Models for the analysis of radon-exposed populations.

Radon-222 is a radioactive decay product of radium-226 and uranium-238, which are found throughout the crust of the earth. Studies of underground miners clearly show that exposure to radon and its decay products increases the risk of developing lung cancer. Data on standardized mortality ratios from eight cohort studies indicate that the radon-lung cancer relationship is statistically homogeneous, even though cohorts are from different types of mines and from different countries. Regression methods for cohort data based on a Poisson probability model permit a thorough consideration of risk patterns. In this report, we review these methods, wherein the disease rate in each cell of a multi-way table is modeled as a function of the cross-classifying variables. The National Academy of Sciences' Committee on the Biological Effects of Ionizing Radiation uses the Poisson regression approach to develop a model for age-specific lung cancer risk which depends on cumulative exposure, age at risk, and time since exposure. This model is reviewed and its implications discussed. The most important determinant of lung cancer is cigarette smoking. This paper discusses relative risk models for analysis of joint exposure to radon and tobacco products. The review of available studies suggests that the joint relationship of radon and smoking with lung cancer is consistent with a multiplicative model, but a submultiplicative relationship is most likely. An additive model is rejected.