectiveness of the Guide as a persuasive tool, EPA sought to strike a balance in deciding how
much information to add. The Agency believes that the final revised Guide contains a sufficient level
of information while sacrificing as little as possible of the clarity, accessibility, and persuasive strength
of the draft 1990 Guide.
EPA added several new sections to the Guide, such as how to use a test kit (including
appropriate sampling locations and procedures), options for short-term and long-term testing, what
to do with test and retest results, radon and home sales, radon and home renovations, and radon
myths. Several sections also have been clarified and/or expanded, such as how radon enters a home
(including radon in building materials), radon in water, how to test, how to mitigate, and the health
risks associated with radon. The Guide does not include what would be an extremely long list of
radon contractors but suggests that the reader call his or her State radon office for a list of EPA-
approved contractors in the State. The Guide also includes a list of State radon office phone
numbers and hotline numbers, and references additional EPA materials and indicates how they can
be obtained. Also, in revising the Guide, an effort was made to address the assumptions and
uncertainties associated with radon risk in such a way that informs homeowners but does not present
a barrier to action. Based on recommendations made by EPA's SAB, the Guide includes several
statements acknowledging uncertainty in the risk estimates. More specific information on uncertainty
was not included because it was considered too technical. Instead, the Guide references this TSD,
which contains a detailed and expanded discussion of uncertainty in Chapter 2,
5. Personalize the Radon Threat
The commenters agreed that the use of comparative risk charts was an effective way of
conveying personal risks. Several commenters suggested that measures should be taken to decrease
the chance of alienating nonsmokers or allowing smokers to discount their risk due to radon. In
addition, many commenters suggested that the x-ray comparison is misleading and frightening and
should be replaced by a more tangible analogy, such as the risks associated with accidents in homes
or cars, or radiation doses from airplane flights.
The Agency's primary goal in designing a risk chart for the updated Guide was that it
characterize risk as accurately as possible without alienating either smokers or nonsmokers or allowing
them to deny their risk. To do this, the final version of the Guide presents two separate risk charts.
The Agency believes that this approach allows both smokers and never smokers to accurately
characterize their risk without alienating either group.
EPA had spent a considerable amount of effort prior to the release of the draft revised Guide
identifying an effective and appropriate comparison to the health risks associated with radon. The
Agency found in extensive communications tests that the comparison to x-rays did not frighten
people. In fact, the risk communication research showed that people found the x-ray comparison
clear and easy to understand. Nevertheless, in response to the concerns expressed by commenters,
the Agency conducted additional communication studies on four possible comparisons: x-rays,
illnesses, accidents (e.g., airplane crashes, home fires, etc.), and an approach that combined several
comparisons. EPA found in this later research that none of the sets of risk comparisons was clearly
better than another at producing appropriate responses to the risks from radon (Weinstein and
Sandman 1991). In light of this finding and the concerns of commenters, EPA chose to compare
radon risk to the risk of being killed in different types of accidents.
Chapter 6: Risk Communication Page 6-11
-------
6. Stress that Radon Problems Can Be Corrected without Overstating the Ease of
Fixing
Several cominenters addressed the costs of mitigation as an issue in the draft revised Guide.
Most of these commenters recommended that the Guide should be more explicit and provide specific
remediation costs for different remediation methods, rather than just provide a range. A few
commenters suggested the Guide provide examples of home repairs that are of comparable costs to
radon mitigation, with one believing that the draft Guide exaggerated its claim that a home can be
mitigated for the same cost as other home repairs.
The final Guide discusses remediation techniques and costs in the "How to Lower the Radon
Level in Your Home" section. The Agency revised the range of $5GO-$2,000 estimated to fix a home
to $50G-$2,500, based on data collected by EPA's Office of Research and Development. These data
augmented the Private Sector Radon Mitigator Survey that surveyed 340 radon mitigation firms on
their radon reduction activities (U.S. EPA/Radon Division 1990a), The survey found that, within this
range, approximately $1,200 (in 1989 $) was the average charge to mitigate a detached house. The
Guide also provides examples of home repairs (i.e., painting or installing a new water heater) that are
comparable to the costs of radon mitigation.
Section V: Summary and Conclusion
Risk communication research has made an extremely important contribution in the evolution
of the Citizen's Guide. Over the past six years, a number of researchers have analyzed key issues
relating to radon risk communication, and their findings were incorporated into the Fall 1990 draft
Guide. Studies of the Fall 1990 version of the Guide showed the document to be extremely effective
in terms of risk communication. When the draft Guide was distributed for review, commenters were
able to identify areas in which they believed that other concerns should be considered and factored
in with the risk communication findings. Commenters noted, for example, that while the "intruder"
concept may be effective in stimulating testing, some people may find it offensive. Also, as State
representatives pointed out, an effective level of information from a risk communication standpoint
may result in State radon offices being inundated with calls from people asking for more information.
Thus, the commenters helped EPA refine its use of the six key risk communication findings and
develop a final version of the Citizen's Guide that both incorporates the main findings of risk
communication and addresses the commenters' practical concerns.
Chapter 6: Risk Communication Page 6-12
-------
CHAPTER 7
RATIONALE FOR 1992 CITIZEN'S GUIDE
This chapter summarizes the main conclusions from the previous chapters and briefly describes
how those conclusions were used to develop the 1992 Citizen's Guide. It outlines the rationale
underlying EPA's approach in the revised Guide for accomplishing two main goals: (1)
communicating radon risks to the public; and (2) recommending a radon testing and mitigation
strategy for homeowners. The chapter also discusses the interrelationship of these two goals and how
EPA considered them together in shaping the 1992 Guide.
Section I: Risk Communication
Developing the final approach for risk communication in the revised Citizen's Guide required
the consideration of two distinct issues: what to tell the public about radon risks and how to
communicate that information. The rationale for the revised Guide's approach with respect to each
of these issues is outlined in separate sections below.
What to Tell the Public about Radon Risks
As presented in Chapter -2, radon is one of the most extensively studied pollutants in the
world. Radon risks have been well documented by numerous scientific and public health authorities,
and are based on extensive studies of thousands of underground miners. This research provides
strong scientific evidence that exposure to radon can cause lung cancer in humans. Further research
since the original Citizen's Guide was published in 1986 has helped to improve EPA's projection of
lung cancer risk to the general population due to radon exposures in the home. EPA now estimates
that the number of lung cancer deaths per year in the U.S. due to residential radon exposure is
approximately 14,000, with an uncertainty range of 7,000 to 30,000. Even using the lower end of this
uncertainty range, this makes indoor radon the second leading cause of lung cancer in the U.S., after
smoking. Continuing scientific evaluation of radon hazards serves primarily to refine EPA's estimate
of the annual number of radon-induced lung cancer deaths, not to determine whether indoor radon
exposure poses a serious public health problem.
In addition, in order to understand the risks of indoor radon, the public needs information
on the relationship between radon and smoking risks and the radon risks to children. Key findings
related to each of these issues are summarized below:
• Although the risk of lung cancer from radon exposure appears to be enhanced by
smoking tobacco, a person does not have to be a smoker to be at risk from radon.
EPA estimates that the radon risk for current smokers is close to 20 times the risk for
never smokers and the risk to former smokers may be over 8 times greater than the
risk for never smokers.
» Although there is information that indicates that children may be at greater risk than
adults from some kinds of radiation exposure, there is no direct evidence that children
are at greater risk than adults from radon.
Chapter 7: Rationale for 1992 Citizen's Guide , Page 7-1
-------
Based on these conclusions, the revised Guide warns that an individual's lung cancer risk is especially
high if he or she is a smoker, and provides risk charts that allow readers to identify their radon risk
according to what smoking category they fall into. The revised Guide also clarifies that "children have
been reported to have greater risk than adults of certain types of cancer from radiation, but there
are currently no conclusive data on whether children are at greater risk than adults from radon,"
How to Communicate Radon Risk Information
As discussed in Chapter 6, extensive risk communication research since the 1986 Guide was
published has provided useful insight into why the public remains largely apathetic about indoor radon
and suggestions for overcoming that apathy. In developing the revised Guide, EPA applied six key
findings that have emerged from this research: (1) be prescriptive rather than simply informative; (2)
streamline guidelines on testing and mitigation so they do not present barriers to public action; (3)
overcome public denial through the use of "persuasive appeals" such as concern for the family; (4)
provide an appropriate level of radon information, since too much or too little information can result
in an undesired effect; (5) personalize the radon threat with tangible, relevant comparisons to familiar
risks; and (6) stress that radon problems can be corrected but do not overstate the ease of fixing
them.
In updating the Citizen's Guide, it was important for EPA to balance the results of this risk
communication research with other practical considerations. For example, some of the risk
communication research suggested that the revised Guide should provide only a minimal amount of
technical information on radon, since readers can use additional information to create excuses for
inaction. This finding, however, had to be balanced against concerns that State offices could be
inundated with public requests for more information if the Guide did not provide enough detail to
answer homeowners' questions. This balancing of the risk communication research with practical
limitations led to the following design features with respect to the six key findings listed above:
» The 1992 Citizen's Guide was designed to be prescriptive. It provides brief, clear, and
easy to follow directions on what to do (e.g., how to test and when to mitigate), rather
than simply providing information and allowing readers to come to their own
conclusions.
* EPA's Radon Program experience and risk communication research indicate that
many people drop out of the testing and mitigation process before they fully comply
with the 1986 Guide's recommendation to conduct a long-term follow-up
measurement prior to reaching a mitigation decision. Based on this finding and EPA's
detailed analysis of various testing options, the revised Guide attempts to streamline
the testing guidelines by: (1) emphasizing the benefits of long-term testing; but also
(2) allowing people to choose to conduct either a long- or short-term follow-up test.
• The 1992 Guide was designed to function more as a persuasive document than the
1986 version. It utilizes the "protect your family" theme found to be effective in risk
communication testing, while avoiding other emotional appeals that might compromise
the scientific credibility of the message.
Chapter 7: Rationale for 1992 Citizen's Guide Page 7-2
-------
The revised Guide contains a sufficient level of technical information to educate
homeowners, while sacrificing as little as possible in terms of clarity, accessibility, and
incentive to test,
EPA designed the revised Guide to help personalize the radon threat by providing
tangible, relevant comparisons to familiar risks. Specifically, the revised Guide
provides risk charts tor smokers and never smokers that characterize each group's risk
as accurately as possible without alienating them or allowing them to deny their risk.
The Guide also compares radon risk to other risks, such as drunk driving, drowning,
fires, airline and car crashes, and violent crimes.
The revised Guide puts radon mitigation in proper perspective by describing
remediation techniques, providing realistic estimates of remediation costs based on
EPA's Private Sector Radon Mitigator Survey and research by EPA's Office of
Research and Development, and providing examples of other home repairs that are
of a comparable cost.
Section II: Testing and Mitigation Advice
The Citizen's Guide provides specific advice to homeowners on how to test for radon and
when they should mitigate. EPA's rationale for this advice, drawing on the main conclusions from
the previous chapters of this document, is summarized below.
How to Test
Chapter 3 identifies several major attributes that are used to define the recommended radon
testing strategy, including: (1) a testing location, (2) ventilation conditions, and (3) a testing duration
before reaching a mitigation decision. EPA's recommendations and rationale with respect to each
of these attributes are summarized below, followed by an explanation of the basis for the overall
testing strategy recommended in the revised Citizen's Guide.
1, Testing Location
EPA is revising the recommended test location for the initial short-term radon measurement
from the "lowest livable" level, recommended in the 1986 Guide, to the "lowest lived-in" level. There
is a difference between these two locations. The lowest livable level is the lowest area of a home that
is used or has the potential to he used as a living space, whereas the lowest lived-in level is the lowest
level of a house that is used regularly. In both the original and revised Guide, however, the
recommended location for decision-making tests (i.e., tests on which to base a mitigation decision)
is the lowest lived-in level.
Roughly 50 percent of the homes nationwide have basements; however, only about half of
these homes, or almost 25 percent of the national total, have basements that are used as a lived-in
level (U.S. EPA/Office of Radiation Programs 1991 a). Accordingly, a recommendation that devices
be placed in the lowest livable level would result in 50 percent of the devices being placed in the
basement. A recommendation that devices be placed in the lowest lived-in level would result in only
25 percent of devices being placed in the basement, and the remaining 75 percent being placed on
the first floor.
Chapter 7: Rationale for 1992 Citizen's Guide Page 7-3
-------
Keeping the short-term test in the lowest livable area as compared to moving it to the lowest
lived-in area would reduce false negative results by a factor of 1.5; however, the number of false
positive results would increase by a factor of 2. Measurements taken in the lowest lived-in area strike
a more equitable balance between false positives (public money spent with more limited benefits in
many cases) and false negatives (public health protection) than do lowest livable area measurements.
Most importantly, lowest lived-in measurements are the most representative of human exposure to
radon. Recent research by Harley et al. (1991) suggests that basement measurements overstate
personal exposure by a factor of 3 to 5, while first floor measurements are, on average, only 30
percent greater than occupant exposure (for short-term measurements). Since the goal of radon
measurement on which a mitigation decision relies is to assess occupant exposure (U.S. EPA 1991a),
lived-in level measurements are better predictors of risk than livable level measurements. Based on
this finding and the recommendation of the EPA Science Advisory Board (U.S. EPA 1992b), short-
or long-term measurements that are to be used for the purpose of making mitigation decisions should
be made in the lowest lived-in level because it more closely approximates the radon concentration
to which inhabitants are exposed.
2. Ventilation Conditions
EPA examined whether tests should be made under "open-house" or "closed-house"
conditions. In open-house conditions, tests are made with windows and other ventilating passageways
either closed or open as they would be when a test is not being conducted. In closed-house
conditions, windows and other ventilated passageways are closed to the extent possible. EPA stresses
the value of closed-house conditions in the revised Guide, because these conditions provide more
consistent radon measurements. However, the revised Guide makes allowances for practical
limitations in maintaining these conditions. Specifically, the Guide recommends keeping windows and
outside doors closed "as much as possible" during short-term testing.
3. Testing Duration
The 1986 Citizen's Guide recommended that homeowners perform (1) a short-term
measurement, and (2) a confirmatory measurement if the result of the short-term test was above 4
pCi/L. The recommended duration of the confirmatory measurement depended on the result of the
shorl-term test and varied from less than one week to a full year. Mitigation decisions were to be
based on the results of the longer-term confirmatory measurement. Since the original Guide was
published, however, EPA's Radon Program experience and risk communication research have
indicated that many people drop out of this process before obtaining the long-term measurements
needed to reach a mitigation decision. Short-term tests are much more appealing to the public
because they are simpler to make and provide faster results. Therefore, it became apparent that a
recommended testing procedure that does not rely solely on long-term testing may be more effective
in actual risk reduction by removing an existing barrier to radon mitigation.
EPA examined the implications of relying on short-term tests as indicators of annual average
radon levels in developing the 1992 Guide. From knowledge of the relationship between short-term
measurements and annual averages gained from EPA/State Residential Radon Surveys, the Agency
recognized that short-term measurements are imperfect indicators of annual averages, but believed
they potentially could provide homeowners with a reasonably accurate basis for determining whether
they need to mitigate. EPA thus constructed various testing options, including different test durations
prior to reaching a mitigation decision, and analyzed the effectiveness of each option in terms of
misclassification rates and public acceptability. Based on the results of this analysis, described in
Chapter 7: Rationale for 1992 Citizen's Guide Page 7-4
-------
detail in Chapter 3 and summarized in the following section, EPA concluded that a testing strategy
that allows homeowners to reach a mitigation decision based on short-term test results would
maximize public health protection.
4. Rationale for Recommended Testing Strategy
EPA began its analysis of radon testing options by considering a strategy that requires a single
short-term measurement followed by a confirmatory long-term test. This option, defined as Option
A in Chapter 3, parallels the testing procedure recommended in the 1986 Citizen's Guide, except that
both tests would be conducted on the lowest lived-in level. Under Option A, 98 percent of homes
were correctly classified with regard to the need for mitigation. To determine how testing procedures
relying on a short-term confirmatory test would compare to this option, EPA developed and analyzed
five other options, defined as Options B through F in Chapter 3. EPA found that all of these options
produced results yielding at least 94 percent correct classifications.
Given these results, the Agency determined that on the basis of misclassification, none of the
options should be rejected. EPA decided instead that other factors, such as procedure simplicity,
should be weighed with the trade-offs for increased levels of correct classification in selecting the
testing option to recommend in the revised Guide, As noted above, current Radon Program
experience indicates that few people are actually taking long-term follow-up measurements prior to
reaching a mitigation decision, and that most people who do mitigate do so based on a single short-
term test. This finding is supported by EPA's risk communication research that indicates that as little
as 9 percent of the population is willing to conduct year-long tests. These conclusions suggest that
a testing protocol that relies exclusively on long-term confirmatory tests is unlikely to be followed by
most of the public. Therefore, although Options A and D have more desirable error rates than the
other options, compliance with these options is likely to be low. Also, selection of either of these
two options is apt to result in a situation in which decisions are most often made based on a single
short-term measurement. By explicitly providing homeowners a process that calls for more than one
short-term test, all of the other options considered would be more likely to result in people having
better information to use in reaching a mitigation decision.
On the other hand, since long-term measurements are more desirable than short-term
measurements and since there are some individuals who are willing to conduct long-term tests, the
use of long-term tests should not be precluded from the testing protocol. As a result. Options B, E,
and F, which do not include long-term tests, are also undesirable.
Option C offers an effective compromise between these different approaches. It promotes
long-term testing by people who are willing to conduct long-term tests and recommends an effective
short-term test as an alternative for people who are not willing to conduct a long-term test.
Therefore, consistent with the advice from EPA's Science Advisory Board, the revised Guide
recommends Option C because it should maximize the total risk reduction the public would gain
through future testing and mitigation while minimizing error. The Guide also explains the trade-offs
between short-term and long-term testing. It acknowledges that long-term testing is more
representative of annual exposures, but enables the public to have a short-term measurement process
that is sound.
Chapter?: Rationale for 1992 Citizen's Guide Page 7-5
-------
When to Mitigate
EPA is recommending that the action level of 4 pCi/L established in the 1986 Guide be
maintained for several reasons. First, lower action levels introduce more uncertainty in the
measurement results. Measurement device error increases to approximately 50 percent at 2 pCi/L.
