Twenty-first annual Warren K. Sinclair keynote address: challenges faced by studies of nuclear industry workers ^*

Epidemiological studies of large groups of nuclear industry workers offer a significant opportunity to increase our understanding of the long-term effects on health of the protracted accumulation of dose received at a low dose rate from many discrete exposures to ionising radiation. The effects of such extended aggregation of doses form an important part of the everyday concerns of radiological protection against low-level exposures. For more than half a century, databases of nuclear workers have been assembled, and the numbers of workers currently included in studies, together with the numbers of deaths among them that have now occurred, are capable of generating reasonably precise risk estimates that should provide meaningful comparisons with those obtained from other studies, such as of the Japanese atomic bomb survivors. However, constructing, updating and maintaining these large databases, linking workers to occupational dose databases and to registers of deaths (and other outcomes, such as incident cancers) and their causes, and analyses of these data that endeavour to take account of other influential factors (such as smoking) are far from straightforward. A critical review of recent nuclear worker studies illustrates the difficulties in reliably interpreting reported statistical associations between rates of cancer and cumulative occupational doses because of the real possibility of distortions produced by biases, confounding and/or the interplay of radiation with other risk factors. This does not mean that studies of nuclear workers should be abandoned, far from it, but it does mean that appropriate effort needs to be expended on these studies before confident conclusions about the levels of risks from radiation exposure can be drawn from them. Unexpected findings should be examined in depth to gain a proper understanding of their origin, including the impact of doses from intakes of radionuclides upon dose-responses derived using doses from external sources of radiation.