PB8ft-218«79
Estimating the Risk of T,ung Cancer from
Inhalation of Radon Daughters Indoors
Review and Evaluation
Colorado State Univ., Fort; Collins
Prepared for
Environmental Monitoring Systems Lab.
E'ts Vegas, NV
J jri 88
[
iW. ana ii ^^iMriii"jWi';ii iiMifirii nii>i.Mia>qMa&a—i
-------
TECHNICAL REPORT DATA
(Phase rco*J Inst/ui lions on me mtrif tx/oer completing!
. REPORT NO.
tPA/600/6-88/008
3 RECIPIENT S ACCESSION NO
PB8 S - 2 I o o 7 9 /AS
«. TI T tE AND SUBTITLE
ESTIMATING THE RISK OF LUNG CANCER FROM INHALATION OF
RADON DAUGHTERS INDOORS: Review and Evaluation.
S REPORT DATE
June 1988
6. PERFORMING ORGANIZATION CODE
7 AUTHORISI
Thomas B. Borak and Janet A. Johnson
a PERfORMING ORGANIZATION HEPORT NO.
9 PERFORMING ORGANIZATION NAME AND ADDRESS
Department of Radiology & Radiation Biology
Colorado State University
Fort Collins , CO 80523
10 PROGRAM ELEMENT NO
CCVN1A
11. CONTRACT/GRANT NO
CR 813606-01
12 SPONSORING AGENCY N AMc AND ADORESS
Nuclear Radiation Assessment Division
Environmental Monitoring Systems Laboratory
P.O. Box 93478
Las Vegas, NV 89193-3478
13 T>PE OF REPORT ANO PERIOD COVEREO
Final - Oct. 86 to Apr. 88
14. SPONSORING AGENCY CODE
EPA/600/07
15 SUPPLEMENTARY NOTES
Work performed under cooperative agreement between Colorado State University and U.S.
Environmental Protection Agency. The Project Officer was Stuart C. Black, EMSL-LV.
1C. ABSTRACT
A review of the dosimetric models and epidemiological studies with regard to the
relation between indoor radon exposure and lung cancer indicates that the Working
Level is an appropriate unit for indoor radon exposure; that the uncertainty in
applying risk estimates derived from uranium miner data may be reduced by determining
ncse vs. mouth breathing ratios, residential aerosol characteristics, and lung cancer
risk vs. age at exposure; that there is persuasive evidence of an association between
radon exposure indoors and lung cancer; and theat epidemiological studies in progress
may provide a basis for revision or validation of current models but only is
experimental designs are employed that will permit pooling of data to obtain greater-
statistical power.
KEY WORDS AND DOCUMENT ANALYSIS
a DESCRIPTORS
b IDENTIFIERS/OPEN ENDED TERMS
C COSATi 1'ielii'Cpioup
U.S. Dft-'vpnai^ta! F
ill Jr.rormati
V; -p' v*'•»£>-)
fr'1' 'V
5j " {fl* ^
otecficn flgenca
n Resource
j
16. DISTRIBUTION STATEMENT
Release to the pub1ic
19 SECURITY CLASS i fhs% Key on i
Unciassifled
21 no Of- PAGES
no
70 SECURITY CLASS ('/III! ;>«!•/
Unciossified
22 PRICF
P®7
CPA Form 2270-1 4-77) previous edition it oeiounc
-------
gjpp PB38-218^79
/\£CHIVE>
prpA t PA/600/6-88/008
Prn June 1988
^0(7 -
6-
w-
ESTIMATING THE RISK OF LUNG CANCER FROM INHALATION OF
RADON DAUGHTERS INDOORS: Review and Evaluation
by
Thomas B. Borak and Janet A. Johnson
Department of Radiology and Radiation Biology
Colorado State University
Fort Collins, CO 80523
Cooperative Agreement
Number CR813606
Project Officer
cr
o
ft
p-
^cv, Stuart C. Black
¦5- Nuclear Radiation Assess ment Division
PCs
Environmental Monitoring Systems Laboratory
% l.as Vegas , NV 89193-3478
ENVIRONMENTAL MONITORING SYSTEMS LABORATORY
OFFICE OF RESEARG1 AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
LAS VEGAS, NV 89193-3478
REPRODUCED BY
U.S. DEPARTMENT OF COMMERCE
National Technical Information Servic*
SPIlli'JGrlCLD. VA ?2161
-------
NOTICE
The information in this document has been funded in part by the U.S.
Environmental Protection Agency under cooperative agreement number CR 813606
with Colorado State University. It has been subject to the Agency's peer and
administrative review, and it has been approved for publication as an Agency
document. Mention of trade names or commercial products is for illustration
only and does not constitute endorsement or recommendation for use.
-------
ABSTRACT
A review of the dosimetric models and epidemiological studies with regard
to the relaticn between indoor radon exposure and lung cancer indicates that
the Working Level is an appropriate unit for Indoor radon exposure; that the
uncertainty in applying risk estinates derived from uranium miner data may
be reduced by determining nose vs. mouth breathing ratios, residential aero-
sol characteristics, and lung cancer risk vs. age at exposure; that there
Is persuasive evidence of an association between radon exposure indoors and
lung cancer; and that epidemiological studies in progress may provide a basis
for revision or validation of current models but only If experimental designs
are employed chat will permit pooling of data to obtain greater statistical
power.
ill
-------
iv
-------
TABLE OF OONTENTS
Page
ABSTRACT iii
Summary l
INTRODUCTION 5
DOSIMETRY 7
Introduction 7
Dosimetric Models 8
Depos It Ion 9
Lung Morphometry 12
Clearance 13
Location of Radioactivity 14
Location of Target Cells 14
Discussion 15
Target Cells 18
Deposition 19
Clearance 19
Age 20
Aerosol Characteristics 20
Unattached 20
Attached 21
EPIDEMIOLOGY 2 3
Summary of Currently Available Epidemiologic Data 23
Role of Indoor Radon in Lung Cancer Etiology 27
Estimated Risk Coefficients 32
Confounding Variables 32
Ecological Studies 35
Studies in Progress 37
Statistical Power of Studies in Progress 38
RISK MODELS 45
Absolute risk model 46
Additive risk model 46
Proportionate hazards model 47
CONCLUSIONS 55
REFERENCES 57
APPENDIX SPECIAL QUANTITIES AND UNITS
APPENDIX B: SUMMARY OF PUBLISHED STUDIES WITH REGARD TO
INDOOR RADOtl DAUGHTER EXPOSURE AND LUNG CANCER RISK
v
-------
TABLE OF CONTENTS (continued)
APPENDIX C: SUMMARY OK UNPUBLISHED STUDIES WITH REGARD TO
INDOOR RADON EXPOSURE AND LUNG CANCER RISK
APPENDIX D: SUMMARY OF STUDIES IN PROGRESS WITH REGARD TO
INDOOR RADON EXPOSURE AND LUNG CANCER RISK
APPENDIX E: COMPUTATIONS OF LIFETIME LUNG CANCER RISKS
ATTRIBUTABLE TO INDOOR RADON
-------
I, I ST OF TABLES
Page
Table 1 Comparison of Dosimetric Models 10
Table 2 Overall Summary of Results of Published Studies
with Regard to Indoor Hadon and Lung Cancer 26
Table 3 Distribution of Indoor Radon Concentrations In U.S.
Population Adjusted for A pCl t,-' and 2 pCl
Exposure Cutoff Points AO
Table A Estimated Relative Risk for Various Cutoff
Concentrations and Risk Coefficients 42
Table 5 Calculated Number of Cases Required for p = 0.05
Level of Statistical Significance and Statistical
Power 0.80 or 0.90 * AA
Tablo 6 Comparison of Estimated Lifetime Excess Lung
Cancer Risk Due to Indoor Radon Exposure 53
vil
-------
ACKNOWLEDGMENTS
The authors gratefully acknowledge the assistance of the following
individuals:
s.
C. Austin
E.
G. LeTourneau
R.
Brownson
C.
R. Mulrhead
0.
Castren
G.
Pershagen
B.
L. Cohen
P.
Rand
F.
T. Cross
J.
M. "anet
S.
C. Darby
K.
J. schlager
N.
H. Harley
J.
B. Scnoenoerg
W.
Jacobi
J.
H. Ste^oings
A.
C. Janes
H.
G. Stuckwell
J.
B. Klotz
E.
Stranden
E.
Laconbe
C.
Svensson
E.
M. Lanctot
S.
fakacs
vilJ
-------
SUMMARY
Public concern Tor the health effects of exposure to indoor radon has
made It necessary to make risk estimates based on l,T3dequat-? and Incom-
plete data. The challenge to the professional cotnrunity is tc evaluate
available information on occupational exposure and adapt it to non-
occupational environments using basic concepts of radiation dosimetry.
The results must then be validated on the basis of epidemiologic evidence
and data on residential exposures.
Dose conversion factors for inhalation of radon daughters have
appeared in the literature since 1956. They range from 0.7-29 rnOy WLM 1
(Ja87). Recently the range of values has been reduced considerably. The
results shown in Table 1 indicate that the spread between modelf is
greater tiv .1 the conversion from occupational to environmental exposures
within each model. •
Each modc-i. has been formulated by distinguished scientists who have
selected input variables according to their interpretation of available
and often identical data. Ht this time there is no indisputable evidence
that per.nits ranking or elimination of any of the computations.
The average of all three models gives a ratio of dose conversion
factors for residential to occupational exposure of 1.3 t 1.3. The only
conclusion that ca.. be made with confidence Is that the ratio of dose
conversion factors I,-, greater than 1. The added uncertainty of deriving
risk coefficients u-ing data from underground miner? may not be
signi fleant.
1
-------
The concept of cumulative potential alpha energy la juf!iclent for
describing the exposure of individuals and there is no justification for
redefining or modifying the WLM or J h m . However, there are several
important factors which influence the con/ersion from exposure to dose.
Improved data on the following could reduce the uncertainty in the ri3k
estimates to the general public:
• Fraction of time nose breathing vs_. moutn breathing.
• Unattached fraction.
• Aerodynamic median diameter and geometric standard
deviation of attached aeroscls.
• Age dependence.
• Location of radiosensitive targets.
The common denominator for dose conversion factors is cumulative
exposure to potential alpha energy (WLM). Mcst epidemiological studies of
indoor environments measure radon gas oily." It is important to understand
the relationship between radon gas and radon daugnter concentrations.
James (Ja87) has reported that the conversion to dose can be related
directly to radon concentrations indoors. The reason is that for a
constant level of radon the potential alpha energy, WL, increases as the
concentration of room aerosols Increases. However, the availability of
condensation nuclei reduces the unattached fraction, fp. These
cc Tioensating factors tend to dampen variations in the dose con/orslon
factor for a given concentration of radon gas. These concepts should be
carefully evaluated in future studies.
The epidemiologic evldsnce of an association between indoor radon
exposure and lung cancer In the general population is persuasive but by no
means definitive. Twenty-one papers, published in the genera] literature,
2
-------
have been summarized In this report. In addition, the results of five
unpublished studies have been suitmarized.
Two general types of epidemiologic studies are represented In this
body of literature: ecological and case-control. Ecological studies may
have an Inherent systematic- bias towards showing no association between
lung cancer and Indoor radon due to the effect of population migration. A
second source of bias in such studies may involve secondary
characteristics of geographic regions studied which may either dilute or
enhance an apparent association. Due to the inherent problens with
Interpretation of ecological studies they can be weighted less heavily
than case-control studies in the assessment of the strength of the
evidence for a causative role of radon daughter exposure in lung cancer
et lology.
The majority of the case-control studies relied oil surrogate measures
of radon daughter exposure. However, at least these measures were deter-
mined on an individual home basis. The studies are so diverse in design
and execution that the data cannot p-"-led or combined in order to
increase the statistical significance. However, each of the published
studies can bo treated as an independent trial to test the hypothe3i3 of
an association between radon and lung cancer.
Six of the seven published case-control studios have Indicated a
relative risk or odds ratio greater than one. If there 13 no association
between indoor radon and lung cancer and there is no systematic bias among
the studies, it can be assumed that there would be a 50$ chance of finding
a positive association (relative risk or odds ratio greater than one) and
a 50% chance of finding a negative association. Using the binomial
probability distribution, the probability of six of seven 3uch studies
3
-------
showing a positive ascocl.it ion. if, in fact, none exists, Is approximately
0.06. This analysis depends on the assumption that the results of the
published studies represent a random sample from a binomial population of
results of all possible studies. The question of bias In publication of
studies could invalidate this analysis.
The studies in progress are generally of case-control design and will
use actual radon measurements. Several also have common design
features. Collectively, they have the potential to .how an association
between indoor radon exposure and lung cancer which would withstand a more
rigorous statistical analysis if such an association truly exl3ts. It is
much more difficult to provide definitive evidence that an association
does not exist if, in fact, this is the case.
Even under the best circumstances, the exposure data from studies In
progress may not be sufficiently refined to allow for development of ris*
models and risk coefficients independent of the Information already ob-
tained from studies of underground miners. It Is likely that the studies
in progress will provide a means for validating the adaptation of risk
models derived from miner data to non-occupational exposures among the
general population.
4
-------
INTRODUCTION
Inhal3tlon of radon gas was the first situation In which radiation
wan implicated as a cause for cancer. The problem can bo traced back for
more than 400 years. In the sixteenth century an unusual fatal disease
was occurring among underground miners in Bohemia. About 100 years ago
this disease was diagnosed as lung cancer and at that time about 50% of
the miners in the region died from lung cancer.
Around 192'J it was suggested that the high rate of lung cancer may be
attributed to elevated concentrations of t.he radioactive noble gas, radon.
In many ways it was difficult to reconcile the fact that an insoluble gas
could be responsible for the disease. However, in 1950 It was recognized
that the true cause of high absorbed doses to the lung was Inhalation of
the short lived radioactive descendants (daughters) of radon which are
Initially created by the decay of radon in air.
It has recently become evident than this same mechanism could be
responsible for the induct!on of lung cancer in the general public.
Measurements of radon in dwellings indicate that 20-60? of -the dose
commitment from natural background radiation is due to radon. It is
generally more pronounced in regions where dwellings must be closed and
insulated to protect the occupants from the weather.
Over the past several years, energy conservation has developed into a
popular- and patriotic theme. One of the easier ways to accomplish this is
to increase the Insulation In houses and reduce ventilation. This could
result in elevated levels of toxic gases, including radon, and increase
5
-------
the incidence of fatal lung cancers. Thus, an apparent cost effective
means for conserving energy could actually be unacceptably expons've in
terms of lives lost or life shortening. Approaches to resolving this
dilemma will depend on an understanding of the true risk for induction of
lung cancer from inhalation of radon daughters.
The objective of this report is to summarize state of the art
methodologies for deriving risk estimates from this environmental
pathway. It also Includes an evaluation of the uncertainties of each
method and suggestions for improving the risk ¦ estimation process. The
report Is divided into the following major sections:
• DOSIMETRY
s EPIDEMIOLOGY
• RISK MODELS
Sections on dosimetry and epidemiology are Included since each
discipline has contributed to the derivation of risk models employed to
assess public health detriment duo to Indoor radon exposure.
Current risk models are based on epidemiologic data from underground
miners. The intent of this report Is to summarize the epidemiologic data
available from Indoor radon studies and investigate Its usefulness as a
basis for estimating risk coefficients or validating those derived from
miner data. In addition, studies in progress are summarised and their
potential contribution to quantitative risk estimation discussed.
The section describing epidemiologic risk models Is included simply
to enhance the usefulness of this report. No attempt was made to evaluate
the merits and deficiencies of each of the models.
A summary that includes results from all three major topics is
presented and followed by a condensed list of conclusions.
6
-------
DGSIHETRY
INTRODUCTION'
Risk estimates Tor the Induction of lung cancer from occupational
exposure to the short-lived descendents of radon (daughters) have been
derived primarily from epidemiological studies of underground uranium
miners. The risks are related to the total accumulated exposure to
potential alpha energy of radon daughters in air (J h m ^ or WLM; see
Appendix A for a description of quantities and units). However, there are
large uncertainties in this method since the exposure for most miners has
been reconstructed from estimations of the concentration of radon
daughters underground. Nevertheless, there ic a basis for confidence in
the risk estimate for accumulated exposures down to 100 WLM (Th8b).
Classically, the response of biological systems to Ionizing radiation
i3 related to the absorbed dose received by the tissue or cells of
Interest. Since many organs appear to have different sensitivities to
radiation, weighting factors have been proposed to accommodate the vari-
ation in the appearance of late stochastic effects (ICRP77). There is,
therefore, a strong precedent to apply a similar methodology for
describing the induction of lung cancer by the inhalation of radioactive
aerosols such as the short-lived descendants of radon.
It has not been possible to measure the absorbed dose to lung tissue
from Inhalation of radon daughters; It must be calculated using models
which simulate the sequence of events loading to energy deposition. This
7
-------
requires a quantitative description of the physiological properties of the
respiratory system a3 well as the physical and chemical properties of the
Inhaled aerosol.
Cancer of the respiratory tract is one of the most common forms of
fatal cancer In industrialized countries. Exposure to radon daughters in
domestic environments may be an Important, factor for induction of this
disease in the general public (USEPA86). Controversy arises since an
estimate of the rate of incidence Is derived by combining concentration
measurements in dwellings with risk factors obtained from underground
mines (ICRP87).
It has b«.en proposed that principles of radiation dosimetry might be
capable of resolving this dilemma. This requires a hybrid procedure
whereby the occupational exposure is converted into absorbed dose in the
lung using aerosol characteristics and breathing patterns in mines. The
incidence of lung cancer In miners Is then related to dose rather than
exposure. These risk factors would then be applied to dose estimates
computed for the general public based on aerosols and breathing patterns
in domestic environments.
The following sections describe the underlying principles of the dose
models and methods for adapting these to environmental and occupational
situations.
DOSIMETRIC MODELS
There have been a number of attempts to model the absorbed dose to
the lung and portions of the respiratory track from inhalation of radon
daughters. These have been reviewed and summarized by the National Council
8
-------
on Radiation Protection and Measurement (NCRP81) and James (Ja87). Many
of the later calculations incorporated methods or concepts from previous
works. This report will focus on three recent mortals which were developed
by Harley (Ha72, IICRPS^, Ha86), Jacob! (Ja30, OECD83) and James (Ja87,
JaS6, Ja8!0.
Each model includes three basic components: deposition, clearance
and energy absorption. There are subcategories within each that control
the result of the calculation. The assumptions employed by each model are
summarized in Table 1. The objective is to identify which parameters have
the greatest Influence on calculations of absorbed dose and the
implications of converting from occupational to environmental conditions.
