Long working hours can be toxic.

There are complex interconnections of variables such as work hours, occupation type, and socioeconomic status with physiology that can aff ect the risk of diabetes. Results of investigations into associations between long working hours and incident type 2 diabetes have been inconsistent. In The Lancet Diabetes & Endocrinology, Mika Kivimäki and colleagues combine published and unpublished data in a meta-analysis (n=222 120) to show that working long hours is associated with increased risk of incident diabetes in people with low socioeconomic status. 1 Their study design and methods have several advantages: large sample size, consistent exposure defi nition, control for health behaviours and other potential confounders, and analysis to test for bias due to the method used to determine diabetes incidence. Importantly, the investigators enable clear comparisons with previous reports by presenting serially adjusted analyses largely replicating previous fi ndings. However, the investigators' discussions of residual confounders focused primarily on the possibility that confounding might explain the higher diabetes risk in the low socioeconomic groups; whereas, residual confounding by protective factors could lead to an underestimation of the eff ect of the exposure in the workers in the high socioeconomic status group. Workers in this group have additional capital, social, or other resources at their disposal that might mitigate the eff ects of long working hours. For example, they are able to contract out activities such as child care and household chores, thereby buying more weekly hours. Unresolved issues of measurement and residual confounding make us guarded about the study's conclusions, especially the assertion that the lack of an association in workers in the high socioeconomic status group negates the possibility that working long hours per se is toxic. In occupational health studies, work hours are an essential component of exposure assessments. Fundamentally, the acuity of the work exposure above and beyond the duration of habitual paid work hours (eg, intensity of physical or mental work, schedule fl exibility, and commute burden) is often not measured in epidemiological studies. 2 Unmeasured diff erences in how workers of diff erent socioeconomic groupings achieve total weekly hours might aff ect the risk of diabetes. Reviews with national samples show that workers in the high socioeconomic status group have greater schedule control, meaning that they are able to spread their hours throughout the week. 3 Meanwhile, workers in lower socioeconomic status groups might achieve the same number of hours by working overtime, 3 …