This device error in conjunction with the larger fraction of homes (of total homes testing) that have
radon levels around 2 pCi/L would result in a threefold increase in false negatives and a twofold
increase in false positives over those expected at a 4 pCi/L action level. In addition, as outlined in
detail in Chapter 4, the Office of Research and Development's (ORD's) research on mitigation
effectiveness and the Office of Radiation Programs' Private Sector Radon Mitigator Survey suggest
that elevated levels of radon can be reduced to 4 pCi/L more than 95 percent of the time. Results
from the mitigator survey indicate that 2 pCi/L can be achieved about 70 percent of the time, while
the ORD research suggests this estimate may be even higher (U.S. EPA/ORD 1989; U.S. EPA/Radon
Division 1990a). Reducing the action level to 2 pCi/L, therefore, could result in perhaps as many as
30 percent of homes with elevated levels being unable to achieve the action level.
However, EPA recognizes that mitigation down to lower radon levels may be appropriate
because levels below 4 pCi/L still pose a health risk. Furthermore, as mentioned above, mitigation
technology available today permits most homes to be reduced to 2 pCi/L or below, and Congress has
set a long-term goal that indoor radon levels be no more than outdoor levels, which are typically less
than 2 pCi/L. As a result, EPA also closely examined the costs and benefits of selecting an action
level of 2 pCi/L and 3 pCi/L.
The results of this cost-effectiveness analysis, detailed in Chapter 5, show that setting the
action level at 4 pCi/L would result in a cost of roughly $700,000 per lung cancer death averted (i.e.,
per life saved). Lowering the action level would incrementally increase this cost to $1,700,000 per
life saved if 3 pCi/L was used, and to $2,400,000 per life saved if 2 pCi/L was used instead of 3 pCi/L.
All three of the action levels have cost per life saved values that are at the lower end of, or below,
the values that the public is willing to pay to save a statistical life, according to EPA's 1983 Regulatory
Impact Analysis Guidelines. Based on these findings, EPA believes any of the three action levels
considered would provide cost-effective results.
At the selected action level, the Radon Program would be as or more cost-effective than many
other government programs for personal safety and environmental protection. EPA also believes the
recommended testing protocol combined with an action level of 4 pCi/L in the 1992 Guide will be
more cost-effective than that recommended in the 1986 Guide — $700,000 per life saved now versus
the $900,000 per life saved that EPA calculates from program experience based on the 1986 Guide.
EPA's decision to keep 4 pCi/L as the action level in the revised Citizen's Guide is supported
by the cost-effectiveness analysis. However, the revised Guide also notes, although not with the same-
weight as the recommended action level, that homeowners can further reduce their lung cancer risk
by mitigating homes that are below 4 pCi/L. As long as the revised Guide clearly establishes 4 pCi/L
as the recommended action level to avoid confusion and the other problems mentioned above, EPA
believes this discussion of reducing radon below the action level helps to fully inform the reader and
is justified based on: (1) the health risk involved; (2) the effectiveness of available mitigation
technology; (3) cost-effectiveness; and (4) Congressional intent.
Chapter 7: Rationale for 1992 Citizen's Guide Page 7-6
-------
Section III: Conclusions
In developing the 1992 Citizen's Guide, EPA had to balance the findings of its technical
analyses on risk, testing accuracy, mitigation technology, and cost-effectiveness with the information
it was collecting from its risk communication research. For example, EPA and the scientific
community had amassed and analyzed in depth a considerable amount of information on the level and
significance of indoor radon risk since the original Guide was published in 1986. EPA had to convey
that risk in the updated Guide with a message that was strong enough to persuade homeowners to
act, while being careful not to provide too much detail, which could sacrifice accessibility, or make
it too startling, which might compromise scientific credibility and support. Furthermore, although
EPA recognized the technical superiority of long-term versus short-term testing after extensive
evaluation of the issue, it had to accept the compelling practical limitation that the public at large is
more likely to use short-term testing. A lot of "good" testing, after all, will provide greater public
health protection than a more limited amount of "perfect" testing. Finally, emphasis on an action
level that is achievable by the vast majority of homes is better than recommending a lower action
level that pushes the limit of technology. EPA's ultimate objective was to advance a technically well-
supported 1992 Citizen's Guide that takes a pragmatic step in better communicating radon's risks to
the public and promoting broader public action in response to the problem.
Chapter 7: Rationale for 1992 Citizen's Guide Page 7-7
-------
-------
REFERENCES
Akerman, J., Bergman, L. and Johnson, F.R. 1989. Paying for Safety: Voluntary Reduction of
Residential Radon. Stockholm School of Economics, Stockholm, Sweden.
American Cancer Society. 1988. Unpublished Tabulation from Cancer Prevention Study.
Archer, V.E., Roscoe, R.J. and Brown, D.P, 1985. Is Silica or Radon Daughters the Important
Factor in the Excess Lung Cancer Among Underground Miners? In Silica, Silicosis and
Cancer: An International Symposium. Society for Occupational and Environmental Health,
NY, Praeger.
Blot, W., et al. 1990. Indoor Radon and Lung Cancer in China. Journal of the National Cancer
Institute. 82(12): 1025-1030.
Bostrom, A. 1989. Risk Perceptions: An Approach to Perceptions of Risk Processes, a Mental
Models analogy, and an Application to Radon. Center for Risk Perception and
Communication, Carnegie-Mellon University, Pittsburgh, PA
Brown, L. 1983. National Radiation Survey in the U.K.; Indoor Occupancy Factors. Radiation
Protection Dosimetry. 5: 321-325.
Bruskin Associates, 1991. Telephone survey, November 1991.
Bureau of the Census. 1992a. Series P-25, No. 1082. Estimates of Population of the U.S. Through
December 1990. Issued February 1992.
Bureau of the Census. 1992b. Monthly Residential Population Projection. May 1992.
Bureau of the Census. 1991. Characteristics of New Housing: 1990. C25-9013.
Chameaud, L, Masse, R., Morin, M. and Lufuma, J. 1980. Lung Cancer Induction by Radon
Daughters in Rats. In Occupational Radiation Safety in Mining, Volume 1. H. Stokes, ed.,
Canadian Nuclear Association, Toronto, Canada.
Chess, C. and Hance, BJ. 1988. Alerting the ..Arjat]b.gtic_ and Reassuring the Alarmed:
Communicating about Radon Risk in Three Communities. Final Report submitted to U.S.
EPA Office of Policy, Planning, and Evaluation. EPA-230-08-88-036.
Chmelynski, H. 1992. Evaluation of the Performance of Alternative Short-term Radon Testing
Procedures in U.S. Homes. May 1992,
Cohen, B.L. 1987. Reducing the Hazards of Nuclear Power: Insanity in Action. Physics and Society.
16(3): 2-51.
Condon, W., Huang, J., and Rimawi, K, 1990. Survey of Indoor Radon Levels in New York State
and Evaluation of U.S. Screening Protocols. New York State Energy Research and
Development Authority, Report 90-0. June 1990.
References Page R-l
-------
Cross, F.T., et al. 1991. Inhalation Hazards to Uranium Miners. Pacific Northwest Laboratory
Annual Report for 1990 to the DOE Office of Energy Research: Part 1 Biomedical Sciences.
Cross, F.T., et al. 1982. Carcinogenic Effects of Radon Daughters in Uranium ore dust and cigarette
smoke in beagle dogs. Health Physics. 42: 35-42.
Davis, F., Boice, J,, Hrubek, Z. and Monson, R. 1989. Cancer Mortality in a Radiation-Exposed
Cohort of Massachusetts Tuberculosis Patients. Cancer Research. 49: 6130-6136.
Department of Health and Human Services. 1989. Reducing the Health Consequences of Smoking:
25 Years of Progress. A Report of the Surgeon General. Public Health Service, Centers
for Disease Control, Center for Chronic Disease Prevention and Health Promotion, Office
on Smoking and Health, Rockville, MD. DHHS (CDC) 89-8411.
Desvousges, W.H. and Cox, J.A. 1986. Radon Focus Groups: A Summary. Research Triangle
Institute, Research Triangle Park, NC.
Desvousges, W.H. and Kollander, M. 1986. Radon Focus Groups: A Summary. Research Triangle
Institute. Research Triangle Park, NC.
Doll, R. 1978. An Epidemiological Perspective of the Biology of Cancer. Cancer Research. 38:
3573-3583.
Doyle, J.K., et al. 1990. An Evaluation of Strategies for Promoting Effective Radon Mitigation.
University of Colorado, Boulder, CO,
Doyle, J.K., et al. 1989. An Evaluation of Strategies for Solving the Radon Problem. University of
Colorado, Boulder, CO.
Fisher, A. and Johnson, F.R. 1990. Radon Risk Communication Research: Practical lessons.
Fisher, A., et al. 1989. The Value of Reducing Risks of Death: A Note on New Evidence. Journal
of Policy Analysis and Management. Volume 8, No. 1.
Fowler, L,, et al. 1988. Radon Mitigation in Slab-On-Grade Houses in Florida. Presented at the 81st
Annual Meeting of Air Pollution Control Association.
Francis, E.A. 1987. Patterns of Building Occupancy for the General Public. NRPB, Chilton, Didcot,
Oxfordshire, UK. NRPB-M129.
GEOMET. 1981. Comparison of Indoor and Outdoor Air Quality. EPRI Project 1309,
Gaithersburg, MD. EPRI EA-1733.
George, A.C. and Breslin, A.J. 1980. The Distribution of Ambient Radon and Radon Daughters in
Residential Buildings in the New Jersey-New York Area. In Natural Radiation Environment
III. CONF - 780422. T.F. Gessel! and W.M. Lowder, eds., NTIS, Springfield, Va.
Harley, N.H., Chittaporn, P., Roman, M.H., Sylvester, J. 1991. Personal and Home 222Rn and
Gamma-Ray Exposure Measured in 52 Dwellings, Health Physics. 61:737-744.
References Page R-2
-------
Henschel, B. 1987. Radon Reduction Techniques for Detached Houses. Second Edition. U.S. EPA,
Washington, DC. EPA 625/5-87/019.
Henschel, B. and Scott, A, 1987. Testing of Indoor Radon Reduction Techniques in Eastern
Pennsylvania: An Update. In Proceedings of Jhe Air Pollution Control Association
International Specialty Conference - Indoor Radon.
Hopper, R. April 1991. National Ambient Radon Study. Proceedings of the 1991 International
Symposium on Radon and Radon Reduction Technology. U.S. Environmental Protection
Agency, Office of Radiation Programs, Las Vegas Facility.
Howe, G.R., et al. 1986. Lung Cancer Mortality (1950-1980) in Relation to Radon Daughter
Exposure in a Cohort of Workers at the Eldorado Beaver Lodge Uranium Mine: Journal of
the National Cancer Institute. 77: 357-362.
Hubbard, L., et al. 1987. Radon Entry into Detached Dwellings: House Dynamics and Mitigation
Techniques, Submitted to Radiation Protection Dosimetry.
International Agency for Research on Cancer. 1988. Monograph on the Evaluation of Carcinogenic
Risks to Humans, Vplume_43: _ Man-made Mineral Fibers and Radon. IARC, World Health
Organization.
International Commission on Radiological Protection. 1987. Lung Cancer Risk Jrom Indoor
Exposures to Radon Daughters. Pergamon Press, Oxford, U.K. ICRP Publication 50,
Israeli, M. 1985. Deposition Rates of Rn Progeny in Houses. Health Physics. 49: 1069-1083.
James, A.C. 1991. Personal Communication, Anthony C. James, Battelle Pacific Northwest
Laboratories, to Jerome S. Puskin, U.S. EPA.
Johnson, F.R. 1990. Analysis of the Wirthlin Survey Radon Questions. U.S. EPA Office of Policy,
Planning, and Evaluation.
Johnson, F.R. and Fisher, A. 1989. Conventional Wisdom on Risk Communication and Evidence
from a Field Experiment. Risk Analysis. 9(2): 209-213.
Johnson, F.R. and Luken, R.A. 1987. Radon Risk Information and Voluntary Protection: Evidence
from a Natural Experiment. Risk Analysis. 7: 97-107.
Johnson, F.R. and Smith, V.K. Undated. Bad News and Perceived Risk: Homeowner Responses to
Indoor Radon. U.S. EPA Office of Policy Analysis.
Jonassen, W. and Jensen, B. 1989. Radon Daughters in Indoor Air. Final Report to Vattenfall
Technical University of Denmark, Lyng.
Keller, G. and Folkerts, K.H. 1984. Radon-222 Concentrations and Decay Product Equilibrium in
Dwellings and in Open Air. Health Physics. 47: 385-398.
References
-------
KJeinbaum, D., Kupper, L. and Morgenstern, H. 1982. Epidemiologic Research Principles and
Quantitative Methods. Van Nostrand Reinhold Company Inc., New York.
L'Abbe, K.A. et al. 1991. Radon Exposure, Cigarette Smoking, and Other Mining Experience in the
Beaverlodge Uranium Miners Cohort. Health Physics. 60(4): 489-495.
Leferman Associates, Inc. 1990. Communications Test of Alternative Radon Brochures. Prepared
for the U.S. EPA and the Advertising Council.
Lenchek, et al. 1987. Superinsulated Design and Construction. Van Nostrand Reinhold Publishers.
Lubin, J.H. 199L Note from Jay H. Lubin, National Cancer Institute, to Anita L. Schmidt, U.S.
EPA.
Lubin, J.H., Samet, J.M., and Weinberg, C. 1990. Design Issues in Epidemiologic Studies of Indoor
Exposure to Rn and Risk of Lung Cancer. Health Physics. 59(6): 807-817.
Lubin, J.H., et al. 1990. Quantitative Evaluation of the Radon and Lung Cancer Association in a
Case Control Study of Chinese Tin Miners. Cancer Research. 50: 174-180.
Lubin, J.H. 1988. Models for the Analysis of Radon-Exposed Populations. Yale J. Biol. & Med.
61: 195-214.
Makelainen, I. 1980. Preliminary Survey of Radon in Finnish Dwellings. In Radiation in Our
Environment. Nordic Society for Radiation Protection, Oslo, Norway.
Marcinowski, F. 1990. Analysis of the Relationship of Short-Term Measurements to Annual
Measurements in Support of the Citizen's Guide Revision. Unpublished document submitted
to U.S. EPA Radon Division.
Moeller, D.W. and Underbill, D.W. 1976. Final Report on Study of the Effects of Building Materials
and Population Dose Equivalents. Harvard School of Public Health, Contract 68-01-3292.
Morrison, H.I., Semenciw, R.M., Mao, Y. and Wigle, D.T. 1988. Cancer Mortality Among a Group
of Fluorspar Miners Exposed to Radon Progeny. Amer. J. Epidem. 128: 1266-1275.
Muller, J. 1984. Study of Mortality of Ontario Miners 1955-1977, Part I. Pp. 335-343 in Proceedings
of the International Conference on Occupational Radiation Safety in Mining, Volume I. H.
Stacker, ed., October 14-18, Canadian Nuclear Association, Toronto.
National Academy of Sciences, 1991. Comparative Dosimetry of Radon in Mines and Homes.
National Academy Press, Washington, DC,
National Academy of Sciences. 1990. Health Effects of Exposure to Low Levels of Ionizing
Radiation: BEIR V. National Academy Press, Washington, DC.
National Academy of Sciences. 1988. Health Risk of Radon and Other Internally Deposited Alpha-
Emitters: BEIR IV. National Academy Press, Washington, DC.
References . Page R-4
-------
National Academy of Sciences. 1980. The Effects on Populations of Exposure to Low Levels of
Ionizing Radiation: BEIR III. National Academy Press, Washington DC.
National Center for Health Statistics. 1985. U.S. Decennial Life Tables for 1979-1981. Volume 1.
Number 1. Public Health Service, Washington, DC. U.S. Government Printing Office.
DHHS (PHS) 85-1150-1.
National Center for Health Statistics. 1983. Vital Statistics of the United States 1980, Volume II
Mortality. Part A. Public Health Service, Washington, DC. U.S. Government Printing Office.
DHHS (PHS) 88-1101.
National Council on Radiation Protection and Measurements. 1988. Measurement of Radon and
Radon Daughters in Air. Bethesda, MD. NCRP Report 97.
National Council on Radiation Protection and Measurements. 1984. Evaluation of Occupational and
Environmental Exposures to Radon and Radon Daughters in the United States. Bethesda,
MD. NCRP Report .78.
National Council on Radiation Protection and Measurements. 1980. Influence of_Dose and its
Distribution in Time on Dose-Response Relationships for Low-LET Radiations. Bethesda,
MD. NCRP Report 64.
National Health Interview Survey (NHIS). 1985. Unpublished Tabulation.
National Institute of Health. 1985. Report of the National Institute of Health Ad Hoc Working
Group to Develop Radjoepidjsmiological Tables. U.S. Department of Health and Human
Services, Washington, DC. NIH 85-2748.
National Institute for Occupational Safety and Health. 1987. Radon Progeny in Underground Mines.
A Recommended Standard for.Occupational Exposure. Washington, DC, DHHS (NIOSH)
88-101.
New Jersey Department of Environmental Protection. 1989. Highlights of the Statewide Scientific
Study of Radon. Trenton, NJ. #609-987-2101.
Nitschkc, L, et al. 1985. Indoor Air Quality,...In.fiJtration and Ventilation in Residential Buildings.
New York State Energy Research and Development Administration 'Report 85-10.
Oakley, D.T. 1972. National Radiation Exposure in the United States. U.S. EPA, Washington, DC.
ORP/SID 72-1.
Office of Management and Budget. 1991. Budget of the United States Government: Fiscal Year
1992. Chapter IX.C, Reforming Regulation and Managing Risk Reduction Sensibly.
Peake, R.T. and Shumann, R.R. 1990. Regional Radon Characterizations. U.S. Geologic Survey
Bulletin.
Perritt, R. 1990. Radon-222 Levels in New York State Homes. Health Physics. 58: 147-155.
References Page R-5
-------
Pershagen, G., et al, 1992. Residential Radon Exposure and Lung Cancer in Swedish Women.
Submitted for publication in Health JPhysics.
Porstendorfer, J. 1987. Free-Fractions, Attachment Rates, and Plate-Out Rates of Radon Daughters
in Houses. Pp. 285-300 in Radon and Its Decay Products: Occurrence. Properties and
Health Effects. P.K. Hopke, ed., American Chemical Society, Washington, DC, Symposium
Series 331.
Prentice, R.L., Yoshimoto, Y. and Mason, M.W. 1983. Relationship of Cigarette Smoking and
Radiation Exposure to Cancer Mortality in Hiroshima and Nagasaki. Journal of the National
Cancer Institute. 10: 611-622.
