Linear versus Non-linear: a Perspective from Health Physics and Radiobiology

There is a vigorous debate about whether or not there may be a "threshold" for radiation-induced adverse health effects. A linear-no threshold (LNT) model allows radiation protection practitioners to manage putative risk consistently, because different types of exposure, exposures at different times, and exposures to different organs may be summed. If we are to argue to regulators and the public that low doses are less dangerous than we presently assume, it is incumbent on us to prove this. The question is therefore whether any consonant body of evidence exists that the risk of low doses has been over-estimated. From the perspectives of both health physics and radiobiology, we conclude that the evidence for linearity at high doses (and arguably of fairly small total doses if delivered at high dose rate) is strong. For low doses (or in fact, even for fairly high doses) delivered at low dose rate, the evidence is much less compelling. Since statistical limitations at low doses are almost always going to prevent a definitive answer, one way or the other, from human data, we need a way out of this epistemological dilemma of "LNT or not LNT, that is the question". To our minds, the path forward is to exploit (1) radiobiological studies which address directly the question of what the dose and dose rate effectiveness factor is in actual human bodies exposed to low-level radiation, in concert with (2) epidemiological studies of human populations exposed to fairly high doses (to obtain statistical power) but where exposure was protracted over some years.