The risk of leukemia from low doses of low-LET radiation

In this communication, we examine the evidence (if any) for a nonlinear dose response in relation to leukemia mortality in the Japanese A-bomb population. Specifically, we seek an estimate of the probability that, at low doses of radiation, the relative risk (RR) is smaller than one. Using Bayesian bootstrap techniques, we find that there is a 90% probability that—for at least one dose group at or below 200 mSv—there is a reduction of effect relative to the control group. We take this as evidence for radiation hormesis (end point: leukemia mortality) at low doses of radiation.