Puskin, J.S. and Nelson, C.B. 1989, EPA's Perspective on Risks from Residential Radon Exposure.
JAPCA. 39:915-920.
Puskin, J.S. and Yang, Y. 1988. A Retrospective Look at Rn-Induced Lung Cancer Mortality from
the Viewpoint of a Relative Risk Model. Health Physics. 54: 635-643.
Pyle, B., et al. 1988. Radon Mitigation in Crawlspace Houses in Nashville, Tennessee. Presented
at the 81st Annual Meeting of Air Pollution Control Association, June 1988.
Radford, E.P. and St. Clair Renard, K.G. 1984. Lung Cancer in Swedish Iron Miners Exposed to
Low Doses of Radon Daughters. N. Engl. J. Med. 310: 1485-1494.
Radonics Inc. 1991. Unpublished Data.
Ronca-Battista, M. 1990, The Relationship Between Wintertime Screening and Annual Average
Radon Levels in U.S. Homes. Section III-P2 in 1990 International Symposium on Radon and
Radon Reduction Technology: Preprints, Volume 2.
Ronca-Battista, M. 1989. Radon in U.S. homes: A Summary of the Available Literature on Annual
and Screening Radon Concentrations, Seasonal Variations, and Differences Between Floors.
Memorandum from Melinda Ronca-Battista to U.S. EPA Radon Division.
Ronca-Battista, M. and Gray, D. 1988. The Influence of Changing Exposure Conditions on
Measurements of Radon Concentrations with the Charcoal Adsorption Technique. Radiation
Protection Dosimetry. 24: 361-365.
Roscoe, R.J. 1991. Personal Communication, Robert J. Roscoe, NIOSH, to Neal S. Nelson, U.S.
EPA.
Roscoe, R.J., et al. 1989. Lung Cancer Mortality Among Non-Smoking Uranium Miners Exposed
to Radon Daughters. Journal of the American Medical Association. 262(5): 629-633.
Roy, M. and Courtay, C. 1990. Daily Activities and Breathing Parameters for Use in Respiratory
Tract Dosimetry. Commissariat a L'Energie Atornique, France.
Ruosteenoja, E. 1991. Indoor Radon and Risk of Lung Cancer: An Epidemiological Study in
Finland. Finnish Center for Radiation and Nuclear Safety. STUK-A99.
References Page R-6
-------
Samet, J.M. 1989. Radon and Lung Cancer. Journal of the National Cancer Institute. 81(10): 745-
757.
Samet, J.M., Stolwijk, J. and Rose, S. 1991a. Summary: International Workshop on Residential Rn
Epidemiology. Health Physics. 60(2): 223-227.
Samet, J.M. et al. 1991b. Lung Cancer Mortality and Exposure to Radon Progeny in a Cohort of
New Mexico Underground Uranium Miners. Health Physics. 61(6): 745-752
Samet, J.M. and Hornung, R. 1990. Workshop on Indoor Air Quality: Review of Radon and Lung
Cancer Risk. Risk Analysis. 10(1): 65-75.
Schoenberg, J.B., et al. 1990. Case-Control Study oi Residential Radon and Lung Cancer Among
New Jersey Women. Cancer Research. 50: 6520-6524.
Sevc, J., et al. 1988. Cancer in Man After Exposure to Rn Daughters. Health Physics. 54: 27-46.
Sevc, J., et al. 1984. Comments on Lung Cancer Risk Estimates. Health Physics. 46(4): 961-964.
Shimizu, Y., Kato, H. and Schull, WJ. 1988. Life Span Study Report II. Part 2. Cancer Mortality
in the Years 1950-85 Based on the Recently Revised DosesJ^DS 86). Radiation Effects
Research Foundation, Hiroshima. RERF TR 5-88,
Sjoberg, L. 1989. Radon risks: Attitudes, Perceptions, and Actions. Center for Risk Research,
Stockholm School of Economics, Stockholm, Sweden.
Smith, V.K., et al. 1990. Can Public Information Programs Affect Risk Perceptions? Journal of
Policy Analysis and Management. 9(1): 41-59.
Smith, V.K., et al. 1987. Communicating Risk Effectively: A Mid-Course Evaluation. EPA 230-07-
87-029.
Steinhausler, F., Hoffman, W., Pohl, E. and Pohl-Ruhling, J. 1980. Local and Temporal Distribution
Pattern of Radon and Daughters in an Urban Environment. Pp. 1145 in National Radiation
Environment III. T.F. Gesell and W.M. Lowder, eds., NTIS, Springfield, VA. USDOE
CONF-780422
Sterling, R., Meixal, G., Stein, L., Labs, K. and Bligh, T. 1985. Assessment of the Energy Savings
Potentiai of Building Foundations Research'. ORNL/SUB/84-0024/1.
Stranden, E., Berteig, L. and Uglevek, F. 1979. A Study of Radon in Dwellings. Health Physics.
36: 413-421.
Swedjemark, G.A. 1983. The Equilibrium Factor, F. Health Physics. 45: 453-462.
TBWA Advertising. 1990. A Report on ASI Copy Test Results of "X-Ray" Television Commercial.
Prepared for the U.S. EPA and the Advertising Council.
References . Page R-7
-------
TBWA Advertising. 1989, Report on Radon Consumer Concept Focus Groups. Prepared for the
U.S. EPA and the Ad Council.
Thomas, D.C., McNeil, K.G. and Dougherty, C. 1985. Estimates of Lifetime Lung Cancer Risks
Resulting from Rn Progeny Exposures. Health Physics. 49(5): 825-846.
Travis, C.C., Richter Pack, S. and Fisher, A. 1987. Cost-Effectiveness as a Factor in Cancer Risk
Management. Environment International. 13: 369-474.
Turk, B., et af. 1986. Radon and Remedial Action on Spokane River Valley Residences: An Interim
Report. In Proceedings of the Air Pollution Control Association Annual Meeting.
United Nations Scientific Committee on the Effects of Atomic Radiation. 1977. Sources and Effects
of Ionizing Radiation. Report to the General Assembly, New York.
U.S. DOE/Office of Energy Research, 1989. International Workshop on Residential Radon
Epidemiology: Workshop Proceedings. Commission of European Communities, Radiation
Protection Program. CONF-8907178.
U.S. DOE/Office Energy Research. 1988a. Radon Inhalation Studies in Animals. DOE/ER-0396.
U.S. DOE/Office Energy Research. 1988b. Radon Epidemiology: A Guide to the Literature.
DOE/ER-0399.
U.S. EPA. 1992a. Letter from Raymond C. Loehr and Oddvar F. Nygaard to William K. Reilly,
EPA-SAB-RAC-LTR-92-003, January 9, 1992.
U.S. EPA. 1992b. Letter from Science Advisory Board to William K, Reilly. Review of Draft
Revised Citizen's Guide to Radon. EPA-SAB-RAC-LTR-92-OQ5, February 1992.
U.S. EPA. 1991a. An SAB Report: Correlation of Short-Term and Long-Term Test Results for
Indoor Radon. EPA-SAB-RAC-92-008, December 1991.
U.S. EPA. 1991b. Parametric Analysis of the Installation and_ Gjperatmg Costs._pf Active Soil
Depressurization Systems for Residential Radon Mitigation.
U.S. EPA/National Air and Radiation Environmental Laboratory. 1991. Radon Measurement
Uncertainties. Memorandum from Ed Sensintaffar to Frank Marcinowski, U.S. EPA Radon
Division.
U.S. EPA/Office of Radiation Programs. 1991a. National Residential Radon Survey, Statistical
Analysis, National_and Regional Estimates, Volume I.
U.S. EPA/Office of Radiation Programs. 1991h. National Residential Radon Survey, Summary of
Questionnaire Data, Volume II.
U.S. EPA/Office of Water. 1991. Radiorniclides in Drinking Water Factsheet. National Primary
Drinking Water Regulations for Radionuclides Proposed Rule. EPA 570/9-91-700.
References Page R-8
-------
U.S. EPA/Radon Division. November 30, 1991. RCP Interim Radon Mitigation Standards.
U.S. EPA/Radon Division. 1990a. Private Sector Radon Mitigator Survey.
U.S. EPA/Radon Division. 1990b. Technical Support Document for the 1990 Citizen's Guide to
Radon (draft).
U.S. EPA/Radon Division. 1990c. State Radon Assessment Program: Quality..Assurance Project
Plan, prepared by RTI.
U.S. EPA/Radon Division. 1990d. Radon-Measurement Comparison Study: Butte, Montana — 68
Homes. EPA 520/1-89/034.
U.S. EPA/Office of Health and Environmental Assessment. 1989. Exposure Factors Handbook.
EPA 600/8-89-043.
U.S. EPA/Office of Radiation Programs. 1989. Risk Assessments Methodology. Environmental
Impact Statement: NESHAPS for Radipnuclides. Background Information Document,
Volume 1. EPA 520/1-89/005.
U.S. EPA/Office of Research and Development. 1989. Cost of Indoor Radon Reduction Systems.
Memorandum to the U.S. EPA Radon Division Director.
U.S. EPA/Radon Division. (November, 1991 Draft). Indoor Radon and Radon Decay Product
Measurement Protocols.
U.S. EPA. 1988. Letter from Norton Nelson and Williams Schull to Lee M. Thomas. SAB-RAC-88-
042.
U.S. EPA/Office of Public Affairs. 1988. EPA and Assistant General Call for Radon Home Testing.
Environmental News. Monday, September 12, 1988,
U.S. EPA/Office of Health and Environmental Assessment. 1987. The Risk Assessment Guidelines
of 1986. EPA 600/8-87/045.
U.S. EPA/Office of Radiation Programs. 1987. Radon Reference Manual. EPA 520/1-87/20.
U.S. EPA/Office of Research and Development. 1987. Potential Carcinogens for Designation or
Updating. Memorandum from Jim Cogliano to Jack Kooyoomjian, U.S. EPA OERR/ERD.
U.S. EPA/Program Evaluation Division. 1987. Evaluating and I proving EPA's Risk_Advisory
Programs.
U.S. EPA. 1983. Regulatory Impact Analysis Guidelines.
Vanmarcke, H., Reineking, A., Porstendorfer, J. and Rates, F. 1988. Comparison of Two Methods
for Investigating Indoor Radon Daughters. Radiation Protection Dosimetry. 24: 281-284,
References Page R-9
-------
Wagner, D. and Dickson, ML 1991. Radon Media Campaign. U.S. EPA Radon Division, Office of
Air and Radiation.
Walker, M. 1991. Long-Term Confirmatory Testing — Practical Problems. Radon Division, Office
of Air and Radiation, U.S. EPA
Weinstein, N.D. and Sandman, P.M. 1991. Evaluating Risk Comparisons for Use in the_Cffcm'v
Guide to Radon. Rutgers, The State University of New Jersey, Cook College.
Weinstein, N.D., Sandman, P.M. and Roberts, N.E. 1989. Public Response to the Risk from Radon.
1988-1989, Final Report. Rutgers, The State University of New Jersey, Cook College.
Weinstein, N.D., Sandman, P.M. and Roberts, N.E. 1988. Homeowner Radon Mitigation. Summary
Report. Rutgers, The State University of New Jersey, Cook College.
Weinstein, N.D., Sandman, P.M. and Klotz, M. 1987. Public Response to the Risk from Radon,
1986, Final Report. Rutgers, The State University of New Jersey, Cook College.
Wicke, A. and Porstendorfer, J. 1982. Radon Daughters Equilibrium in Dwellings. Pp. 481 in
Natural Radiation Environment. Wiley Eastern Ltd., New Delhi.
Witter, K.S., et al. 1988. Radon Mitigation Choices in the United States - A Comparison of Private
and Public Sector Developments. U.S. EPA, Research Triangle Park, NC.
Woodward, A. et al. 1991. Radon Daughter Exposures at the Radium Hill Uranium Mine and Lung
Cancer Rates Among Former Workers, 1952-87. Cancer Causes and Control. 2: 213-220.
References Page R-10
-------
APPENDIX A
RADON CONCENTRATIONS IN AMBIENT AIR
According to Section 301 of the Indoor Radon Abatement Act, "The national long-term goal
of the United States with respect to radon levels in buildings is that the air within buildings in the
United States should be as free of radon as the ambient air outside of buildings." To ensure that
EPA program goals are consistent with the requirements of this legislation and to confirm reported
outdoor ambient concentrations, EPA reviewed available literature on outdoor radon measurements
and conducted the Ambient Radon Field Study (Hopper 1991).
Individually measured radon concentrations vary widely in the reviewed literature -- from 0.1
to 1.2 picocuries per liter (pCi/L) - while the average concentrations range from about 0,1 to 0.8
pCi/L. In general, concentrations appear to be higher in the western States than in the eastern
States. Articles describing measurements of ambient radon-222 concentrations obtained within one
mile of a uranium mill tailings pile or phosphogypsum stack were not used in this analysis.
The first consideration when measuring the very low radon concentrations expected in
outdoor air is the reliability of the measurement method at such low levels. The methods used to
measure ambient radon concentrations include alpha track detectors, continuous radon monitors,
continuous flow-through monitors (scintillation cells), pump/collapsible bag devices, and electret ion
chambers (EICs). There are very few data establishing accuracy and precision of these methods at
levels of 0.1 to 1 pCi/L. Because of this uncertainty, the data quality objectives for the EPA Ambient
Radon Field Study were:
1. To determine if the majority of the ambient radon levels are < 0.6, < 1.0, or >1.0
pCi/L;
2. To maintain a precision in the measurement that will allow EPA to distinguish
between an ambient air level of 0.2 pCi/L and 0.6 pCi/L; and
3. To assure that the 2-standard deviation error will be no greater than 0.25 pCi/L, at
an ambient radon concentration of 0.25 pCi/L.
The study was designed to measure selected outdoor radon levels across the country. Since
EPA's Office of Radiation Programs (ORP) has Environmental Radiation Ambient Monitoring
System (ERAMS) stations in every State, these stations were selected as the sample sites for making
quarterly outdoor radon measurements. The study was limited in scope and not designed to
statistically represent the distribution of ambient radon concentrations for the United States.
However, estimates of annual average ambient radon concentration and associated error estimates
can be derived at each site since the radon samples were stratified by quarter.
ORP's Las Vegas, Nevada facility (ORP-LVF) conducted the study, but ERAMS station
operators were responsible for deployment and retrieval of detectors. ORP generally relies on State
personnel to operate and maintain these stations. ORP-LVF selected the short-term ETC paired with
a thermolummescent dosimeter (TLD) for this study. The short-term EIC deployed for 90 days
provided adequate sensitivity to meet the data quality objectives to and from the ERAMS stations.
Appendix A Page A-l
-------
Since EICs respond to ionizations both from radon decay within the detector chamber and from
external gamma radiation, TLDs were used to measure and correct for background gamma radiation.
At each ERAMS station, three EICs and three TLDs were placed in ventilated shelters
located approximately one meter from ground level to provide uniformity. The devices were left in
place for 90 days and exchanged each quarter by station operators. The quarterly results were
compared and have been combined into an average annual radon concentration for each site.
The primary goal of the study, to measure outdoor ambient radon concentrations at different
geographic locations in the United States, has been met with good results. For the sites examined,
the mean annual outdoor concentrations ranged from a low of 0.16 at one site to a high of 0.57 pCi/L
at another. The individual concentrations used to calculate mean concentrations ranged from 0.0 to
1.11 pCi/L, with a median of 0.39 pCi/L. In the study to date, field measurements using short-term
EICs have been made with acceptable errors and the devices have exhibited sufficient sensitivity for
measuring ambient levels of radon.
Appendix A Page A-2
-------
APPENDIX B
BACKGROUND INFORMATION ON RADON
Radon is a naturally-occurring, chemically inert, radioactive gas. Because radon is chemically
unreactive with most materials, it is free to travel as a gas. It can move easily through very small
spaces such as those between particles of soil and rock. Radon is odorless, invisible, and without
taste; thus, it cannot be detected with the human senses. Radon is also moderately soluble in water
and, therefore, can be absorbed by water flowing through rock or sand. Its solubility depends on the
water temperature; the colder the water, the greater radon's solubility.
The two natural sources of radon, thorium and uranium, are common, naturally-occurring
elements that are found in low concentrations in rock and soil. Through radioactive decay, both are
constant sources of radon. Radon is produced from the radioactive decay of the element radium,
which is itself a decay product of either uranium or thorium. Radioactive decay is a process in which
an unstable atomic nucleus undergoes spontaneous transformation, by emission of particles or
electromagnetic radiation, to form a new nucleus (decay product), which may or may not be
radioactive. The level of radioactivity is measured in curies, where 1 curie equals 37 billion
disintegrations per second. The time required for a given specific activity of an isotope to be reduced
by a factor of two is called its half-life. A picocurie (pCi) is equal to one-trillionth of a curie.
Specific activity concentrations are typically measured in picocuries per gram (in a solid) or picocuries
per liter (in a gas, such as air).
Uramum-238 decays in several steps to radium-226, which decays into radon-222. Radon-222
has a half-life of 3.8 days and, therefore, has enough time to diffuse through dry, porous soils or to
be transported in water for a considerable distance before it decays. Similarly, thorium-232 decays
into radon-220 (a different radon isotope, also called thoron), which has a half-life of only 55 seconds.
Because of its short half-life and limited ability to migrate into residences, radon-220 is usually a less
important source of radon exposure to humans. The United Nations Scientific Committee on the
Effects of Atomic Radiation has estimated the average exposure from indoor radon-220 decay
products to be about 25 percent of that from radon-222. Only radon-222 is addressed specifically in
the Citizen's Guide because it is the radon isotope of most concern to the public. Although radon-
220, or thoron, has not been measured separately in most homes, radon mitigation will also reduce
exposure to thoron.
Radon-222 is preceded in the uranium-238 decay series by radium-226, which has a half-life
of 1,600 years. Radon-222 decays in several steps to form radioactive isotopes with short half-lives:
polonium-218, lead-214, bismuth-214, and polonium-214. These isotope particles are commonly
referred to as radon decay products, daughters, or progeny. Radon decay products are chemically
reactive and can attach themselves to walls, floors, or airborne particles that are inhaled into the
lungs. Unattached radon decay products also can be inhaled and, subsequently, can become deposited
on lung tissue.
The four radon-222 decay products just mentioned all have half-lives of less than 30 minutes.