DEPOSITION
All models assumed that the air flow in the lower airways is
laminar. For unattached and attached daugtiters, diffusion Is the dominant
mechanism, and expressions exist which compute the deposition due to this
process (G0I9, In75). For the upper respiratory track, turbulence can
exist and the deposition may be enhanced by this secondary flow. When the
aerodynamic median dlaT.eter (AMD) or geometric standard deviation (0 ) of
the attached aerosols becomes large, deposition can also increase due to
gravitational settling and inertlal impaction.
The Jacobi model assumes diffusion based 011 laminar flow and enhanced
deposition in the upper airways based on experiments using a plastic
d tchctorr.ous symmetrical branching device. Harley also assumes that
laminar diffusion is the only mechanism with enhanced deposition in the
9
-------
Table 1. Comparison of Dosimetric Hodels
Category
Lung Morphometry
Bronchial Deposition
Masai Deposition:
unat tached
Attached
>
Clearance:
Mucocllliary Transport
Solubility
Location of Activity
Sensitive Cells
Dose
Jamc
Jacobl
Harley
Yeh & Schum
Weibel A
'JCI
Weibel A
Yeh & Schum
Laminar Diffusion
+ Inpacticn +
Sedi-nentat ion
Laml.iar Diffusion +
Turbulent Enhancement
in Upper Airways
La.ninar Diffusion +
Turbulent Enhancement
in Upper Alrway.s
50?
0%
Model constrained to
keep mucus thickness
constant
10* 7y = 15 mln
30? Ti,J - 10 h
'2
Distributed in
mucus and mucosa
All stem cells in
bronchial epithelium
at each generation
Dose averaged ov.r all
cells in epithelium
and over all bronchial
generations
50?
Model based on mucus
velocity in trachea
in TB region
Attached Ty - 10 h
Unattached 'Tl, - 15 min
7j
Uniformly distributed
in 15um layer of mucus
Basal cells distributed
at variable depth at
each generation
Dose to basal cells
averaged over
generations 2-15
60?
2?
Mucus production constant
over surface of TB
region
Neglected
Uniformly distributed
in 15un layer of mucus
Basal cells located
2?um below mucus layer
in segmented bronchi
Dose to shallow basal
cells in bronchial
generations 2-4
-------
Table 1. Comparison of Dosimetric Model3
Category James jacobi Harley
Breathing Rate:
Occupational
Residential
1.2 m-h"1
0.75 m3h~1
1.2 m3h~1
0.75 m3h~'
1.1
1.1 m3h_1 Active (67?)
0.54 m3h_1 Resting (33?)
Aerosol Characteristics:
Size:
Unattached
0.001ym
Diff. CoeT. =
0.05^ cm2 s"1
Diff. Coef. -
0.0025 cm2 s"1
Attached
Occupational
Res idential
AMD » 0.2pm
A!!D - 0. 1 vim
AMD = 0.2pm
AMD =¦ 0.1 5
AMD ¦ 0.1 7I'm
AMD = 0.12um
Unattached Fraction
Occupat ional
Residential
Dose Conversion
[rnGy/WLM]
Occupational
Residential
fp - 0.03
fp - 0.05
6.3 + 130 r„
10 + HO fl
f = 0.025
fp - 0.03 (ICRP50)
4.6 + 35 f
5.3 + 15 rr
f - 0.01 [1/0.6/0.3/0.2]
fp - 0.017 [1/0.9/0.6/0.4]
3-6
4.2
Scale Factors:
3reathing Rate
Age Dependence
Da a [BR]7*
B
Insignifleant
D Ages 0-10
D Adult
1.5
-------
upper airways determined using casts prepared from autopsy specimens of
the hurnc'.n bronchial tree.
The James model uses diffusion equations for laninar flow without
turbulent enhancement In the upper airways. His justification is basec! on
result.-; of deposition using ventilated pig lungs. The model does however
include gravitational sedimentation and inertlal Impaction for larger
particles.
LUNG MORPHOMSTRY
The geometrical configuratior: and size of the respiratory system
influences deposition of aerosols. Most models initially used airway
dimensions described by thr Wiebel A dlchotomous model (W163). It gives
the diameter and length of bronchial airways and assumes that airways at
each level of branching are Identical. .Jacobl uses thl3 description
exclusively.
Extensive measurements of airway size were reported by Yeh and Schum
(Ye80). They prepared a replica cast by injecting silicone rubber into a
lung In situ In the thorax of a human cadaver. This oroeedure preserved
the in vivo shape of the lung but gave rise to enlargement of some
airways. Harley has adopted a scaled down version of the Yeh-£chur model
that corresponds to the normal functional residual capacity of an adult.
More recently Phaien has reported measurements of airway si7.es from
replica casts of twenty lungs (Ph35). They derived regression formulae to
give the variation of airway diameter arv length as a function of age.
James uses the average of all three lung models 3i»ce "there 13 no over-
riding reason to prefer a particular model" (Ja87).
12
-------
CLEARANCE
In the bronchial region, the aerosols are deposited on the surface of
the mucus. They move from this location either by mucociliary clearance
toward the throat or absorption through the epithelium and elimination
into the blood stream.
In the pulmonary region the daughters are deposited on a thin surface
fluid in close proximity to the blood capillaries. For short-lived radon
daughter.; this clearance mechanism can be neglected since the dissolution
time is longer than the physical half lives.
The model of Harley assumes that both attached and unattached
daughters are insoluble and are cleared by mucociliary transport only.
Jacobi assumes that in addition to mucociliary transport attached
daughters have a solubility characterlzed by a 10 hour half-time while the
unattached daughters are transferred through the epithelium with a half-
time of 15 min.
James has developed a coinpaitmentalI zed model for clearance where 60%
of the radon daughters are Insoluble, 10® have a rapid clearance through
the mucosa with a half-time of 15 min. and 30° enter a compartment of pro-
tracted retention with a half-time of 10 hours. He does not distinguish
between attached and unattached daughters.
The mucociliary transport velocities are similar for all models with
values ranging from -10 mm min ^ in the trachea to -0.01 ram min 1 in
generation 1H.
13
-------
LOCATION OF RADIOACTIVITY
The location of the radioactivity on the walls of the airways
following deposition Is critical to dosimetry since the range of the alpha
particles is similar to the dimensions of the material surrounding the
cells. Harley and Jocobl assumed that the activity is uniformly distri-
buted In the mucus layer which is 15vim thick. James assumes deposition In
a thin layer of mucus gel only 7iim thick and that the activity penetrates
through the mucous and enters the mucosa containing the epithelium,
basement membrane and lamina propria. He assumes a concentration gradient
which falls to zero at the base of the mucosa where blood capillaries are
found and that the epithelium occupies the top half of this mucosal layer.
LOCATION OF TARGET CELLS
Lung cancers observed In uranium miners are primarily bronchial in
origin and usually localized In the first few generations of the bronchial
tree. It is generally assumed that the target cells are undifferenti-
ated stein cells located in the bronchial epithelium.
Harley identifies the targets as basal cells attached to the basement
membrane of the epithelium. Her model assigns a fixed depth for these
cells at 22um below the mucus-eplthellum interface for generations 1-9 and
10pm after the ninth generation. However, the model focuses upon the dose
to these shallow basal cells in the segmented oronchl, specifically
generation 4.
Jacobl also assumes that the targets are basal cells of the bronchial
epithelium. However, his model uses a distribution of depths below the
base of the cilia which decreases as the generation number Increases. The
14
-------
model computes the dose to bai'.il cells at a mean depth for each generation
and then averages over generation 2-]'4 to derive a 3ingle value for the
bronchial region.
James postulates that the radiosensitive cells responsible for the
Induction of lung cancer are distributed throughout the bronchial airways.
He assumes that these cells are not restricted to basal cells attached to
the basement membrane, but extend over the entire thickness of the
epithelium. Thus, the dose Is computed to all cells in the epithelium
whic'". has a thickness determined by measurements of clinical biopsy
specimens made by Cestineau (Ga69). The final result Is the mean dose to
all epithel-.il cells in each generation and averaged over all generations.
Discussron
In any theoretical exercise such as this it is Important to recognize
the difference between postulates which are axiomatieally true without
need for proof and assumptions which are educated guesses. Unfortunately,
there are precious few axioms in radiation biology.
The conjecture that lung cancer is directly related to average energy
deposited per unit mass is a clear example of this. Harley states " li>at
the underlying risk factor for radon daughter Induced lung cancer "ought"
to be the alpha dose to target cell3 (Ha8'i). The OECD states that an
increased risk of bronchogenic cancer is the "expected" consequence of
absorbed dose In bronchial tissues (0SCDS3). The NCR? takes a firmer
stance saying that absorbed dose to cells in the epithelium of the upper
15
-------
airways in the tracheobronchial tree "is" the significant dose for cancer
induction V'iCRPS'l). Those arc at best intelligent speculations.
A popular description of carcinogenesis 13 that radiation is
responsible for initiation of the disease which remains dormant until
acted upon by one or mot -- promoters (We83). There have been several
studies of oncogenic transformation of mammalian cells In vitro. Lloyd er.
al. reported no excess transformation in mouse embryo fibroblasts for
doses less than 20 rad when irradiated with a particle;, having a LET of 85
KeV/ym (L179). Robertson et al. shows an excess transformation frequency
in mouse fibroblasts of 2x10-14 per irradiated cell at 25 rad using alpha
particles with a mean LET of 150 KeV/pm (Ro83). Hleber et al. obtained a
transformation frec/iency of 1.6x10 14 per Irradiated cell for an o dose of
25 rad (HI87). They also report that the effect is not dependent on dose
rate down to 200 mrad/min.
A linear interpolation of this data yields the transformation
frequency at low doses of - 1x10 -'/rad cell. Similar experiments with
human epithelial cells have not been successful Ir generating enough
transformation to obtain an estimate of the frequency at low doses.
However, since the mouse cell lines already have one damaged lccus, it
might be necessary to have two independent events to produce a
transformation in normal Gills (I.e. [1x10 ~* ] x [1x10 J] ¦= 1x10
transformation/rad. rell).
T'ne approximate number of basal cells at risk in bronchial
ri p 2
generations 2-!i Is - 5x10 (airway area 5 cm-; 1 basal cell/1000|im ).
An indoor exposure to 'I pC1 1 ^ (0.02 WL; 0.8 WLM a 0.5 rad WLM for
50 years yields - 10"^ oncogenic transformed, ion. This example Is flawed
16
-------
for many reason;-; however If each transformation develops into a cancer,
the ristt estimate i3 similar to that obtained by other methods. It
Illustrates the Importance of understanding the transformation process in
order to assign reliable risk factors to absorbed energy.
Therefore, risk estimates from dosimetric considerations must Include
the human experience. At this time the most complete information comes
from cancer incidence in underground uranium mine'.",. Dosimetry cannot
include carcinogenic co-factors such as diesel fume3 and dust, but on the
other hand houses are not necessarily as pure as one would like to believe
(Ga85). Thus, It Is recommended that dosimetry should focus upon scaling
factors that reflect both the physical properties of the aerosols in each
environment and the physiological factors associated with respiration,
deposition and clearance.
Quality factors are used to account for the relative biological
effectiveness of different types of radiations. In general, the exposure
rates from external gamma rays in domestic environments are low a:id can be
neglected. Gan.-na exposure rates in underground mines are higher than
houses, but there is a la°ge uncertainty In the exposure to the population
of miners currently used to obtain risk estimates. Host probably the
accumulated gamma dose to the lung was only a small fraction of the alpha
dose from inhaled radon daughters.
Weighting factors have been derl/ed by the ICHP to adjust for
differences in the sensitivity of organs with regard to the dc-veloppent of
cancer following an absorbed dose of ionizing radiation (ICRP77). This
has been extended to include separate weighting factors for the pulmonary
and bronchial regions of the lung (OECD33, ICRP81). Since the toxic agent
-------
is alpha particles from radon daughters and the biological end point Ls
strictly bronchogenic cancer, it ls not necessary to Include these
adjustments.
The concepts of nicrodosimetry can be used to obtain i •.'ormalion
concerning mechanisms on the subcellular level. These could intimately
lead to an improved understanding of the Initiation processes associated
with carcinogenic transformation. At present this experimental
methodology is restricted primarily t<) lnvltro Investigations.
Since the objective Is to derive risks for the general public based
on exposure of uranium miners, absorbed dose ls a sufficient basis of
comparison. The subcellular mechanisms of radiation action and
macroscopic weighting factors for dose equivalent are similar for both
groups and do not need to be included. The models should be restricted to
physical and physiological properties which can be verified with measure-
ments whenever possible.
TARGET CELLS
There Is a general consensus that transformed 3tem cells or their
differentiated progeny do not migrate large distances within the respira-
tory system. Since most, of the observed primary tumors are located in the
upper regions of the bronchial tree, that Is where the dose should be
calculated. Combining the dose to basal cells over all generations (2—1 it)
does not produce a large effect compared to considering only generations
However, the dose ls increased If the deposited activity migrates
below the mucus into the epithelium according to the James model.
18
-------
DEPOSITION
It is reasonable to assume that there Is some turbulence In the upper
airways. Cohcr: (CcS6) reports a larger deposition correction factor In
the trachea than previously suggested by Jacobi (Ja80). By not using
correction factors the deposition is shifted toward lower generations
which increases the dose to the segmental bronchi.
Increasing the breathing rate will Increase the intake of radon
daughters. However, there is a corresponding Increase in flow rate Ir the
bronchial tree which decreases the fractional deposition. Although these
two effects are not completely compensating the effect on dose Is small
and James suggests a scaling factor depending on the square root of the
breathing rate.
The most important factor controlling deposition Is nose vs. mouth
breathing. Miners are generally involved in light to heavy activity which
could result in intermittent or continuous mouth breathing. .A large
fraction of the exposure in indoor environments occurs when people arc
sleeping and therefore nose breathing. The nasal passage is an effective
filter for unattached daughters which preferentially deposit in the first
few generations of the tracheobronchial tree. This factor must be
understood to effectively compare absorbed do3e in the two environments.
CLEARANCE
Changes in the rale of mucociliary clearance do not have a large
effect on absorbed dose. Assumptions about solubility and transfer into
the blood make an appreciable change in absorbed dose to epithelial cells.
Jacobi assumes a rather high solubility and correspondingly short half
19
-------
life in the mucus which tends to lower the dose to epithelial cells.
Harley concludes that the effects of the clearance mechanism are small,
and the James model Is somewhere In between these two extremes.
AGE
It Is Important to recognize that Indoor exposure la not limited to
healthy middle-aged males. Children spend large amounts of time indoors
especially In the winter months. The OECD reports that the dose to the
tracheobronchial tree is about a factor of 1.5 higher for children than
adults (OECD83). James concludes that the mean bronchial dose la only
marginally increased in young children and can be regarded as insignifi-
cant (Ja87). Hofmann has computed age dependent modifying factors for
alpha dose rates to the respiratory track (Ho79). He obtains values of
1.9 for Infants with a maximum value of 2.4 at age 6.
The uncertainty is a result of assumptions on the thickness of the
mucus lay-?»' and epithelium in children. This issue needs to be resolved
along with determinations of nose _vs. mouth breathing for the general
publIc.
AEROSOL CHARACTERISTICS
Unattached
All of the unattached daughters that penetrate the nasal passage will
be deposited in the tracheobronchial tree. There I3 some difference of
opinion as to the size distribution of these particles. The Harley model'
2
uses a diffusion coefficient of 0.0025 cm /3 corre3po.n1ing to a particle
-------
size of 0.005ym (Kn83). Jacob! and James use a diffusion coefficient of
0.05^ corresponding to a particle size of 0.001ym.
The nasal passage will filter out '10-60 percent of the unattached
radon daughters independent of size range between 1 and 5 nm. However,
the smaller particles will have an increased deposition in the trachea and
thus a smaller fraction will be available for deposition in the following
generations. This size effect is, however, not large and does not appreci-
ably alter the do3e conversion factor for unattached radon daughters.
The major factor 13 not so much the size of the unattached particles,
but the quantity. All dose models are sensitive to fp, the fraction of
potential alpha energy which Is unattached. Although only 3 to 5? of the
potential alpha energies is unattached it accounts for up to 50? of the
absorbed dose to the bronchial epithelium. Measurements of this quantity
are essential for comparing mine and residential atmospheres.
Attached
Radon daughters attached to condensation nuclei or other aerosols are
responsiole for the largest contribution to potential alpha energy. How-
ever, they are not aercdynamically suitable for efficient deposition In
tne upper airways of the human respiratory tract. In general less than 2%
of these particles are filtered by the nasal passage and about 3-10? are
deposited in the tracheobronchial region.
Deposition In the first few generations depends on the aerodynamic
diameter of the carrier aerosols. Underground mines are dusty and can
have high concentrations of fumes from internal combustion engines.
Measurements indicate an AMD ranging from 0.1 - 0.3l>m in mines.
21
-------
Atmospheric conditions in dwellings tend to produce smaller aerosols:
measurements range from 0.03 - 0.1pm. The AMD can change rapidly
depending on the activities of the occupants.
Deposition by diffusion Increases in the upper airways as the AMD of
the attached aerosol decreases. However, as particles become larger or
the distribution broader (i.e., a large) there is an Increase in depo-
S
sltlon due to Impaction. The models do not specifically address depo-
sition at bifurcations which can be enhanced (Ma?2, Co37).
Harley indicates that the dose conversion factor can be a factor of ^
higher for aerosols having an AMD of 0.03pm compared to an AMD of 0.12ptn
(Ha86). It has also been mentioned that aerosols might grow after entering
the humid airways of the respiratory tract. This would decrease deposition
In the tracheobronchial region.
The type and concentrations of condensation nuclei can also affect
the mixture of suspended radon daughters. However, changes in the
daughter ratios do not have a large influence on the dose conversion
factors based on potential alpha energy.
22
-------
EPIDEMIOLOGY
Exposure to radon daughters has . been generally accepted,, as a
causative factor In the observed excess risk of lung cancer among under-
ground miners. Epidemiologic studies spanning three decados have been
reported in the literature with relatively good agreement among their; as to
risk coefficients (Th85). The data are continually being updated and
reanalyzed as the follow-up period for miners Increases.
Due to basic differences between miners and members of the general
public In terms of lung morphometry, breathing patterns and the aerosol
characteristics of their environment, the applicability of the miner
derived risk coefficients to the general public has been questioned. Is
described in the previous section, dosimetric analyses have been used to
adjust the coefficients. However, validation of the dosimetric models
using epidemiologic data is at least desirable if not essential.
Since it is generally accepted that radon daughters cause lung
cancer, the concern of indoor radon epidemiology need not be to prove the
causal relationship. The objectives should be to determine if the risk is
significant under the conditions and levels associated with residential
exposure and to develop or validate risk coefficients.
