Pooled Bayesian analysis of twenty-eight studies on radon induced lung cancers.

The influence of ionizing radiation of (222)Rn and its progeny on lung cancer risks that were published in 28 papers was re-analyzed using seven alternative dose-response models. The risks of incidence and mortality were studied in two ranges of low annual radiation dose: 0-70 mSv per year (391 Bq m(-3)) and 0-150 mSv per year (838 Bq m(-3)). Assumption-free Bayesian statistical methods were used. The analytical results demonstrate that the published incidence and mortality data do not show that radiation dose is associated with increased risk in this range of doses. This conclusion is based on the observation that the model assuming no dependence of the lung cancer induction on the radiation doses is at least ∼90 times more likely to be true than the other models tested, including the linear no-threshold (LNT) model.