This short half-life is significant since, once deposited on lung tissue, the radon decay products can
undergo considerable decay before the action of mucus in the bronchial tubes can clear these
radioactive particles. Two of the short-lived decay products, polonium-218 and polonium-214, emit
Appendix B Page B-l
-------
alpha particles during the decay process. An alpha particle is a subatomic particle that has two
protons and two neutrons and has a double positive electrical charge. It is identical to a helium
nucleus.
Radon-222 is found virtually everywhere in at least small amounts because its predecessor,
radium-226 (or, more distantly, uranium-238), is found in all rock and soil. In outdoor air, radon
concentrations are usually less than one picocurie per liter (pCi/L) (see Appendix A for a summary
of the results of EPA's Ambient Radon Field Study). Higher concentrations of radon outdoors may
be observed during brief periods, such as during a temperature inversion, when a warm air mass traps
a colder one beneath it. The highest individual concentration measured outdoors in EPA's Ambient
Radon Field Study was 1.11 pCi/L. Indoor air concentrations, in contrast, can vary from around 0.5
pCi/L to over 2,000 pCi/L, with results from EPA's National Residential Radon Survey indicating that
over 6 percent of all homes nationwide have average annual indoor radon levels above 4 pCi/L. Most
indoor radon comes from the rocks and soil around a home, although other, usually less significant,
sources of indoor radon are water and some construction materials.
Appendix B Page B-2
-------
APPENDIX C
COMMONLY USED RADON'MEASUREMENT DEVICES
Activated Charcoal Adsorption Devices - Activated charcoal adsorption devices (AC) are
passive devices requiring no power to function. The passive nature of the activated charcoal allows
continual adsorption and desorption of radon. During the measurement period, the adsorbed radon
undergoes radioactive decay. Therefore, the technique does not uniformly integrate radon
concentrations during the exposure period. As with all devices that store radon, the average
concentration calculated using the mid-exposure time is subject to error if the ambient radon
concentration adsorbed during the first half of the sampling period is substantially higher or lower
than the average over the period. For a 2- to 7-day exposure period, the lower level of detection
(LLD) should be 0.5 Pci/L or less (U.S. EPA/Radon Division 1989). This level of sensitivity can
normally be achieved with a counting time of up to 30 minutes. This LLD should be calculated using
the results of charcoal background determinations. The coefficient of variation should not exceed
10 percent (1 sigma) at radon concentrations of 4 pCi/L or greater (U.S. EPA/Radon Division 1989).
This precision should be monitored using the results of duplicate canister analyses. ACs can achieve
an average coefficient of variation of less than 5 percent at concentrations of 4 pCi/L or greater.
Alpha Track Detectors - An alpha track detector (AT) consists of a small piece of plastic or
film enclosed in a container with a filter-covered opening. Radon diffuses through the filter into the
container and alpha particles emitted by radon and its decay products strike the detector and produce
submicroscopic damage tracks. At the end of the measurement period, the detectors are returned
to a laboratory. Plastic detectors are placed in a caustic solution that accentuates the damage tracks
so they can be counted using a microscope or an automated counting system. The number of tracks
per unit area is correlated to the radon concentration in air, using a conversion factor derived from
data generated at a calibration facility. The number of tracks produced per unit time is proportional
to the radon concentration, so an AT functions as a true integrating detector and measures the
average concentration over the measurement period. The LLD (sensitivity) and precision of an AT
system is dependent upon the tracks counted, and therefore, the area of the detector that is analyzed.
With present ATs, routine counting can achieve an LLD of 1 pCi/L-month, and an LLD of 0.2
pCi/L-month may be achieved by counting additional area. The coefficient of variation (precision)
should be monitored using the results of duplicate detectors. The coefficient of variation should not
exceed 20 percent (1 sigma) at radon concentrations of 4 pCi/L or greater (U.S. EPA Measurement
Protocols 1989).
Electret Ion Chamber Radon Detectors - Electret ion chamber radon detectors (EICs) require
no power and function as true integrating detectors, measuring the average concentration during the
exposure period. EICs contain a permanently charged electret (an electrostatically charged disk of
Teflon1), which collects ions formed in the chamber by radiation emitted from radon decay products.
When the device is exposed, radon diffuses into the chamber through filtered openings. Ions that
are generated continuously by the decay of radon and radon decay products are drawn to the surface
of the electret and reduce its surface voltage. The amount of voltage reduction is directly related to
the average radon concentration present during the exposure period. There are both short-term (2-
to 7-day) and long-term (1- to 12-month) EICs that are currently marketed. The thickness of the
electret affects the usable measurement period. For a 7-day exposure period using a short-term EIC?
as well as for a long-term EIC, the LLD (sensitivity) is about 0.3 pCi/L (U.S. EPA/Radon Division
1989). The coefficient of variation should not exceed 10 percent (1 sigma) at radon concentrations
Appendbc C Page C-l
-------
of 4 pCi/L or greater (U.S. EPA/Radon Division 1989). This precision should be monitored by using
results of duplicate detector analyses.
Continuous Radon Monitors - There are three types of continuous radon monitors (CRs).
In the first type, ambient air is sampled for radon in a scintillation cell after passing through a filter
that removes radon decay products and dust. Alpha particles (produced by radon decays) strike the
zinc sulphide coating of the cell, yielding scintillations that are detected by a photo-multiplier tube
in the detector. A second type of CR operates as an ionization chamber. Radon in the ambient air
diffuses into the chamber through a filtered area so that the radon concentration in the chamber
follows the radon concentration in the ambient air with some small time lag. The third type of CR
functions by allowing ambient air to diffuse through a filter into a detection chamber. As radon
decays, the alpha particles are counted using a solid state silicon detector. Most CRs are capable of
an LLD (sensitivity) of 0.5 pCi/L or less (U.S. EPA/Radon Division 1989). Special cells are available
for some CRs that have LLDs of 0.1 pCi/L. The precision of most CRs can achieve a coefficient of
variation of less than 10 percent (1 sigma) at 4 pCi/L or greater (U.S. EPA/Radon Division 1989).
Appendix C Page C-2
-------
APPENDIX D .
COVERAGE OF RADON TESTING POLICY
EPA's estimation of the housing stock and residents covered by its radon testing policy
required the determination of how many units were covered in each category of structures (e.g.
single-family homes) and the occupancy rates of these units. It was assumed that the units covered
would be homes that were intended for regular (year-round) use. Units that were seasonally or
occasionally used (by Census definitions), or could not be linked to regular usage (the Census
classifies them as "other" vacancies) were assumed to not be covered by EPA's testing policy.
The coverage estimates are based on 1990 Census data. Many of these data were available
through 1991 Census press releases. The remainder was made available to EPA from internal Census
documents that are normally used to answer specific inquiries.
Adjustments of the Census data were necessary because the specific statistics needed were
not available (e.g., number of group quarters units), or because EPA's testing policy only covered a
segment of a housing group (e.g., units below the third floor in multi-unit structures.) Some statistics
were available nationally; others had to be constructed from State-level data. The results of the
analysis appear in Exhibit D-l. A summary of how the estimates of housing units and residents
covered for each major housing group were developed follows. >
Housing Units
» Single-Family Homes: The 1990 Census collected data on the number of single-family
homes. This figure was then adjusted to reflect only homes intended for year-round use.1
* Mobile Homes: The number of mobile homes placed on permanent foundations was assumed
to be five percent of all mobile homes. This assumption was based on a review of new
construction reports published by the Census and information collected from trade
associations.
• Multi-Units: The 1990 Census collected data on the number of multi-units. This figure was
then adjusted to reflect only multi-units intended for year-round use, using the same approach
used for single-family homes. The number of apartments in multi-unit structures that are
below the third floor was assumed to he two-thirds of the total number of apartments in
multi-unit structures. This was derived from an assessment of the relationship between multi-
unit structures and numbers of stories in these buildings.
» Group Quarters: The number of group quarters was calculated by dividing the number of
people living in group quarters (also taken from the 1990 Census data) by the national
*A vacancy adjustment factor was calculated by dividing the number of housing units used occasionally by the total
number of housing units and adding that figure to the adjustment factor for homes occupied by residents that "usually
reside elsewhere" (LIRE) (e.g., travellers). The Census Bureau provided a national URE factor of approximately 2
percent. The vacancy adjustment factor was then used to determine the number of units that were not intended for
regular use.
Appendix. D Page D-l
-------
occupancy rate for multi-units.2 An assumption was made that these structures have
comparable occupancy rates given their similarity. The number of group quarters below the
third floor was assumed to be two-thirds the total number of group quarters.
Residents in Housing Units
The population affected by the testing and mitigation program includes all residents of single-
family homes, multi-units, mobile homes, and group quarters that should test.
• Single-Family Homes: Based on information from the Census Bureau, the average occupancy
rate of single-family homes that are intended for year-round use was determined to be about
2.8 persons per home. This occupancy rate was then multiplied by the number of single-
family homes intended for year-round use.
» Mobile Homes and Multi-Units: The occupancy rate for mobile homes and multi-units was
calculated by taking the total national.population, subtracting out the number of people living
in single-family homes and group quarters, and dividing the remainder by the number of
mobile homes and multi-units. This occupancy rate was then multiplied by the number of
mobile homes and multi-units that should test.
• Group Quarters: The number of people residing in group quarters was taken from State-by-
State results from the 1990 Census. The number was multiplied by two-thirds to calculate the
number of residents in group quarters that should test. This was done based on the
assumption that group quarters have the same distribution of stories and building geometry
as multi-unit structures.
2The national occupancy rate of multi-units was assumed to be comparable to the occupancy
rate of group quarters.
Appendix D Page D-2
-------
EXHIBIT D-1 ESTIMATION OF 1990 U.S. POPULATION IN DIFFERENT TYPES OF HOUSING
REQUIRING RADON TESTING AND POTENTIAL MITIGATION BY STATE
(IN THOUSANDS)
STATE
Alabama
:A'lasKaHw:x:;;xt;B«xffxx
Arizona
iftKarjMK" xXxrx'xxXx;:;
California
Connecticut
ifefaMwllxltllii
District of Columbia
FtQf j£j«k-x'X:x-:-: xXXX;-: x>::X-:;v
Georgia
:la'^sp:lfl|if|xfp
Idaho
Indiana
:iJ*PllSSil!ilyt
Kansas
:J*feiiMlWxS:Sx:i:Pi5?-?
Lousiana
Maryland
:MSsK^aS^;;x!ix:
Michigan
^ih^ipWJfSSxilxIs?;
Mississippi
•W&xltii^^'-^-***^
Montana
iietiS^llliliS;
Nevada
iKfSiyi:BaSsp8|iri::Si|i;
New Jersey
^BWxM^^SxxScsS
New York
North Carolina
jfllilMiiill®
Ohio
l|;|||p«;|Il|I;||
Oregon
'j§jMJjM§jijjl
Rhode island
«^tiS^|irSjSa;i?:p::i
South Dakota
f;s^K?S*|SI&lll|l
Texas
^^ftx|siiliillli
Vermont
^fai|8lxili;l;li|i|?
Washington
i^ixMiPisii
Wisconsin
[$$^^S!KS'SCK?.' ';•
-.-:•;• x- -:• -.-'•;•.- -Xv"v:-.-:vX :•, •;•:-:•.-:
4,400
?|||||§|||j||i|
2,130
|iSff||v;£|§g:?
3J32
iilillliBx^i;
3,569
JfSxt;lx;S;;?:?*l??S:'
7,234
pR|li|i;||||
1,963
lilllilli^lll
'. 636
705
iiliilllixlliliss
4,963
;iix:!;i:;:Sl:i|ilo®
8,597
5,037
xliiliilllPil
8,278
2,129
612
-:•:-:•:•:•;•'•;•-.:-:•: : ••:'i-''i':>> -:•:; i'ii'ii.''-*
:;rc vX;,';,";." -;-:-,• •;;:-; -:;>>.';:-^ •'j4oS' •' "
552"
Sxllilllliiftlii
12,349
>lllsfllSi31|?
404
IMM;m$si&$f
3,521
3,534
174,891
MOBILE
HOMES
230
253
654
i; Xv •>:•-'•,'::- og'.';.
:'•-: :':':-.•;•;• >;-"-:•; .':•;
30
3
fliilii
317"
IllflfSf
ss"
170 "
!|H;;M|
76
SxAi|???H|
209
i?iili^
54
IslflsSf
266"
||ii|3|
144
||;||;|i|
54"
lliiilllll
72
iklli|;;M-
73
3!p8|
288
429
iiiiip
239
-139
IIH^I
8
illfsss:;
33'
PHHl!
604
23
ifffimj^i:
197"
|iixi|i;
117
8,067
RESIDENTS OF
MOBILE
HOMES
389
428
v.v. -:--.;-:-:•, •'-:• :•:••-:•. ...;.,.. ,:.-•;.•.
••>••'•:•:• ;>'v:*'-> •'•", •:*'':-"i'7^-""
1,653
x«;!*::x5:-:sJ|lf
llliI51iii:;;
5
iiilis;i!*|!
59l"
||:;|||;j|p|:||;|l;
114
PSiSiiifiSS*
281
iiliiili:ils|
79
l:ifl;Jpi'x!P^i
383"
fisiiiiiiiil
110
•>;< X-' •:•',",' ••:-:•;->:-; .• "•." :1 Q Sli-:
•-;-;.•. .-.v-'.--:..-;.-.-"- --.- .;.;.- -. .-.-.
512
lliiilllll
285
71
140
llsililxiiiefei;;
165"
670
679
lllllllpl^i
" 429
lliilllli
237
17
illllliiiSx
47
lliliiif
1,113
42
357
239
•x.:.x-:::x:iC'XK:-i:x ;:»?::
14,526
MOBILE HOMES
WITH
PERMANENT
FOUNDATIONS
11
13
-:-:•;•••_-: :••:•:••••-;•:•:-: '••'•;'•::-'. •'••'--^' •'•
/•,"xvy''-v>:':'x: '-'.'"•r-.:.:.':''?'-:-
33
1
Plilllli§:i2l
0
xlxliili:>i|S
Ye"
i§s|||ii§lpil
3
s"
iilillllillK
4
Iililiii
10"
p|'?pi|||l;Si:
3
illxixfxill??;:
13
iililiifilif
7
xSKpix*!*?;:^!:
3
'"•'/: •""•": ';•'"' :": ;'":'; •:''•:•:' i-x':-:* i '
4
4"
•ilillllll
14
21
l|p|l||l||i:
12'
;l||lli|ll;;|:
7
0
llpfllif
2
Ittxlll'tl
30
;;||Illi|||l|i;
1
ililliiiii^
10
|||||||;;s|||;S;
e'
iiHiimn
403
RESIDENTS OF
MOBILE HOMES
WITH PERMANENT
FOUNDATIONS
19
21
xlixiiii:i:J;vi|
83
lilllslilllli;
3
x:::xjxj:x;;:^;x;}:Jj:4x:::;|f
0
Sii;:?S:::llPeJx:
30"
ffii|||l|||||l|;:*:S
6
" 14
lillllllllllll
4
pii^llillllslSil
19"
!|Pll!!!!S;|5i
5
llll-l;;flx;5;|:f!|:Sji
26
liisiiliiilfiitiSIl;
14
l||i|i|iilllliJ
4
l-lii:lll!:;ll;i?
7'
s
jx:5;;;?s-;:¥wxSxS¥::rK:i*rt'
fexX&JxxS*Sx:x.x¥axX.;
34
34
¥:::;:" ,:::"':- ,:.::::-:'.<": : ' -.::: :' ~'f .'•:'•:-: :: :: :';:^f:
•:•>:-'''• ,:*-":):::--:-:v';":-:"-"">"-'" :*'•'• :x:":;-: •:••":
21
IllgiSllllll;
12"
Itellllllllfil:
1
Pl||flli;li:||;;:
2
|f||||||||||?|1|i
56
iiiiiilixiilif^
2"
illllliiixil^
18
12"
726
MULTI-UNITS
254
375'
:-:•:-: :-• x-:-:.:-.-..:-:-:-:^^:.:-j
• x_ •:•:•: : /iy:-'-X'i'g*Z:Xt
3.437 I
*f;lll"x|7v8i;|
458
|llplf:;M3
165
578
Ifffflijp
54"
422'
|||||1|||'|
178
:;:|3p;S|||||
323
SSftlpf
483
;i?:l:ll:;;a3S||
"696
|l|ll5;!.4|i|
129
Ixiilll^l
"~ 52'
:;:;:;:;xp:ii:;:|^:;:
173
fxxfl:vS;::i:1S2?;
1 ,067
Htllil
3,509
434
jjSSijiji.
1,050
240
167
*flll|S:S2|B;f
48
liJllJsM:
1,700
piii|p|l||l
53
Illliixi!
'• 480
lllllllll
484
lllllllll
26,569
Appendix D
PageD-3
-------
EXHIBIT D-1 ESTIMATION OF 1990 U.S. POPULATION IN DIFFERENT TYPES OF HOUSING
REQUIRING RADON TESTING AND POTENTIAL MITIGATION BY STATE
(IN THOUSANDS)
STATE
Alabama
S^ilffxSMlif;!
Arizona
||||j^||||||||;;||;
California
xiiiiiiiiii
Connecticut
||||il||||:;ill|
3istriot o< Columbia
Georgia
|S^|i||||::l|!||||
daho
1^iiSlxWs;x-slite5«x
ndiana
^|;;Si;||||;|;||||;;;|
g46l;
283
IPIsiipllill
120
Ilflxil||ts|
216"
mips? l^ll
323
IlliilPlI
466
86
l?S;PfiS§iil
35
^x:-'iS:-''.":5^?>:;:'':'-;^r:^5
lie"
715
illlflilll
2,351
291
Ilillllll
703
161
lllllli
112
tlliixlli
""" 32"
lillsiilPli
1,139
llltlitti
3S"
illllixli
322
324
17,801
RESIDENTS OF
MULTI-UNITS
THAT SHOULD
TEST
289
;SSs*£?
""424"
5,818
620
iiilllilli
182
722
IllllilliiPi
74
469
124
397
liiliiillfj
662
lillllfitll
ass"
171
46
223
1,628
fgiSxIfSllxxj-gip
5,479
461
1,259
Illlillillll
275
225
;:::::::::::::::":^K¥x::;J!CI*:;
^xxx-^XxxSxx^xx—
47
Sl-SIP:^!!:
2,098
llilllliili
64'
.... ••---——'
660
35,239
GROUP
QUARTERS
47
slliixlT<|:;
41
V^SMgjK
380
51
21
88
iiiiiiilii
n
82
tllliii
42
illiii
57"
mmmi&f
:-x¥:;:;:::.:;)::;;:;;;::;,:ifSS>
58"
ilillliil
107
35
Pwlxiai
12
12.
