Information about http://www.cheec.uiowa.edu/misc/rd_review.pdf
REVIEWS ON ENVIRONMENTAL HEALTH PREPRINT …
Tags: case control, college of public health, compelling evidence, connecticut, d college, epidemiology, evidence suggests that, field ph, iowa city iowa, iowa iowa city, lung cancer, methodology, public health department, radon exposure, risk, southern idaho, study evidence, study iowa, university of iowa, unpublished study,Pages: 17
Language: english
Created: Tue Nov 6 11:59:21 2001
Display cached document
Page 1
Page 2
Page 3
Page 4
Page 5
Page 6
Page 7
Page 8
Page 9
Page 10
Page 11
Page 12
Page 13
Page 14
Page 15
Page 16
Page 17
REVIEWS ON ENVIRONMENTAL HEALTH PREPRINT VOLUME 16, NO. 3, 2001
A Review of Residential Radon Case-Control Epidemiologic
Studies Performed in the United States
R. William Field, Ph.D.
College of Public Health, Department of Epidemiology
University of Iowa, Iowa City, Iowa, U.S.A.
CONTENTS and Missouri (two studies). The methodology
from an unpublished study covering Connecticut,
Summary Utah, and Southern Idaho is also presented.
Introduction Overall, the higher categorical risk estimates
Summary of Residential Radon Case-Control for these published studies produced a positive
Studies association between prolonged radon exposure
New Jersey Residential Radon Study and lung cancer. Two studies (Missouri-II and
MissouriI Residential Radon Study Iowa) that incorporated enhanced dose estimates
MissouriII Residential Radon Study produced the most compelling evidence suggesting
Iowa Residential Radon Study an association between prolonged residential
Connecticut/Utah/S. Idaho Combined radon exposure and lung cancer. The prevailing
Residential Radon Study evidence suggests that the statistically significant
Discussion findings may be related to improved retro-
References spective radon exposure estimates. The general
findings from the U.S. studies, along with
extrapolations from radon-exposed underground
SUMMARY miners, support the conclusion that after
cigarette smoking, prolonged residential radon
Lung cancer is the leading cause of cancer exposure is the second leading cause of lung
death in the United States for both men and cancer in the general population.
women. Although most lung cancer deaths are
attributable to tobacco usage, even secondary
causes of lung cancer are important because of KEYWORDS
the magnitude of lung cancer incidence and its
poor survival rate. This review summarizes the radon, lung cancer, United States, case-control,
basic features and major findings from the epidemiology
published U.S. large-scale residential radon case-
control studies performed in New Jersey, Iowa,
INTRODUCTION
_________________________
Reprint requests to: R. William Field, College of Public
Lung cancer is the leading cause of cancer
Health, Department of Epidemiology, N222 Oakdale Hall,
University of Iowa, Iowa City, Iowa, 52242, USA; e-mail: death for both men and women in the U.S.
bill-field@uiowa.edu population, accounting for an estimated 157,400
© Freund Publishing House Ltd. 2001 151
152 R.W. FIELD
deaths in the United States during 2001 /1/. In fact, associated with the extrapolations led researchers
lung cancer (including bronchus) in the United to investigate directly (without extrapolation)
States is responsible for 31% and 25% of all whether residential radon exposure is associated
cancer deaths in males and females, respectively with an increased lung cancer risk in the general
/1/. The majority of these lung cancer deaths are population. Since 1981, over 20 ecologic /79/ and
attributable to the voluntary practice of cigarette 12 major case-control studies /1022/ have been
smoking. Because of the magnitude of lung cancer published examining the association between
incidence and poor lung cancer survival rates, residential radon exposure and lung cancer.
however, even secondary causes of lung cancer The ecologic studies generally attempted to
present a major public health concern. Radon-222 correlate geographic-based lung cancer rates with
(radon), a naturally occurring radioactive noble the mean radon concentrations from a geographic
gas generated by the radioactive decay of radium- area. The aggregate measures of radon concentration
226, is a known occupational lung carcinogen used in these studies were obtained from a limited
/2, 3/. It was discovered in the 1970s that radon number of short-term radon measurements. Because
gas from subsoils, and to a lesser degree from ecologic studies lack information at the level of the
groundwater sources, can enter a home and individual, the study design is limited to formulating
accumulate to relatively high concentrations. causal hypotheses. Alternatively, case-control
Radon gas undergoes radioactive decay to a series studies can establish risk factors for groups of
of decay products called radon progeny or radon individuals by collecting information at the level of
daughters. These inhaled solid radon decay the individual and controlling for potential con-
products deliver the radiologically significant dose founders that may affect the risk estimates.
