General Commentary: Protocols for Epidemiological Studies of Radiation Workers

Colin R Muirhead^*

• Independent researcher, Newcastle upon Tyne, United Kingdom

There is a large literature on epidemiological studies of radiation workers, with hundreds of studies published over the past century, dating back to the first studies of radiologists (1). More recent studies have looked at the health of workers in the nuclear industry, such as the INWORKS (2) and Mayak worker studies (3,4), as well as workers in medical professions (e.g. ( 5)) and other occupations (1). Some of these studies were based around protocols that were published at the outset of the research; for example, a prospective cohort study of radiation workers in Korea (6). There are many benefits to researchers, funders and especially the workers themselves of having a clear idea at the outset of what the proposed study involves. Consequently, protocols such as those of Qin et al (7) for their study of radiation workers in Chongqing (China) are to be welcomed. Nevertheless, there are some important points to bear in mind when preparing these protocols.First, it is valuable to refer to internationally agreed guidelines for standard protocol items in observational studies. As part of the EQUATOR network for enhancing the quality and transparency of health research (https://www.equator-network.org/), the SPIROS guidelines aim to facilitate the drafting of high-quality observational study protocols (https://www.maastrichtuniversity.nl/spiros-2024-responsible-reporting). Complying with guidelines of this type might be seen by some as simply a 'tick-box' enterprise. I disagree; I believe that these guidelines can help researchers think more deeply about some critical aspects of study design. Qin et al have addressed many of the standard protocol items on the SPIROS checklist in their protocol, including the study design and setting, inclusion and exclusion criteria, a sample size calculation, ethical approvals and consent. However, it is unclear how potential confounding 37 will be For example, when assessing possible associations between 38 radiation and loss, workers' to noise would be a key 39 consideration. of the feasibility study might well be very limited. 53Thirdly, there needs to be to be a clear and robust basis for any sample size calculation. Sadly, I believe that this is a key deficiency of Qin et al's protocol. They have based their 55 sample size calculation on the 'average white blood cell count for the population in the 56Chinese radiation group', based on a meta-analysis of a few hundred Chinese radiation 57 workers (8). However, there is no discussion as to whether the radiation doses received 58 by these workers would be similar to those received by future workers in the Chongqing 59 study. Also, Qin et al's calculation is based on simply looking for a difference between 60 groups in white blood cells, whereas their aim is to examine exposure-lag relationships. 61 This is a much more complex task, as demonstrated -for example -in an analysis of 62German uranium miners (9) that workers receiving less than 20 mSv per year would be unlikely to receive skin 91 injuries or incur hearing loss as a consequence of their radiation exposure (11). 92Qin et al are to be congratulated for publishing the protocol for their study. However, it 93 is vitally important to take adequate account of internationally agreed guidelines and 94 findings from previous research when designing a study such as this. In this way, 95 epidemiological studies can play a valuable role in protecting the health of workers. 96