Yuan and colleagues (1) conducted a systematic review and meta-analysis of studies that have investigated the association between night shift work and the risk of cancer in females. Evidence about shift work and cancer risk is highly inconsistent, so a systematic review and meta-analysis on this topic has the potential to bring some clarity to what is an important public health question.
Systematic reviews and meta-analysis are widely perceived to be at the top of the evidence pyramid and have high prestige and influence (2). It is therefore vital they are conducted appropriately. An important step is to ensure only studies that are directly relevant to the research question are included. Unfortunately, we believe this is not the case in the review by Yuan and colleagues (1). After careful review, we believe more than half (31) of the 61 articles listed in Table 1 should not have been included in this review. Specifically:
Many of the included studies did not assess the exposure of shift work. Nine studies investigated sleep duration/disruption [refs. 46, 59, 64, 70, 73–76, and 80 in Yuan and colleagues' work (1)], two studies investigated light at night (refs. 43 and 58), and six studies investigated specific occupations (refs. 36, 60–63, and 77). Although these exposures are arguably related to shift work and/or have been proposed as possible mechanisms through which shift work may influence cancer risk (3), they are not equivalent to being “exposed to shift work” (1; p. 26) and Yuan and colleagues (1) provide no rationale for why they should be.
In three studies, the outcome was not cancer risk (adherence to cancer screening in ref. 71; cancer mortality in refs. 28 and 34).
Risk estimates from seven meeting abstracts are included (refs. 32, 35, 65, 66, 68, 72, and 82), despite “Meeting abstracts, reviews, and letters” being one of the reasons for study exclusion in Figure 1 (1). One of these abstracts (ref. 82) has been withdrawn from publication.
In several instances, multiple risk estimates from the same study population are included in the same cancer-specific meta-analysis (e.g., refs. 35 and 53; refs. 38, 39, 67, and 68; refs. 41 and 51; refs. 50 and 78; and refs. 56 and 70).
Exclusion of these studies would have a major impact on the results. For example, only one of the five studies in the skin cancer meta-analysis would remain. As such, we believe this review does not accurately synthesize the evidence about night shift work and cancer risk, and that the results and conclusions are inaccurate and misleading.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Acknowledgments
T. Boyle is supported by a Fellowship from the Australian National Health and Medical Research Council (#1072266). L. Fritschi is supported by Fellowships from the Australian National Health and Medical Research Council (#37614900) and Cancer Council Western Australia.