to the lung tissues. The potential adverse health In the United States, five major case-control
effects from radon progeny exposure prompted the studies have been performed (Table 1) to assess
United States Environmental Protection Agency to the lung cancer risk posed by prolonged residential
adopt a 150 Bq m-3 (4 pCi L-1) indoor action level radon exposure. The New Jersey /19/, Missouri II
for radon. Comparative risk analyses performed in /21/, Iowa /22/, and the combined states study
the United States by numerous states and the (Connecticut, Utah, and Southern Idaho) /23/
federal government ranked residential radon examined the risk posed to a mixed sample of
exposure as one of the most serious environmental smokers and non-smokers; the Missouri-I study
hazards /4/. Lung cancer risk projections, extrapo- /20/ examined the lung cancer risk posed by radon
lated from case-control epidemiologic studies of only among ex-smokers and never smokers. This
radon-exposed underground miners, attribute paper describes the methodologies and results of
around 18,600 lung cancer deaths per year (range these studies when available.
3,000 to 41,000) in the United States population to
residential radon exposure /5/.
Because of inherent differences between miners SUMMARY OF RESIDENTIAL RADON CASE-
and the general population, in addition to the CONTROL STUDIES
differences between mine and home environments,
extrapolations from miners to the residential New Jersey Residential Radon Study
population are uncertain /5/. In fact, the validity of
even extrapolating from high dose to low dose The New Jersey residential radon study was
effects has been questioned /6/. The uncertainty the first large-scale study that was based on actual
RESIDENTIAL RADON CASE-CONTROL STUDIES IN THE UNITED STATES 153
TABLE 1
Attributes of United States residential case-control studies
Primary
Enrollment Period Number of subjects
reference
Study Location
Cases Controls
New Jersey Phases I+II /19/ 19821983 480 women 442 women
Missouri-I /20/ 19861991 538 women 1,183 women
Missouri-II /21/ 19931994 512 women 553 women
Iowa /22/ 19931996 413 women 614 women
Multi-State Study /23/ 19891993 Total = 1,474 Total = 1,811
Connecticut 963 men and 949 men and
women women
Utah, Southern Idaho 511 men and 862 men and
women women
radon concentrations in the home and detailed files for controls to match deceased cases. Living
smoking histories for individual subjects. Papers controls were frequency matched to living cases by
reporting the preliminary /2426/ and primary /18/ race and age. For deceased cases, the controls were
findings on the initial phase I study for the New individually matched to cases by race, age, and date
Jersey Radon Study were published in 1989 and of death. Potential control death certificates that
1990. The phase I radon study was a sub-study of a reported any respiratory disease were not selected.
previous investigation examining the variation in Phase I included 433 cases and 402 controls. Forty-
smoking-related cancers /26/. The initial study /26/ three percent of the case respondents and forty-seven
included all female residents of New Jersey with percent of the control respondents were proxies.
lung cancers diagnosed between August 1982 and Face-to-face interviews collected information
September 1983. The controls were population- on the following information:
based and selected by random digit dialing from · lifetime smoking history by brand of cigarette,
New Jersey women, using either (a) drivers' · smoking by other household members,
license files for those under age 65, (b) Health · lifetime history of towns lived in,
Care Financing Administration files for those 65 · occupational history, and
years of age old and older, or (c) death certificate · dietary history of foods containing vitamin A.
154 R.W. FIELD
TABLE 2
Radon dosimetry type and placement for residential case-control studies performed in the United States
Study location Ref. Dosimetry Test Location of placement
type1 duration
New Jersey /18, 19/ ATD 1 year ATDs placed in living area and basement.
CC 4 days Short-term CCs placed in some living areas and
basements
Missouri-I /20/ ATD 1 year ATDs placed in the kitchen and bedroom
Missouri-II /21/ ATD 1 year ATDs placed in kitchen and in subject's bedroom
RRD 60 days RRDs placed on selected household glass items.