:||l;:l|:iSgli
87"
27(5"
113
132"
|f||!!!|||
33
19
13
199
i"i
iiiiiim
"""" 61
iiiiiii
e'f
SSSSSSSSBBJ
!«p;»«
3,383
RESIDENTS
OF GROUP
QUARTERS
92
|:5;xfxii;:iSil
81
Ililllll^:?
752
ijflllllll
i'bi"
•;•:- :•:• :••• ;•;•:•' -x-x- •••-•.•-••:- •
•:-•-. •'••'•-'-- l-~.-''-;':'--''--''-''-£fl- '•
42
174
iiiiiiiiiii
21
iillliffisSIS:?!
162
ISIIlliibiiJi
83
liillipl
113
::;:::::::v::'::;X;X::,>:;ij^«:*-
x-:x"-:-'x^xx::::-;-;x-:-^x?:
114
212
70
24
iiisiiiii
24"
171
SfSxlSsSil^K
545
224
lllHBlii
261
Illi|i;||p?
66 '
SiivlllSiSx"
39
26
"393
llllllllps
22
121
134
6,698
GROUP
QUARTERS
THAT SHOULD
TEST
31
Kliiislt;
27
Pllffi-li'lsii
'254
ii|il|s!|||||s
34
|;:s|;;:s;x|;;xsK;;|>:
14
lllllliiMj:
59"
8|:||||:|::1;|||
7
55
;iilll;ll»;-
28
38
;:: :.: :"::':' -i"xv>:'< '::::i:i«i:'.':
":^'->:-; •-•;•-: ;:^::S-;-:--:::::. ^::-'
39
||l||Cll^;j
72"
24
8
iliilli
" 8
Illlllllll^
""""'" '"sa"
i,?SSSx£>J>S5ij;jjr:
185
76
|||||||||||?
88"
22
;S:;I;|li;is|||x
13"
•9
133
;.;.;.*. v,-.v. v.v.-.y.;, .•.'•: .•£-•_'
w.x-: •-:-:-: •;'.>-: v. •. : : '.-.'-'10:--.
"""""" " 7"
41
liidiiiili
'""'""""'"""""45
2,267
RESIDENTS OF
GROUP QUARTERS
THAT SHOULD
TEST
62
x-.x- ;X_: :;:.::--. :;-x.x:-::x.:;V ^.x
.v'jXv-?>xxSA '&••< ixxjftx
54"
A|M^P;?511!^:'
504
§illll|||llil:§3;:
""'es'
l;i:lfilxi;Ex*li5A
28
Illllllplplll;!
116
14
l*lf«xiiillls§2l
109
55
iiiixllllitli^i
75"
XvX-XvXv; ;::;y ::>;-.>->:"y"O"i';:
;;%x:::;::;;>;;s:;;:;:-y;;:^:::v;:._.;::-;.?r:;;
76
llllll!:lls-:lS
142
IllillliilMl
47
:xi»!tx«?S;:?fiw;i
16
Ilillliil^
""". IB'
|||ifS|s:;||||l|^|
•••••'••••••••••••-^•g-
ilSiS S;:x!xfS: xix IjktflS
365
150
iliillliil
175
||||||||||||rra|
44
ilSslllllilllll
26
•iiiiiiiiiiijiiSi
17
264
llllllPlllii
15
81
iiSaiaKSsssKxS&W
90"
4,487
ALL UNITS
THAT SHOULD
TEST
1,330
SxM'-xKxxiTS:?
1^192
vSi:S:;^8S:l!:
9,258
||i:;|:fi;I||s|
i,128
228
liliii^l
2^118
318
;;ix:xvJx:sSfZ96^
1,918
zlllSiilil^Ss:
912
|||!:w||i2|;||
1,384 '
•••:-:v;v;-.-.vX-x::-:^-v-v-:-:
;:??-i-;-;Hs;v-:3^*s
1.639
Illlxl^gip
3,"l3S
818
,.;•;-. -.•:-:.-. .;-:-:-;•:%-.'-.- ..^-.':
':v':v "',•'• ':"-""'"S':'J"*'(BS8"i"
273
379
llli-llitps-S
2.549"
:8iK:4y:x;:x::W:474;i:
5.620
2.187
3,760
llllliPPI
951
":::'::-':;x:.::;:-''-:r^'^'1:o'@;''
;,•:;" -,-.-.-,•, ••;-:•-"*-*•- -v.?;^
344
|l|H|l;llp|i!;
240
ixlisilllfsi
5,713
Ilillilll9*§
188
lilllliPI
1 ,630
1.637
82,932
^ — ' ™ — ,
RESIDENTS OF
ALL UNITS
THAT SHOULD
TEST
3,498
:|x •••::":'.x.x""".d^7'.
:x .:. . . -- . Vx-Xx.v:::??/.: V.
3^023
x^lill^iii::;
25,076
2.891
:^xx:xv:-^x^^-^xix.:•.
X:":xxx::::.xx:;xg^6:;:
499
pxSf3§fi:i
5,604'
i|lii|£|;||^!;;:
855
ISxfeSlBi?*
4,992
2.314
lllli^lli;:
',3.623
:••'•-/'•<:'.. ->:':": ~::-;i:':iiij V.' "
•••::':^; ' :'-:- :::;:7:'^^/:.;.
4,313"
f$'v,''-*'-:t'sl:•- .-• -•- --. .- ;..
•'., -' : ; '. •:•'. ', •'•'^-•C^'fi''"
--, ;-:-: .:. ':•'-''.'-.': '• . '• s *.—& -.--'•'
2,195
701
951
|||;::|i||:x;:;|Sg|
6,714
sisjiysst^gSji
14,475
5,682
f||p;|fsSi|
9,734
lillllijiii*
2.460
864
Illi*ii
619
14,766
ISllllxfl
484
Siilliplii
4,202"
miiitiiii
4,296
215,344
Appendix D
Page D-4
-------
APPENDIX E
RISK ANALYSIS METHODOLOGY
The assessment of reductions in risks that result from the public following the testing
procedure advice in the final revised Citizen's Guide ("Testing Option C" in Chapter 3) for
alternative action levels of 2 pCi/L, 3 pCi/L, and 4 pCi/L had two major components:
• Estimation of the number of residents in homes that have testing results above the
action level and should mitigate their radon levels,
• Calculation of the risk reductions that should result when these mitigations occur.
Each of these components of the analysis is discussed in the sections that follow.
Residential Population that Should Mitigate
To determine the size of the residential population that would be mitigating their homes, it
was necessary to (1) estimate the residential coverage of EPA's radon testing policy and (2) predict
the testing results that this group would have if it followed EPA's advice in the revised Citizen's
Guide. Appendix D explains how the Agency estimated the population covered by EPA's testing
policy. Based on 1990 Census data, EPA estimates that about 215 million people are covered by the
testing policy. The remainder of this discussion explains how EPA predicted the results of this
group's radon testing (using Option C).
The public is advised to conduct an initial short-term test. If the results are above the action
level, Option C gives the public the choice of (1) conducting a follow-up long-term test and fixing
the house if the results are above the action level (like Option A); or 2) conducting a second short-
term test and fixing the house if the average of the two results is above the action level (like Option
B), EPA believes that under Option C, the public will actually pursue Option A testing about 9
percent of the time and Option B testing about 91 percent of the time. EPA used a statistical model
to predict the radon test results from using Testing Option C for alternative action levels by
independently looking at the predicted outcomes for Option A and Option B for each action level
(Chmelymki 1990). EPA weighted the results by the expected usage of each testing approach by the
public and then combined them to provide Option C results.
The statistical model used to predict testing results provided the joint probability distribution
for the results of short-term testing given the actual iong-lerm radon levels houses were likely to
have. Exhibits E-l through E-3 show these distributions for each action level examined. Each exhibit
provides results of Option A, Option B, and Option C, which is the combination of the preceding two
options. Summations of the appropriate portions of each of these distributions leads to estimates of
the fraction of homes that will need to be fixed. Multiplying these fractions by the residential
population covered in EPA's testing policy yields the number of persons affected. For instance, the
fraction of homes that have less than 1 pCi/L of radon and that will have false positive radon testing
results under Option B is the summation of the exhibit entries in the row "0-1" that begins from "4-6"
to ">20." This fraction multiplied by 215 million people provides the population that lives in homes
that are below 1 pCi/L and that would get false positive results, if all homeowners used Option B to
test their homes for radon.
Appendix E Page E-l
-------
EPA generated the distribution tables for each action level based on testing devices used,
number of home floors, and testing conditions.1 EPA assumed there was an equal chance of
homeowners using short-term devices like charcoal canisters and alpha track detectors. For charcoal
canisters, EPA assumed that all tests were done under closed-house conditions.2 For alpha track
detectors (ATDs), EPA assumed that 50 percent of tests would occur under open-house conditions
and that 50 percent would occur under closed-house conditions. Based on data from the 1990 Census
Data and the Characteristics of New Housing: 1990 Current Construction Reports, the weights for
single floor vs. multi-floor homes were determined to be 46 and 54 percent, respectively. (The same
approach that EPA used to derive the mitigation classification rates in Chapter 3 was also used here,
except the focus was on options that had different action levels using the same Option C testing
approach.)
Risk Reductions Due to Mitigation
EPA assumed for the Chapter 5 analysis that all homeowners of houses testing above the
action levels in long- or short-term testing installed and operated mitigation systems. The Agency
used the radon testing results shown in the first three exhibits to develop estimates of homes at
various radon levels that mitigated, as well as the exposure reductions they would receive, for each
of the selected pCi/L intervals in these exhibits. To estimate the exposure reductions received in
each interval, EPA subtracted the level of radon existing after mitigation from the radon levels
estimated to exist within each interval before mitigation. The average pre-mitigation radon levels in
the intervals were taken from results of the EPA National Residential Radon Survey. Exhibit E-4
provides these average levels. Based on the mitigation research and experience that is summarized
in Chapter 4, EPA assumed that homes that were above 2 pCi/L would have their levels reduced to
2 pCi/L (on average) and homes below that level would not have any radon reductions.3 The same
type of approach was used to estimate the risk reductions that were not gained ("lives lost") when
mitigation of homes above the action levels did not occur due to false negative testing results.
To facilitate the assessment of risk reductions that result from exposure reductions, EPA
developed multipliers that provided the lung cancer deaths averted per million persons receiving the
exposure reduction in each interval. The multipliers were based on EPA's central estimate of the
risk factor (i.e., that there will be 43.2 lung cancer deaths annually per million persons annually
exposed to one pCi/L). Exposure reductions (in pCi/L) for each interval were multiplied by the risk
factor to arrive at each interval's own multiplier. Exhibit E-4 shows the multipliers that were used
in all the analyses of the action levels.
EPA independently calculated risk reductions gained (or lost) for homes that had false
positive test results, true positive test results, and false negative test results. Exhibjts E-5 through E-7
show these results for each action level The final results that appear in Chapter 5 are the Option
C results in each exhibit. The intermediate results for Options A and B are also shown in the
exhibits. These results assume that the entire testing population (215 million people) follows that
Distribution tables were generated using An evaluation of the, performance of alternative shortTterm radon testing
procedures in homes with pending real estate transactions. Chmelynski, H. 1990. Submitted to EPA.
2For a more detailed explanation of radon tests, please refer to Chapter 3.
3There will often be some risk reduction that will occur for homes below 2 pCi/L Due to the lack of data on exactly how
much reduction, however, EPA made a conservative assumption that there would be no reduction.
Appendix E Page E-2
-------
testing option. The results from Options A and B are then weighted by the portion of the population
that follows each option to derive the results for Option C. Exhibit E-8 helps explain the reasons
for the differences in the levels of risk reduction that can occur given the exposure distribution that
exists in the U.S. housing stock.
Appendix E Page E-3
-------
OPTION A
Exhibit E-l
Weighted Testing Distributions
at Action Level 4pCi/L
Average
(pCi/L)
0-1
1-2
2-3
3-4
4-6
6-8
8-10
10-20
>20
TOTAL
OPTION
Short-term Measurement (pCi/L)
0
0-1
.4628973 !
0.07260808
0.00935460
0.
0
0
0
0
0
B
,00196702
,00068686
.00005831
.00001098
.00000251
0
,54758572
Two refunds of testing for
Average
(pCi/L)
0-1
1-2
2-3
3-4
4-6
6-8
8-10
10-20
>20
TOTAL
OPTION
Average
(pCi/L)
0-1
1-2
2-3
3-4
4-6
6-8
8-10
10-20
>20
TOTAL
1-2
0.09305886
0,06399632
0.01951012
0.00694166
0.00364974
0.00047387
0.00011760
0.00003546
0.00000019
0.18778386
positives (one
2-3
0.03151984
0.03126357
0.01472083
0.00740183
0.00532080
0.00095279
0.00028938
0.00010599
0.00000091
0.09157597
3-4
0.01374314
0.01589892
0.00945063
0.00588678
0.00529788
0.00121286
0.00043318
0.00018565
0.00000217
0.05211125
4-6
0.01085996
0.01373066
0.00979078
0.00734195
0.00824628
0.00245525
0.00106341
0.00055858
0.00000943
0.05405634
6-8
0.00384689
0.00507591
0.00421989
0.00372795
0.00515586
0.00197302
0.00103796
0.00068153
0.00001679
0.02573583
8-10
0.00168459
0.00221464
0.00201153
0.00196585
0.00312115
0.00142079
0.00086413
0.00068512
0.00002330
0.01399113
10-20
0.00187647
0.00232588
0.00231958
0.00254703
0.00480042
0.00274733
0.00206755
0.00231919
0.00015659
0.02116006
>20
0.00031058
0.00028592
0.00032199
0.00041995
0.00102097
0.00080572
0.00081574
0.00162592
0.00039030
0.00599713
TOTAL
0.61979767
0.20739993
0.07169998
0.03820005
0.03730000
0.01209997
0.00669997
0.00619998
0.00059971
0.99999731
round of testing for negatives)
Short-term Measurement (pCi/L)
0
0
0-1
.46289731
.07260808
0.00935460
0
.00196702
0,00068686
0.
0
.00005831
.00001098
0.00000251
0
C
0
.54758572
1-2
0.09305886
0.06399632
0.01951012
0.00694166
0.00364974
0.00047387
0.00011760
0.00003546
0.00000019
0.18778386
2-3
0.03930476
0.03809300
0.01772150
0.00874512
0.00613777
0.00106081
0.00031517
0.00011330
0.00000094
0. 1 1 149240
3-4
0.01827882.
