The radon issue is extremely important as Ruth
Weiner stated and it is worth to be discussed. This is true especially when the
Iowa Radon Study is praised by EPA for its quality and as a landmark study by
Don Smith and therefore used to refute COHEN's and other results from radon
studies not in accordance with LNT.
FIELD claims somewhat presumptuous that his study
"...incorporated a unique combination of study design and enhanced dosimetric
techniques....". It is not disputed that great efforts where made to
measure smoking history and radon exposure as precise as
possible.
Before discussing the influence of smoking and
its control in the Iowa Study a question regarding a possible selection
bias has to be raised. The study started with 1,974 female lung
cancer cases and included in the end 283 cases ALIVE = 14,3% of all cases
and 47% cases that met all 4 eligible
criteria.
As BEIR VI stated, the lung cancer risk from
smoking amounts to 10-20 and that from indoor radon to 0,2-0,3 only. When
case-control studies are discussed, BEIR VI stated: " The detection of an excess risk of lung cancer is potentially
complicated also by an inability to control completely for other lung-cancer
risk faktors, paticularly cigarrette-smoking, which has an RR (relative risk) of
10-20". Due to the overwhelming risk level for lung cancer from smoking, the
contribution of each single factor has to be determined, when radon studies are
conducted with mostly smokers among cases and controls. A strong confounder as
smoking can be fully controlled in individual studies without error when
correctly measured. FIELD measured the influence from smoking with an elaborated
questionnaire precisely. But what he measured precisely was what cases and
next of kind reported.
Even with an underreporting of one cigarette per
day, the relative error with the estimation of the lung cancer risk from
radon comprises 13%. Underreporting of cigarettes smoked results always in a
higher risk estimate for radon. Lung cancer cases tend to minimise their
possible own contribution to their deplorable fate by reducing cigarettes smoked
and even deny their habit (CONRADY et al. 1999). This is common for all
life-style related diseases. As further examples see PEREZ-STABLE et al. 1990
and OFFER et al. 2000.
FIELD could not establish a health hazard
from radon when ALL cases (413) where analysed (Table 4) because the odds ratios
where not significant. Only when a subgroup with only 283 cases ALIVE was
used, a significant result for one exposure category from four could
be established. No explanation has been given why less ALIVE cases (48) in
the highest exposure category deliver a significant result versus 67 cases for
ALL with a not significant result.
What could be a possible explanation for this
surprising change? The just diagnosed ALIVE cases were interviewed.
They were stressed by the evidence and feeling of guilt for their possible
contribution to their lung cancer by smoking so they minimise their smoking
habit and even deny it (smokers became non-smokers). This results in a higher
risk for lung cancer from radon (OR 2,14). Possibly the next of kin reported
more accurately about the smoking history of the cases. This results in a
smaller risk estimate from radon (OR 1,79). Additionally it is impossible to
report the true amount of cigarrettes smoked over a period of 20 to 30
years because such a strong confounder invalidates any risk
estimate for lung cancer from the weak cause of indoor radon when only one
cigarrette per day is underreported.
When power is discussed, FIELD avoided any
estimates. This is easily understood when from own estimates for ALL cases
(mobility included) a power of 40,1% and for ALIVE 50,8%
resulted. For a "landmark study" besides significant results with narrow
confidence intervals a power >90% is needed. There is no case-control
study yet that fulfills these statistical
criteria for the quality of a study.
Conclusion: Only studies with highly exposed
validated non-smokers with a wide range of exposure from 50 Bq/m³ to > 3.000
Bq/m³ can result in valid risk estimates from indoor radon for lung cancer.
Thuch studies are possible in regions with high indoor radon exposure included
in a cancer registry (for example the Schneeberg region in Germany, CONRADY et
al. 1999). The pooling of biased data is not the solution. When biased data,
especially due to data from smoking history are used, the result would be
biased, too even when some thousend of cases are added up in such a
study.
References:
Conrady J, Martin K, Poffijn A, Tirmarche M,
Lembcke J, Do Minh Thai, Martin H: High residential radon health effects in
Saxony (Schneeberg Study). Contract N° FI4P-CT95-0027, European Commission, DG
XII, Nuclear Fission Safety Programme. PreCura Institute for Preventive
Medicine, Schneeberg, August 1999 (www.precura.de)
Perez-Stable EJ, Marin BV, Marin G, Brody DJ,
Bennowitz NL: Apparent underreporting of cigarette consumption among Mexican
American smokers. Am J Public Health 80 (1990), pp 1057-1061
Offer D, Kaiz M, Kenneth I, Bennet H, Bennet E: The
altering of reported experiences. J Am Acd Child Adolesc Psychiatry 39(6), 2000,
pp 735-742
|