Iowa /22/ ATD 1 year ATDs placed in subject's bedroom, historic bedroom,
home work area, and each level of the home and outside
the home
RRD 1 year RRDs placed in bedrooms and living area
Connecticut, /23/ ATD 1 year ATDs placed in subject's bedroom, lowest living level,
Utah, S. Idaho and basement.
1
ATD = Alpha Track Detector; RRD = Retrospective Radon Detector; CC = Charcoal Canister
Information was not collected concerning the residency period requirement and included subjects
percent of time spent in the home. Phase I subjects who had been living in one or more `index
were included into the study only if they had been residences' during the 5- to 30-year period before
living in a single `index' residence for at least 10 diagnosis or selection. Phase II added both the
years during the 10 to 30 years before diagnosis individuals who had been ineligible for phase I and
for cases or selection for controls. The authors additional houses for the phase I subjects. Subjects
point out that in this study, the overall distribution were included in the combined phase I and II
of radon concentrations in the home was relatively analysis only if the radon measurements or the
low /24/. Homes of only 36 subjects (24 cases and estimates covering at least 9 years of the 25-year
12 controls) out of the 835 phase I subjects had index period were available. Thirty-five percent of
living area radon concentrations greater than 73 the subjects had been living at the index
Bq m-3. An updated version of phase I, called residence(s) for the full 25-year period. The
phase II, was published as part of a proceedings median residency in the index homes was 22 years
/19/ and included additional subjects that were for both cases and controls. The combined phase I
ineligible under the phase I study. Phase II of the and II New Jersey Study included 480 cases and
New Jersey Study /19/ relaxed the 10-year index 442 controls (Table 1).
RESIDENTIAL RADON CASE-CONTROL STUDIES IN THE UNITED STATES 155
TABLE 3
Basement and Living Area Radon Concentrations for U.S. Residential Radon Studies
Geometric Mean (Geometric Standard Deviation)
Study Location (Bq m-3)1
Basement Level 1 Level 2 Reference
New Jersey 19 (2.3) 26 (2.2) /24/
Missouri-I 89 (1.9) 44 (2.2) 44 (2.2) /20/
Missouri-II 89 (2.0) 44 (2.2) 44 (2.2) /21/
Iowa 170 (2.2) 93 (2.2) 74 (2.1) /22/
Connecticut, Utah 56 (2.7) 19 (2.8) 15 (2.7)
Southern Idaho2 67 (2.3) 44 (2.2) 37 (2.2)
1
Summary data represent those homes that were measured with no imputed values added.
2
Shore D, Personal Communication
For the combined phase I and II study, year- interval was used. The analysis using cumulative
long alpha-track detector (ATD) measurements for radon exposures assumed a living-area radon
radon were performed for the living area of the concentration of 22 Bq m-3 for missing gaps.
home for 77% and 76% percent of the case and Information on the time the subject spent in the
control homes, respectively (Table 2). Living-area home was not collected.
estimated radon concentrations were made from Based on the combined phase I and II data, the
ATDs placed in the basement for 5% of cases and respective adjusted odds ratios (ORs) for the 4
for 6% of controls. Living-area estimated radon time-weighted average exposure categories 148 Bq m-3 0.71 Exposure based on time-weighted indoor air track-
(0.31.3)** etch radon detectors. Test for trend in odds ratios with
and increasing radon concentration (continuous p = 0.79)
-3
>148 Bq m 3.33 Exposure based on time-weighted-average indoor
(1.57.5)** exposure CR-39 surface measurements. Test for
trend in odds ratios with increasing radon
concentration (continuous p = 0.02)
Iowa /22/ >16.95 WLM5-19 1.79 Exposure for all cases and controls based on
(0.993.26) ** cumulative radon exposures using a complex
retrospective exposure algorithm Test for trend in
odds ratios with increasing radon concentration
and (continuous p = 0.14; categorical p = 0.05).
>16.95 WLM5-19 2.14 Exposure for all live cases and controls based on
(1.124.15)** cumulative radon exposures using a complex
retrospective exposure algorithm Test for trend in
odds ratios with increasing radon concentration
(continuous p = 0.03; categorical p = 0.01).