SUMMARY OF CURRENTLY AVAILS IDEMIOLOGIC DATA
The emphasis on energy conservation during the previous decade led to
a concern about radon daughter exposures in residences. As a consequence,
a significant role In che etiology of lung cancer In the general
23
-------
population lias been proposed for this agent. From 5,000 to 20,000 lung
cancer deaths per year are postulated due to radon daughter exposure from
Indoor radon (USEP«66). However, much of the epidemiologic data with
regard to non-occupational radon daughter exposures has only recently been
published and at this time the Information 13 still relatively sparse.
Summaries of the Individual studies published in the open literature to
date are given in Appendix B.
Thero are some studies which have been completed but are as yet
unpublished. Summaries of unpublished studies are given in Appendix C.
In addition to the unpublished completed studies, some preliminary results
are available from pilot studies and studies in progress. Where this
Information ha3 been published in the open literature it is Included in
Appendix B.
The i .iformat Ion in the appendices is organized to give a brief
summary of each study with regard to basic method and results. Very few
of the studies are quantitative with respect to radon or radon daughter
exposure. All of the studies which include quantitative data suffer from
a lack of statistical power due to low numbers of lung cancers Included.
The studies are too varied in design, type of surrogate for radon daughter
exposure used, method and type of data collection and reporting, and
control for confounding variables to allow for pooling of results by any
reasonable statistical netnod for the purpose of examining exposure-
response quantitatively. This is unfortunate since, collectively, they
provide persuasive evidence of an etiologic role of indoor radon in lung
cancer in the general population.
-------
No estimate of risk coefficients can be made on the basi3 of these
studies taken together. However, several individual studies provided
enough Information on which to base an estimate of risk coefficients.
These estimates are noted in the appendices.
The Judgment as to whether excess risk of lung cancer due to indoor
radon exposure is demonstrated by a study depends on the ma.^iitude of the
point estimate of relative or absolute risk, the reliability of the data,
the degree to which confounding variables were taken into account and the
statistical significance of the results. In each appendix, the statement
of whether excess risk was demonstrated by each study 13 the opinion of
the authors of this report and 13 based on the above mentioned
considerations. The conclusions of the investigator are also explicitly
stated. In most cases the investigators for the individual studies used
conservative statistical requirements (I.e. 95% confidence limits) in
postulating an effect of Indoor radon on lung cancer risk. Since the
objective of this report is to look more generally at the evidence, an
element of judgment was u?ed in deciding whether an excess risk was
demonstrated. As stated previously, the methodological variations among
the studies precluded any statistical pooling of the data to obtain
results wil i greater statistical significance.
Some of the studies listed In the appendices were more general than
others and covered cancers other than lung cancer and risk factors other
than indoor radon. The appendices include only those results pertinent to
lung cancer risk from indoor radon. Table 2 is a brief summary of all of
the published studies.
25
-------
Table 2. Overall Summary of Results of Published Studies
with Regard to Indoor Radon and Lung Cancer
Author (date) Results
Approx. Range
Rn Daughter
Concentration
(WL)
Approximate
Re]ative
Risk
Estimated
Risk
Coefficient
CASE CONTROL STUDIES
Axelson (Ax79)
+ +
na
1.8-5.4
na
Ouimette (Ou83)
0
na
1.2
na
Pershagen (Pe84)
+ /0
na
na
na
Edling (Ed8H, Ed86)
+ +
0.011-0.OUG
1.2-5.1
5-7 E-6
per PY-WLM
Damber (Da86)
+
na
T.4-2.0
na
Lees (Lee87)
+
na
1 .4-2.il
na
Svensson (Sv87)
+ +
na
2.2
na
ECOLOGICAL STUDIES
Bean (Be82)*
+ +
na
1.3-1.7
na
Dousset (Do85)
0
na .
na
na
Forastiere (FoS5)
+ /o
na
1.2
na
Hofmann (Ho85, H086)
0
0.2-0.4 WLM/a
na
na
Archer (Ar87)
+ +
na
na
na
Fleischer (F181)
+ +
na
na
na
Edling (Ed82)
+ +
na
na
na
Hess (He83)
+ +
na
na
na
Letourneau (Let8'j)
0
na
na
na
Fleischer (F186)
+ +
na
na
na
Walter (Wa86)
0
na
na
na
Stranden (Str86, Str87)
+ +
na
na
RR coeff.
0.003-0.009
per WLM
Castren (Ca87))
0
na
na
na
OTHER
Simpson (Si83)
0
na
na
na
++ — Significant positive association
+ — Posit've association r.ot significant
0 — No association
+/0 — Equivocal
na — Mot Applicable
* The surrogate for indoor radon used in this study, radium concentration in
water, has not been correlated with indoor radon.
26
-------
In general, epidemiologic studies arc considered "positive" only if
they show a statistically significant effect. Studies which show an
increased risk of disease which is not statistically significant are
considered "inconclusive". However, such studies should not be considered
"negative" as that term implies a finding of no effect. For that reason,
the term "positive" as user1 in the context of Table 2 indicates only that
the study showed an increased risk of lung cancer with increased indoor
radon concentration, or its surrogate, as indicated by a point estimate of
the relative risk greater than 1.0 or a positive correlation.
ROLE OF INDOOR RADON IN LU'lG CANCER ETIOLOGY
By itself, no single published or unpublished study reviewed provides
definitive or even persuasive evidence of an association between indoor
radon exposure and lung cancer. However, taken collectively, they
constitute persuasive evidence of such an association. In Table 2, nearly
all of the case-control studies show an effect of indoor radon exposure on
the measure of lung cancer risk even though in only three studies were the
effects statistically significant. One study that showed no effect was a
part of a general study of cancer in Mesa County, Colorado, and involved
homes contaminated with mill tailings (Ou33). Any effect of indoor radon
exposure would have been masked by the presence of large numbers of
retired uranium miners. Among the case-control studies, it should be
noted that only two (EdBt, Lee87) used actual radon measurements. The
other case-control studies relied on surrogate measures.
Results of the ecological or geographic studies are equivocal with
little more than half of the studies showing a significant positive
27
-------
association between radon daughter- exposure and lung cancer incidence or
death rates. In some of these studies, migration may have played a large
role In exposure misclassIficatlon which diluted the observed effect of
indoor radon. In other studies, the statistical power was just not great
enough to have shown an effect even if It did exist.
As noted previously, all but two of the case-control investigations
used surrogate measures for radon daughter exposure such as area geology,
housing characteristics or background gamma radiation. Such surrogate
measures are not a good substitute for real data. However, it is
interesting to note that studies using independent measures show a similar
association between lung cancer and the exposure surrogate. Surrogate
measures used In these epidemiologic studies are, presumably, independent
factors which may be associated with Indoor radon concentration.
In assessing the strength of the evidence for a true association
between indoor radon and ]ung cancer, it is essential to consider the
potential for systematic bias among the studies. A possible source of
such bias is the tendency for positive ">tudios to be submitted and
accepted for publication, whereas negative or inconclusive results are
often considered uninteresting and never published. Ho other systematic
bias among the case-cor.trol studies is apparent.
Retrospective or case-control epidemiologic studies can 3how an
assocl -.'tion of a particular agent with a specific disease state but do not
necessarily establish causation. However causation can be inferred from
epidemiologic studies on the basis of a set of criteria which historically
has been applied for this purpose (ScS2). These criteria are by no means
intended to be a "checklist". Under soue conditions, such as low levels
28
-------
of association, they may he irrelevant. However, such criteria can
provide a systematic basis for examining causal inference, which is "at
best tentative and still a subjective process" (R086).
1i Temporal Sequence— In order for causation to be
inferred, the temporal sequence of the exposure and the
disease must be reasonable. Basically, that is, the
period of exposure must precede the onset of the disease.
In the case of lung cancer and radon daughter exposure,
the latent period must also be taken into account. The
designs of all of the studies reported in Appendices B
and C are In accordance with thi3 criterion. However,
the relatively long latent period for lung cancer (> 5
yr) and the difficulty of estimating past exposure levels
from current measurement data inject a degree of
uncertainty In this regard.
2. Cons i stenuy — The association must be observed under a
variety of conditions. Repetition in epidemiologic
studies by different researchers using various
populations provides support for this criterion. As with
the temporal seque.ice, the epidemiologic studies summar-
ized in Table 2, for the most part, show the same effect
even though the magnitude of the effect cannot be
compared among the studies. The majority of the studies
showed an association between lung cancer and indoor
29
-------
radon (or its surrogate) even though, In some cases, that
association was not statistically significant at the 95?
confidence level. In general, the studies showing no
effect were "ecological" studies which are subject to
exposure misclassIflcation that tends to bias results
towards the null or have very low statistical power.
Animal studies and epidemiologic studies cf uranium
miners consistently 3how a causal association between
radon daughter exposure and lung cancer.
3. Strength of Association — The greater the magnitude of
the observed effect the more likely it is to be caus-
ative. This Is not always the case as an observed
association may be due to a second factor whic'-. is the
true cause. The observed association would then depend
on the magnitude of the effect of the true causative
agent. In the case of the indoor radon studies, the
strength of the association is variable and is obviously
dependent on the Indoor radon daughter concentrations for
the populations studied.
1i. Biological Gradient — An obvious dose-response effect is
good evidence of causation. Several of the indoor radon
studies showed an exposure-response effect (EdS 4,
Lee37). However, the majority of the studies used
surrogate measures or indoor •-adon daughter exposure so
30
-------
an exposure-response demonstrate on was often not
appropriate. At low levels of exposure, the exposure-
response may be difficult to discern. Therefore, lack of
an exposure-response gradient should not be considered a3
evidence against a causal association.
5. Specificity of Effect — The issue of specificity of
effect is a questionable criterion in the case of
radiation induced cancer. This criterion is met if the
factor (radon daughter exposure) always produces the same
effect (lung cancer) and If the effect disappears when
the factor is removed. Obviously the etiology of cancer
in general and lung cancer in particular is complex and
this criterion cannot be met. Therefore, it Is
reasonable to conclude that specificity would strongly
support a causative inference but that lack of
specificity should not be a reason for concluding that an
exposure i3 not causative in the presence of other
evidence or conformity with the other criteria.
6. Biological Plausibility — The effect should be a logical
consequence of the exposure In te-nc of what is known
about biological processes and the results of collateral
studies. This criterion is often ignored In the asser-
tions of causation with regard to radiation exposure.
However, in the case of Indoor radon, the studios of
31
-------
occupational exposure to radon ^ujhters and animal
research support the Inference of causation. There is
little doubt that radon daughter exposure causes an
Increased risk of ' lung cancer In miners. Ti.e major
question with regard to Indoor radon exposures Is whether
that effect occurs at much lower levels of exposure, I.e.
is t-iere an effective threshold,
ESTIMATED RISK COEFFICIENTS
While ii mcy he concluded from the epidemiologic data available that
indcor radc.i exposure Is a causative factor for lung cancer, the magnitude
of the effect (risk coefficient) Is very much In question. Only two of
the published studies provided sufficient information with which to
estimate risk coefficients and even , in those instances many assumptions
had to be made with regard to occupancy, equilibrium fractions, and other
critical factors (Ed86, Str86). It is Interesting to note that
considering the Inherent problems In making these estimations, the
estimated risk coefficients from these two studies were within the ran^e
of the risk coefficients determined for underground miners.
C0NF0U11DIHG VARIABLES
In almost all of the studies described in Appendix B the
Investigators tried to take into account confounding variables to the
extent possible. In order to be considered a confounding variable the
factor must bo associated with both the disease and the exposure of
interest but not directly caused by the exposure. Confounding variables
-------
must be taken into account in any epidemiologic study. Other variables
which arc known to be associated with the disease but not the exposure
need not necessarily be taken into account as they ;nay be assumed to be
randomly distributed among all groups if there is no bias in selection of
study subjects. Some of the variables potentially confounding in both
published studies and studies in progress are tobacco use, diet (vitamin A
consumption), socioeconomic status, occupational exposures, and urban vs.
rural environment.
Socioeconomic Status
Socioeconomic status is known to be associated with lung cancer, that
is, the risk is greater with low socioeconomic 3tatus (Wy77). This may be
a function of the prevalence of cigarette smoking and other factors
affecting health status. Socioeconomic status may also influence indoor
radon exposure through housing characteristics such as living in an
apartment versus single family dwelling, degree of home insulation, type
of construction materials and method of heating and/or air conditioning.
In addition, the exposure and consequent dose to individuals in lower
socioeconomic classes may be affected by the proportion of time spent
outdoors as well as breathing characteristics associated with manual
labor.
Smoking
Cigarette smoking is the most common variable accounted for in these
indoor radon studies. However, in many studies this was not possible.
This would not be a major problem if smoking were not a true confounder,
that 13 associated with both indoor radon concentration and lung cancer.
33
-------
30
Smoking is unquestionably associated with lung cancer (Wy83). Smoking may
directly influence the dose to the lung from indoor radon exposure due to
many factors including airborne particulate concentration affecting the
unattached fraction and the effect of smoking on the thickness of the
mucus 1ining the upper respiratory tract and to a lesser extent the
clearance rate. It also may be associated with indoor radon concentration
through a common association with socioeconomic status. Lung cancer risk
is inversely associated wil.h socioeconomic status (Wy77). Indoor radon
may also be associated with socioeconomic status through housing
characteristics as described previously. In one situation, smoking, low
socioeconomic status Increases the risk; in the other, indoor radon
concentration, low socioeconomic status may tend to decrease the risk due
to the higher probability of living in an apartment building or an older,
less energy efficient home. None of the published studies referenced had
enough data to contribute to an understanding of the relationship between
smoking risk and radon risk. The question of whether an additive
relationship exists or the effect is multiplicative for non-occupaMonal
exposure to radon daughters is not adequately addressed by the published
stud les.
Diet
None of the published studies took into account the role of diet in
lung cancer risk. Dietary Vitamin A has been suggested to reduce the risk
of lung cancer (S184). As with smoking, diet Is associated with
socioeconomic status and thus may be a confounding variable.
34
-------
Ambient air quality
There has been some suggestion that outdoor air pollution has an
etiologie role In lung cancer (Ve82). Since air quality Is related to the
urban characteristic of an area, living in an urban area may increase the
risk of lung cancer from outdoor air pollution. However, an urban-rural
gradient In Indoor radon concentrations has been Indicated with rural
homes having higher concentrations than urban homes (Co87a). In addition,
urbari dwellers are more likely to live in apartments above the ground
floor also indicating lower average Indoor radon concentrations.
Therefore, as with socioeconomic status, urban air pollution could tend to
weaken any observed effect of radon.
^ Other variables which have been associated with lung cancer may be
considered In future studies. The number of confounding variables
included in an analysis will affect the statistical power, therefore, for
greatest efficiency without sacrificing accuracy, that number should be
kept as low as reasonable.
ECOLOGICAL STUDIES
Many of the published studies are "ecological" studies. That is, the
assignment of exposure status Is based on local conditions as opposed to
individual measures. Geographic or "ecological" studies lack sensitivity
and can lead to erroneous results due to what Is known ^s the "ecological
fallacy." The "ecological fallacy" is a well known problem in
epidemiologic studies which compare community disease rates with mean
community levels of the exposure of interest. The communities may differ
in many ways other than just the factor being studied. A difference in
-------
community disease rates may be attributed to tne exposure of interest
when, in Tact, Lt is actually due to one or more of the other factors.
Another major problem with this type of study when it is applied to a
disease with a long latent period such as lung cancer is the effect of
migration. In such studies, finding no effect of indoor radon on lung
cancer ris'< is not good evidence that indeed there is no real effect.
Studies of this type which do show a statistically significant effect are
likely to underestimate the risk (P08O).
The greater the rate of migration, the more bias that is introduced
into geographic studies. Using as large an area as practical for the
geographic unit reduces the source of bias; however, when too large an
area is used exposure mlsclassificatlon is more likely. Knowledge of
migration rates for the geographic unit can be used to estimate the effect
of this factor.
In general, ecological studies are less likely to show a :;tatlsti-,
cally significant effect of a geographically related exposure when one
truly exists than case-control studies which use an individual measure of
exposure. Therefore, it is not surprising that the case-control studies
consistently show an association between ,'ndoor radon and lung cancer
whereas the ecological studies are equivocal. Case-control studies can be
subject to the same bias towards the null when inadequate measures of
exposure are used, i.e. recent exposures a3 opposed to effective lifetime
exposure.
36
-------
STUDIES IN PROGRESS
The concern with the potential effects of radon daughter exposjre
from indoor radon on the general population and the inadequacy of the
currently available data have emphasized the need for new research
projects. While a review of the studies in progress cannot add to our
present knowledge with regard to radon risk coefficients, it can give us
an estimate of the potential for improving those coefficients on the basis
of epidemiology. Studies in progress are summarized in Appendix D. These
studies are generally funded by state and federal agencies. Depending on
how the studies are designed, the information derived from them has the
potential to enhance the ability to make reasonable risk estimates or to
validate the risk estimates derived from the underground miner data.
Several of the studies in progress have similar characteristics and,
in contrast to previously published studies, the data may be ammenable to
pooling by statistical methodology. This is an advantage derived from
communication among investigators In the field and direction from sone
funding agencies with regard to study design.
The results of most of these studies should be available within the
next four to five years. Until that time, however, risk estimates for
presentation to the general public must be based on the risk coefficients
derived from occupational exposure studies with appropriate adjustment for
the factors which affect the dosimetry.
Since studies of uranium miners have established a causative role of
radon daughter exposure for lung cancer, the major benefit which can be
derived from these studies is a better understanding of the quantitative
37
-------
risk coefficlents. Therefore, It is essential that careful assessment be
made of the exposures of all study subjects.
STATISTICAL POWER OF STUDIES IN PROGRESS
In order to assess the potential of studies In progress with regard
to improving the current risk coefficients, it is necessary to examine
their statistical power (i.e. the probability of finding a statistically
significant effect if, in fact, one exists).
The a priori determination of sensitivity or statistical power of an
epidemiologic study Is an important consideration in study design. Several
of the studies in progress Involve relatively small case and control
numbers (<500) due to the relative rarity of lung cancer and the
constraints of geographic area. It is useful to know in advance if these
studies have the potential to show a statistically significant effect of
indoor radon exposure when one truly exists. This a priori determination
of statistical power should not be confuted with the determination of
statistical significance of a completed study. Studies which do not have
great statistical power (C0.80) may still show a statistically significant
effect.
