Predictability of the national psychological stress screening for subsequent long-term psychiatric sick leave among employees: a multi-center nested case-control study

Objective The aim of the study is to predict employees’ long-term sick leave due to psychiatric disorders using the national psychological stress screening program. Methods University employees who took long-term psychiatric sick leave in 2016–2018 were assigned as cases. Those who were present at work and matched for sex, age, and occupation type were assigned as controls. Answers in a 57-item questionnaire were analyzed by multivariable regression, and a prediction model was developed. It was validated in cases and matched controls in 2019. Results Six items were identified as independent predictors by multivariable regression and included in a prediction model. The area under the receiver-operating characteristics curve was 0.768 (95% confidence interval: 0.723–0.813). This finding was similar to that in the validation sample. Conclusions The performance of the prediction model was modest and the national Stress Check Program should be further refined.