1
Results have not yet been reported for the joint study performed in Connecticut, Utah and Southern Idaho.
*90% Confidence Interval; **95% Confidence Interval
cautiously interpreted because the upper exposure based on the 90% confidence interval (Table 4),
category contained only five cases and one control. using only 4 cases and 1 control. The trend in ORs
Again using the combined phase I and II data, the with increasing cumulative radon was not
respective adjusted ORs for the 4 cumulative radon statistically significant (1-sided p=0.06). The
exposure categories 16.95 WLM5-
or to proxy respondents (generally next-of-kin), 19 were 1.00, 1.34, 1.73, 1.62, and 1.79. The OR
with a follow-up, face-to-face facilitation and for the highest exposure category for cumulative
interview. The questionnaires provided information exposure was nearly statistically significant, based
on family health history, demographics, personal on the 95% confidence interval (Table 4). The
health history, occupational exposures, dietary trend in ORs with increasing cumulative radon was
factors, retrospective personal mobility, and housing not statistically significant for continuous trend
characteristics. (p=0.14), but was statistically significant for the
A radon dosimetry assessment included the categorical analyses (0.05).
following: The Iowa authors presented separate findings
· an on-site residential assessment, excluding deceased subjects to minimize possible
· on-site radon measurements (both ATD and biases or exposure misclassification that might be
RRD), associated with the second-hand information from
· regional outdoor radon measurements, relatives. When the analysis focused on the living
· assessment of subject's exposure while in cases, the OR for the highest exposure category
another building, and was statistically significant (Table 4). In addition,
· linkage of historical subject mobility with the tests for trend were statistically significant for
residential, outdoor, and other building radon both the continuous (p=0.03) and the categorical
concentrations. analyses (p=0.01). Excess odds were calculated for
an average exposure of 11 WLM5-19 (approxi-
Yearlong ATD measurements were performed mately equal to a 15-year residential exposure at
on each level of the home and in the current and 148 Bq m-3). After adjustment for age, smoking,
historical master bedroom. Overall, 97% of all and education, statistically significant excess odds
alpha track detectors were retrieved 1 year later. of 50% and 83% were found using categorical
Cumulative radon exposures, accounting for all radon exposure estimates for all cases and live
RESIDENTIAL RADON CASE-CONTROL STUDIES IN THE UNITED STATES 161
cases, respectively. Slightly lower excess odds of were also imposed, including (a) the duration of
24% and 49% were noted for all subjects and live adult residence in study states, (b) the number of
subjects, respectively, using a continuous analysis. lifetime residences, (c) employment in mining, and
Among the different histologic types, large-cell (b) the ability to complete the interview.
carcinoma exhibited both a statistically significant Of the 1,474 cases and 1,811 controls (Table
continuous and categorical trend. The differences 1), 8% of the cases and 14% of the controls never
in the linear excess odds between histologic types smoked. Telephone and in-home interviews
were not statistically significant, however. A collected information on residential history,
reanalysis of the Iowa data, incorporating glass- education, medical history, and lung cancer risk
based retrospective reconstruction detectors, is factors. As part of the residential history, subjects
planned /4244/. Supplementary papers concerning provided information about each home that had
the Iowa study are available elsewhere /4551/. been occupied for one year. The information
included the number of hours spent on each floor
Connecticut/Utah/Southern Idaho Combined of the home, the location of bedroom where they
Residential Radon Study slept, and whether they worked outside the home.
Forty-nine percent of the case respondents and one
Sandler et al. /23/ performed a combined case- percent of the control respondents were proxy
control study of subjects residing in Connecticut, respondents. Radon measurements were attempted
Utah, and Southern Idaho. The findings of the in each home where the subject had lived since the
paper have not yet been published, but the study age of 25, as well as the longest childhood
methodology has been described /23/. The lung residence. In all homes measured, yearlong alpha
cancer cases were male and females aged 40 to 69 track measurements were placed in the subject's
years and diagnosed between 1989 and 1993 from bedroom, in another room on the lowest living
the cancer registries and medical records located in level where significant time was spent, and in most
their respective states. A screening telephone basements (Table 2). In a sub-sample of multi-
interview was used to select subjects according to level homes, a detector was placed on each level.
