Cigarette use and the estimation of lung cancer attributable to radon in the United States.

Residential exposure to radioactive radon and its decay products has been estimated to account for 10-12% of all lung cancer deaths in the U.S. It has been difficult to evaluate fully the impact of cigarette smoking, the most important cause of lung cancer, on this estimate, because factors for patterns of tobacco use have not been included in the risk models, since risk models are derived from studies of underground miners exposed to radon and detailed data on smoking are limited. Lung cancer risk estimates for exposure to radon progeny in smoker and non-smoker populations are obtained by applying the same risk model to each population group, thereby assuming the joint effects of smoking and exposure to radon progeny are multiplicative. However, in miners, joint relative risks (RR) for the two exposures are most consistent with an intermediate relationship between multiplicative and additive, so that the present approach likely results in an overestimate of risk in smokers and an underestimate of risk in nonsmokers. Lubin et al. (National Institutes of Health Publication No. 94-3644, 1994) present an ad hoc approach for adjusting risk models to incorporate smoking status. The approach is based on the relative magnitude of the effects of radon progeny in smokers and nonsmokers and therefore may not be applicable to non-miner populations if the proportion of smokers and the RR for smoking differ. We show that the modification can be derived explicitly by assuming an arithmetic mixture model for the joint RR for smoking and exposure to radon progeny. In this way, smoking parameters in the population of interest (the proportion of smokers and the RR of smoking) can be used directly to adjust radon progeny risk models and obtain risk estimates that are specific for smokers and nonsmokers. With an intermediate RR relationship for smoking and radon progeny, the attributable percentage of lung cancer deaths from residential radon may be twofold greater in non-smokers than in smokers.