Background In epidemiological research, shift work systems are commonly assessed using self-reports. The aim of the current study was to evaluate the validity of generally used questions on shift systems by comparing them to objective data.

Material and methods We matched the working hour characteristics of 3 preceding months based on the pay-roll based registry data to the questionnaire date in 2014. The data comprised of all (n = 11,052 ) employees in a prospective cohort with a work contract during the questionnaire excluding only on-call workers. 81% (n = 8896) had at least 31 work shifts during the 3 months period.

Results Using objective data as the gold standard, questions on “shift work with night shifts” and “permanent night work” showed high sensitivity (i.e., the proportion of true shift/night workers that are correctly identified; 96% and 90%) and specificity (the proportion of true non-shift/night workers that are correctly identified, 92% and 97%). Self-reported “regular daywork” showed low sensitivity (73%), but high specificity (99%). “Shift work without night shifts” showed both low sensitivity (62%) and low specificity (87%). In analysis of associations between “shift work without night shifts” and health outcomes in the prospective data, age- and sex-adjusted odds ratios (multivariate logistic regression) were lower for subjective compared to objective assessment (e.g. for fatigue during free-time compared to day work 1.21, 95% CI: 0.78–1.87 versus 1.89, 95% CI: 1.06–3.35). Non-responding (n = 2156) to the questionnaire was not associated to the objective shift system but 55% of the dayworkers had at least one year of earlier shift work experience.

Conclusions These findings suggest that the validity of self-reported assessment of shift work varies depending on the shift system. Exposure misclassification was most common in self-reported shift work without night shifts and regular day work, contributing to bias towards the null in analyses of health effects.