Issues in risk modeling

The long-term follow-up of atomic bomb survivors conducted by the Radiation Effects Foundation (Life Span Study, LSS) has been the major source of information for risk estimates that have been used for setting up radiological protection standards by the International Commission on Radiological Protection (ICRP) and other organizations. This is mainly because various features of the LSS make risk estimates from this cohort study desirable, including the large population size consisting of relatively non-selected individuals of both sexes in a broad age range, systematic dosimetry providing individual dose estimates, and virtually complete mortality follow-up complemented by high-quality tumor registry-based cancer incidence data. There are several areas of uncertainty in estimating risk from the LSS data. The most important source of uncertainty at this time relates to the fact that a large proportion of the survivors are alive, thus necessitating projection beyond the observation. As of 1994, more than half of the subjects in the LSS are alive and about 85% of those exposed to the bomb early in life, e.g., age 30, are alive. Other sources of uncertainty include dosimetry errors, transfer of risk between populations, diagnostic errors, and extrapolation of risk from high dose to low dose and dose rates. For most of these, the magnitude of errors due to uncertainty can be estimated and corrected, while more effort is needed in some areas. In general, contribution of uncertainty from these sources seems relatively small, and current estimates seems justifiable. (author)