0.02211290
0.01379046
0.00871796
0.00771167
0.00165693
0.00055988
0.00022744
0.00000241
0.07305851
4-6
0.00393081
0.00664640
0.00636274
0.00570512
0.00703149
0.00199744
0.00075293
0.00031344
0.00000277
0.03274317
6-8
0.00132012
0.00236203
0.00278552
0.00314840
0.00529483
0.00224083
0.00113674
0.00062719
0.00000856
0.01892425
8-10
0.00051290
0.00087742
0.00114636
0,00146495
0.00297480
0.00164993
0.00106112
0.00077200
0.00001597
0.01047548
10-20
0.00045174
0.00066808
0.00096730
0.00139973
0.00344802
0.00256784
0.00225052
0.00278277
0.00016090
0.01469695
>20
0.00004232
0.00003566
0.00006135
0.00011005
0.00036477
0.00039397-
0.00049497
0.00132583
0.00040793
0.00323690
TOTAL
0.61979768
0.20739994
0.07169998
0.03820005
0.03729999
0.01209996
0.00669995
0.00619996
0.00059972
0.99999728
Short-term Measurement (pCi/L)
0-1
0.46456937
0
0
0
0
0
0
.07260808
.00935460
.00196702
,00068686
.00005831
.00001098
0.00000251
0
0
,54925778
1-2
0.09305886
0.06612329
0.01951012
0.00694166
0.00364974
0.00047387
0.00011760
0.00003546
0.00000019
0.18991083
2-3
0.03860412
0.03747835
0.01913118
0.00862422
0.00606425
0.00105109
0.00031285
0.00011264
0.00000093
0.11137967
3-4
0.01787061
0.02155364
0.01339987
0.00990340
0.00749443
0.00161697
0.00054848
0.00022367
0.00000239
0.07261350
4-6
0.00357704
0.00604823
0.00579009
0.00519166
0.00840968
0.00181767
0.00068517
0.00028523
0.00000252
0.03180731
6-8
0.00120131
0.00214945
0.00253482
0.00286504
0.00481829
0.00288534
0.00103444
0.00057074
0.00000779
0.01806726
8-10
0.00046674
0:00079846
0.00104318
0.00133310
0.00270707
0.00150144
0.00149201
0.00070252
0.00001453
0.01005908
10-20
0,00041108
0.00060795
0.00088024
0.00127375
0.00313769
0.00233673
0.00204798
0.00306066
0.00014642
0.01390256
Appendix E
>20
0.00003851
0.00003245
0.00005582
0.00010015
0.00033194
0.00035851
0.00045042
0.00120650
0.00042490
0.00299925
TOTAL
0.61979768
0.20739994
0.07169998
0.03820005
0.03729999
0.01209996
0.00669996
0,00619997
0.00059972
0.99999728
Page E-4
-------
Exhibit E-2
Weighted Testing Distributions
at Action Level 3pCi/L
OPTION A
Average
(pCi/L)
0-1
1-2
2-3
3-4
4-6
6-8
8-10
10-20
>20
TOTAL
OPTION
Short-term Measurement (pCi/L)
0-1
0.46289731
0
,07260808
0.00935460
0
0
0
0
0
0
B
.00196702
.00068686
.00005831
.00001098
.00000251
0
.54758572
Two rounds of testing for
Average
(pCi/L)
0-1
1-2
2-3
3-4
4-6
6-8
8-10
10-20
>20
TOTAL
OPTION
Average
(pCi/L)
01
1-2
2-3
3-4
4-6
6-8
8-10
10-20
>20
TOTAL
1-2
0.09305886
0.06399632
0.01951012
0.00694166
0.00364974
0.00047387
0.00011760
0.00003546
0.00000019
0.18778386
2-3
0.03151984
0.03126357
0.01472083
0.00740183
0.00532080
0.00095279
0.00028938
0.00010599
0.00000091
0.09157597
3-4
0.01374314
0.01589892
0.00945063
0.00588678
0.00529788
0.00121286
0.00043318
0.00018565
0.00000217
0.05211125
4-6
0.01085996
0.01373066
0.00979078
0.00734195
0.00824628
0.00245525
0.00106341
0.00055858
0.00000943
0.05405634
6-8
0,00384689
0,00507591
0.00421989
0,00372795
0.00515586
0.00197302
0.00103796
0.00068153
0.00001679
0,02573583
8-10
0.00168459
0.00221464
0.00201153
0.00196585
0.00312115
0.00142Q79
0.00086413
0.00068512
0.00002330
0.01399113
10-20
0.00187647
0.00232588
0.00231958
0.00254703
0.00480042
0.00274733
0.00206755
0.00231919
0.00015659
0.02116006
>20
0.00031058
0.00028592
0.00032199
0.00041995
0.00102097
0.00080572
0.00081574
0.00162592
0.00039030
0.00599713
TOTAL
0.61979767
0.20739993
0.07169998
0.03820005
0.03730000
0.01209997
0.00669997
0.00619998
0.00059971
0.99999731
positives (one round of testing for negatives)
Short-terra Measurement (pCi/L)
0-1
0.46289731
0
0
.07260808
.00935460
0.00196702
0
0
0
0
0
c
.00068686
.00005831
.00001098
.00000251
0
.54758572
1-2
0.10153744
0.06937046
0.02100491
0.00739270
0.00384442
0.00049249
0.00012124
0.00003631
0.00000020
0.20380020
2-3
0.04280681
0.04403728
0.02111349
0.01041371
0.00716892
0.00119839
0.00034771
0.00012226
0.00000097
0. 12720958
3-4
0.00575363
0.00917330
0.00693803
0.00467816
0.00401302
0.00074225
0,0002 1113
0.00006876
0.00000038
0.03157871
4-6
0.00445658
0.00812751
0.00806765
0.00729849
0.00893349
0.00251960
0.00094417
0.00038884
0.00000343
0.04073978
6-8
0,00136478
0.00251395
0.00302242
0,00342878
0.00574629
0.00241887
0.00122360
0.00067140
0.00000920
0,02039933
8-10
0.00050742
0.00088221
0.00117354
0.00150674
0.00306243
0.00169167
0.00108561
0.00078679
0.00001633
0.01071278
10-20
0.00043364
0.00065262
0.00096456'
0.00140473
0.00347800
0.00258330
0.00225984
0.00279025
0.00016108
0.01472806
>20
0.00003993
0.00003448
0.00006072
0.00010969
0.00036654
0.00039505
0.00049565
0.00133281
0.00040808
0.00324298
TOTAL
0.61979758
0.20739994
0.07169996
0.03820005
0.03730000
0.01209997
0.00669997
0.00619996
0.00059970
0.99999718
Short-terra Measurement (pCi/L)
0
0
0-1
.46580626
.07260808
0.00935460
0
0
0
0
.00196702
.00068686
.00005831
.00001098
0.00000251
0
0
.55049467
1-2
0.10077437
0.07244466
0.02087038
0.00735210
0.00382690
0.00049081
0.00012092
0.00003623
0.00000020
0.20591660
2-3
0.04179098
0.04288765
0.02306845
0.01014264
0.00700259
0.001 17629
0.00034246
0.00012080
0.00000096
0. 12653285
3-4
0.00523581
0.00834770
0.00631361
0.00622718
0.00365185
0.00067544
0.00019213
0.00006257
0.00000035
0.03070668
4-6
0.00405548
0.00739603
0.00734156
0.00664162
0.01061731
0.00229283
0.00085920
0.00035384
0.00000312
0.03956104
6-8
0,00124195
0,00228769
0,00275040
0.00312018
0,00522912
0.00315652
0.00111348
0.00061098
0,00000837
0.01951874
8-10
0.00046175
0.00080281
0.00106792
0.00137113
0.00278681
0.00153942
0.00155328
0.00071597
0.00001486
0.01031401
10-20
0.00039461
0.00059389
0.00087775
0.00127831
0.00316498
0.00235081
0.00205645
0.00308416
0.00014658
0.01394758
Appendix E
>20
0.00003633
0.00003138
0.00005525
0.00009982
0.00033355
0.00035950
0.00045104
0.00121285
0.00042523
0.00300498
TOTAL
0.61979759
0.20739994
0.07169996
0.03820005
0.03730000
0.01209997
0.00669997
0.00619996
0.00059970
0.99999719
Page E-5
-------
OPTION A
Exhibit E-3
Weighted Testing Distributions
at Action Level 2pCi/L
Average
(pCi/L)
0-1
1-2
2-3
3-4
4-6
6-8
8-10
10-20
>20
TOTAL
0-1
0.46289731
0.07260808
0.00935460
0.00196702
0.00068686
0.00005831
0.00001098
0.00000251
0
0.54758572
Short-term Measurement (pCi/L)
1-2 2-3 34 4-6 6-8 8-10 10-20 >20 TOTAL
0.09305886 0.03151984 0,01374314 0.01085996 0.00384689 0.00168459 0.00187647 0.00031058 0.61979767
0.06399632 0.03126357 0.01589892 0.01373066 0.0050759! 0.00221464 0.00232588 0.00028592 0.20739993
0.01951012 0.01472083 0.00945063 0.00979078 0.00421989 0.00201153 0.00231958 0.00032199 0.07169998
0.00694166 0.00740183 0.00588678 0.00734195 0.00372795 0.00196585 0.00254703 0.00041995 0,03820005
0.00364974 0.00532080 0.00529788 0.00824628 0.00515586 0.00312115 0.00480042 0.00102097 0.03730000
0.00047387 0.00095279 0.00121286 0.00245525 0.00197302 0.00142079 0.00274733 0.00080572 0.01209997
0.00011760 0.00028938 0,00043318 0.00106341 0.00103796 0.00086413 0.00206755 0.00081574 0.00669997
0.00003546 0.00010599 0.00018565 0.00055858 0.00068153 0.00068512 0,00231919 0.00162592 0.00619998
0,00000019 0.00000091 0:00000217 0.00000943 0.00001679 0.00002330 0.00015659 0.00039030 0,00059971
0.18778386 0.09157597 0.05211125 0.05405634 0.02573583 0.01399113 0.02116006 0.00599713 0.99999731
OPTION B
Two rounds of testing for positives (one round of testing for negatives)
Average Short-term Measurement (pCi/L)
(pCi/L) 0-1 1-2 2-3 3A 4-6 6-8 8-10 10-20 >20 TOTAL
0-1 0.46289731 0.12680067 0.01613261 0.00694534 0.00476502 0.00133799 0.00048029 0,00040174 0.00003646 0.61979747
1-2 0.07260808 0.08653552 0.02265543 0.01235090 0.00916006 0.00255170 0.00086859 0.00063603 0.00003357 0.20739993
2-3 0.00935460 0.02569338 0.01237851 0.00971940 0.00923427 0.00311838 0.00-117925 0.00096163 0.00006051 0.07169997
3-4 0.00196702 0.00877038 0.00597427 0.00659872 0.00831560 0.00353820 0.00151918 0,00140689 0.00010977 0.03820007
4-6 0.00068686 0.00441416 0.00364563 0.00561705 0.01005635 0.00591306 0.00309478 0.00350249 0.00036960 0.03730000
6-8 0.00005831 0,00054525 0.00048325 0.00103758 0.00281113 0.00247601 0.00170199 0,00259007 0.00039636 0.01209999
8-10 0.00001098 0.00013130 0.00011404 0,00029464 0.00104921 0.00125052 0.00109080 0.00226217 0.00049627 0.00669998
10-20 0.00000251 0.00003855 0.00003140 0.00009533 0.00042971 0.00068433 0.00078908 0.00279173 0.00133729 0.00619997
>20 0 0.00000020 0.00000012 0.00000053 0.00000379 0.00000941 0.00001640 0.00016107 0.00040815 0.00059970
TOTAL 0.54758572'0.25292944 0.06141529 0.04265953 0.04582517 0.02087963 0.01074041 0.01471386 0.00324803 0.99999711
OPTION C
Average Short-term Measurement (pCi/L)
(pCi/L) 0-1 1-2 2-3 3-4 4-6 6-8 8-10 10-20 >20 TOTAL
0-1 0.46864304 0.12376390 0.01468067 0.00632026 0.00433617 '0.00121757 0.00043707 0.00036559 0.00003318 0.61979749
1-2 0.07260808 0.09087859 0.02061644 0.01123932 0.00833566 0.00232205 0.00079042 0.00057879 0.00003055 0.20739993
2-3 0.00935460 0.02513689 0.01511961 0.00884466 0.00840318 0,00283772 0.00107312 0,00087508 0.00005507 0.07169997
3-4 0.00196702 0.00860579 0.00543659 0.00864105 0.00756720 0.00321976 0.00138245 0,00128027 0.00009989 0.03820007
4-6 0.00068686 0.00434537 0.00331752 0.00511152 0.01211798 0.00538088 0.00281625 0.00318726 0.00033633 0.03730000
6-8 0.00005831 0.00053883 0.00043975 0.00094420 0.00255813 0.00329427 0.00154881 0.00235697 0.00036069 0.01209998
8-10 0.00001098 0.00013007 0.00010378 0.00026812 0.00095478 0,00113797 0.00158405 0.00.205857 0.00045161 0.00669998
10-20 0.00000251 0.00003827 0.00002858 0.00008675 0.00039103 0,00062274 0.00071806 0,00309506 0.00121693 0.00619997
>20 0 0.00000020 0.00000011 0.00000048 0.00000345 0,00000856 0.00001492 0,00014658 0.00042537 0.00059970
TOTAL 0.55333145 0.25343794 0.05974308 0.04145639 0.04466761 0.02004156 0.01036520 0.01394420 0.00300966 0,99999713
Appendix E
Page E-6
-------
Exhibit E-4
Radon Exposure and Deaths Per Million
In Each Interval
pCiA
Interval
0-1
1 -2
2-3
3-4
4-6
6-8
8-10
10-20
>20
Arithmetic
Mean Exposure
In Interval
0.440
1.400
2.440.
3.430
4.942
6.900
8.950
12.890
29.250
Deaths per
Million in
Interval
NONE
NONE
19
62
127
212
300
470
1,177
Appendix E
Page E-7
-------
Option A
Exhibit E-5
Risk Calculations for Existing HOOKS Using
Distribution for All Homes that Should Test
at Action Level 4pCi/L
Radon
pCi/1
Interval
0-1
1-2
2—3
3-4
4-6
6-8
8-10
10-20
>20
4-6
6-8
8-iO
10-20
>20
Test
Status*
FP/TN
FP/TN
FP/TN
FP/TN
Tp
TP
TP
TP
TP
FN
FN
FN
FN
FN
Fraction of
Population
0.6198
0.2074
0.071?
0.0382
0.0223
0.0094
0.0058
0.0059
0.0006
0.0150
0.0027
0.0009
0.0003
0.0000
Radon
Reduction
Level
NONE
NONE
NONE
NONE
2
2
2
2
2
NONE
NONE
NONE
NONE
NONE
Annual Deaths
Averted per
Million Persons
NONE
NONE
NONE
NONE
127
212
300
470
1,177
127
212
300
470
1,177
Annual Lives
Saved
(Lost)
0
0
0
0
612
429
378
595
151
(409)
(123)
(55)
(33)
(1)
False
Positive
0
True
Positive
2,164
False
Negative
(622)
Option B
Radon
pCi/1
Interval
0-1
1-2
2-3
3-4
4-6
6-8
8-10
10-20
Test
Status*
FP/TN
FP/TN
FP/TN
FP/TN
TP
TP
TP
TP
>20 1 TP
4-6
6-8
8-10
10-20
>20
FN
FN
FN
FN
FN
Fraction of
Population
Radon
Reduction
Level
0.6198 | NONE
0.2074
0.0717
0.0382
0.0191
0.0089
0.0057
0.0058
0.0006
0.0182
0.0032
0.0010
0.0004
0.0000
NONE
2
2
2
2
2
2
2
NONE
NONE
NONE
NONE
NONE
Annual Deaths
Averted per
Million Persons
NONE
NONE
19
62
127
212
300
470
1,177
127
212
300
470
1,177
Annual Lives
Saved
(Lost)
0
0
46
157
523
403
368
590
151
(498)
(148)
(65)
(38)
(1)
False
Positive-
204
True
Positive
2,036
. False
Negative
(750)
Appendix E
Page E-8
-------
Exhibit E-5 (continued)
Option C
Radon
pCi/1
Interval
0-1
1-2
2-3
3-4
4-6
6-8
8-10
10-20
>20
4-6
6-8
8-10
10-20
>20
Test
Status*
FP/TN
FP/TN
FP/TN
FP/TN
TP
TP
TP
TP
TP
FN
FN
FN
FN
FN
Fraction of
Population
0.6198
0.2074
0.0717
0.0382
0,0194
0.0089
0.0057
0.0058
0,0006
0.0179
0.0032
0.0010
0.0004
0.0000
Radon
Reduction
Level
NONE
NONE
NONE/2**
NONE/2**
2
2
2
2
2
NONE
NONE
NONE
NONE
NONE
Annual Deaths
Averted per
Million Persons
NONE
NONE
17
56
127
212
300
470
1,177
127
212
300
470
1,177
Annual Lives
Saved
(Lost)
0
0
42
143
531
406
369
590
151
(490)
(146)
(64)
(38)
(1)
False
Positive
185
True
Positive
2,047
False ;
Negative
(738)
*True Positive (TP), False Positive (FP), True Negative-(TN), and False Negative (FN). Assumes 100% testing
and mitigation. Column entries may not sum due to rounding of estimates.
**There is no radon reduction level for TN homes, there is a radon reduction level of 2 for homes that test FP.
Appendix E
PageE-9
-------
Option A
Exhibit E-6
Risk Calculations for Existing Homes Using
Distribution for All Homes that Should Test
at Action Level 3pCi/L
Radon
pCi/1
Interval
0-1
1-2
2—3
3-4
4-6
6-8
8-10
10-20
>20
3-4
4-6
6-8
8-10
10-20
>20
Test
Status*
FP/TN
FP/TN
FP/TN
TP
TP
TP
TP
TP
TP
FN
FN
FN
FN
•FN
FN
Fraction of
Population
0.6198
0.2074
0.0717
0,0219
0.0276
0.0106
0.0063
0.0061
0.0006
0.0163
0.0097
0.0015
0.0004
0.0001
0.0000
Radon
Reduction
Level
NONE
NONE
NONE
o
2
2
2
j*j
2
NONE
NONE
NONE
NONE
NONE
NONE
Annual Deaths
Averted per
Million Persons
NONE
NONE
NONE
62
127
212
300
470
1,177
62
127
212
300
470
1,177
Annual Lives
Saved
(Lost)
0
0
0
291
757
484-
406
614
152
(217)
(264)
(68)
(27)
(15)
(0)
False
Positive
: - °
•' v' ''.True:: '-•,;•
Positive
: ., 2,703
;- ;• ; . .•' .
False
Negative
° •
(591)
- -.. .
Option B
Radon
pCi/1
Interval
0-1
1-2
2-3
3-4
4-6
6-8
8-10
10-20
>20
3-4
4-6
6-8
8-10
10-20
>20
Test
Status*
FP/TN
FP/TN
FP/TN
TP
TP
TP
TP
TP
TP
FN
FN
FN
FN
FN
FN
Fraction of
Population
0.6198
0.2074
0.0717
0.0184
0.0256
0.0096
0.0060
0.0060
0.0006
0.0198
0.0117
0.0025
0.0007
0.0002
0.0000
Radon
Reduction
Level
NONE
NONE
2
2
2
2
2
2
2
NONE
NONE
NONE
NONE
NONE
NONE
Annual Deaths
Averted per
Million Persons
NONE
NONE
19
62
127
212,
300
470
1,177
62
127
212
300
470
1,177
Annual Lives
Saved
(Lost)
0
0
83
245
701
438
389
605
152
(263)
(320)
(114)
(45)
(23)
(0)
False
Positive
83
True
Positive
2,529
False
Negative
(765)
Appendix E
Page E-10
-------
Exhibit E-6 (continued)
Option C
Radon
pCi/1
Interval
0-1
1-2
2-3
3-4
4-6
6-8
8-10
10-20
>20
3-4
4-6
6-8
8-10
10-20
>20
Test
Status*
FP/TN
FP/TN
FP/TN
TP
TP
TP
TP
TP
TP
FN
FN
FN
FN
FN
FN
Fraction of
Population
0.6198
0.2074
0.0717
0.0187
0.0258
0.0097
0.0060
0.0060
0.0006
0,0195
0.0115
0.0024
0.0007
0.0002
0.0000
Radon
Reduction
Level
NONE
NONE
NONE/2**
2
2
2
2
2
2
NONE
NONE
NONE
NONE
NONE
NONE
Annual Deaths
Averted per
Million Persons
NONE
NONE
17
62
' 127
212
300
470
1,177
62
127
212
300
470
1,177
Annual Lives
Saved
(Lost)
0
0
75
249
706
442
.390
606
152
(259)
(315)
(109)
(43)
(23)
(0)
False
Positive
75
True
Positive
2,544:
False
Negative
.. . .' (750)
*True Positive (TP), False Positive (FP), True Negative (TN), and False Negative (FN). Assumes 100% testing
and mitigation. Column entries may not sum due to rounding of estimates,
**There is no radon reduction level for TN homes, there is a radon reduction level of 2 for homes that test FP.
Appendix E
Page E-ll
-------
Option A
Exhibit E-7
Risk Calculations for Existing Homes Using
Distribution for All Homes that Should Test
at Action Level 2pCi/L
Radon
pCi/1
Interval
0-1
1-2
2-3
3-4
4-6
6-8
8-10
10-20
,>20
2-3
3-4
4-6
6-8
8-10
10-20
>20-
Test
Status*
FP/TN
FP/TN
TP
TP
TP
TP
TP
TP
TP
FN
FN
FN
FN
FN
FN
FN
Fraction of
Population
0.6198
0.2074
0.0428
0.0293
0.0330
0.0116
0.0066
0.0062
0.0006
0.0289
0.0089
0.0043
0.0005
0.0001
0.0000
0.0000
Radon
Reduction
Level
NONE
NONE
2
2
2
2
2
2
2
NONE
NONE
- NONE
NONE
NONE
NONE
NONE
Annual Deaths
Averted per
Million Persons
NONE
NONE
19
62
127
212
300
470
1,177
19
62
127
212
300
470
1,177
Annual Lives
Saved
(Lost)
0
0
175
390
902
527
425
624
152
(118)
(119)
(119)
(24)
(8)
(4)
(0)
False
Positive
0
True
Positive
3,196
False
Negative
(392)
Option B
Radon
pCi/1
Interval
0-1
1 —"Z
2-3
3-4
4-6
6-8
8-10
10-20
>20
2-3
3-4
4-6
6-8
8-10
10-20
>20
Test
Status*
FP/TN
FP/TN
TP
TP
TP
TP
TP
TP
TP
FN
FN
FN
FN
FN
FN
FN
. Fraction of
Population
0.6198
0,2074
0.0367
0.0275
0.0322
0.0115
0.0066
0.0062
0.0006
0.0350
0.0107
0.0051
0.0006
0.0001
0.0000
0.0000
Radon
Reduction
Level
NONE
NONE"
2
2
2
2
2
2
2
- NONE
NONE
NONE
NONE
NONE
NONE
NONE
Annual Deaths
Averted per
Million Persons
NONE
NONE
19
62
127
212
300
470
1,177
19
62
127
212
300
470
1,177
Annual Lives
Saved
(Lost)
0
0
150
365
881
524
424
624
152
(143)
(143)
(140)
(28)
(9)
(4)
(0)
False
Positive
0
True
Positive
3,121
False
Negative
(467)
Appendix E
Page E-12
-------
Exhibit E-7 (continued)
Option C
Radon
pCi/1
Interval
0-1
1-2
2-3
3-4
4-6
6-8
8-10
10-20
>20
2-3
3-4
4-6
6-8
8-10
10-20
>20
Test
Status*
FP/TN
FP/TN
TP
TP
TP
TP
TP
TP
TP
FN
FN
FN
FN
FN
FN
FN
Fraction of
Population
0.6198
0,2074
0.0372
0.0276
0.0323
0,0115
0.0066
0.0062
0.0006
0.0345
0.0106
0.0050
0.0006
0.0001
0.0000
0.0000
Radon
Reduction
Level
NONE
NONE
2
2
2
2
2
2
2
NONE
NONE
NONE
NONE
NONE
NONE
NONE
Annual Deaths
Averted per
Million Persons
NONE
NONE
19
62
127
212
300
470
1,177
19
62
127
212
300
470
1,177
Annual Lives
Saved
(Lost)
0
0
152
368
883
524
424
624
152
(141)
(141)
(138)
(27)
(9)
(4)
(0)
False
Positive
0
True
Positive
3,127
False
Negative
(460)
*True Positive (TP), False Positive (FP), True Negative (TN), and False Negative (FN). Assumes 100% testing
and mitigation. Column entries may not sum due to rounding of estimates.
