Background We investigated whether changes in alcohol use predict changes in the risk of sickness absence in a case-crossover design.

Methods Finnish public sector employees were surveyed in 2000, 2004 and 2008 on alcohol use and covariates. Heavy drinking was defined as either a weekly intake that exceeded recommendations (12 units for women; 23 for men) or having an extreme drinking session. The responses were linked to national sickness absence registers. We analysed the within-person relative risk of change in the risk of sickness absence in relation to change in drinking. Case period refers to being sickness absent within 1 year of the survey and control period refers to not being sickness absent within 1 year of the survey.

Results Periods of heavy drinking were associated with increased odds of self-certified short-term (1–3 days) sickness absence (multivariable-adjusted OR 1.21, 95% CI 1.07 to 1.38 for all participants; 1.62, 95% CI 1.19 to 2.21 for men and 1.15, 95% CI 1.00 to 1.33 for women). A higher risk of short-term sickness absence was also observed after increase in drinking (OR=1.27, 95% CI 1.07 to 1.52) and a lower risk was observed after decrease in drinking (OR=0.83, 95% CI 0.69 to 1.00). Both increase (OR=1.38, 95% CI 1.21 to 1.57) and decrease (OR=1.27, 95% CI 1.19 to 1.43) in drinking were associated with increased risk of long-term (>9 days) medically certified all-cause sickness absence.

Conclusion Increase in drinking was related to increases in short-term and long-term sickness absences. Men and employees with a low socioeconomic position in particular seemed to be at risk.