Background Epidemiological studies of shift work are often biassed by exposure misclassification and lack of sufficient control for time-invariant factors. We utilised hospital employees’ 6-year follow-up from 2008 to 2014 to analyse whether changes in sleep and fatigue were related with exposure to shift work.

Material and methods We linked questionnaire data on sleep and fatigue in 2008, 2012 and 2014 (n = 11 133) to pay-roll -based records of objective working hour data from previous 3 months to analyse whether changes in sleep and fatigue were related with exposure to shift work compared to day work (n = 925). We used fixed effects time-dependent logistic modelling to analyse records of shift work in relation to changes in sleep length, insomnia (Jenkins scale) and fatigue.

Results An increase in fatigue during work and free-time from 1–3 days/month to 2 days a week was associated with exposure to shift work (OR 1.34, 95% CI: 1.01–1.74 and OR 1.45, 95% CI: 1.07–1.96, respectively) compared to day work. Increase of insomnia (but not problems to fall asleep) according to the Jenkins scale was less likely related with exposure to shift work (OR 0.72, 95% CI: 0.56–0.92). The shortening of sleep length was related with exposure to shift work only among the <40 age group (OR 1.92, 95% CI: 1.02–3.62 for a change from ≥8-hours to 7–7.5 hours) but not among the older age groups (40–49 year old group: OR 0.40, 95% CI: 0.16–0.98).

Conclusions The results indicate that in relation to changes between day- and shift work, preceding exposure to shift work is related to increased fatigue but not to increased insomnia.