current and past tobacco usage (never smokers, An average of 4 homes were reported for the
non smokers, and current smokers). In addition to time window of interest (age 25 up to 5 years
the reported primary lung cancer diagnosis, for the before diagnosis for cases or before interview for
majority of cases, tissue slides were reviewed for controls). Yearlong radon measurements were
histologic verification. Controls were identified by available for 57% of the eligible dwellings in
random telephone screening. For Utah and Idaho, Connecticut and 60% in Utah/S. Idaho (Table 3).
controls over the age of 64 years were also Seventy-nine percent and eighty-three percent of
identified from Health Care Finance Administra- the subjects in Connecticut and Utah/S. Idaho,
tion files. Randomized recruitment was used to respectively, had dosimetry coverage (one or more
select controls matched on smoking status 10 years radon measurements) in at least 50% of their
before interview, age, and gender. In most instances, homes for the time window of interest. Overall,
one control was chosen for each case, except in 62% of subjects had complete exposure
Utah where two controls were selected for never information for the period from 5 to 25 years
and non-smokers. Additional selection criteria before diagnosis for cases or interviewed controls.
162 R.W. FIELD
IOWA (all subjects) 8.7
IOWA (live subjects)
MISSOURI - II (CR-39)
MISSOURI - II (ATD)
NEW JERSEY - I and II
4
MISSOURI - I
Odds Ratio
3
2
1
0
0 50 100 150 200 250
Radon Gas Concentration (Bq m-3)
Fig. 1: Plot of odds ratios versus estimated radon concentrations for the various exposure categories for each study.
The radon gas concentration point estimates were constructed by using the midpoint of the exposure category
or the lower limit in the case of the highest exposure category from each of the study's publications.
Confidence intervals for the point estimates are not presented on the plot. The reference line represents an odds
ratio of 1.0 or no increased risk. Odds ratios are presented for both types of radon measurements used in the
Missouri-II Study. Findings for both all cases and controls and the live cases and controls are presented for the
Iowa Study.
Average exposures to radon from age 25 to 5 years DISCUSSION
before diagnosis or interview were calculated as a
time-weighted average of both the amount of time The residential radon case-control studies
in each residence and the proportion of time spent performed in the United States had not only many
on each level of the home. Mean radon similarities in study design but also numerous
concentrations from measured homes were used to factors that varied among the studies, including
impute radon concentrations for similar homes that differences in state residential radon concentration
could not be measured. Similar residences were distributions (Fig. 1; Table 3) and study designs.
identified by use of regression trees that included The studies designs varied by case selection
categories like the level of home, housing method, subject residency requirements, exposure
characteristics, geological characteristics, and windows of interest, dosimetry methods, and
others /52/. The authors indicate /23/ that they will analytical analyses. For example, the first exposure
estimate lung cancer risk associated with category for the combined New Jersey study ended
cumulative radon exposure for specific time at 37 Bq m-3, whereas the first exposure category
windows. for the Missouri-I Study ended at 29 Bq m-3.
RESIDENTIAL RADON CASE-CONTROL STUDIES IN THE UNITED STATES 163
Because the first category is the referent category lifetime. Although studies of radon-exposed under-
for the calculation of the ORs for the remaining ground miners report that the 15 to 20 years before
categories, comparisons among the studies become the development of lung cancer is the biologically
more involved. Therefore, cautiously interpreting important period to assess exposure, radiation
the comparative findings between the studies and exposures occurring at younger ages may also
carefully weighing the strengths and weaknesses carry increased risk. Attempts to find an in vivo
of each study is prudent, while considering the marker, such as measurement of polonium-210 in
following observations. bone, to predict the lifetime cumulative radon
Elevated risks (ORs) were noted for the exposure have been somewhat limited for
highest exposure categories for all of the published residential studies by their limited level of
studies, except for the Missouri-II analysis that sensitivity and the confounding by other sources
relied on current home contemporary radon gas (such as cigarettes) of polonium-210 deposition
measurements (Table 4, Fig. 1). The combined /54, 55/.
phase I and II New Jersey Study produced the Residential radon case-control studies have
highest categorical OR (OR=8.7), which was also some advantages over other types of case-control
statistically significant at the 90% confidence studies in determining retrospective exposure
level. This result requires a cautious interpretation, because a significant proportion of the radon
however, because that category contained a very exposure occurs in the home and the radon
limited number of subjects. In the Missouri-II concentrations can be measured at some later date.