All but two of the investigations in progress are case-control
studies. The simplest form of this type of epidemiologic study compares
the fraction of cases exposed to the factor of interest to the fraction c"
controls exposed using a d'chotomous exposure classification. The statis-
tical power calculation requires estimates of the relative risk and the
fraction of the control population exposed. For indoor radon the relative
risk can be estimated from the relative risk coefficients de- ived from the
-------
underground miner studies and the mean radon daughter exposures for the
"exposed" and "unexposed" classification. The fraction of controls ex-
posed can be- estimated on the basis of indoor radon concentration distri-
butions measured on a random basis.
In order to make a generic estimate of statistical power for case-
control studies, the following assumptions are made and are illustrated in
Table 3:
a. The distribution of residential radon concentrations in
the populations under study is represented by the random
measurements made by Dr. B. Cohen, University of
Pittsburgh ^adon Project, as reported by J. Stolwijk
(Sto87), and shown in the first two columns of Table 3.
b. The designation of "exposed" vs "unexposed" is based on
either a ? pCi 2.-1 (columns 3 and 4) or 4 pCi J,-1
(columns 5 and 6) cutoff. Since it is unlikely that many
individuals remain in the same residence for a lifetime,
a weighted mean radon concentration was calculated for
the "exposed" population based on 15 years of residence
in the high radon concentration residence and 35 years of
residence at the average radon concentration for the
"unexposed" population.
c. The excess lifetime relative risk from exposure to radon
daughters is \% to per WLM lifetime exposure.
The weighted mean lifetime residential radon concentration for
"exposed" individuals with a 2 pCi SI-1 cutoff is 3.1* pCi The
weighted mean concentration for the measurements greater than 2 pCi is
9.1 pCi t"1. Therefore, the weighted mean lifetime concentration for the
exposed population, assuming 15 years residence at the mean concentration
39
-------
Table 3. Distribution of Indoor Radon Concentrations In U.S.
' Population Adjusted for 1 pCi I 1 and 2 pCi I 1
Exposure Cutoff Point3
Original
Data
2 pci r1
Cutoff
1 pCi
r1
Cutoff
Frac.
Pop.
Rn Cone.
pci r1
Frac.
Pop.
Rn C.onc.
pci r1
Frac.
Pop.
Rn Cone.
pCi i_1
0.05
0.1
0.05
0.1
0.05
0.1
0.10
0.3
0.10
0.3
0.10
0.3
0.10
0.5
0.1-0
0.5
0.10
0.5
0.20
0.8
0.20
0.8
0.20
0.8
0.20
1.1
0.20
1.1
0.20
1.1
0.20
3.2
0.10
1.8*
0.20
3.2
0.10
6.2
0.10
3.2
0.05
3.6*
0.10
6.2
0.05
6.2
0.03
15.0
0.03
15.0
0.03
15.0
0.0125
j2.0
0.0125
32.0
0.0125
32.0
0.005
18.0
0.005
18.C
0.005
18.0
0.0025
100.0
0.0025
100.0
0.0025
100.0
Mean =
» 3.
1
Mean unexp
. = 0.9
Mean unexp
. = 1 .5
Wtd. mean exp
. = 3.1**
Wtd. mean
exp
. = 5.8**
Frac. exp
. = 0.25
Fract.
exp
. = 0.10
* Partitioned on the basis of a log-normal plot of the data.
** Assuming 15 years exposure at concentrations greater than the cutoff and
35 years exposure at mean concentrations for unexposed.
40
-------
greater than the 2 pCi cutoff and 35 years residence at the mean
concentration lower than the 2 pCi J."1 cutoff is as follows:
[ (15y) (9.1 pCi e~1) + (35) (0.9 pCi J."1) ] / [50y] = 3-1 pCi SI'1
For a 1 pCi 2,"^ cutoff the mean lifetime residential radon concentration
is 5.8 pCi The mean radon concentrations for "unexposed" individuals
with a 2 pCi £~1 cutoff is 0.9 pCi 8.-1, with a 4 pC i 2,"1 cutoff, 1.5 pCi
. The percent of population exposed with a 2 pCi I ^ cutoff is 25J,
and for a pCi H_1 cutoff, 10?.
The estimated relative risks with a specified risk coefficient and
exposure cutoff points are shown in Table 4.
The number of casss required (assuming an equal number of controls)
for a given level of statistical significance (a) and power (1-fc) can be
calculated from- the following equation (Sc7^).
N = (7.a V 2u(1-u) + Zg \/ f(l-f) + P3q3)2/(f-p3);?
where:
f = fraction exposed among controls
R = relative risk
P3 = probability of exposure among cases
Q3 = 1~P3
u = 0.5f 1+ R/(1+f(R-1))
P3 = fR/(1+f(R-1))
a = 0.05 (one sided)
8 = 0.10 or 0.20
Z = standard normal deviate = 1.645 for ci = 0.05 (one-sided)
a
Z = standard normal deviate = 1.28 or 0.82'4 for 6 = 0.10
and 6 = 0.20
41
-------
Table 1. Estimated Relative Risk for Various Cutoff
Concentrations and Risk Coefficients
Relative Risk
Coefficient
(?/WLM)
Exposure Cutoff
(pCi r1)
Estimated
Relative Risk
1
2
1.21
2
2
1.1(8
3
2
1.72
H
2
1.96
1
4
l.i)?
2
if
1.81
3
J)
2.26
i)
'1
2.68
Sample Calculation:
Lifetime excess exposure at 2 pC i 5."^ cutoff:
(3.pci a"1 - o.9 pci r1) = 2.5 pci r'
(2.5 pCi r1)(0.5)(87*0 hr/a) (0.75)(50a)
- .
COO pCi l~ -WLM170 hr/a)
Relative risk:
R = 1 + (0.01)(2D) = 1 .24
WL
Equilibrium factor = — = 0.5 (See Appendix A)
rPCl T I
1 100 J
Fraction of time spent indoor3 = 0.75
42
-------
The calculated numbers of cases required for a given statistical
power are shown in Table 5.
These calculations represent'the simplest, ideal condition with well
defined exposure and no confounding variables. This Is not generally the
case In real epidemiologic investigations. Smoking or other factors
potentially associated with both lung cancer and indoor radon concen-
trations would reduce the theoretical power of the study as would uncer-
tainty In exposure. In addition, the a3sunptions regarding mean exposure
for "exposed" and "unexposed" are arbitrary and may not be a true re-
flection of exposure distribution for Individual studies. Therefore, the
estimated numbers of cases required represent a lower limit-..
Several of the proposed studies will use more than one control per
case, Increasing the statistical power. Most of the studies planned or in
progress will have more than 300 cases. Therefore, they have a fair chance
of showing an association between lung cancer and indoor radon concentra-
tions if the random measurenents taken by Cohen and used in this analysts
are representative of the areas in which the studies take place and If the
current relative risk coefficients (USEPAS6) are realistic. For areas
where exposures are higher, the statistical power could be greater.
The statistical power of each of the studios in progress depends on
the conditions in the geographic areas, migration patterns and the extent
of stratification into appropriate subgroups and must be calculated on an
individual basis as several of the researchers have done. The generic
calculation gives a crude indication of the potential of these studies i'n
general to produce statistically significant results.
A3
-------
Table 5. Calculated Number of Cases Required for p - 0.05 Level
of Statistical Significance and Statistical Power 0.80
or 0.90
Relative
Exposure Risk
Cutoff Coefficient R f p, N N
(pCi r1) (%/WLM) (1-&)=0.90 f1-B>=0.80
2
1
1,2>i
0.25
0.29
1900
i100
2
2
1.18
0.25
0.33
550
100
2
3
1 .72
0.25
0. 36
280
200
2
1
1.96
0.25
0.10
180
130
980
290
150
100
R = estimated relative risk
f = fraction of controls exposed
Pj = fraction of cases exposed
N = nurber of cases required assuming equal numbers of controls
1
1
1 .12
0.10
0.14
1100
1
2
1.81
0.10
0.17
110
1
3
2.26
0. 10
0.23
2i a
1
1
2.68
0.10
0.23
110
A A
-------
Several of the studies In progress have sufficient 3i.nl laritie3 to
allow for eventual pooling of the data which may result In greater
statistical power and t.he potential for some quantitative determination of
exposure-response. The studies already published do not provide
sufficient basis for such an analysis but do suggest a positive
association between indoor radon and lung cancer. Therefore, the main
contribution of the studies In progress will be to quantify the exposure-
response relationship if pooling of the data from several of the studies
can be accomplished. Most do not have enough cases to allow for such
refined analyses on an individual basis.
RISK MODELS
The data from studies of lung cancer in underground miners have been
used to develop various risk models, /it the present time these models are
the basis for the projected risk to the general population from exposure
to radon daughters. Ps noted in the section on dosimetry, the risk
coefficients derived from these models may need adjustment to account for
differences between miners and members of the general population as well
as differences between the mine atmosphere and residential envlromnents.
However, the basic forms of the models are applicable to either
situation. The risk models are summarized in thi3 report simply to add to
its usefulness. No analysis or discussion of the relative merits or
deficiencies of each type of model Is Included.
45
-------
Three basic types of models have been used to predict the risk of
radon induced lung cancer in the presence of other risk factors: absolute,
additive, and multiplicative or proportionate hazards models.
Absolute risk model
A (t) ° A (t) + 3W
rx x
where:
X (t) = lung cancer risk at age t with specific risk factors x
and radon, r.
A (t) = age specific lung cancer risk with no excess radon
daughter exposure
3 = risk coefficient for radon daughter exposure
W = cumulative radon daughter exposure
The absolute risk model assumes the risk of lung cancer from radon
daughter exposure is independent of both the baseline cancer risk and all
other risk f^cto^s.
Additive risk model
A (t) =¦ A (t) (1 ~ BW «• YX)
r x o
where:
A (t) <= baseline age specific lung cancer risk with
no excess radon daughter exposure or exposure to other
risk factors
46
-------
X = risk factors other than radon daughter exposure 3uch as
smoking
1 = risk coefficients for other risk factors
The additive risk model assumes that the radon daughter induced lung
cancer risk i« dependent on the baseline or age dependent lung cancer risk
but independent of all other risk factors.
Proportionate hazards model
The proportionate hazards model predicts the risk of indoor radon as
a function of the baseline risk and all other lung cancer risk factors.
X (t) - » (t) e (#w * W
rx o
This equation represents the general form of the proportionate hazards
model. The exponential or log-linear form is commonly used in epidemi-
ology. However, the exponential function can be replaced by any other
function such that the risk function is equal to 1 when ail exposures or
all risk coefficients are zero. The linear multiplicative model is the
form of the proportlonate hazards model most commonly used in radon risk
estimatIon.
\ (t) = X (t) (1 + BW) (1 + TfX)
rx o
The proportionate hazards models, including the multiplicative risk
model, take Into account interaction between radon daughter exposure and
other risk factors.
47
-------
The most commonly used models Tor projecting the It . of Indoor
radon on public health are applications of these basic form3. Some
examples are as follows.
BEIR III (BEIRSO)
The BEIR III risk estimates are based on an age dependent absolute
risk model. That is, the risk coefficient is a function of the age at
diagnosis for lung cancer: 10 per million person-years-WLM for ages 35-
U9, SO per million person-years-WLM for ages 50-65, and 50 per million
person-years-WLM for ages over 65. The term person-years refers to tne
population and period at risk following exposure. No interaction with
t
smoking or other risk factors is assumed for these risk coefficients.
BEIR III Is equivocal, stating that if the lung cancer risk after
radiation Is consistent with a multiplicative effect of radiation on
smoking induced lung cancer, then the excess risk for smokers would be
increased by about 50* and decreased by a factor of six for nonsmokers.
BEIR III states that a purely multiplicative relationship between lung
cancer risks for smoking and radiation is unlikely.
NCRP 78 (NCRPPJO
The b^.sic risk moUel used in NCRP 78 is a modification of the
absolute risk nodel
A(t|t0) = RC P(t|tQ) e~k(t_to)
where:
48
-------
A(t|tQ) - attributable annual tumor rate at ago t from one WLM
exposure at time t
RC = absolute risk coefficient
P(t|tQ) = probability that an individual who survives to age tQ
will be alive at age t.
k = rate of risk expression due to cell death,
repair or other mechanism
An absolute risk coefficient of 10 per million person-years per WLM
was used in the NCRP 73 risk calculations.
This model does not take into account any Interaction with other risk
factors such as smoking, but does correct for survival and latent period.
When used to calculate risk from chronic exposure the model takes into
account a five yea>" latent period. In contrast to the ICRP50, BEIR III
and EPA models, the NCRP78 model accounts for reduction in risk over time,
due to cell death or repair, with a 20 year half-time. In their analysis
of the U.S. miner data, Hornung and Meir.hardt (Ho87) found a decrease in
risk per WLM with time outside the mine. Thomas and McNeill (Th85)
reported an initial increase in relative ri3k coefficient vs. time since
first exposure, followed by a decrease.
EPA (EPA35)
X (t) = X (t) (1 + BW)
rx x
49
-------
Thi3 Is a simple linear multiplicative risk model. The relative risk
coefficient 13 assumed to be between and per WLM.
ICRP 50 (ICRP37)
ICRP 50 gives a proportionate hazards moael (linear multiplicative)
to express risk of lung cancer from chronic exposure to radon daughters.
The multiplicative risk model can be represented as follows:
ASr(t) = A (t) (l + S(t)} (1 + r E (t-T))
where:
ASr (t) = the risk of lung cancer with radon exposure and
smoking status at age t
A (t) = the baseline lung cancer risk at age t
S(t) = the risk factor for smoking at age t
r » the mean relative ^isk coefficient for exposure tc
radon daughters
E = the mean annual exposure to radon daughters
T = latent period for expression of lung cancer following
exposure
50
-------
The estimated relative risk coefficients for radon daughter exposure are
0.019/WLM for exposure to radon daughters up to age 20, and 0.0064/WLM for
exposures at ages greater than 20. These coefficients were derived from
i
uranium miner risk coefficients adjusted for various factors such as
breathing rate and the presence of other carcinogenic factors in the mine
environment.
BEIR rJ
The BEIR IV (NAS 80) model Is a proportionate hazards
(multlplicdtive) model with an adjustment for age at risk and time since
exposure (TSE).
A (t) = A (t)[1 + 0.025Y(t)(W, ~ 0.5 W,)]
rx x 12
(The symbols have been changd to be consistent with those used for
previously defined models in this report.)
A (t) = lung cancer mortality rate from all causes
rx
A^t) = ago specific background lung cancer mortality
i ate (all causes except radon)
Y(t) = age specific adjustment to radon risk coefficient
T(t) = 1.2 for t <55 years
Y(t) = 1.0 for t = 55 - 64 years
Y(t) = O.'l f<£xr t >64 years
-------
W1 = cumulative exposure Incurred between 5 and 15 years
before age t
= cumulative exposure incurred 15 years or more
before a?e t
This model weights exposures received more than 15 years in the past
by a factor of 0.5 to account for the decrease in risk per WLM observed in
miners with time out of the mine or time since first exposure (Ho87,
Th85). The authors of the BEIR IV repjrt avoided speculation regarding
the biological reasons for the decrease citing a need for further clinical
and laboratory investigations.
The age specific risk coefficient takes into account the observed
/
decrease in excess relative rij-.k with age at which the risk is evaluated.
Table 6 shows a comparison cf estimated lifetime risk of lung cancer
from indoor radon exposure at the presumed U.S. average (1 pCi t, ^ ) and
tne EPA guideline (4 pCi S."1) calculated on the basis of each of t-he
models. The risk calculations are shown in Appendix E.
Several models have been developed on the basis of the U.S. miner
data. Whittemore and McMillan (VJh83) concluded that a multiplicative
linear model fit the data better than an additive model and that combining
the additive and multiplicative models did not improve the fit of the
data. Hornung and Meinhardt (HoSY) used an exponential proportionate
hazards model and a power function model. The power function model
appeared to provide the b^st fit to the data.
In both cases only the U.S. miner data was used. A large fraction of
this cohort of miners is still lining. There is some suggestion that with
-------
Table 6. Comparison of Estimated Lifetime Excess Lung
Cancer Kisk Due to Indoor Radon Exposure
Mean
Indoor
Radon
Concentration
Lifetime
Exceao
Risk (?)
Smoking
Status
MOD EI.
BEIH III
MCRP
ICRP
EPA
BEIR IV
1 pci r1
Never Smoked
1
0.2
0.1
0.1-0.3
0.1
Ex-smoker
1
0.2
0.2
0.2-0.5
0.3
Chronic Smoker
(1 pack/day)
1
0.2
0.4
0.H-1
1
t pCi f1
Never Smoked
l)
0.7
0.3
0.3-1
0.5
Ex-smoker
4
0.7
0.7
0.7-2
1
Chronic Smoker
H
0.7
2
2-5
5
53
-------
increased follow-up, the data will show closer correspondence to an addi-
tive risk model (Ar87b).
The principal cause of lung cancer in the general population is
cigarette smoking and it Is generally the major factor taken into consid-
eration when the risks due to indoor radon are projected. Tobacco smoke
is traditionally treated a3 a single variable. However, tobacco smoke
contains a variety of known or suspected carcinogens includLng Po-210
(PHS82^. It is likely that so~e components have a multiplicative inter-
action with radon daughter exposure and others such as the Po-210 an
additive relationship. Therefore, it may be assumed that the interaction
of tobacco smoke with radon daughters is a complex function, neither
strictly additive nor strictly multiplicative. A model that is
intermediate between additive and multiplicative might better represent
the true condition.
Whittemore and McMillan («h83) compared a mixture model with the
linear multiplicative risk model and found that it was not a significant
improvement when the U.S. miner data were examined. However, this may
have been a result of too short a follow-up period In this cohort. Some
other miner studies with longer follow-up show a risk relationship best
represented by the additive model.
The indoor radon epidemiologic data published to date is inadequate
to test the fit of these models to non-occupational exposures. Therefore,
at this time, the risk mcdol which best fits the miner data nay have to be
used to project risk of non-occupational exposure.
-------
CONCLUSIONS
A review of the dosimetric models and the existing epidemiologic
studies with regard to lung cancer and indoor radon exposure leads to the
following conclusions:
Dosimetric analyses that take Into account differences between
underground miners and members of the general public, in terms of
lung morphometry, breathing patterns and environmental aerosol
characteristics, indicate that the dose per unit exposure to radon
daughters may be marginally higher for nonoccupational exposures than
for miners. Therefore, there is no apparent rationale for redefining
the Working Level (WL) for indoor radon exposure simply on the basis
of the reduced volume of air inhaled per unLt time.
The uncertainty in applying risk estimates derived from studies of
underground miners to the general public may be reduced by
determining the fraction of the time persons inhale through the nose
vs. the mouth, the physical characteristics of residential aerosols
which would influence the unattached fraction, and the relationship
between lung cancer risk and age at exposure.
