Variations with time and age in the relative risks of solid cancer incidence after radiation exposure

The Japanese atomic bomb survivor cancer incidence dataset and data on five groups exposed to radiation for medical reasons in childhood are analysed and evidence found for a reduction in the radiation-induced relative risk of cancers other than leukaemia with increasing time since exposure and age at exposure. The rate of the reductions in relative risk with time since exposure are not significantly different for those exposed in childhood and for those exposed in adulthood, if adjustment is made for the effects on the relative risk of age at exposure. For those irradiated in childhood, there is a statistically significant annual reduction of 5.8% (95% CI 2.8, 8.9) in excess relative risk, and there are no strong indications of inter-cohort heterogeneity in the speed of reduction of relative risk. After adjustment for the effects of age at exposure, there is a significant annual reduction of 3.6% (95% CI 1.6, 5.6) in excess relative risk in all age-at-exposure groups. There are significant reductions of 5.2% (95% CI 3.7, 6.8) in excess relative risk per year of age at exposure. There are statistically significant ( P = 0.04) interactions between the exponential adjustments to the excess relative risk for age at exposure and time since exposure in the Japanese data, but no indications ( P = 0.38) of such interactions when powers of time since exposure and attained age are used to adjust the excess relative risk, so that the fit of the model with power adjustments is to be preferred to that of the model with exponential adjustments. There is no significant heterogeneity by cancer type within the Japanese atomic bomb survivor data in the variation of relative risk with time and age. There are indications that the model which assumes a variation in risk dependent only on attained age fits all the datasets analysed together significantly worse ( P = 0.11) than one which allows for different rates of change of risk as a function of time since exposure and age at exposure. The lack of fit of the attained-age model is much more striking in the Japanese bomb survivor incidence data ( P = 0.02), which has much the greatest power of any of the datasets analysed to differentiate time and age trends in excess risk. A variety of relative risk models are used to evaluate solid cancer mortality and incidence risks for a current UK population. High dose-rate solid cancer mortality risks are estimated to lie in the range 6.9% - 13.8% Sv-1. High dose-rate solid cancer incidence risks are calculated to lie in the range 12.1% - 22.0% Sv-1. The preferred generalised relative risk model, with adjustments to the excess relative risk proportional to a product of powers of time since exposure and attained age, predicts solid cancer mortality risks of 9.0% Sv-1 and solid cancer incidence risks of 15.0% Sv-1.