Background Shift work has been linked with the increased risk of major chronic diseases including breast cancer and cardiovascular diseases, but less is known how shift work associates with sickness absence (SA). So far the use of subjectively assessed work schedules have not permitted detailed evaluation of shift work characteristics.

Aim To investigate if incident SA spells can be predicted by preceding shift work characteristics. In specific, we investigated if the seven days preceding and following a SA spell regardless of length were different in terms of shift work characteristics.

Sample and methods The pay-roll based data of working hour characteristics (from 2008 to 2013) was based on employers’ records of six hospitals of the Finnish Public Sector study. The sample (n = 12747) was restricted to those with shift work and at least one SA spell (length ≥ 1 day). The statistical analyses included testing the difference of the seven days preceding and following a SA spell by t-test for weekly working hours,% of night work and% of <11 h shift intervals, and by chi2 tests for the timing of SA in terms of weekday and shift type.

Results Weekly working hours preceding SA were higher than following SA (mean difference 1.8 h, p < 0.001), but the% of night work (−0.3%) and% of <11 h shift intervals (−11%) were lower (p < 0.001). The comparison of weekdays and work shifts preceding and following SA indicated differences only in weekdays: Thursday preceded SA most frequently (19%). The older 10-year age groups (+40 years) differed from those younger, but no sex differences were detected.

Conclusions These preliminary results suggest that overall work load in terms of weekly working hours might predict SA. However, effect of the working hour characteristics in shift work for SA will be analysed in detail for the conference presentation.