The results of epidemiologic studies dealing with indoor radon
provide persuasive evidence of an association between Jung cancer and
residential radon exposure. However, these data are not sufficient
to allow derivation of quantitative risk estimates specific for
55
-------
indoor radon or validation of risk estimates derived from underground
miner data.
Epidemiologic research in progress may provide a basL3 for revision
or validation of current risk models and coefficients. This i3 feasible
only if the individual investigations employ designs which allow for
pooling of data to obtain greater statistical power than that possible for
any single study.
56
-------
REFERENCES
Ar87a Archer, V. E. Association of lung cancer mortality with
precatnf plan granite. Arch. Env. Health '42:87 — 91 ; 1 987.
Ar87b Archer, V. E. University of Utah, Personal Communication, 1987.
Ax79 Axel3on, 0.; Edling, C.; Kllng, H. Lung cancer and residency: A
case-referrent study on the possible impact of exposure to radon
and its daughters in dwellings. Scand. J. Work Environ. Health
5:10-15; 1979.
Be82 Bean, .J. A.; Isacson, D.; Hohns, R. M. A.; Kohler, J. Drinking
water and cancer incidences in Iowa: II. Radioactivity in
drinking water. Am. J. Epidemiol. 116:924-932; 1982.
Ca87 Castren, 0. Dealing with t adon in dwellings. In: Proceedings of
Second International Specialty Conference on Indoor Radon. New
Jersey. Air Pollution Control Association; 1987.
C0873 Cohen, B. L. University of Pittsburgh. Paper presented at the
32nd Annual Health Physics Society Meeting, Salt Lake City, UT,
July 9, 1937.
C087 Cohen, B. S. Deposition of ultrafine particles in the human -
tracheobronchlal tree. In: "Radon and its decay products."
Hopke, P., ed.; Proceedings of the American Chemical Society;
1987.
Da86 Damber, L. Lung cancer in males: An epidemiologic study in
northern Sweden with special regard to smoking and occupation.
Urnea University Medical Dissertations. New Series Mo. 167:113"
125; 1986.
D085 Dousset, M. ; Jammet, H. Comparison de la nortalite par cancer
dans le Limousin et le Poitou-Charentes. Radioprotection 20:61-
67; 1985.
Ed82 Edling, C.; Comba, P.; Axelson, 0.; Flodin, U. Effects oT low-
dose rate radiation — A correlation study. Scand. J. Work
Environ. Health 8, Suppl. 1:59-6ll; 1982.
Ed84 Edling, C.; Kling, H. ; Axelson, 0. Radon ir. homes — a possible
cause of lung cancer. Scand. J. Work Environ. Health 10:25~3lt;
1 98't.
F181 Fleischer, R. L. A possible association between lung cancer- and
phosphate mining and processing. Health Phys. 41:171-175; 1931.
F186 Fleischer, R. L. A possible association between lung cancer and
a geological outcrop. Health Phys. 50:823-827; 1986.
57
-------
Fo85 Forastlere, F.; Valesini, S.; Arco, M. ; Magllola, M. E.;
Miehelozzi, P.; Tasco, C. Lung cancer and natural radiation in
an Italian province. Sci. Total Environ. ^5:519-526; 1985.
Ga72 Gastlnoau, R. M. ; Walsh, P. J.; Underwood, N. Thickness of
bronchial epithelium with relation to exposure to radon. Health
Phys. 23:857; 1972.
Ga85 Gamage, R. B. ; Kaye, S. V. Indoor Air and Human Health. Oak
Ridge, .TN:Lewis Publishing; 1985.
Go19 Cormley, P. G.; Kennedy, M. Diffusion from a stream flowing
through a cylindrical tube. Proc. Irish Acad. Section A. 52:163;
1919.
rla72 Harley, N. !i. ; Pasternack, B. S. Environmental radon daughter
alpha dj.se factors in a five lobed human lung. Health Phys.
"!2 :-"71 -732 ; 1 972.'
Ha81 Irnrley, N. H. Comparing Radon Daughter Dose: Environmental
V".-su3 Underg—jnd Exposure. Radiation Prot. Dos. 7:371-375;
19814.
Ha86 Harley, N. H.; Cohen, B. S. Updating radon daughter bronchial
dosimetry. In: "p^1on ^,iq it3 decay products." Hopke, P., ed.,
Proceed! gs or the American Chemical Society, pp.119-129; 1986.
He83 Hess, C. T. ; Weifenbach, C. V.; Norton, S. A. Environmental
radon and cancer correlations in Maine. Health Phys. 15:339-313;
1983.
Hi87 Hieber, L. ; Pansel, G.; Roos, H. ; l-'enn, S. ; Fronke, E. ; Kellercr,
A. M. Absence of a dose-rate effect in the transformation of C3
lOlVj cells by o-rays. Institul. fur Meaizinisc'ne Strahlcnkurule;
Universitat Wiir/.burg, Versbacher Strasse 5, D-8700 Wiirzburg, West
Germany; 1987.
Ho79 Hofmann, W. ; Steinhausler, F. ; Pohl, E. Dose calculations for
the respiratory track from inhaled natural radioactive nuclides
as a function of age-I. Health Phys. 37:517-532; 1979.
Ho85 Hofmann, W. ; Katz, R. ; Zhang, C. Lung cancer Incidence in a
Chinese high background area — epidemiological results and
theoretical interpretation. Sci. Total Environ. 15:5 27-5 3 -'1;
1985.
Ho86 Hofmann, W.; Katz, R.; Zhang, C. Lung cancer risk at low doses
of alpha particlos. Health Phys. 51:157-168; 1986.
Ho87 Hornung, R. V/.; Mcinhardt, T. J. Quantitative risk assessment of
lung cancer in U.S. uranium miners. Health Phys. 52:117-130;
1 937.
58
-------
ICRP77 International Conmission on Radiological Protection. Recommen-
dations of the ICRP; Report No. 25; Oxford, England:Pergamon
Press; 1977.
ICRP81 International Commission on Radiological Protection. Limits for-
Inhalation of Radon Daughters by Workers; Report No. 32; Oxford,
England: Pergainon Press, 1931.
ICRP87 International Commission on Radiological Protection. Lung cancer
risk from indoor exposures to radon daughters. New York:Pergamon
Pres3; 1987.
In75 , Ingham, D. B. Diffusion of aerosols from a 3tream flowing
through a cylindrical tube. Aerosol Sci. 6:125; 1975.
Ja80 Jacobi, W.; Eisfeld, K. Dose to tissues and effective dose
equivalent by inhalation of radon 222 and . adon 220 and their
short lived daughters. GSE-Report-S-626 ; Munich, Neuherberg,
West Germany; 1980.
Ja84 James, A. C. Dosimetric Assessment of Risk From Exposure to
Radioactivity in Mine Air. Occupational Radiation Safety in
Mining; Stocker, H., ed., Canadian Nuclear Association, ill
Elizabeth Street, Toronto, Ontario, Canada, pp. 415-^2*7; 198^.
Ja86 James, A. C. A reconsideration of cells at risk and other key
factors in radon daughter dosimetry. In: "Radon and its decay
products." Hopke, P., ed , Proceedings of the American Chemical
f 13ty, pp. 'IGO-^18; 1986.
Ja87 ,' A. C. Lung dosimetry for radon and thoron daughters: A
re1, .ew and reassessment with emphasis on domestic exposure. In:
Nazaroff, W. W. and Nero, A. V., eds.; "Radon and Its Progeny in
Indoor Air;" Hew York:Wile/ Interscience; 1937.
Kn83 Knutson, E. 0.; George, A. C. Koh, B. R. Radon daughter plateau
II, Prediction model. Health Phys. '15 ; '4 4 5 - ^ 5 2 ; 1983.
La85 Lanctot, E. M. Ra'ion in the dOMesu'c environment and its
relationship to cancer: An epidemiologic study. Master's Thesis,
State University of t;ew York at Stony Brook. Maine Geological
Survey, Dept. of Conservation Publication 85-88; 1985.
La82 Lanes, S. F. Lung cancer and environmental radon exposures: A
case-control -t.fL.dy. Doctoral Disser tation, University of
Pittsburgh; 1982.
Le87 Lees, R. E. M.; Steele, R.; Roberts, J. H. A case-control study
of lung cancer relative to domestic radon exposure. Int. J.
Epid. 16:7-12; 1937.
59
-------
Le83 Letourneau, E. G.; Mao, Y.; McGreggor, R. G.; Semenolw, Ft. ;
Smith, M. H.; Wlgle, D. T. Lung cancer mortality and indoor
radon concentrations in 18 Canadian cities. In: Epidemiology
Applied to Health Physics. Proceedings of the Sixteenth Midyear
Topical Meeting of the Health Physics Society, Albuquerque, NM;
1983.
L179 Lloyd, E. D. ; C-emmell, M. A.; Henning, C. B. ; Gemmell, D. S. ;
Zabransky, J. B. Transformation of mammalian cells by alpha
particles. Int. J. Radiat. Biol. 3&. :^67—Jl78; 1979.
Ma72 Martin, D.; Jacobi, W. Diffusion deposition of small sized
particles In the bronchial tree. Health Phys. 23:23-29; 1972.
NAS80 National Academy of Sciences, Committee on the Biological Effects
of Ionizing Radiations. The effects on populations of exposure
to low levels of ionizing radiation. Washington DC:Natlonal
Academy Press; 1980.
NAS88 National Academy of Sciences, Committee on the Biological Effects
of Ionizing Radiations. Health risks of radon and other
Internally deposited alpha emitters. Washington DC:Natlonal
Academy Press; 1988.
NCRP84 National Council on Radiation Protection and Measurements. NCRP
Report No. 78: Evaluation of occupational and environmental
exposures to radon and radon daughters in the United States.
Bet'nesda, MD: NCRP ; 1984.
Ne8r: Nero, A. V.; Schweher M. B. ; Nazaroff, W. W. ; Revzan, X. L.
Distribution of Airborn Radon-222 Concentrations in U. S. Homes,
Science 23!l: 992-997; 1 986.
OBr86 O'Brien, T. I?.; Decoufle, P.; Rhodes, P. R. Selected cancer
mortality in Florida counties where radium water levels exceed
the EPA limit. Report to Florida Department of Health and
Rehabilitative Services; 1986.
OECD33 Organization for Economic Cooperation and Development. Dosimetry
aspects of exposure to radon and thoron daughter products.
Report by a Group of Experts; OECD, 2 rue Andre Pascal, 75775
Paris, Cedex 16, France; 1983.
Ou83 Ouimette, D.; Ferguson, S. W. ; Zoglo, D.; Murphey, S.; Alley, S.;
Bohler, S. An epidemiologic study of selected malignant
neoplasms in Mesa County, Colorado, 1970-1°79. Final Report,
Denver C0:Colorado Department of Health; 1953.
Pa85 Phalen, R. F. ; Oldham, M. J.; Beaucage, C. B. ; Crocker, T. T. ;
Mortensen, J. D. Postnatal enlargement of human tracheobronchial
airways and Implication for particle deposition. flnat. Rec.
212:368; 1985.
60
-------
Pe84 Pershagen, G.; Damber, L.; Falk, R. Exposure to radon In
dwellings and lung cancer: A pilot study. In: Proceedings of the
Third International Conference on Indoor Air Quality and Climate,
Vol. 2. Stockholm, Sweeden, 73-78; 1984.
P08O Pollsar, L. The effect of migration on comparison of disease
rates in geographic studies in the United States. Am. J.
Epidemiol. 111:175-182; 1980.
Ro83 Robertson, J. B.; Koehler, A.; Georgo, J.: Little, J. B.
Oncogenic transformation of mouse BALR/3T3 cells by Pu-238 alpha
particles. Radiat. Res. 96:261-274; 1933.
Ro86 Rothman,. K. J. Modern Epidemiology. Boston- Little, Brown and
Company; 1986.
Sc74 Schlesselman, J. J. Sample size requirements in cohort and case-
control studies of disease. Am. J. Epidemiol. 99:331 —384; 1974.
ScS2 Schlesselman, J. J. Case-control studies: Design, conduct,
analysis. Second printing. New York:Oxford University Press;
1982.
S184 Siegler, R. G.; Mason T. J.; Stemhagen, A.; Hoover, R.;
Sc'noenberg, J. B. ; Gridley, G.; Virgo, P. W. ; Altman, R.;
Fraumeni, J. F. Jr. Dietary carotene and vitamin A and risk of
lung cancer among white men in New Jersey. J.N.C.I. 73:1^29—
1435; 1984. .
Si83 Simpson, S. G.; Comstock, G. W. Lung cancer and housing
characteristics. Arch. Environ. Health 3S:24S-251; 1983.
Str36 Stranden, E. Radon in Norwegian dwellings and the feasibility of
epidemiologic studies. Radiat. Environ. 3iophys. 25:37-42; 1986.
Str87 Stranden, E. Radon-222 in Norwegian dwellings. In: Hopke, P. K.,
ed. Radon and Its Decay Products: Occurrence, properties, and
health effects. ACS Symposium Series 331:70-893; 1957.
Sto87 Stolwijk, J. A. J. Yale University, Testimony before
Subcommittee on Environmental Protection, Senate Committee on
Environment and Public Works, April 24, 1987.
Sv87 Svensson, C.; Eklung, G. ; Pershagen, G. Indoor exposures to
radon from the ground and bronchial cancer in ucrriiin. Intl. Arch.
Occup. Environ. Health 59:123-131; 1937.
Th85 Thomas, 0. C. ; McNeill, KK. G. ; Dougherty, C. Estimates of
lifetime lung cancer risks resulting from Rri progeny exposure.
Health Phys. 49:825-846; 1985.
USEPA78 U.S. Environmental Protection Agency. Indoor radon exposure dun
to Radium-226 in Florida fitiosp'nate Lands. EPA 520/4-78-01 3; 1978.
61
-------
USEPA36 U.S. Environmental Protection Agency. A Citizens Guide to Radon.
OPA-86-OOH; 1986.
USPHS82 U.S. Public Health Service. The health consequences of smoking.
Cancer: A report of the Surgeon General. Rockville MD:USHPS;
1982.
Ve82 Vena, J. E. Air pollution as a risk factor in lung cancer. Am.
J. Epidemiol. 116:42-56; 1962.
Wa86 Walter, S. D. ; Meigs, J. W. ; Hestor., J. F. The relationship of
cancer incidence to terrestrial radiation and population density
in Connecticut 1935-1974. Am. J. Epidemiol. 123:1-14; 1986.
We63 Weibel, E. R. Morphometry of the Human Lung. Orlando,
FL:Academic Press; 1963.
We83 Weinsteln, I. B. ; Horowitz, A.; Jeffery A. and Ivanovic, V.
Cellular Events in Multistage Carcinogenesis, In Cenes and
Proteins in Oncogeneses (I. B. Weisstein and H. J. Vogel, eds).
Acadmic Press: 99-110; 1983.
Wh83 Whittemore, A. S. ; McMillan, A. Lung cancer mortality among U.S.
' uranium miners: A reappraisal. J. Natl. Cancer Inst. 71:489—^99;
1983.
Wy83 Wynder, E. L. ; Goodman, M. T. Smoking and lung cancer: Some
unresolved issues. Epidemiologic Reviews 5:177-207; 1933.
Wy77 Wynder, E. L. ; Stellman, S. D. Comparative- epidemiology of
tobacco-related cancers. Cancer Res. 37:4608-4622; 1977.
Ye80 Yeh, H. C.; Schum, G. M. Models of human lung airways and their
application to inhaled particle deposition. Bull. Math. 3io.
12:461; 1930.
62
-------
APPENDIX A
SPECIAL QUANTITIES AND UNITS
The concentration of radon gas in air 13 expressed in terms of
radioactivity per unit volume. The most common quantities are pCi I ^ and
Bq m J where,
1 pCi f1 - 37 Bqm"3
When radon gas decays it initiates the appearance and eventual decay of
the short-lived daughters, 21®Po, 21l4Bi, 21Vb and 21i|Po. The health risk is
directly associated with radon daughters that remain suspended in air and
eventually Inhaled.
Normally, the concentration of daughters i3 not expressed in ' erms of
radioactivity per volume of air. This departure Is based on the premise- that
once daughters are deposited in the respiratory system they remain there until
the decay sequence of short-lived daughters is completed. The concept of
potential alpha energy was developed to accommodate the principle. It is a
measure of the total kinetic energy released by alpha particles for any
mixture of radon daughters in air that proceeds through the entire serial
pin
decay sequence down to Pb. Potential alpha energy is expressed in units of
J m~3 or MeV 2,-1. A working level, WL, is defined as
WL = 1.3 x 105 MeV 2,"1 = 2.1 x 10"5 J m"3
These seemingly peculiar numbers are obtained from the situation where
all of the short-lived daughters are in radioactive equilibrium at
100 pci r1.
-------
The concept of potential alpha energy does not require that the daughters
C .1
have equal concentrations. Any mixture that yields 1.3 x 10^ MeV I la equal
to 1 WL. The concentration of radon gas does not provide a_ priori information
on the concentration of potential alpha energy and vice versa. Independent
measurements are required. However, it is frequently convenient to generalize
or estimate one from the other. For this purpose the equilibrium factor (EF)
is defined as the ratio of the potential alpha energy actually present to the
value that would be present if the daughters were in radioactive equilibrium
with the existing concentration of radon gas. For example
Rn(pCl a"1)
100
The potential alpha energy concentration of any mixture of radon
daughters can also be expressed in terms of the equilibrium equivalent
concentration (EEC). The EEC of a mixture of radon daughters in air is that
777
activity concentration of Rn in radioactive equilibrium with its short-
lived daughters which has the same potential alpha energy concentration as the
actual fixture.
The EEC can be expressed in terms of the equilibrium factor
EEC = EF x CRn
where Cpn Is the activity concentration of radon gas.
The amount of alpha energy deposited in the lung is related to the
potential alpha energy concentration and the duration of time that a person 13
-------
exposed to this concentration. This can be expressed as J m~3 h or
Bq m 3 h when EEC is used to express the concentration of radon daughters.
