Seasonal patterns of long sickness absence due to 411 diagnostic groups: a nationwide register-based study in Finland during 2020-2023.

Aim: Seasonal patterns in sickness absence (SA) are little studied but crucially important to understand in order to design preventative measures and allocate resources. We aimed to identify seasonal patterns in long-term SA, that is, absences longer than 10 working days, due to different diagnostic groups.

Method: Long-term SA recipients on a monthly basis from 2020 through 2023 were analyzed (2,257,011 long-term SA recipients in total). Monthly relative deviations from the expected SA recipient numbers given no seasonality were calculated for each diagnostic group defined by digits of ICD-10 codes. The seasonal deviations in 411 different diagnoses were used as input in an unsupervised learning method, the K-means clustering algorithm, to identify specific diagnoses susceptible to seasonal variation.

Results: The number of long-term SA recipients was lowest in the summer, and reached three peaks in February-March, October, and December. We identified three seasonal patterns by diagnostic group. A winter and autumn peaks cluster (6% of SA recipients) consisted of 42 diagnostic groups, such as sleep disorders. A spring high cluster (81%) included mainly mental and musculoskeletal diagnoses. An autumn high cluster (13%) consisted of a mixed set of 262 diagnostic groups, including stress-, injury-, and musculoskeletal disorder-related diagnoses. These clusters differed in terms of the age and gender of the recipients.

Conclusions: There is substantial potential to reduce SA by addressing its seasonal determinants. The identified patterns could be used to design the optimal provision of preventative measures throughout the calendar year in health policies, occupational health care, and workplaces.