**There is no radon reduction level for TN homes, there is a radon reduction level of 2 for homes that test FP.
Appendix 'E
Page E-13
-------
t
I
tn
I
70
60
50
40
30
20
10
0
DISTRIBUTION OF HOMES AND TOTAL EXPOSURE AT SELECTED
RADON LEVELS FOR ALL HOMES
64%
Percentage of Homes
Percentage of Exposure
TolaB wi» not sum to 100% due to rounding errors
22%
22%
20%
20%
5%
11%
IS
0.06%
^•— ^t~
0-1
1-2
2-3 3-4 4-8
PicoCuries Per Liter
8-20
>20
-------
APPENDIX F
RADON MITIGATION COST MODEL
EPA has a cost model for estimating the total system costs for radon mitigation in single-
family detached homes. For any specified action level, the model can predict the total costs of
reducing radon levels for homes above that level down to a specified "target level," an average level
all homes will reach when mitigation systems are installed. The model calculates a weighted average
cost (WAC) of mitigation, which can be multiplied by the number of housing units that EPA
estimates will require radon mitigation to arrive at total mitigation costs. The costs are for
contractor-installed systems (versus homeowner efforts). The estimates are provided as present value
estimates-all unit costs are summed and presented in 1991 dollars based on a 3 percent discount rate.
Background
When the model was first built in 1987, the overall average target level of radon reduction
was 4 pCi/L (equal to EPA's action level). For ease of development, the model's unit costs were
developed using 4 pCi/L as a reference level for overall reduction. Sealers were devised to multiply
against the reference unit costs for achieving a reduction down to 4 pCi/L to consider higher, or
lower mitigation target levels. For instance, in some cases installing a mitigation system that would
get homes above 4 pCi/L down to 2 pCi/L was estimated to be twice as expensive as a system that
gets homes above 4 pCi/L down to 4 pCi/L. The model has been updated over time as EPA has
learned more about mitigation experiences through the U.S., but it has kept this reference cost and
sealer (multiplier) approach to estimating individual unit costs of various mitigation activities.
Recent Model Update
The model was updated in the Fall of 1991. EPA reviewed recent radon mitigation literature
and consulted mitigation experts throughout the United States during its revision of the model.
EPA's recently completed Parametric Analysis of the Installation and Operating Costs of Active Soil
Depresmrization Systems for Residential Radon Mitigation (EPA, October 1991) was relied on quite
heavily for updating the costs of active subslab dcpressurization, which is the technology that EPA's
expects the vast majority of homeowners to use. To a lesser degree, EPA's Private Sector Radon
Mitigator Swvey (June 1991) which reports on data collected in 1989 was also used,
Radon mitigation has been an active business for less than 10 years. There is not extensive
published information on the usage rates of different systems and cost differences geographically.
To overcome this lack of data, EPA consulted with expert radon mitigators throughout the country
to develop information on what types of systems were being used and the costs of their usage in
different parts of the country . These experts had significant experience and were mitigation
instructors at EPA's Regional Radon Training Centers. They had substantial knowledge of practices
in the Northeast, South, West, Pacific Northwest, and Central U.S. Additionally, a senior EPA
mitigation expert, who has directed its extensive mitigation research activities for existing homes, and
the principal developer of EPA's radon mitigation training course, who has substantial national
experience in reducing radon levels in homes, were also consulted. From these experts' input and
Appendix F Page F-l
-------
recognition of where most mitigation activity in the country would be-occurring1, cost functions and
estimates of technology usage throughout the nation were prepared.
Model Components
The model has two major components. The first combines unit costs of different aspects of
radon mitigation to derive total upfront and lifetime system costs for selected mitigation technologies
that are most prevalently used in the U.S. The second component uses estimates of how often
various technologies are used to provide weighted average costs of radon mitigation for upfront and
lifetime expenditures. Each of these components is explained below.
Unit and Total Costs of Mitigation Systems
The model examines unit costs in five major cost categories: (1) diagnosis, (2) installation,
(3) post-mitigation, (4) regular operation, and (5) repairs. It is supplied unit costs at a level of detail
that is sensible to consider cost variations for four major factors: (1) foundation type, (2) mitigation
technology, (3) initial radon levels, and (4) degree of difficulty in installing a system (i.e., whether
homes are easy or hard to fix). For installation, it makes sense to provide unit costs for each
combination of these factors. Therefore, average unit costs for all appropriate situations were
developed (see Exhibit F-2). For diagnosis and post-mitigation, the best way to examine unit costs
was on the basis of whether the home was easy or hard to fix (see Exhibit F-3). For regular
operation costs (shown in Exhibit F-4) and repairs (shown in Exhibit F-5), the best way to examine
unit costs was on the basis of the mitigation technology used.
Each of the cost categories used in the analysis is discussed below.
* Diagnosis - This category covers the average costs for a radon mitigator to provide
a limited assessment of the extent of a radon problem in a home and to estimate the
cost to fix it — it is essentially the cost of a price estimate. It is delineated by whether
the home will be easy or difficult to fix (see Exhibit F-3).
• Installation - This category covers the costs of materials and labor for initially
installing the system (see Exhibit F-2). Unit costs were estimated for every mitigation
situation (i.e., combination of factors) that the Agency believed should occur at levels
that are not insignificant. It was assumed that any practice that would be used less
than 5 percent of the time was insignificant. Note that the unit costs for homes
between 2 and 4 pCi/L only apply if the action level considered in the cost analysis
is either 2 or 3 pCi/L. For all technologies, it is assumed some sealing occurs in areas
of the foundation where it makes sense to do so. (Sealing/Plugging as a mitigation
technology category here means that its the only method used.)
* Post-mitigation - This category covers the contractor returning to the home shortly
after system installation to check its operation and make any necessary adjustments.
It is delineated by whether the home was easy or difficult to fix (see Exhibit F-3).
'Exhibit F-l shows by EPA Region the percentage of homes that are above EPA's action level (4 pCi/L) based on
EPA's report on the National Residential Radon Survey.
Appendix F Page F-2
-------
* Regular Operation - This category estimates the normal operating expenses incurred
by homeowners during mitigation system operation (see Exhibit F-4). For all
technologies, it is assumed to be at least $35 every 2 years (or about an average of
$18 a year) based on the assumption that all homes will biennially test their radon
levels (as recommended in the RCP Interim Radon Mitigation Standards, 1991) using
an alpha track detector. Additional annual operating costs for each technology
include:
Active subslab depressurization - there are costs for fan electricity and
heating/cooling losses of conditioned air.
Simple ventilation - for this cost category, there are averaged costs of using
ventilation fans with open windows and simply opening windows. The costs
are for fan electricity (where a fan is used) and heating/cooling losses.
Heat recovery ventilation - this technology has added costs for fan electricity,
heat loss, annual filter replacement, and annual unit inspection.
Depressurization in crawl space - there are costs for fan electricity and the
added space heating requirements resulting from this system.
« Repairs - This cost category gives the costs of labor and materials for repairing
mitigation systems, or replacing materials over time (see Exhibit F-5). The repair
costs vary by mitigation technology:
Active subslab depressurization - For all foundations, it is assumed that cracks
will be resealed in years 2 and 20 after system installation to fix problems
caused by caulks degrading or house settling. This resealing is estimated to
cost $100 each time. Fans are replaced every 10 years at a cost of $150 each
time and the alarm device that indicates fan failure is replaced for $95 in year
37 (half way through system life.) For crawl space homes, the membrane
liner is replaced in year 37 (halfway through system life.)
Sealing/plugging - initial sealing is repaired every five years at 40 percent of
the original cost of installation.
Simple ventilation - repair and replacement is averaged between using fans
with open windows and open windows only. For all foundations, assumes
cracks will be resealed in years 2 and 20. Basement and slab-on-grade homes
that use fans replace them every 20 years for $230 each time. There are no
replacement costs associated with open windows other than sealing foundation
cracks.
Heat_recgyery ventilation - costs are twice the ASD costs above, since the
systems use two fans.
The unit costs are meant to reflect national average unit costs. This is after compiling the
unit cost estimates mitigators provided from different regions of the country and considering where
Appendix F • ^_ Page F-3
-------
in the country certain mitigation practices in certain types of houses (by foundation) were likely to
occur. As mentioned earlier, there was also consideration of areas of the country that will have the
largest percentages of homes to fix.
In each case, the "average" size house that was considered had 1,900 square, feet. Only major
mitigation technologies that would often be used were considered ~ active subslab depressurization
(ASD), sealing/plugging, simple ventilation (using fans with open windows or just properly opened
windows), heat recovery ventilation (HRV), and space depressurization for crawl space homes.
Whether a house was easy or hard t'o fix related to its structural design, materials the home
foundation rested on, and interior space use. The definition reflected the level of effort (and
materials to some degree) a mitigator would spend on installing a system. Effort was always linked
to whether areas in the house that required mitigator attention were readily accessible or not. For
instance, whether cracks in the foundation that needed sealing were easy to reach or not. Often it
would also depend on how much effort and materials were needed to install a system that would be
fully effective in a particular situation. For example, where the area under a foundation does not
provide for easy passage of soil gas, the resulting "poor communication" will lead to the placement
of more than one suction pipe in an active subslab depressurization system. A house such as this
would be "hard" to fix.
Initial radon levels, in pCi/L, were selected based on radon ranges that could lead to different
selections of various types of mitigation systems and different overall levels of effort (although to a
lesser degree). It is important to recognize that although mitigators are fixing homes with levels
above 4 pCi/L, it is uncommon for homes initially testing between 2 to 4 pCi/L to be mitigated.
Therefore, the expenses for mitigating homes in this range are based on theoretical assumptions of
what the costs should be and not practical experience. The housing foundation types selected
represent the three dominant types that exist (note that partial and full basement homes are simply
considered as basement homes). It was assumed that 47 percent of the homes have basements, 26
percent have slab-on-grade, and 27 percent have crawl space. These estimates were taken from the
American Housing Survey for 1989 published by the Census Bureau.
The costs are estimated from the vantage point of the homeowner. For diagnosis, installation,
post-mitigation, and repairs, the costs are the amounts homeowners would pay commercial contractors
to perform the work. For the operating expenses, the costs reflect what homeowners would pay
testing companies and utilities for electricity and space conditioning. It is important to consider that
the first three costs elements together (diagnosis, installation, and post-mitigation) should sum to the
total price mitigators would charge for mitigation system installation. It is assumed that mitigators
are using practices EPA recommends for installing mitigation systems in its guidance materials and
training programs. It is also important to recognize that the costs in each category are meant to
provide average unit costs that can be combined to provide overall system costs. The costs considered
in the analysis are for 74 years, the assumed average lifetime of a house.
The total costs for upfront expenses and system life are calculated for each type of mitigation
system. The estimates are the summation of the appropriate unit cost components as shown in
expert mitigators were queried on whether it was reasonable to assume that the costs of getting the average
house between 2 and 4 pCl/L down to 2 pC/I was the same as getting the average house between 4 and 8 pCi/L down to
4 pCi/L. They all believed such an assumption was reasonable.
Appendix F Page F-4
-------
Exhibits F-2 through F-5. Exhibit F-7 shows the total lifetime costs. An example of how these unit
costs are combined is provided below:
For basement homes that are greater than 20 pCi/L and easy to fix, their average cost is the
summation of diagnosis ($80), installation ($1,000), post-mitigation ($70), annual operation
($110 per year, or $3,265 in present value terms), and repairs ($565 in present value terms),
or $4,971. The initial costs that a mitigator would want to charge someone, $1,150, are
captured by diagnosis, installation, and post-mitigation.
Mitigation Technology Usage
For each set of homes that is expected to exist in various pCi/L intervals, the model uses
estimates of the percentage of time various technologies would be used in easy and hard situations.
Exhibit F-6 shows the assumptions made by each type of home foundation. These percentages were
arrived at by using the judgments of the expert mitigators mentioned above.
The left-hand portions of Exhibit F-7 show the cost results of the total lifecycle analysis in
present value dollars for the case of all homes with an annual average level above 4 pCi/L reducing
their radon levels to an overall annual average of 2 pCi/L. These initial results are used to develop
a weighted average cost (WAC) for each foundation type and ultimately a WAC for the whole
housing stock. This is calculated using information on the percentages of homes that are found in
each range of initial radon levels and the percentages of homes with each foundation type.
Exhibit F-8 shows for each action level examined in Chapter 5 what percentage of homes
were in different initial radon ranges. (EPA assumed an even distribution of range levels by
foundation type.) In the cases where the action level is 4 pCi/L, the system costs for 2-4 pCi/L homes
would not enter the analysis. The right-hand portion of Exhibits F-6 and F-7 show the case where
the action level could be 2 or 3 pCi/L and homes between 2 to 4 pCi/L reduce their overall annual
average to 2 pCi/L. As Exhibit F-8 shows, each of the action levels that was examined has different
WACs for the upfront and total lifetime costs due to the variability in percentages of homes that are
fixed at different levels. The lower WACs for lower action levels are due to an increasingly larger
percentage of homes at lower radon levels that enter into the calculation of the WAC. In the
majority of situations (especially in homes using ASD), the costs of reducing radon levels down to
2 pCi/L is about the same. However, mitigations in the smaller set of homes that use sealing and
ventilation technologies cost less at the lower action levels because the mitigation systems need to
provide lower percentage reductions in radon levels.
Appendix F Page F-5
-------
EXHIBIT F-1
ESTIMATED NUMBER OF HOUSING UNITS WITH
RADON LEVELS ABOVE 4pCi/L, 1989-1990
REGIONS
REGION 1
REGION 2
REGION 3
REGION 4
REGION 5
REGION 6
REGION 7
REGIONS
REGION 9
REGION 10
TOTAL
SINGLE-FAMILY
HOMES
ABOVE 4pCi/L*
124
189
526
555
1,122
2Q5
596
• - 407
56
33
3,812
MOBILE
%OF I HOMES %OF
TOTAL LABOVE4pCi/L' TOTAL
3%| 8 2%
. • 5%j 14 3%
14%! 51 11%
15%
29%
' . 5%
16%
'. :.11%
1%
. 1%
120, 26%
MULTI-UNITS
ABOVE 4pCi/L'
80
\ 179
173
1 0P;
99 22%) 424
32 7%
6§
60 13%) 158
•••- . ' . . 57 13%
! 141
7 2%| 29
453
1 11
% OF
TOTAL
5%
12%
12%
* 13%|
29%
5%
TOTAL
UNITS
ABOVE 4pCi/L*
212
382
750
867
1 644
3m
11% 813
10%.
2%
1°A
1,453
€04,
92
56-
TOTAL
4%
7%
13%
15%!
29%
5%
14%
11%
2%
1%
5,719
"Number of homes in thousands.
Source: National Residential Radon Survey
Appendix F
Page F-6
-------
EXHIBIT F-2
RADON MITIGATION SYSTEM INSTALLATION COSTS BY BUILDING FOUNDATION,
SOURCE STRENGTH, AND DEGREE OF DIFFICULTY TO FIX
3
SOURCE STRENGTH 20
$1,000
1,500
1,250
0
0
0
$1,200
1,700
0
0
0
0
$1 ,300
1,650
0
0
0
450
4 pCi/L TO
8-20
$1,000
1,300
1,100
3,000
0
2,000
$1,200
1,600
600
3,000
0
0
$1,300
1,650
1,250
0
Q
450
4 pCi/L
BETWEEN 2-4 pCi/L TO 2 pCi<
4-8'. | . 2-4
j
$1,000
1,300
500
2,750
115
2,000
$1,200
1,600
600
1,250
115
0
$1,300
1,650
500
1,250
0
450
$1,000
1 ,300
500
2,750
115
0
$1,200
1,600
600
1,250
115
0
$1,300
1,650
500
1,250
400
450
MULTIPLIERS FOR HOMES ABOVE 4 pCi/L
ACHIEVING 2PCI/L LEVELS (MULTIPLIED AGAINST COSTS FOR REACHING 4 pCi/L)
ACT.SUBSL. DEPPRESURIZATION - EASY:
ACT.SUBSL. DEPPRESURIZATION - HARD:
SEALINGSPLUGGING -
EASY AND HARD:
SIMPLE VENTILATION (BASEMENT AND S-
O-G):
SIMPLE VENTILATION (CRAWLSPACE):
HEAT RECOVERY VENTILATION:
DEPRESSURIZATION:
1.00
1.25
1.50
2.00
1.00
1,75
1,00
HOMES BETWEEN 2-4 pd/L REACH 2 WITH SAME ASSUMED COST AS HOMES BETWEEN 4-8pCl/L REACHING 4 pCI/L
* Simple ventilation and depressurization (in the crawlspace) are assumed to b* easy and heat recovery ventilation is assumed to be hard to do.