A special quantity, the working level month (WLM), was defined for
expressing occupational exoosure to radon daughters. A working level month is
equivalent to exposure at 1 working level for 170 hours (NCRPS4). Although
this was based on working schedules for miners it applies without modification
to environmental or Indoor exposures:
WLM = WI* X hP
170
T'.ie WLM should not be confused with a calendar month since exposure to 1 WL
for 170 h is the same as 10 WL for 17 h or 0.2 WL for 850 h. Relationships to
other quantities are as follows:
1 WLM = 170 WL h
1 WLM =3.5x10"3Jm3h
1 WLM = 6.3 x 105 Bq m"3 h
-------
APPENDIX B
SUMMARY OF PUBLISHED STUDIES WITH REGARD TO
INDOOR RADON DAUGHTER EXPOSURE AND LUNG CANCER RISK
The published epidemiologic studies involving non-occupational
exposure to radon daughters are summarized in this appendix. Host
studies listed involved several different analyses of the data including
separate analyses for males and females and various levels of adjustment
for confounding variables. The results given in this table generally
represent the analyses with the greatest degree of adjustment for those
factors. The analysis with the greatest degree of adjustment for
confounding variables was used as the basis for determining whether a
study indicated an excess risk of lung cancer with radon daughter
exposure. That determination is the judgement of the authors of this
report. In some cases the researcher used a more conservative standard
for concluding whether an excess risk was demonstrated. The
researcher's conclusions are also stated in such cases.
-------
B 2
Author: Axelson
Publication date: 1979
Country: Sweden
Type of study: Case-control
Cases: 37 lung cancer cases from rural population greatei than 40
years old
Controls: 178 from death register three positions before and after each
case.
Control or adjustment for confounding variables: none
Determination of radon daughter exposure:
Surrogate - type of housing
wood, no basement - 0
stone with basement - 2
all others - 1
Results:
Exposure category Odds ratio 90% confidence limits
0 vs 1+2 1.8 (0.99,3.2)
0 vs 2 5.4 (1.5, 19)
Etiologic fraction = 29%
Significant exposure-response trend (p<0.05)
Excess risk indicated: yes
Estimated risk coefficient: not applicable
Reference:
Axelson, 0., Edling, C., Kling, H. Lung cancer and residency: A
case-referrent study on th<» possible impact of exposure to radon and its
daughters in dwellings. Scand. J. Hork Environ. Health 5: 10-15; 1979.
-------
B 3
Author: Ouimette
Publication date: 1983
Country: USA (Hesa County, Colorado)
Type of study: Case-control
Cases: All lung cancer cases - Hesa County, 1970-1979 - 273
Controls: Colon, stomach and pancreatic cancers occurring during the
same time period - 275 controls
Control or adjustment for confounding variables:
Uranium mining history, smoking, sex, vital status, years of
residence in Mesa County, age at diagnosis
Determination of radon daughter exposure:
Ever lived in Type A home (home contaminated with mill tailings)
Results: Lung cancer cases no different from controls with respect to
residence in Type A homes
Crude odds ratio = 0.98
Adjusted odds ratio = 1.23 (p = 0.66)
Excess risk indicated: no
Estimated risk coefficient: not applicable
Reference:
Ouimette, D., Ferguson, S. W., 2oglo, D., Hurphey, S., Alley, S.,
Bohler, S. An epidemiologic Study of Selected Malignant Neoplasms in
Mesa County, Colorado, 1970-1979. Final Report; June 1983.
-------
B 4
Author: Pershagen
Publication date: 1984
Country: Sweden
I
Type of study: Case-control (pilot studies)
Cases: Lung cancer cases among twins - 53
Lung cancer patients from northern Sweden - 30
Controls: Twins of lung cancer cases - 53
Unrelated controls from northern Sweden - 30
Control or adjustment for confounding variables: smoking
Determination of radon daughter exposure:
Lifetime residence histories obtained for cases and controls.
Dwellings categorized by several fajtors: information on basement,
building material, type of house, ventilation, year of
construction. Radon level in each dwelling estimated on the basis
of results from nationwide measurements.
Results: No difference in radon exposure for cases and controls in twin
study for either smokers or non-smokers.
For the study of lung cancer in northern Sweden, cases who
smoked showed significantly higher radon exposures than
controls who smoked. No difference between non-smoking cases and
controls.
Excess risk indicated: equivocal
Estimated risk coefficient: not applicable
Reference:
Pershagen, G., Damber, L., Falk, R. Exposure to radon in
dwellings and lung cancer: A pilot study. In: Proceedings of the
Third international Conference on Indoor Air Quality and Climate, Vol.
2. Stockholm, Sweden, 73-78; 1984.
-------
B 5
Author: Edling
Publication date: 1334, 1986
Country: Sweden
Type of study: Case-control
Cases: Lung cancer deaths meeting specified criteria - 23 (22
controlled for smoking, 19 where actual radon measurement
was used)
Controls: From death register - 202 (178 controlled for smoking, 159
where actual radon measurement used)
Control or adjustment for confounding variables: smoking, age, sex
Determination of radon daughter exposure:
Visual classification of house
0 - wood house without basement on normal ground
2 - wood house with basement on alum shale ground
stone, brick or plaster house with basement on any ground
1 - all others
Categorization by actual radon measurement - cellulose nitrate film
Results:
Adjusted for smoking - actual radon measurement
Exposure cat.
Odds ratio (90% CL)
Mean
Measured cone
0 vs 1+2
2.7 (1.1,6.4)
0
0.11 WL
0 vs 2
5.1 (1.4,18.5)
1
0.15 WL
0 vs 1
2.3 (0.9.5.9)
2
0.46 WL
1 + 2
0.25 tfL
Adjusted for smoking - visual calssification
0 vs 1+2 1.8 (0.9,3.9)
0 vs 2 3.5 (1.3,9.2)
0 vs 1 1.2 (0.5,3.0)
Significant exposure-response trend
Smoking nulltiplicative with radon
Excess risk indicated: yes
Estimated risk coefficient: 5 - 7/1E6 person-yrs-fcJLM for non-smokers
Re \.erenr,e:
Edling, C., Kling, H., Axelson, 0. Radon in homes - a possible
cause of lung cancer. Scand. J. Work Environ. Health 10: 25-34, 1984.
Edling, C., Hingren, G., Axelson, 0. Quantification of lung cancer
risk from radon daughter exposure in dwellings - an epidemiological
approach. Env. Int. 12:55-60; 1986.
-------
B 6
Author: Damber
Publication date: 1986
Country: Sweden
Type of study: Case-control (pilot)
Cases: Hale lung cancer patients, 1972 - 1977, from the three
northernmost counties - 604 cases
Controls: Deceased controls drawn from the National Registry matched for
sex, year of death, age, and municipality; for cases born
after 1900, a living control selected, matched for sex, year
of birth and municipality
Control or adjustment for confounding variables:
Smoking, occupation
Determination of radon daughter exposure:
Surrogate - years of living in non-wooden house
Results:
No radon tffect on lung cancer risk from smoking seen, implying no
multiplicative effect
Ho significant difference in odds ratios after adjusting for
smoking only
Results of analysis for persons without occupational risk after
adjusting for smoking
Model I - Cases and matched deceased controls
Years in non-wooden houses OR (95% CI)
1-20 1.36 (0.83,2.25)
>20 1.53 (0.75,3.24)
Model II - Cases born after 1900 and both living and
deceased controls
1-20 1.46 (0.91,2.34)
>20 2.01 (1.01,4.03)
Excess risk indicated: yes (marginally significant for one analysis)
Estimated risk coefficient: not applicable
Reference:
Damber, L.: Lung cancer in males: An epidemiological study in
northern Sweden with sperial regard to smoking and occupation. Umea
University Medical Dissertations. rJew Series No. 167: 113-125; 1986.
-------
B 7
Author: Lees
Publication date: 1987
Country: Canada (Port Hope, Ontario)
Type of study: Case-control
Cases: Lung cancer cases living in Port Hope 1969 - 1979, living in
Port Hope at least 7 years prior tc diagnosis - 27 cases
Controls: Two controls for each case; one deceased and one living for
deceased cases, two living controls for living cases, matched
on sex and date of birth - <1 control":
Control or adjustment for confounding variables: smoking, duration of
residence in Port Hope
Determination of radon daughter exposure:
Prior radon or radon daughter measurement; estimated background
radon daughter exposure subtracted, cumulative radon daughter
exposure calculated for each subject
Results:
Conditional logistic regression analysis controlled for smoking:
OR = 2.36 (p = 0.057, one-sided) for erposed vs unexposed
(CULM - bkg =0)
Not controlled for smoking:
OR - 1.55 (p = 0.19, one sided)
Radon daughter exposure treated as continuous variable with smoking
controlled in analysis - positive correlation (p = 0.014)
Excess risk Indicated: yes (Lees, et al considered the results
inconclusive)
Estimated risk coefficient: not applicable
Reference:
Lees, R. E. H., Steele, R., Roberts, J. H. A case-control study of
lung cancer relative to domestic radon exposure. Int. J. Epid. 16:
7-12; 1987.
-------
B 8
Author: Svensson
Publication date: 1987
i
Country: Sweden (Stockholm)
Type of study: Case-control
Cases: Female unspecified epithelial lung cancers, principally small
cell anaplastic and large cell, diagnosed between 1972 - 1980,
232 cases
Controls: Population based controls matched by year of birth and alive
at the time of diagnosis of the case - 584
Control or adjustment for confounding variables: none
Determination of radon daughter exposure:
Radon + high radon geological type ground, ground level dwelling
Radon - above ground dwelling
10% of sample measured - all homes designated Radon +, and a
random sample of 110 homes designated Radon -
Results:
For both matched and unmatched analyses RR = 2.2 (p = 0.01)
95% Confidence limits (1.2,4.0)
Excess risk indicated: yes
Estimated risk coefficient: not applicable
Reference:
Svensson, C., Eklung, G., Pershagen, G. Indoor exposure to radon
from the ground and bronchial cancer in women. Int. Arch. Occup.
Environ. Health 59: 123-131; 1987.
-------
B 9
Author: Bean
I
Publication date: 1982
Country: USA (Iowa)
Type of study: Ecological
Exposed: High Ra-226 concentration in Mater, 2 categories
Unexposed: Low Ra-226 concentration in water
Hethod: Age adjusted cancer rates determined £or 22 small
municipalities (1000 - 10000) with wells >500 ft deep as sole
public water supply. Counties divided into 3 groups according to
Ra-226 concentration in water. Cancer rates compared for each
group
Control or adjustment for confounding variables:
Smoking patterns for municipalities, median income, percent of
manufacturing workers, percent of agricultural workers, fluoride
levels in water.
Determination of radon daughter concentration:
Surrogate - Ra-226 concentration in water
(Mo correlation has been found between radium in water and radon
in indoor air.)
Results:
Significant relationship between Ra-226 concentration in water and
male lung cancer (p = 0.028)
Exposure-response trend noted
Excess risk indicated: yes
Estimated risk coefficient: not applicable
Reference:
Bean, J. A., Isacson, D., Hohne, R. H. A., Kohler, J. Drinking
water and cancer incidence in Iowa: II Radioactivity in drinking water.
Am. J. Epidemiol. 116:924-32; 1982.
-------
B 10
Author: Dousset
Publication date: 1985
Country: France
Type of study: Ecological
Exposed: Limousin region - high gamma and radon
Unexposed: Poitou-Charentes - low gamma and radon
Control or adjustment for confounding variables: tobacco consumption
the same for the two regions studied
Determination of radon daughter exposure:
Surrogate - geographic location (background radiation)
Results:
Lung cancer rates for males and females no different
High background - Hale 52 E-5/a; female 6.8 E-5/a
Low background - Male 53.8 E-5/a; female 6.8 E-5/a
Ex^-S3 risk indicated: no
Estimated risk coefficient: not applicable
Reference:
Dousset, H., Jammet, H. Comparaison de la mortalite par cancer
dans le Limousin et le Poitou-Charentes. Radioprotection 20:61-67;
1985.
-------
B 11
Author: Forastiere
Publication date: 1985
Country: Italy (Viterbo Province)
Type of study: Ecological
Exposed: High background area (volcanic soils)
Unexposed: Low background area (non-volconic soil)
Method:
Comparison of lung cancer mortality rates for 1969 - 1978 for
population 35 - 74 years old
Control or adjustment for confounding variables:
age, degree of degree of urbanization, cigarette sales
Determination of radon daughter exposure:
Surrogate - volcanic content of soils, background radioactivity
Results:
Risk ratio males - 1.22 (p = 0.22)
Risk ratio females - 1.24 (p = 0.37)
Risk ratio total - 1.20 (p = 0.22))
Excess risk indicated: equivocal
Estimated risk coefficient: not applicable
Reference:
Forastiere, F., Valesini, S., Arco, H., Hagliula, H. E.,
Miehelozzi, P., Tasco.. C. Lur.g cancer and natural radiation in an
Italian Province. Sci. Total Environ. 45: 519-526; 1985.
-------
B 12
Author: Hoffman
Publication date: 1985, 1986
Country: China, Austria
Type of study: Ecological
Exposed: China - High background area - 0.38 HLM/a
includes both radon and thoron daughters
Austria - Badgastein - 0.4 ULM/a
Unexposed:
China - Low background area - 0.16 WLH/a
includes both radon and thoron daughters
Austria - Salzburg - 0.2 WLH/a
Control for factors other than radon: For China, smoking habits and sex
ratios comparable for both populations
Determination of radon daughter concentration:
Actual measurements of radon and thoron concentration
Results:
No difference in lung cancer rates batween exposed and unexposed
for both Austria and China
Excess risk indicated: no
Estimated risk coefficient: not applicable
Reference:
Hofmann, W., Katz, R., Zhang, C. Lung cancer incidence in a
Chinese high background area - epidemiological results and theoretical
interpretation. Sci. Total Environ. 45: 527-534; 1985.
Hofmann, W., Katz, R., Zhang, C. Lung cancer risk at low doses of
alpha particles. Health Physics 51:457-468; 1986.
-------
B 13
Author: Archer
Publication date: 1987
Country: USA (Reading Prong)
Type of study: Ecological
Exposed: 16 counties in NY, NJ, and PA associated with Reading Prong
Reading Prong (RP) counties = 7
Fringe counties (F) = 9
Unexposed: Counties adjacent to fringe counties = 17
Method: Lung cancer rates for whites compared for the three groups
of counties
Control or adjustment for confounding variables:
none, (urban-rural, socioeconomic factors considered but not in
analysis)
Determination of radon daughter exposure:
Surrogate - association with Reading Prong
Exposure-response effect observed
Population growth highest in RP counties indicating greatest
degree of migration
Excess risk indicated: yes
Estimated risk, coefficient: not applicable
Reference:
Archer, V.E.: Association of lung cancer mortality with
precambrian granite. Arch. Env. Health 42: 87-91; 1987.
Results:
RP
F
C
31.32 (30.52,32.12)
27.49 (26.80,28.08)
23.91 (23.37,24.45)
-------
B 14
Author: Fleischer
Publication date: 1981
Country: USA
Type of study: Ecological
Method:
Comparison of counties with high lung cancer risk with geographic
location
' 1. Counties in top decile of lung cancer rate - statistically
significant difference from national average
2. Counties where 95% confidence interval on the lung cancer
rate does not overlap 95% confidence interval for national
rate
3. Counties with high lung cancer rates but overlapping
confidence intervals
Determination of radon daughter exposure:
Surrogate - Geographic location (counties with phosphate mines,
deposits or processing plants)
Control or adjustment for confounding variables: population
Results:
Comparisons of observed coincidences between phosphate counties
and highest and significantly high lung cancer counties with
expected coincidences showed obs/eip >1 (p<0.01) for males and
females. When adjusted for population, effect was seen for all
males (p = 0.01 - 0.08) and females in the most highly populated
areas (p<0.015)
Excess risk indicated: yes
Estimated risk coefficient: not applicable
Reference:
Fleischer, R.L.: A possible association between lung cancer and
phosphate mining and processing. Health Physics 41: 171-175; 1981.
-------
B 15
Author: Edling
Publication date: 198?.
Country: Sweden
Type o£ study: Ecological
Number of locations: 24 counties
Control or adjustment for confounding variables: none
Determination of radon daughter exposure:
Surrogate - background gamma radiation
Results:
Excess risk shown for lung cancer
Kales r = 0.46 (p = 0.012)
Females r = 0.55 (p = 0.003)
Excess risk indicated: yes
Estimated risk coefficient: not applicable
Reference:
Edling,C., Comba, P., Axelson, 0., Flodin, 0. Effects of lov-dose
radiation - A correlation study. Scand. J. Work Environ. Health 8;
suppl 1: 59-64; 1982.
-------
B 16
Author: Hess
Publication date: 1983
Country: USA (Maine)
Type of study: Ecological
Nuraber of locations: 16 counties
Control for factors other than radon: none
Determination of radon daughter exposure:
Surrogate - Radon concentration in water
Results:
Males, r = 0.46 (p<0.10)
Females, r = 0.65 (p<0.01)
Average, r =0.56 (p<0.05)
Excess risk Indicated: yes
Estimated risk coefficient: not applicable
Reference:
Hess, C.T., Weifenbach, C. V., Norton, S. A. Environmental radon
and cancer correlations in Maine. Health Physics 45: 339-348; 1983.
-------
B 17
Author: Letourneau
Publication date: 1983
Country: Canada
Type of study: Ecological
Number of locations: 18 cities
Control or adjustment for confounding variables: smoking for 14 cities
Determination of radon daughter exposure:
Geometric mean of the measured radon daughter concentrations
Results:
No significant correlation between lung cancer rates and geometric
mean radon daughter concentration
Multiple linear regression on smoking and radon daughter
concentration showed no effect of radon
Excess risk indicated: no
Estimated risk coefficient: not applicable
Reference:
Letourneau, E.G., Hao, Y., HcGreggor, R. G., Semenciw, R., Smith,
M. H., Wigle, D. T. Lu"} cancer mortality and indoor radon
concentrations in 18 C". .adian cities. Proceedings of the Sixteenth
Midyear Topical Meeting of the Health Physics Society, Epidemiology
Applied to Health Physics, pp 470 - 483; 1983.
-------
B 18
Author: Fleisher
Publication date: 1986
Country: USA (Reading Prong)
Type of study: Ecological
Hethod:
Comparison of high lung cancer risk counties with geographic
location
Risk groups:
1. Counties in top decile of lung cancer rates -
statistically significant difference from national rates
2. Counties where 95% confidence interval on lung cancer rate
does not overlap 95% confidence interval for national rate
3. Counties with high lung cancer rate but with 95%
confidence interval overlaping 95% confidence interval for
national rate
Control or adjustment for confounding variables: none
Determination of radon daughter exposure:
Geographic location
1. >50% within Reading Prong - 3 counties
2. <50% within Reading Prong - 10 counties
Results:
Comparison of observed coincidences between Reading Prong counties
and highest, significantly high and not significantly high lung
cancer counties with expected coincidences showed obs/exp > 1
(p = 0.1) for coincidences between highest and significantly high
lung cancer counties and >50% Reading Prong counties
Excess risk indicated: yes
Estimated risk coefficient: not applicable
Reference:
Fleischer, R.L. A possible association between lung cancer and a
geological outcrop. Health Physics 50: 823-827; 1986.
