Design issues in epidemiologic studies of indoor exposure to Rn and risk of lung cancer.

Recent data on indoor air quality have indicated that Rn (222Rn) and its decay products are frequently present in domestic environments. Since studies of Rn-exposed miners have established that Rn decay products are a lung carcinogen, their presence in indoor air raises concerns about an increase in lung cancer risk for the general population. To directly evaluate lung cancer risk from domestic exposure to Rn and its decay products, as well as to evaluate risk assessments derived from studies of Rn-exposed underground miners, several epidemiologic studies of indoor Rn exposure have been initiated or are planned. This paper calculates sample sizes required for a hypothetical case-control study to address several important hypotheses and shows the impact of several difficult problems associated with estimating a subject's Rn exposure. We consider the effects of subject mobility, choice of the exposure response trend which is used to characterize an alternative hypothesis, and errors in the estimation of exposure. Imprecise estimation of Rn exposure arises from errors in the measurement device, exposure to Rn decay products from sources outside the home, inability to measure exposures over time in current as well as previous residences, and the unknown relationship between measured concentration and lung dose of alpha energy from the decay of Rn and its progeny. These methodological problems can result in large discrepancies between computed and actual study power. Failure to anticipate these problems in the design of a study can result in inaccurate estimates of power. We conclude that case-control studies of indoor Rn and lung cancer may require substantial numbers of subjects in order to address the many questions of importance that burden current risk assessments with uncertainty. We suggest pooling data from studies with the largest numbers of cases and with the most precise estimates of Rn exposure as the best approach for meeting present research needs.