study, statistically significant upper exposure Uncertainty in the estimating retrospective radon
categories were also found for the glass-based CR- exposures increases, however, when certain time
39 detector analysis (OR=3.3) and for the live case periods in the 15 to 20-year time period before
subset analysis of the Iowa study (OR=2.1). A study enrollment are missing. For example,
nearly statistically significant odds ratio of 1.8 was consider the conflicting results from the Missouri-
also noted for the upper category of the overall II study, which used two methods to estimate past
analysis for the Iowa study. Statistically significant radon residential radon concentrations. The first
tests for trend in ORs with increasing cumulative technique used by the Missouri-II study relied on
radon were noted for both the Iowa Study and the current radon gas concentrations in the current
Missouri-II Study (glass-based CR-39). home to predict past exposures. In some cases, the
In reviewing the findings from the residential current home was occupied for only a few years,
radon case-control studies performed in the U.S., which likely resulted in poor overall retrospective
the studies with the more advanced dosimetric radon concentration estimates. The inability to
approaches have indicated a statistically signi- account for missing time periods increases the
ficant association (95% confidence level) between likelihood of exposure misclassification in a study,
prolonged residential radon exposure and lung which in turn decreases a study's power to detect
cancer. In fact, Field et al. /22/ and Alavanja et al. an association if one exists. To capture an average
/53/ suggested that the inability to detect an integrated radon exposure over a longer period, the
association in certain studies may have been due to Missouri-II study used CR-39 (RRD) measure-
poor retrospective radon exposure assessment /53/. ments from glass items that were located for many
One of the major challenges in performing a case- years in the current home and in previous homes.
control study is the ability to assess retrospectively The first technique did not find any association
the exposure to an agent accurately over a person's between the radon gas measurement and risk of
164 R.W. FIELD
lung cancer, whereas the second technique found a Missouri studies. Following the intercalibration of
statistically significant association. The positive the detector, the Iowa researchers will be
findings using the more advanced measurement analyzing the results of the glass-based
technique are attributable either to decreased retrospective radon detector measurements that
exposure misclassification or to some unknown have already been performed in the Iowa study
systematic bias. Additional work to determine the homes. A pooling of the glass-based results from
validity of this new dosimetry method is currently the Iowa and Missouri studies is planned /56/. In
underway /56/. addition, the data from all the residential radon
The findings of the Iowa Radon Lung Cancer studies that have been performed in North America
Study also suggest that improved retrospective are being pooled. Once the North American
radon measurement enhances the ability of a study pooling is complete, the data will be pooled with
to detect an association between radon exposure the data from the on-going European pooling.
and lung cancer. The Iowa study limited the In summary, the general findings from the
enrollment only to subjects who had been living in United States studies, along with extrapolations
their current home for at least the previous 20 from radon-exposed underground miners, support
years. This feature of the study prevented gaps in the conclusion that prolonged exposure to
the radon measurement data and allowed the residential radon may contribute to a significant
investigators to focus their radon measurements on increase in lung cancer risk.
a single home per subject. The Iowa study also
collected information on where the subject spent
time within the home, as well as the time spent REFERENCES
outside the home and in another building. This
information on personal mobility was linked to the 1. Greenlee RT, Hill-Harmon MB, Taylor M, Thun
estimated radon concentrations for each area to M. Cancer statistics. 2001. Cancer J Clinicians
determine a cumulative retrospective exposure 2001; 51: 1536.
2. International Agency for Research on Cancer.
estimate for the 20-year period before study
IARC Monographs on the Evaluation of Carci-
enrollment. The Iowa authors also performed an nogenic Risks to Humans. Man-Made Mineral
error analysis suggesting that alternative, less Fibres and Radon, Vol. 43. Lyon, France: IARC,
rigorous methods, which failed to link either 1988; 1300.
mobility or all the radon measurements in the 3. Agency for Toxic Substances and Disease
home, produce lower risk estimates. Several of the Registry. Toxicological Profile for Radon. (Final
Report, ATSDR/TP-90/23). Atlanta, Georgia,
studies (New Jersey and Connecticut) had
USA: ATSDR, Public Health Service, U.S. Dept.
relatively low residential radon concentra-tions, of Health & Human Services, 1990; 1170. NTIS
which reduced the overall power of the studies to Accession No. PB91-180422.
detect an association. Other methodologic 4. Johnson B. A review of health-based comparative
challenges associated with residential radon risk assessments in the United States. Rev Environ
studies are presented elsewhere /45/. Health 2000; 15: 273287.