-------
B 19
Author: Halter
Publication date: 1986
Country: USA (Connecticut)
Type o£ study: Ecological
Number of locations: 169 towns
Control or adjustment for confounding variables: socioeconomic status,
population density
Determination of radon daughter exposure: none
Background radiation was the exposure variable
Results:
Mo significant correlation between lung cancer rates and background
radiation
Excess risk indicated: no
Estimated risk coefficient: not applicable
Reference:
Halter, S. D., Meigs, J. H., Heston, J. F. The relationship of
cancer incidence to terrestrial radiation and population density in
Connecticut 1935 - 1974. Am. J. Epidemiol. 123: 1-14; 1986.
-------
B 20
Author: Stranden
Publication date: 1986, 1987
Country: Norway
Type of study: Ecological
Number of sites: 75 locations measured; 20 hones/location
Control or adjustment for confounding variables: smoking
Determination of radon daughter exposure:
Activated charcoal and TLD; '.wo per home
Results:
Significant correlation (95* confidence level) found between lung
cancer incidence and mean radon concentration in grouped locations
categorized by radon concentration*
Excess risk indicated: yes
Estimated risk coefficient:
Excess relative risk: 0.001 - 0.003/Bq-m-3 for radon
0.002 - 0.006/Bq-m-3 for progeny**
0.003 - 0.009/HLM***
Reference:
Stranden, E.: Radon-222 in Norwegian Dwellings. Radon and Its
Decay Products; Occurrence, Properties, and Health Effects. ACS
sypmosium Series 331, Hopke, P.K. ed. p 70 - 33; 1987.
Stranden, E.: Radon in Norwegian dwellings and the feasibility of
epidemiologic studies. Radiat. Environ. Biophys. 25: 37-42; 1986.
* It is unclear whether the data was analyzed individually by location
or whether locations ware grouped by mean radon concentration and
grouped data compared to lung cancer risk.
** Paper states radon progeny risk factor as 0.002 - 0.06
It wa3 assumpd that the 0.06 was a typographical error that
should have been 0.006.
*** (0.002/Bq m-3)(3.7 E3 Bq m-3/HL) = 7.4/UL
(1 WLM8760 hr/a/170 hr/mo)(0.8) = 40 HLH/HL a
{(7.4/ML)/(40 WLM/HL a)){60 a) = 0.003/WLM
-------
B 21
Author: Castren
Publication date: 1987
Country: Finland
Type of study: Descriptive
Method: Comparison of geographical distribution of lung cancer and
elevated radon concentrations
Control or adjustment for confounding variables: none
Determination of radon daughter exposure: alpha track detectors in
hones
Results: No observed resemblance between high lung cancer rates and
high radon concentrations for males; some indication of resemblance
in distribution for rural women
Excess risk indicated: no
Estimated risk coefficient: not applicable
Reference:
Castren, 0. Dealing with radon in dwellings. -Second International
Specialty Conference on Indoor Radon. Air Pollution Coitrol Association.
New Jersey, April, 1987.
-------
B 22
Author: Simpson
Publication date: 1983
Country: USA (Maryland)
Type of study: Cohort
Control or adjustment for confounding variables:
age, sex, many other variables; all housing variables studied
except the one of interest
Determination o£ radon daughter exposure:
Surrogate - housing characteristics (basement construction,
building material of walls , heat source, cooking fuels
Results: No difference in lung cancer rates with housing
characteristics
Excess risk indicated: no
Estimated risk coefficient: not applicable
Reference:
Simpson, S. G., Cornstock, G. W. Lung cancer and housing
characteristics. Arch. Environ. Health 38:248-251; 1983.
-------
APPENDIX C
SUMMARY OF UNPUBLISHED STUDIES WITH REGARD TO
INDOOR RADON EXPOSURE AND LUNG CANCER RISK
The studies summarized in Appendix C are completed but as yet
unpublished. One of the studies is the subject of a paper in press
(Stocltwell, American Journal of Epidemiology). Results of the other
studies were reported in Masters or PhD theses, as oral presentations at
meetings, as a special report or were obtained by personal communication
with the researcher. As with the summaries.of the published studies,%
the results shown are generally those for the analyses with the greatest
degree of adjustment or control for confounding.
-------
C 2
Author: Cohen
Country: USA
Type oE study: Ecological
Method:
Correlation between lung cancer mortality rates and average radon
exposure in various counties in the U. S. (310 counties)
Results:
Weak negative correlation observed.
Excess risk demonstrated: no
Estimated risk coefficient: not applicable
Reference:
Cohen, B. L. University of Pittsburgh. Personal communication:
October, 1986, September, 1987.
-------
C 3
Author: Lanes
Country: USA (Pennsylvania)
Type of study: Case-control
Cases: Lung cancers from 1/1/61 to 12/31/79 in Cannonsburg
and Houston Boroughs - 50 cis?s
Controls: Sequential arteriosclerotic heart disease deaths
from death records - 48
Control or adjustment for confounding variables: Socio-economic
status, smoking (by assumption that all lung cancer cases were
smokers), no stratification by sex, no adjustment for age
Determination of radon daughter exposure:
Track etch detectors (Temdex) summer and winter
Results:
Mean concentration ir homes of cases was compared to mean
concentration in homes of controls
No significant difference between case and control homes for
geometric mean radon concentration
Borderline significant difference between case and control
homes for arithmetic mean (p = 0.09, summer measurement; p =0.13
for winter measurement)
Excess risk indicated: equivocal
Estimated risk cofficient: not applicable
Reference:
Lanes, S. F.: Lung cancer and environmental radon exposures: A
case control study. Doctoral Dissertation, University of Pittsburgh;
1982.
-------
C 4
Author: Stockwell
Country: USA (Florida)
Type of study: Case-control
Cases: All cases of lung cancer among Florida residents from 1981 -
1983 - 25,398 cases
Controls: Individuals with cancer of colon or rectum - approximately
22,000
Control or adjustment for confounding variables:
Smoking status considered in the analysis
Determination of radon daughter exposure:
Surrogate: residence in central Florida where phosphate deposits
are located
Results:
Two-fold increase in lung cancer risk among non-smoking males
living in study area.
Slight, but not significant, increase in risk among smokers.
No significant elevation in risk among women.
Residents of Tampa, not a high phosphate area, also showed
increased risk of lung cancer.
Excess risk indicated: yes
Estimated risk coefficient: not applicable
Reference:
Stockwell, H. University of South Florida, Tampa, FL. personal
communication. September, 1987. (paper in press)
-------
Author: Stockwell
Country: USA (Florida)
Type of study: Case-control
Cases: All cases of carcinoma of the lung first diagnosed between 1931
and 1983 among residents of 53 Florida counties - 25,266 cases
Controls: All cases of colon and rectal cancers among residents of the
sane counties during the same time period
Control or adjustment for confounding variables: age, sex, race and
tobacco use
Determination of radon daughter exposure:
Residence in county classified as having elevated radon levels
based on statevide radon mapping study
3 counties - 15% or more of measured homes above 4 pCi/1
15 other counties with elevated radon levels
35 counties with no evidence of elevation in indoor radon
potential
Results:
Males: significant elevation in odds ratio for highest three
counties (All cell types: white, OR = 1.3, 95% CI (1.1,1.6);
non-white, OR = 2.7, 95% -CI (1.4,5.2)
Females: no significant increase in lung cancer risk
Odds ratios adjusted for age, sex, race and tobacco use:
Three highest counties: OR = 1.25, 95% CI (1.09,1.43)
Remaining 15 counties: OR = 0.88 95% CI (0.84,0.92)
Excess risk indicated: yes (among males only)
Estimated risk coefficient: not applicable
Reference:
Stockwell, H. paper presented at the Fourth International
Symposium on The Natural Radiation Environment in Lisbon Portugal,
December, 1987.
-------
C 6
Author: Lanctot
Country: USA (Maine)
Type of 3tudy: Case-control pilot study
Cases: Lung cancer cases in Maine at least 10 years using water from
privately owned drilled well - 35
Controls: Other cancers - 118 self-selected
Non cancer patients - 174 self-selected
All controls using water from privately owned drilled well
Control or adjustment for confounding variables:
Analysis for smoking, sex, age, residence history, education,
occupation
Determination of radon daughter concentration:
Alpha track placed in kitchen - 2 month exposure
Results:
Significant excess risk of lung cancer for men under the age of 65
with radon concentration greater than 3 pCi/L in home
Excess risk indicated: yes
Estimated risk coefficient: not applicable
Reference:
Lanctot, E.M.: Radon in the domestic environment and its
relationship to cancer: An epidemiologic study. Masters Thesis, Stata
University of New York at Stony Brook. Maine Geological Survey,
Department of Conservation Publication 85-88; 1985.
-------
C 7
Author: Austin
Country: USA (Uravan, Colorado)
Type of study: Cohort
Method: Lung cancer rate for women living in Uravan, CO for at least
one year were compared to expected rate (U.S. population, age
adjusted; Colorado, age adjusted)
Control or adjustment for confounding variables: none
Determination of radon daughter exposure:
Measurement of individual residences by alpha track
Mean = 0.02 WL
Results: 6 cancers observed vs 3 expected (Colorado cancer rate)
1 SIR » 2.0 95% confidence limits (0.73,4.36)
Compared to U. S. cancer rate:
SIR = 1.15 95% confidence limits (0.42,2.51)
38,000 person-years of follow-up
Excess risk indicated: yes, but not statistically significant and only
when results are compared to Colorado cancer rate.
(Austin considers results inconclusive.)
Estimated risk coefficient: not applicable
Re ferenee:
Austin, S., Fort Collins, CO, personal communication, September,
1987.
-------
APPENDIX D
SUMMARY OF STUDIES IN PROGRESS BITH REGARD TO
INDOOR RADON EXPOSURE AND LUNG CANCER RISK
The following summaries of studies in progress are based in most
cases on personal interviews with the principal investigator or other
researchers involved as well as written study protocols. For three
studies (Stebbings, Stockwell, Stockwell), the information was obtained
from telephone communication.
The procedures described are those planned as of the date of the
interview and may be changed before or during the studies.
-------
0 2
Country: Sweden
Principal investigator: Pershagan
Type: Case-Control
Females
Cases: Females admitted to hospital with suspect lung or bronchus
nalignancy
Controls: Control born on the same day determined £rom population
registry
Estimated numbers: Cases - 200
Controls - 400
Control or adjustment for confounding variables: Diet (Vitamins A and C)
Passive smoking, active smoking
Determination of radon daughter exposure:
1. Questions asked about all residences lived in aore than two
years
location, type of house, building year, type of heating
system, type of ventilation, type of building materials
2. Radon exposure estimated from geological conditions and
building characteristics
3. Measurement of radon gas by track etch - two week measurement
"high risk" homes in Stockholm area
random sample of "low risk" homes in Stockholm
4. Measurement of radon gas by track etch - one year - 1500 case
and control homes
Expected date for preliminary results: 1987
Expected date for final results: 1988
Reference:
Pershagan, G. National Institute of Environmental Medicine,
Stockholm, Sweden. Personal communication; Hay, 1987.
-------
D 3
Country: Canada (Winnipeg, Manitoba)
Principal investigator: LeTourneau, Health and Welfare Canada
Type: Case control
Male and female
Cases: Lung cancers diagnosed - living at time of interview
(no surrogate interviews)
Controls: No cancer, selected at random, matched for age, sex and
occupation
Estimated numbers: Cases - 700 (200/yr)
Controls - 700
Control or adjustment for confounding variables:
Questionnaire requests information on marital status, ethnicity,
education, education, employment, vitamin use, health history,
smoking, passive smoking, use of hair dye, income, use of oral
contraceptives and hormones
Determination of radon daughter exposure:
Radon exposure to be determined for all current and previous
residences using alpha track detectors, soil sampling and grab
sampling
Expected date for preliminary results: none
Expected date for final results: 1990
Reference:
Letourneau, E. Health and welfare Canada, Ottawa, Ontario, Canada.
Personal communication; June, 1987.
-------
D 4
Country: Hungary (Miskolc)
Principal investigators: Takacs and Paripas; Regional Health
Department, Miskolc, Hungary
Type: Ecological
Indoor radon concentrations compared for two towns, one with high
lung cancer (5 E-4/a) and one with low lung cancer rate (2 E-4/a)
Estimated numbers: not known
Control or adjustment for confounding variables: no information on
smoking
Determination of radon daughter exposure:
Radon measurements made in approximately 100 residences in each
town using alpha track detectors
Thoron measurements made (estimate from total alpha minus radon)
External gamma measurements made using TLDs
Preliminary results:
No significant difference in radon concentrations between the two
towns
Gamma exposure approximately 20% higher in high lung cancer rate
town
Thoron concentration significantly higher in high lung cancer rate
town
Final results: Study is continuing; no specific date for final results
Reference:
Takacs, S., Regional health department, Miskolc, Hungary. Personal
communication; August, 1987.
-------
D 5
Country: England (Cornwall and Devon)
Principal investigators: Doll, Darby; Imperial Cancer Research Fund
NRPB
Type: Case-control
Cases: Patients under 75 years of age admitted to hospital with
presumptive diagnosis o£ lung cancer
Controls: Hatched sample of patients admitted to same hospital with
presumptive diagnosis of conditions unrelated to smoking
Hatched sample of healthy individuals randomly selected from Family
Practitioner Committee lists for Devon and Cornwall
Estimated numbers: Cases - 500 - 1000
Controls - 1000 - 2000
Control or adjustment for confounding variables:
Smoking, occupation, sex, age
Use of presumptive diagnosis and later rejection of cases with
other confirmed diagnosis provides control group free of interview
bias
Determination of radon daughter exposure:
Measurement of atmospheric radon In random sample of homes
stratified by area and length of time in which individuals
inhabited them
Expected date for preliminary results: unknown (interviews
to be completed in two years)
Expected date for preliminary results: unknown
Reference:
James, A. National Radioligical Protection Board. Personal
communication; April, 1987.
Darby, S. C. Personal communication; April, 1987.
Huirhead, C. R. Personal communication; April, 1987.
-------
0 6
Country: Norway
Principal investigator: Stranden, Norwegian Staten3 Institutt for
Stralenhygiene in collaboration with:
NPRB, United Kingdoa
Norwegian Cancer Registry
Type: Correlation
Hethod: Lung cancer incidence by municipality as determined from Cancer
Registry data compared to mean radon concentration in dwellings
Control or adjustment for confounding variables: smoking
Determination of radon daughter exposure:
Stratified random sample of dwellings in each municipality; number
to be proportional to population except in two largest cities.
(Sample stratified by type of housing) - 10,000 dwellings
NRPB dosemeters in each dwelling for six months in main bedroom
Expected date for preliminary results: End of 1988
Expected date for final results: unknown
Reference:
Stranden, E. Norwegian Statens Institutt for Stralenhygiene, Oslo,
Norway. Personal communication; Hay, 1987.
-------
0 7
Country: Finland (Uusimaa, Kymi)
I
Principal investigators: Castren, Ruosteenoja
Finnish Centre for Radiation and Nuclear Safety
Finnish Cancer Registry
Type: Case-control
Cases: Hale lung cancer cases diagnosed 1980 - 1985 in study area
Controls: Population based - random sample of nen living in study area,
stratified by age, from Finnish Population Register Center
Estimated numbers: Cases - 300
Controls - 1500
Control or adjustment for confounding variables: smoking, rural
residents only used
Determination of radon exposures:
Radon daughter level measured in all long-term residences from 1950
- 1980
Alpha track film in living room for two months
Expected date for preliminary results: unspecified
Expected date for final results: December, 1987
Reference:
Castren, 0. Finish Centre for Radiation and Nuclear Safety,
Helsinki, Finland. Personal communication; Hay, 1987.
-------
D 8
Country: USA (Pennsylvania)
Principal investigator: Stebbings, Argonne National Lab
Type: Case-control
Cases: Female lung cancer cases - women born in state and dying as
resident of state, case series defined by lung cancer cell type
Controls: undetermined
Estimated numbers: 2,000 cases
Control or adjustment for confounding variables: Smoking
Determination of radon daughter exposure:
Alpha track measurement in major living areas
Expected date for preliminary results: none given
Expected date for final results: none given
Reference:
Stebbings, J. Argonne National Laboratory. Personal
communication; July, 1987.
-------
D 9
Country: USA (Missouri)
Principal investigator: Brownson, Missouri Dept. Health
Type: Case-control (population-based)
Cases: Non-smoking female incident lung cancer cases determined from
Missouri cancer registry
Controls: Random sample of non-smoking female Missouri population,
frequency matched to overall case series by age, race and smoking
status.
Estimated numbers: Cases - approximately 28Q
Controls - approximately 560
Control or adjustment for confounding variables:
All cases and controls non-smoking
Questionnaire Mill obtain information with regard to residential
history, passive smoking, family history of cancer, nonmalignant
respiratory disease, hormonal factors and menstrual history, use of
space heatinq and cooking, dietary history, occupational exposures
and history
Data will be analyzed by family history of lung cancer and by lung
cancer cell type
Determination of radon daughterexposure:
Radon measurement made by alpha track in homes of each case and
control occupied during the past thirty years. Two detectors per
home to be left in place for one year
Expected date for preliminary results: 1988 - 1990
Expected date for final results: 1990
Reference:
Brownson, R. Missouri Department of Health, Columbia, Ho.
Personal communication, June, 1987.
-------
D 10
Country: USA (Maine, New Hampshire)
Principal investigator: Rand, Haine Hedical Center
Type: Case-control
male and female
Cases: Incident cases of lung cancer contacted soon after diagnosis,
pathology report required, must have lived in area for 5 years,
must be able to measure 80% of past exposure (5 - 35 years in the
past)
Controls: Population based £rcm drivers licenses under age 65, Health
Care Financing records over age 65; frequency matched for age and
sex - same criteria as for cases
Estimated numbers: 500 female cases and controls
500 male cases and controls
Control or adjustment for confounding variables: smoking, occupational
exposures, house construction and health
Determination of radon daugter exposure:
Alpha track detectors will be placed in living room and bedroom of
all residences cases and controls lived in during the period 5-35
years prior to diagnosis - only cases and controls for whom 80% of
the prior exposure can be determined will be included in the study
Expected date for preliminary results: unknown
Expected date for final results: unknown
Reference:
Rand, P., Lacombe, E. Maine Hedical Center, Portland, HE.
Personal communication; June, 1987.