Nota: Zero entries in ins installation table indicates that the technology shown would not be applied in that particular situation.
Appendix F
Pom F-7
-------
EXHIBIT F-3
COMMON COSTS ASSOCIATED WITH PROPER MITIGATION SYSTEM INSTALLATION
FOR ALL FOUNDATIONS BY DEGREE OF DIFFICULTY TO FIX*
MULTIPLIERS FOR ACHIEVING 2 PCi/L
FOR HOMES REDUCED TO 4 PCi/L FOR HOMES INITIALLY ABOVE 4 pCI/L"
EASY HARD
DIAGNOSTICS $80 $120 1.0 - Same cost for homes
(PRICE ESTIMATE) going down to 2 pCi/L
POSTMITIGATION $70 $280 Multiply by 1.0 for easy homes
and 1.5 for hard homes going
down to 2 pCi/L
'Note that simple ventilation and crawlspace depressurization is always considered easy to install and heat recovery is always
considered hard to install.
* 'Homes between 2 and 4 pCi/L that reduce to 2 pCi/L are assumed to have the same costs as homes above 4 pCi/L reducing down
to 4 pCi/L (i.e., the multiplier is 1.0).
Appendix F ' Page F-8
-------
EXHIBIT F-4
ANNUAL OPERATING COSTS FOR RADON MITIGATION SYSTEMS
All Foundations
SEALING/PLUGGING COSTS (FOR BIENNIAL TESTING)
FOR HOMES TO REACH 4 OR 2 pCi/L $35 (or $18 per year)
FOR HOMES INITIALLY ABOVE 4 pCi/L
Basement &
Slab-on-grade Crawlspace
ACTIVE SUBSLAB DEPRESSURiZATION
FOR HOMES AT 4 PCi/L: $110 $110
FOR HOMES AT 2 PCi/L 110 110
(Includes annual testing, energy penalty, and fan electricity)
SIMPLE VENTILATION
FOR HOMES AT 4 PCi/L: $335 $120
FOR HOMES AT 2 PCI/L: 670 120
(Includes annual testing, energy penalty, and fan electricity)
DEPRESSURIZATION
FOR HOMES AT 4 PCi/L: - $185
FOR HOMES AT 2 PCi/L: - 185
(Includes annual testing, energy penalty, and fan electricity)
HEAT RECOVERY VENTILATION
FOR HOMES AT 4 PCi/L: $240
FOR HOMES AT 2 PCi/L: 480
(Includes annual testing, energy penalty, and fan electricity)
FOR HOMES BETWEEN 2 AND 4 pCi/L REACHING TARGET LEVEL OF 2 pCi/L
Assumed that costs are the same as homes above 4 pCi/L reaching 4 pCi/L target level (i.e., multiplier of 1.0)
Note: Operating costs are assumed constant across homes with different source strengths (i.e., initial radon levels) and dsgres of difficulty to fix.
Appendix F Page F-9
-------
EXHIBIT F-5
REPAIR COSTS FOR COMPONENTS OF RADON MITIGATION SYSTEMS
REDUCING HOMES ABOVE 4 pCi/L DOWN TO ALTERNATIVE TARGET LEVELS
FOR HOMES ACHIEVING 4 pCi/L
ACTIVE SUBSLAB DEPRESSURIZATION: REPLACING FANS EVERY 10 YEARS
AT A COST OF $150 EACH TIME
WARNING DEVICE FAILURE AT YEAR 37: $95
SUBMEMBRANE DEPRESSURIZATION IN CRAWLSPACE: 112 INSTALLATION (EASY) AT YEAR 37
WARNING DEVICE FAILURE AT YEAR 37: $95
SEALING: RESEALING EVERY 5 YEARS AT
40% OF INSTALLATION COST
SIMPLE VENTILATION: REPLACING FANS EVERY 10 YEARS FOR $230
HEAT RECOVERY VENTILATION: DOUBLING THE COST OF ASD ABOVE
(SYSTEMS USE TWO FANS)
FOR HOMES ACHIEVING 2 pCi/L
REPLACEMENT COST MULTIPLIERS FOR HOMES ACHIEVING 2 pCi/L LEVELS
(MULTIPLY THEM AGAINST ABOVE COSTS OF GETTING HOMES DOWN TO A TARGET LEVEL OF 4 pCi/l}
ACTIVE SUBSLAB DEPRESSURIZATION - EASY: 1.00
ACTIVE SUBSLAB DEPRESSURIATION - HARD: 1.25
SEALING/PLUGGING: 1-50
SIMPLE VENTILATION (NON-CRAWLSPACE): 2.00
HEAT RECOVERY VENTILATION: 1.75
CRAWLSPACE DEPRESSURIZATION: 1.00
CRAWLSPACE SIMPLE VENTILATION (Only 2-4 Range): 1.00
FOR HOMES BETWEEN 2 TO 4 pCi/L ACHIEVING 2 pCi/L
COSTS ARE THE SAME AS HOMES FROM 4 TO 8 pCi/L REACHING 4 pCi/L
(i.e., the multiplier is 1.0)
Note: Replacement costs are assumed constant across homes with different source strengths (i.e., initial radon .levels) and degrees of difficulty to fix.
Appendix F Page F.-10
-------
EXHIBIT F-6
MITIGATION TECHNOLOGY USE BY BUILDING FOUNDATION TYPE,
SOURCE STRENGTH, AND DEGREE OF DIFFICULTY TO FIX
.S<&l3gStJ*^T^
DIFFICULTY
BASEMENT
ACTIVE SUBSLAB DEPRESSUREATION
SEALING/PLUGGING
SIMPLE VENTILATION
HEAT RECOVERY VENTILATION
SLAB-ON-GRADE
ACTIVE SUBSLAB DEPRESSURIZATION
SEALING/PLUGGING
SIMPLE VENTILATION
HEAT RECOVERY VENTILATION
CRAWLSPACE
ACTIVE SUBMEMBRANE DEPRESSURIZATION
ISOLATION
DEPRESSURIZATION
SIMPLE VENTILATION
EASY
HARD
EASY
HARD
(EASY)
(HARD)
EASY
HARD
EASY
HARD
(EASY)
(HARD)
EASY
HARD
EASY
HARD
(EASY)
(EASY)
65%
35%
0%
0%
0%
0%
65%
35%
0%
0%
0%
0%
55%
35%
0%
0%
10%
0%
65%
30%
0%
0%
0%
5%
65%
35%
0%
0%
0%
0%
55%
35%
0%
0%
10%
0%
50%
20%
10%
5%
10%
5%
60%
30%
5%
0%
5%
0%
55%
30%
5%
0%
10%
0%
45%
15%
15%
5%
20%
0%
60%
30%
5%
0%
5%
0%
35%
15%
5%
10%
10%
25%
Note: Technologies with less than 2.5% usage were given 0%; all %'s rounded to nearest 5%,
Appendix F
Page F-ll
-------
EXHIBIT F-7
SAMPLE MODEL OUTPUT
RADON MITIGATION SYSTEM COSTS BY PICOCURIE/LITER RANGE
(PRESENT VALUE 1991$)
REDUCING HOMES
TO2pCi/L
^^^K^i^^^^^&i^isi^mmim
BASEMENT
ACTIVE SUBSLA8 DEPRESSURiZATiON
SEALING/PLUGGING
SIMPLE VENTILATION
HEAT RECOVERY VENTILATION
SLAB-ON-GRADE
ACTIVE SUBSLAB OEPRESSURIZATION
SEALING/PLUGGING
SIMPLE VENTILATION
HEAT RECOVERY VENTILATION
CRAWLSPACE
ACTIVE SUBSLAB DEPRESSURIZATION
ISOLATION
DEPRESSURIZATION
SIMPLE VENTILATION
DIFFICULTY
TO FIX
EASY
HARD
EASY
HARD
EASY
HARD
EASY
HARD
EASY
HARD
EASY
HARD
$4,971
6,377
0
0
0
0
$5,171
6,627
0
0
0
0
$5,509
6,862
0
0
6,640
0
$4,971
6,127
0
0
0
19,840
$5,171
6,502
0
0
0
0
$5,509
6,862
0
0
6,640
0
$4,971
6,127
3,078
14,249
21,091
19,840
$5,171
6,502
3,557
0
21,091
0
$5,509
6,862
3,078
0
6,640
0
liliiillll
$4,971
5,521
2,280
9,717
10,620
0
$5,171
5,821
2,599
0
10,620
0
$5,509
6,109
2,280
4,925
6,640
4,539
EXHIBIT F-8
DISTRIBUTION OF HOMES INITIALLY TESTING ABOVE SELECTED ACTION LEVELS
AND WEIGHTED AVERAGE COSTS OF MITIGATION
ACTION LEVEL
2 pCi/L
3 pCi/L
4 pCi/L
DISTRIBUTION OF HOMES ABOVE THE ACTION LEVEL IN pCI/L*
>20 8-20
0.0156 0-1258
0.0257 0.2073
0.0390 0.3119
4-8 2-4 TOTAL
0.3349 0.5237 1.00
0.5047 0.2623 1.00
0.6491 0.0000 1.00
WEIGHTED AVERAGE COST
(1991$)
UPFRONT LIFETIME
$1 ,366
1,442
1,520
$6,126
6,359
6,566
'Rows may not sum due to rounding.
Appendix F
PaseF-12
-------
APPENDIX G
SENSITIVITY ANALYSIS OF THE
COST-EFFECTIVENESS OF THE REVISED CITIZEN'S GUIDE
This appendix presents the results of a sensitivity analysis of the most significant parameters
used in the cost-effectiveness analysis in Chapter 5. It only examines the option that EPA has
decided to adopt in the revised Citizen's Guide, which recommends that the public fix ail homes with
radon levels above 4 pCi/L. Exhibit G-l shows the results of the analysis for this option or Base
Case. To facilitate evaluation of the sensitivity of the cost-effectiveness results to the assumptions
that were used for the most critical parameters, the base case results appear at the top of the exhibit.
Each major parameter that EPA used was analyzed independently with regard to the relevant
range of values it potentially could have. All the other values remained as they were in the base case
(i.e.. were held constant) while each parameter was examined. In the base case, the critical
assumptions were:
• radon's risk factor is equal lo EPA's central estimate;
• 100 percent of the public will follow EPA's testing and mitigation advice;
» 91 percent of the public will conduct a short-term follow-up test, and the rest
will conduct a long-term follow-up test;
• the time period of analysis is 74 years;
• on average, all homes installing radon mitigation systems will reduce radon
levels to 2 pCi/L;
• the radon testing policy covers all single family homes, multi-unit and group
quarters below the third floor, and mobile homes on permanent foundations;
* the social discount rate1 is 3 percent;
» mitigation systems have an average present value cost of $6.566 (about 25
percent of the total expenses are initial costs and about 75 percent of the
costs are operating and maintenance (O&M) costs); and
• smoking habits will continue unchanged.
The following assumptions were analyzed in the sensitivity analysis.
* Risk factor - Used EPA's lower and upper bound estimates of the radon risk
factor. (See Chapter 2 for values used.)
!The opportunity cosi of money used by homeowners to test, fix, and operate their homes.
Appendix G Page G-l
-------
Public Response rates - Examined changes in the expected level of radon
testing and the expected level of mitigation that would occur after testing.
Testing choices - Analyzed the public's use of a long-term follow-up test
(Option A) in all cases or use of a short-term follow-up test (Option B) in all
cases. EPA recommends in the 1992 Guide that the public use either test.
Time period of analysis - Examined shorter time periods for the analysis that
reflect other relevant time periods EPA could have considered. Five years
represents the lowest available estimate of how often homes are sold on
average. Thirty years is the life of a typical mortgage.
Effectiveness of systems - Selected a level of 3 pCi/L and 1 pCi/L as
alternatives to EPA's assumption that radon mitigation will attain an annual
average reduction to 2 pCi/L. Alternatively, it was assumed that homes
between 1 and 2 pCi/L that have false positive results could get down to 1
pCi/L and homes above 2 pCi/L would get down to that level. In a
subsequent run, it was assumed that homes with false positives between 1 and
2 pCi/L could get down to 1 pCi/L and homes between 0 and 1 pCi/L could
reduce their radon levels by 50 percent if mitigation systems are installed
while homes above 2 pCi/L are still reduced down to 2 pCi/L.
Coverage of testing policy - Excluded from consideration'all housing units that
were not single family homes. Single family homes were assumed to average
2.8 residents per unit as opposed to 2 residents per home in other types of
housing units.
Discount rate - Examined other discount rates that have been used by the
federal governmenf in preparing cost analyses.
Initial mitigation system costs - Varied by 50 percent the upfront costs of
installing radon mitigation systems (including the costs of diagnosis,
installation, and post mitigation follow-up).
Mitigation O&M cost - Varied by 30 percent the costs of operating and
repairing radon mitigation systems over their entire lives.
Smoking Habit Changes - Analyzed two types of changes in smoking patterns.
First, EPA examined changes in habits that reduced the level of current
smokers (by 20 to 50 percent) and assumed thai this share of people entered
the "former smoker" category. Second, EPA analyzed changes in smoking
habits that altered the distribution of "current smokers" and "never smokers."
In this case, assumed that 20 to 50 percent of today's current smokers are
instead persons who have never smoked.
Appendix G Page G-2
-------
EXHIBIT G-l
SENSITIVITY ANALYSIS OF THE
COST-EFFECTIVENESS OF THE REVISED CITIZEN'S GUIDE
Case
BASE CASE
RISK FACTOR
EPA Lower Bound Risk Factor
EPA Upper Bound Risk Factor
PUBLIC RESPONSE RATE
50% Testing/100% Mitigation
10% Testing/100% Mitigation
100% Testing/25% Mitigation
100% Testing/10% Mitigation
TESTING CHOICES
100% Option A
100% Option B
TIME PERIOD OF ANALYSIS
5 Years
30 Years
EFFECTIVENESS OF SYSTEMS
Average Reduction to 3 pCi/L
Average Reduction to 1 pCi/L
Homes Between 1 and 2 Can Reduce to 1 pCi/L
Homes Between 1 and 2 Can Reduce to 1 pCi/L
and Homes Below 1 €an Reduce by >50%
COVERAGE OF TESTING POLICY
Single-Family Homes Only
ASSUMED DISCOUNT RATE
0% Discount Rate
5% Discount Rale
7% Discount Rate
10% Discount Rare
INITIAL MITIGATION SYSTEM COSTS
50% Cost Reduction
50% Cost Increase
MITIGATION SYSTEM O&M COSTS
30% Cost Reduction
30% Cost Increase
DIFFERENCES IN SMOKING HABITS
20% Current Smokers Quit
50% Current Smokers Quit
20% of Current Smokers Never Smoked
50% of Current Smokers Never Smoked
Annual
Lives Saved
2,230
1,400
5,900
1,120
220
560
220
2,160
2,240
2,230
2,230
1,710
3,000
2,270
2,280
2,120
2,230
2,230
2,230
2,230
2,230
2,230
2,230
2,230
2,060
1,790
1,930
1,490
Annualized Cost
(Million 1991$)
S 1,500
1,500
1,500
750
150
440
230
930
1,560'
3,930 '
1,730
1,500
1,500
1,500
1,500
1,240
1,240
1,730
1,980
2,360
1,340
1,670
1,180
1,830
1,500
1,500
1,500
1,500
Cost Per
Life Saved
(Thousand 199 IS)
$ 670
1,070
260
670
670
800
1,040
430
700
1,760
780
880
490
660
660
590
560
780
890
1,060
600
750
530
820
730
840
780
1,010
Appendix G
Page G-3
-------
APPENDIX H
COST-EFFECTIVENESS OF RADON ACTION
LEVELS GREATER THAN 4 pCi/L
EPA initially examined five different action levels for the revised Citizen's Guide. It quickly
became apparent that the higher action levels did not reduce nearly as much of the risk from radon
as did the lower action levels. Also, the lower action levels were found to be incrementally cost-
effective. Reducing radon levels in the additional homes covered by lower action levels appeared to
be a good risk reduction purchase for the public to make if practical programs for doing so could be
offered.
Exhibit H-l provides a summary of the results of EPA's analysis of action levels greater than
and lower than 4 pCi/L. The same approach that is explained in Chapter 5 and supporting
appendices was used to provide these results. For the action levels of 8 and 20 pCi/L, EPA estimates
that 2.1 million homes and 0.2 million homes, respectively, would be mitigated if a 100 percent public
response rate were obtained. This compares with 6.4 million homes that would be mitigated when
the action level is 4 pCi/L.
EXHIBIT H-l
COST PER LIFE SAVED
UNDER ALTERNATIVE TESTING AND MITIGATION PROGRAMS
Action Level
2pCi/L
3 pCi/L
4 pCi/L
8pCi/L
20pCi/L
Number of Lives
Saved Annually
3,100
2,600
2,200
1,100
220
Annualized Cost
(1000s of 1991$)
3,421,000
2,181,000
1,504,000
501,000
116,000
Average Cost
per Life Saved
(1000s of 1991$)
$1,100
800
700
400
500
Incremental Cost
per Life Saved
(1000s of 1991$)
$2,400
1,700
900
400
500§/
- Based on assumption that "no action" was the alternative EPA had to this action level.
Note: The central estimate of the radon risk factor is used in this analysis, rather than the upper bound risk estimate.
The upper bound estimate would have increased all the risk estimates by about 2.5 times and reduced the cost-
effectiveness estimates by about 60 percent.
Appendix H
Page H-l
-------
-------
United States
Environmental Protection Agency
(ANR-464)
Washington, DC 20460
Official Business
Penalty for Private Use
$300
EPA/400-R-92-011
May 1992
------- |