5. National Research Council. Health Effects of
Additional research in the area of residential
Exposure to Radon, BEIR VI, Committee on
radon epidemiology is currently underway in the Health Risks of Exposure to Radon (BEIR VI),
U.S. Investigators from the Iowa and Missouri Board on Radiation Effects Research, Com-
studies are calibrating the glass-based detectors mission on Life Sciences, Washington, DC, USA:
(RRDs) that were used in both the Iowa and the National Academy Press, 1998.
RESIDENTIAL RADON CASE-CONTROL STUDIES IN THE UNITED STATES 165
6. Pollycove M. Nonlinearity of radiation health cancer among New Jersey women. Cancer Res
effects. Environ Health Perspect 1998; 106(Suppl 1990; 50: 65206524.
1): 363368. 19. Schoenberg JB, Klotz JB, Wilcox HB, Szmaciaz
7. Neuberger JS. Residential radon exposure and SF. A Case-Control Study of Radon and lung
lung cancer: An overview of published studies. cancer among New Jersey women. Twenty-Ninth
Cancer Detect Prev 1991; 15: 435441. Hanford Symposium on Health and the Environ-
8. Stidley CA, Samet JM. A review of ecologic ment, Indoor Radon and Lung Cancer: Reality or
studies of lung cancer and indoor radon. Health Myth? United States Department of Energy and
Phys 1993; 65: 234251. Battelle, Pacific Northwest Laboratories (sponsors).
9. Neuberger JS, Lynch CF, Kross BC, Field RW, Columbus, Richland, Washington, USA: Battelle
Woolson RF. Residential radon exposure and Press, 1992; 905918.
lung cancer: Evidence of an urban factor in Iowa. 20. Alavanja MCR, Brownson RC, Lubin JH, Berger
Health Phys 1994; 66: 263269. E, Chang J, Boice JD. Residential radon exposure
10. Blot WJ, Xu Z, Boice JD, Zhao D, Stone BJ, Sun, and lung cancer among nonsmoking women. J
J, et al. Indoor radon and lung cancer in China. J Natl Cancer Inst 1994; 86: 18291837.
Natl Cancer Inst 1990; 82: 10251030. 21. Alavanja MC, Lubin JH, Mahaffey JA, Brownson
11. Pershagen G, Liang Z-H, Hrubec Z, Svensson C, RC. Residential radon exposure and risk of lung
Boice JD. Residential radon exposure and lung cancer in Missouri. Am J Pub Health 1999; 89:
cancer in Swedish women. Health Phys 1992; 63: 10421048.
179186. 22. Field RW, Steck DJ, Smith BJ, Brus CP,
12. Pershagen G, Akerblom G, Axelson O, Clavensjo Neuberger JS, Fisher EL, et al. Residential radon
B, Damber L, Desai G, et al. Residential radon gas exposure and lung cancer: The Iowa radon
and lung cancer in Sweden. N Engl J Med 1994; lung cancer study. Am J Epidemiol 2000; 151:
330: 159164. 10911102.
13. Ruosteenoja E, Makelainen I, Rytomaa T, 23. Sandler DP, Weinberg CR, Archer VE, Rothney-
Hakulinen T, Hakama M. Radon and lung cancer Kozlak L, Bishop M, Lyon JE, Stolwijk J. A case-
in Finland. Health Phys 1996; 71: 185189. control study in Connecticut and Utah. Pro-
14. Auvinen A, Makelainen I, Hakama M, Castren O, ceedings of the American Statistical Association
Pukkala E, Reisbacka H, Rytomaa T. Indoor Conference on Radiation and Health: Indoor
radon exposure and risk of lung cancer: A nested Radon and Lung Cancer Risk. Radiat Res 1999;
case-control study in Finland. J Natl Cancer Inst 151: 103105.