-------
D 11
Country: USA (Maine) - Pilot study
I
Researcher: Bogdan, Maine Department of Human Services
Type: Case-control
males and females
Cases: Lung cancer cases living or recently deceased served by a
privately-owned drilled veil for at least ten years
Controls: Individuals with other cancers
Individuals with no cancer
Estimated numbers: Cases - 100
Controls - 150, 250
Control or adjustment for confounding variables: smoking, occupational
exposures; medical history
Questions asked about house construction, water use, occupancy
habits
Determination of radon daughter exposure:
Terrradex Track Etch dosimeters on refrigerator in kitchen
Radon in water
Erpected date for preliminary results: Preliminary results reported in
MS thesis, E. H. Lanetot (see Appendix C, Unpublished studies)
Expected date for final results: 1987
References:
Lanctot, M. Haine Geological Survey, Augusta, HE. Personal
communication; September, 1987.
Rand, P., Lacombe, E. Maine Medical Center, Portland, ME.
Personal communication; June, 1937.
-------
D 12
Country: USA (New Jersey)
Principal investigator: Klotz, New Jersey Department of Health
Type: Case-control
Females only
Cases: Primary and histologically confirmed cancers,
females, 8/82 - 9/83
Controls: Stratified by age, race
Interviewed cases under age 65 - controls selected from drivers
license files
Cases interviewed over age 65 - controls from HCFA files
Next of kin interviewed cases - controls from death certificates
matched by age and date of death
Estimated numbers: Cases - 994
Controls - 995
Control or adjustment for confounding variables: smoking, diet
Determination of radon daughter exposure:
Radon measurement in residence of longest duration over 10 - 30
years prior to diagnosis (greater than 10 years residence)
Charcoal canisters in basement and master bedroom
(4 days) 10/86 - 4/87, repeat Fall 87 - Winter 88
Alpha track (Terradex) 2 or 3 per house, 1 year
Expected date for preliminary results: none
Expected date for final results: Fall 1988
Reference:
Klotz, J., Schoenberg. J. New Jersey Department of Health,
Trenton, NJ. Personal communication; June, 1987.
-------
D 13
Country: USA
Principal investigator: Cohen
Type: Ecolocical
Method: Lung cancer incidence by county compared to mean measured
indoor radon concentrations
Control ok adjustment for confounding variables: unknown
Determination of radon daughter exposure:
Charcoal canister
Expected date for preliminary results: ongoing study, results reported
periodically
Expected date for final results: none given
Reference:
Cohen, B. University of Pittsburgh, Pittsburgh, PA. Personal
communication; October, 1987.
-------
D 14
Country: USA - Florida
Principal investigator: Stockweil, University of South Florida
Type: Pilot study - Case-control
Cases: Females with lung cancer
Controls: Females, randomly selected, matched for age, race and general
geographic area
Estimated numbers: Cases - 80
Controls - 80
Control or adjustment for confounding variables:
Smoking, diet, occupation controlled in analysis
Determination of radon daughter exposure:
Alpha track detectors (3 months) in homes lived in 10 years or
longer
Expected date for preliminary results: none
Expected date for final results: Dec. 1987
Reference:
Stockweil, H. University of South Florida, Tampa, Florida.
Personal communication; September, 1987.
-------
D 15
Country: USA - Florida
Principal investigator: Stockwell
Type: Case-control
Cases: Non-smoking women living in Florida 10 years or longer, newly
diagnosed lung cancer
Controls: Non-smoking women randomly selected, matched for age, race,
geographic location, living in Florida at least 10 years
Estimated numbers: Cases - 300
Controls - 600
Control or adjustment for confounding variables:
Control in analysis for effects of factors such as passive
smok.Vng, occupation, diet, family history
Determination of radon daughter exposure:
Alpha track detectors in home for one year. Current home if lived
in 10 or more years; up to 4 previous Florida homes, lived in at
least 5 years
Expected date for preliminary results: none given
Expected date for final results: 1992 or 1993
Reference:
Stockwell, H. University of South Florida, Tampa, Florida.
Personal communication; September, 1987.
-------
APPENDIX E
COMPUTATIONS OF LIFETIME LUNG CANCER RISKS
ATTRIBUTABLE TO INDOOR RADON
Estimates of lifetime lung cancer risk due to indoor- radon exposure were
computed using models developed by:
0 The National Academy of Sciences Committee on the Biological Effects
of Ionization Radiation, BEIR III (NAS30) and BEIR IV (NAS83)
• The National Council on Radiation Protection and Measurements,
NCRP78 (NCRP3U)
• The International Commission on Radiation Protection, ICRP50 (ICRP87)
• The U.S. Environmental Protection Agency, (USEPA86)
Exposure to indoor radon was calculated for mean annual indoor radon
concentrations of 1.0 pCi 5. \ the presumed U.S. average concentration (Ne86),
and 'J.O pC 1 H"1, the EPA guideline whe-» some remedial action is recommended
(USEPA86). The fraction of time spent indoors was assumed to be 75%, with a
mean outdoor concentration of 0.2 pCi 8, \ and an equi1ibrlum factor of 0.5
for both indoors and outdoors.
The annual exposure at 1.0 pCi ^ (18.5 Bq m^ EEC) i.3:
-------
E 2
[ (1 .0 pCi £~1)(0.75) + (0.2 pCl n"1) (0. ?'3) ][o .5 ] (8760 h a"1 ]
[ 100 pCi 2,"1Wl"1] [ 170 WL h WLM_1 ]
» 0.21 WI.M a-1 (1.3x105 »q h m"3)
The annual exposure at 1.0 pCi I-1 (71 Bq/m3 EEC) is:
[(1.0 pCl if1) (0.75) + (0.2) (0.25)] [p.5} [8760 ha"1]
[100 pCi S,~1WL-1] [ 170 WL h WLM~1]
- 0.79 WLM a"1 (5.0x10^ Bq h m~3)
BEIR
The 3EIR III risk coefficients are:
10x10~6 WLM-1 a"1 ages 35-19
20x10~6 WLM-1 a"1 ages 50-65
50x1O-6 WLM-1 a"1 ages 65-75
Assuming a minimum 5 year latent period and a mean lifespan of 75 years the
excess lifetime risk can be calculated.
The average number of years of exposure for the age group 35-19 is 12
years. Taking Into account a latent period of 5 years, the mean effective
exposure time Is 37 years. Therefore, the mean effective currmulat i ve exposure
at 1 pCl J."1 and 1 pCi SI-1 indoor3 Is
(37a) x (0.21 WLM a-1)
(37a) x (0.79 WLM a"1)
=7.8 WLM
= 29 WLM
-------
E 3
Repeating this for the 50-65 age group:
(52.5a) (0.21 WLM a"1) * 11.03 WLM
(52.5a) (0.79 WLM a-1) - 11.18 WLM
and the 66-75 age group:
(65.5a) (0.21 WLM a"1) = 13.76 WLM
(65.5a) (0.79 WLM a"1) - 51.75 WLM
The lifetime excess lung cancer risks were obtained by multiplying the
risk coefficients by the number of years at risk and the mean effective
cuiwnulative exposure. The number of years at risk is the total number of
years In the age range to which the risk coefficient applies up to age 75. At
1 pC i r1 :
OOxlO"6 a"1 WLM-1) (15 a) (7.77 WLM) - 1.2xlO"3
(20x10~6 a"1 WLM"1) (16 a) (11.03 WLM) = 3.5x10/*
(50xl0~6 a-1 WLM"1) (10 a) (13.76 WLM) - 6.9x10~3
Total 1.2x10"2
At i pci r1
(10x10~6
a"1
WLM"1)
(15
a)
(29.23 WLM)
= 1.1x10~3
(20x10"^
a"1
WLM"1 )
(16
a)
(11.18 WLM)
= 1.3x10"^
(50x10-6
a"1
WLM"1)
(10
a)
(51.75 WLM)
- 2.6x10~2
Total
1. 1x10"2
-------
NC.RP
From !JCSP7d, Table 10.2, the lifetime excess lung cancer risk for a
llfetlrre exposure of 1.0 WLM a ' Is 9.1x10 3
Thus, for 1 pCl the lifetime risk la:
(0.21 WLM a"1) (9.1x10~3 a WLM~1) - 1.9x10~3
For t.O pCl 2, 1 ths llfetlr.e risk Is:
(0.79 WLM a"1) (9.1x10~3 a WLM-1) - 7.2x10~3
ICRP
The model used by the ICRP Is a proportionate hazards model with a
relative excess risk coefficient dependent on age at exposure. The lifetime
risk is:
0.019 WLM-1 for ages 0-20a
0.006U WLM-1 for ages >?0a
The average relative excess risk coefficient adjusted for age at exposure,
assuming a lifespan of 75 years, is 0.0098 WLM"1.
The ICRP model takes Into account the effect of premature death due to
indoor radon exposure. However, the relative excess risk coefficient remains
constant for annual exposures less than 3 WLM.
-------
E 5
Proportionate hazards v,. relative risk models can be applied to
populations using the "observed" lung cancer risk corrected for the population
mean radon risk component:
R = R + r E R
o o
where: R = the observed lung cancer risk
r » mean relative risk coefficient for radon daui hter exposure
E - mean lifetime radon daughter exposure
R = baseline (no radon daughter exposure) lung cancer risk,
o
R
R =
0 1 + r E
The ICRP uses a baseline lifetime lung cancer risk of 0.6* for nonsmokers
without radon exposure. This was derived by subtracting 10$ from the
calculated nonsmoker lifetime lung cancer risk to account for the radon
daughter contribution.
The excess lung cancer risk from radon daughter exposure at an average
indoor radon concentration of 1.0 pCi (0.21 WLM a-1) would be:
(0.21 WLM a-1) (70a) (0.006) (0.0098 WLM-1) - 0.00096
For heavy smokers (>2 packs/day) the lung cancer mortality Is 15 to 25
times that for nonsmokers (USPHS82). However, due to premature death from
other smoking related dlseasta, the ratio of the lung cancer risk due to radon
for smokers versus nonsmokers is not a linear function. For annual Indoor
radon Daughter exposures less than an EEC of 10*5 Bq h m - (1.6 WLM) the ratio
-------
E 6
of lifetime lung cancer risk duo to radon for one pack/day smokers vs.
nonsmokers is approximately 5 (ICRP87). For heavy smokers (>1 pack/day) that
ratio la unlikely to exceed 7.
For people who have stopped smoking for at lea3t 15 years, the mortality
rate for lung cancer is about twice that for nonsmokers (USPHS82).
Therefore, for smokers the lifetime lung cancer risk attributable to
radon daughters at a mean indoor radon concentration of 1.0 pCI S."1 is:
(5) (0.00036) - 0.001)3
For ex-smokers at 1.0 pCI t"1
(2) (0.00->06) - 0.0017
At an average indoor radon concentration of 't.O pCI I ^ the lifetime excess
lung cancer risk for a nonsmoker is:
(0.79 WLM a"1) (70a) (0.006) (0.0093 WLM-1) - 0.0033
For a smoker at 1 pCi i 1:
(0.0033) (5) - 0.016
For an ex-s.nok<;r at 'I pCi i
(0.0033) (2) - 0.0060
-------
E 7
EPA
The EPA uses a relative risk model with risk coefficients of 0.01 WLM-1
to 0.04 "WLM-1. For the relative risk coefficient of 0.01 WLM-1 the baseline
lung cancer risk, used for the ICRP risk calculation, 0.65, is appropriate
since the ICRP relative risk coefficient is nearly the same.
For an assumed relative risk coefficient of 0.04 WLM-1, the baseltne lung
cancer risk must be adjusted to account for the greater contribution of indoor
radon at mean concentration levels to the observed lung cancer rate.
r = —5
o 1 + rE
R - R0 (1 + rE)
Assuming the relative risk coefficient is 0.00^8 WLM-1 (ICRP87)
R = 0.6J [1 + (0.0093 WLM-1) (14.7 WLM)]
R = 0.692
Assuming the relative risk coefficient is 0.04 WLM-1
C 69
Ro = 1 + (0.04) (14.7) = °*'<3
Assuming a lifespan of 75 years and a latent period of 5 year?, at a mean
indoor radon daughter concentration of 1.0 pCl/l the lifetime lung cancer
risks are as follows:
-------
E 8
Nonsmoker
(0.01 WLM-1) (0.006) (0.21 WLM a-1) (70a) » 0.0009
(0.0H WLM"1) (O.OOH3) (0.21 WLM a"1) (70a) - 0.0025
Using the increased baseline risks for ex-smokers and chronic smokers
described in the section regarding the ICRP model, the lifetime attributable
lu. oancer risk for ex-smokers is:
(0.01 WLM-1) (2) (0.006) (0.21 WLM a"1) (70a) ¦= 0.0013
(0.0H WLM-1) (2) (0.0043) (0.21 WLM a"1) (70a) = 0.0051
For chronic smokers:
(0.01 WLM-1 ) (5) (0.006) (0.21 WLM a-1) (70a) = 0.00U4
(0.01 WLM"1) (5) (0.0013) (0.21 WLM a"1) (70a) = 0.013
At a mean Indoor radon daughter concentration of '4 pCi I 1 the following
lifetime risks are calculated:
Nonsmoker:
(0.01 WLM"1) (0.006) (0.79 WLM a"1) (70a) = 0.0033
(0.01 WLM-1) (0.00H31 (0.79 WLM a"1) (70a) = 0.0095
Ex-smoker:
(0.01 WLM-1) (2) (0.006) (0.79 WLM a"1) (70a) = 0.0066
(0.01 WLM"1) (2) (0.00M3) (0.79 WLM a"1) (70a) = 0.019
-------
E 9
Chronic Smoker:
(0.01 WLM-1) (5) (0.006) (0.79 WLM a"1) (70a) =• 0.017
(0.0H WLM"1) (5) (0.00i»3) (0.79 WLM a"1) (70a) » 0.048
The EPA (USEPA86) based their risk estimates on a mean population lung
cancer risk. This risk Is heavily weighted by smoking related lung cancers.
Approximately 85? of lung cancers occur in smokers (USPHS86). Therefore,
while this method is appropriate for estimating the effect of indoor radon
exposure on the entire population, the resulting estimates should not be
interpreted as representing individual risks. Srsoking experience is the
overwhelming factor in determining an individual's lung cancer risk. Other
conditions such as diet, occupational exposures, air pollution and genetic
make-up may be also critical factors in lung cancer risk for some
indivJduals. Therefore, the projected excess lung cancer risk for an
individual 's subject to considerable uncertainty.
EPA also adjusts the radon daughter exposure to account for the
difference In breathing rates between occupationally exposed individuals,
principally miners, and Individuals exposed at home or in sedentary
occupations. The calculations included in this Appendix and shown in Table 6
use WLM without adjustment.
BEIR IV
The excess lung cancer risks due to Indoor radon exposure were estimated
from the risk ratios given in Table 2-1 of the BEIK IV Report (N'AS88). The
lifetime baseline lung cancer risks for smokers and nonsmoker3 were assumed to
be the mean of those for males and females. The baseline risk for ex-smokers
was assumed to be twice that for nonsmokers.
-------
E 10
In order to be consistent with the calculation of the risk estimates
baaed on the EPA model, the baseline risks were adjusted for average radon
exposure. (BEIrt IV estimates do not include this adjustment since It was
considered to be insignlfleant compared to other uncertainties Inherent in the
analysis.)
Estimated lifetime risk of lung cancer from BEIR IV (NAS88).
Males
Smokers 0.123
Nonsmokers 0.0112
Females
Smoker3 0.058
Nonsmokers 0.0060
The lifetime risks were adjusted for mean radon daughter exposure by
dividing by a factor of 1.18, the estimated lifetime risk ratio for exposure
to 0.2 WLM/yr. The estimated mean baseline lifetime risks for a mixed
population, 50$ male and 50$ female, reduced to account for average radon
daughter exposure are 0.077 for smokers and 0.0073 for nonsmokers.
The risk ratios, Re/R0> (lifetime lung cancer rl3k for exposed vs.
lifetime lung cancer risk for unexposed) estimated In BEIR IV (NAS88) are as
follows:
-------
E 11
&'.2 WLM/year 0.8 WLM/year
Smokers
Males 1.16 1.62
Femnles 1.18 1.69
Mean 1.17 1.66
Nonsmokers
Males 1.19 1.73
Females 1.18 1.73
Mean 1.19 1.73
Therefore, the excess lifetime risk at 0.2 WLM/year (1 pCi 8, ^) is as
follows:
Smokers
(0.17) (0.077) = 0.013
Nonsmokers
(0.19) (0.0073) = 0.001^
Ex-smokers
(0.19) (0.0073) (2) = 0.0028
at 0.8 WLM/yr (U pCi if1 ):
Smokers
(0.66) (0.077) - 0.051
Nonsmoker3
(0.73) (0.0073) = 0.0053
Ex-smokers
(0.73) (0.0072) (2) = 0.011
-------
E 12
The BEIR IV risk estimates are based on a multiplicative model. The
authors of BEIR VI (NAS88) found evidence in Colorado miner data to support a
3ubmultiplIcative interaction between smoking and radon daughter exposure but
the analysis was not sufficiently persuasive to abandon the nore conservative
multiplicative model.
-------
mm isiiiiii
- -V .-..*1 -H ¦' if.
/•'••.-.V • -•' ••- • ':"•<• \-v V*
I. .¦11:">':. - i '•'"•• ¦ '.'3
&'.r.'^-: ¦•=5-"" i.'-v:-vv-g
£ ..*•?.
e'j.ri: ::^v*?.-
• • •• . . . f . ' Y ..:
>rt-n
r:^~.
Ja
!:••'•>< --vj
k
¦ -:-u
• • ;';J
¦ri
,C ' - '<¦
¦n v^;-w'-^:.:v =* - '
'•.-'-;v;'~';-J.f•¦" '" -.*fi *•- *'
.yiJlHU "«l '-^ '•».- _".\i I - '•? ' " ", ¦ "¦¦',"¦"1
4®;
V , -.:
f'."'*v '- * ¦ "",'1' - r^
r-^-'-vs
4v;
vv vo- •',;;
:;V;
v 1
'r, _
, '
*¦ s
r-i -'
» ", <
vw* .
' '* *'* ,w 'I :/-V ' -•'.V
: - • 1' • ', .> v -\. ;;. i -£.¦ - ;, ,i, ¦, . , . ' 1 •-»•¦- -m v • "• i
:. :,.: ¦ •••<•.•.• .v- :;, •: • ~:.:\ -,•:. >,:.- v1-',:¦¦ ••:.•:..•*.<-V-d
•• /:- -.;v. -' • • '• • V:-:*•;<¦. - f-iv—: 'Q
~ •' 1 " •••• .''-, •..' ••-. • V,V- .J 4".-; v-;v-<^4v':.:>>;';:'L-- :-V:;'"J
' V'. - •—1¦¦'•'^.-! -\-"., 1.^'"-- '¦•>:•-/¦:"*• ''t*:> f1, f',.;.:'-:-'!V:'-^~-V '•$<
------- |