Working hour patterns and risk of occupational accidents. An optimal matching analysis in a hospital employee cohort

•Utilization of working hour records for studies of irregular working hours and health- or safety-related consequences has increased recently.•Optimal matching with cluster analysis enables to capture the patterns of working hours in irregular shift work for a data-driven examination of the associations between the patterns and safety-related outcomes, such as occupational accidents.•Data-driven analysis identified six different patterns of weekly working hours.•Working late and long with insufficient rest during a period of a week was associated with increased probability of occupational accident among hospital employees.•To assure safe working conditions, regularity and sufficient rest should be considered in planning of work shifts for employees in 24/7 care of hospitals.