1996; 88: 966972. 24. New Jersey State Department of Health, Division
15. Darby S, Whitley E, Silcocks P, Thakrar B, Green of Epidemiology and Disease Control, Division of
M, Lomas P, et al. Risk of lung cancer associated Occupational and Environmental Health, A case-
with residential radon exposure in South-West control study of radon and lung cancer among
England: A case-control study. Brit J Cancer 1998; New Jersey women. New Jersey, USA: NJSDH
78: 394408. Technical ReportPhase I, 1989.
16. Kreienbrock L, Kreuzer M, Gerken M, Dingerkus 25. Schoenberg JB, Klotz JB, Wilcox HB, Gil-del-
G, Wellmann J, Keller G, Wichmann HE. Case- Real MT, Stemhagen A, Nicholls G. Lung cancer
control study on lung cancer and residential radon and exposure to radon in women in New Jersey.
in western Germany. Am J Epidemiol 2001; 153: Morbid Mortal Weekly Report 1989; 42: 715
4252. 718.
17. Letourneau EG, Krewski D, Choi NW, Goddard 26. Schoenberg JB, Wilcox HB, Mason TJ, Bill J,
MJ, McGregor RG, et al. Case-control study of Stemhagen A. Variation in smoking-related lung
residential radon and lung cancer in Winnipeg. cancer risk among New Jersey women. Am J
Manitoba, Canada, Am J Epidemiol 1994; 140: Epidemiol 1989; 130: 688695.
310322. 27. Klotz JB, Schoenberg JB, Wilcox HB. Relation-
18. Schoenberg JB, Klotz JB, Wilcox HB, Nicholls ship among short- and long-term radon measure-
GP, Gil-del-Real MT, Stemhagen A, Mason T J. ments within dwellings: Influence of radon
Case-control study of residential radon and lung concentrations. Health Phys 1993; 65: 367374.
166 R.W. FIELD
28. Alavanja MC, Brownson RC, Benichou J, Swanson Brownson RC, Alavanja MC. Dietary hetero-
C, Boice JD Jr. Attributable risk of lung cancer in cyclic amines and the risk of lung cancer among
lifetime nonsmokers and long-term ex-smokers Missouri women. Cancer Res 2000; 60: 3753
(Missouri, United States). Cancer Causes Control 3756.
1995; 6: 209216. 40. Bennett WP, Alavanja MC, Blomeke B,
29. Alavanja MC, Brownson RC, Benichou J. Vahakangas KH, Castren K, Welsh JA, et al.
Estimating the effect of dietary fat on the risk of Environmental tobacco smoke, genetic suscepti-
lung cancer in nonsmoking women. Lung Cancer bility, and risk of lung cancer in never-smoking
Suppl 1996; 1: S63S74. women. J Natl Cancer Inst 1999; 91: 20092014.
30. Brownson RC, Alavanja MC, Chang JC. Occu- 41. Sinha R, Kulldorff M, Curtin J, Brown CC,
pational risk factors for lung cancer among Alavanja MC, Swanson CA. Fried, well-done red
nonsmoking women: A case-control study in meat and risk of lung cancer in women (United
Missouri (United States). Cancer Causes Control States). Cancer Causes Control 1998; 9: 621630.
1993; 4: 449454. 42. Lively RS, Steck DJ. Long-term radon concen-
31. Alavanja MC, Brown CC, Swanson C, Brownson trations estimated from 210Po embedded in glass.
RC. Saturated fat intake and lung cancer risk Health Phys 1993; 64: 485490.
among nonsmoking women in Missouri. J Natl 43. Field RW, Steck DJ, Parkhurst MA, Mahaffey
Cancer Inst 1993; 85: 19061916. JA, Alavanja MCR. Intercomparison of
32. Brownson RC, Alavanja MC, Caporaso N, Berger retrospective radon progeny measurement
E, Chang JC. Family history of cancer and risk of devices. Environ Health Perspect 1999; 107:
lung cancer in lifetime non-smokers and long- 905910.
term ex-smokers. Int J Epidemiol 1997; 26: 256 44. Steck DJ, Field RW. The use of track registration
263. detectors to reconstruct contemporary and
33. Brownson RC, Loy TS, Ingram E, Myers JL, historical airborne radon (222Rn)