Stochastic and nonstochastic concepts: is revision needed?

Stochastic effects of radiation, specifically cancers and hereditary defects, were defined in ICRP Publication 26 as having two constraints: No threshold in dose and no change in the severity of the induced effect with increasing dose. Nonstochastic effects were narrowly restricted to be the exact opposite, namely having a threshold and changing in severity with dose. Initially, these definitions seemed logical. It now appears, however, that no threshold may exist for the reduction of intelligence by irradiation of the fetal brain at critical ages. Otake et al. (in press) have shown, for Japanese A-bomb survivors irradiated at all fetal ages, that the average intelligence quotients decreased approximately linearly without threshold at 11 +/- 2 IQ points per Gy of fetal dose, based on either the T65D or the DS86 dosimetry systems. The decrease in intelligence quotient (IQ) was most severe in those irradiated at fetal ages of 8-25 wk. There is good reason to suspect that additional "nonstochastic" effects, such as cataracts, impaired fertility, and growth retardation might have a nonthreshold component to their dose-response at least in highly susceptible individuals. It was formerly believed that stochastic effects could be limited to acceptable levels, while nonstochastic effects could be completely prevented. However, to prevent entirely the special class of nonstochastic effects that have no true threshold, zero dose would be required. Zero dose is an unrealistic limit that would prevent most of the beneficial uses of radiation. Therefore, the realistic goal of radiation protection should be to limit all radiation effects to acceptable levels.