Interpretation of risk for low dose radiation exposures: inconsistencies, philosophical problems and fallacies

Low doses of radiation are used in diagnostic tests such as X-rays or CT exams. A Linear No Threshold model (LNT) is used to assess radiation risk. According to this model, risk decreases linearly with a decreasing dose, with no lower threshold. However, there is no definitive evidence to support risk at very low doses and the model relies on extrapolation to estimate risk at doses below a few 10 s mSv. The simplicity of the LNT model can disguise the more complex questions that arise when the facts are uncertain. There is a considerable debate over how best to interpret and communicate these risks for low dose exposures. This paper criticises arguments on both sides of this debate. In discussing the case for asserting risk we argue that it is only as a scientific model that LNT should be used to make predictions that can be tested. Taking the further step of using these calculations to state how many cancers are caused by the use of X-rays requires a strong scientific model and the disputed status of the LNT model and the consequent uncertainty as to whether low dose radiation definitely causes cancer, is the major obstacle to making these types of claims. In discussing arguments which tend to dismiss risk, we identify the hidden premises needed to support these claims. Differences between a prudent model and a scientific model are considered. A 'reductio ad absurdum' objection, and fallacies of ambiguity, where 'insignificant risk' may be confused with 'inconsequential risk', are also discussed. Disagreement over the interpretation of risk has the potential to disrupt the optimum use of medical imaging technology. While communication of risk need not address the underlying philosophical complexities, advice from the radiation protection community must be rooted in an awareness of these issues if inconsistencies are to be avoided.