Background Previous studies have found that sickness absentees have an excess risk of premature death. Most of the studies compared sickness absentees with non-sickness absentees, with 1 report including the total population. However, no study to date has investigated in detail how the risk of premature death varies over the time course of a sick-leave spell.

Methods A population-based cohort of all 5.6 million people aged 16–64 years was followed up for 6 years. Hazard rates with 95% CI for mortality in the total population, sickness absentees with sick-leave spells exceeding 30 days and non-sickness absentees, were estimated over time and were standardised so that the distribution of background variables in all 3 groups was as close as possible to that in the total population.

Results The standardised hazard rate of premature death at the first day under risk for the sickness absentees was 13.58 (95% CI 13.26 to 13.95) deaths per 1000 person-years, 2.29 (2.24 to 2.35) in the total population and 0.90 (0.87 to 0.94) among non-sickness absentees. The standardised hazard rates of premature death at 6 years was 3.42 (3.22 to 3.62) among sickness absentees, 3.48 (3.43 to 3.52) in the total population and 2.61 (2.57 to 2.66) among the non-sickness absentees. The hazard rates varied greatly across sick-leave diagnoses.

Conclusions Right after start of a sick-leave spell exceeding 30 days, sickness absentees were at higher risk of premature death than the total population. The difference in risk, however, decreased over time and vanished completely after 6 years of follow-up.