On the estimation of radiation-induced cancer risks from very low doses of radiation and how to communicate these risks.

The article is intended to give a short overview of epidemiological data on cancer risks associated with very low absorbed doses of ionising radiation. The linear no-threshold (LNT) approach to estimate cancer risks involves the use of epidemiological data at higher doses (>100 mSv), but is supported by data from lower exposure of more sensitive population groups like fetuses and children and the presence of rare types of cancer. The International Commission on Radiological Protection (ICRP) concludes that the LNT model, combined with a dose and dose-rate effectiveness (reduction) factor (DDREF) of 2 for extrapolation from high doses, should be used. The numerical value of the DDREF is challenged by the findings from some recent epidemiological studies demonstrating risks per unit dose compatible with the risks observed in the higher dose studies. In general there is very limited knowledge about the cancer risk after low absorbed doses (10-100 mSv), as most of epidemiological studies have limitations in detecting small excess risks arising from low doses of radiation against fluctuations in the influence of background risk factors. Even if there may be significant deviations from linearity in the relevant dose range 0-100 mSv, one does not know the magnitude or even the direction of any such deviations. The risks could be lower than those predicted by a linear extrapolation, but they could also be higher. Until more results concerning the effects of low-dose exposure are available, a reasonable radiation protection approach is to consider the risk proportional to the dose.