Association between Sleep Duration and Incident Diabetes Mellitus in Healthy Subjects: A 14-Year Longitudinal Cohort Study

Background: This study aimed to investigate whether sleep duration and/or quality are associated with incident diabetes mellitus (DM). Methods: A total of 8816 of 10,030 healthy participants were enrolled in a prospective cohort study. Sleep duration and quality questionnaires were completed. Sleep quality was assessed using the Epworth Sleepiness Scale (ESS), which measures excessive daytime sleepiness in individuals. Results: During the 14-year follow-up period, 18% (1630/8816) were diagnosed with DM. A U-shaped relationship was observed between sleep duration and incident DM, with the highest risk observed when sleep duration was >= 10 h/day (hazard ratios (HR) 1.65 [1.25-2.17]). This group exhibited decreased insulin glycogenic index, a marker of insulin secretory function, during the study period. Among study participants who slept less than 10 h/day, the risk of incident DM increased when the ESS score was >10. Conclusions: We found that the association between sleep duration and incident DM was U-shaped; both short (<= 5 h) and long (>= 10 h) sleep durations were associated with an increased risk for the occurrence of incident DM. When sleep duration was 10 h or longer per day, there was a tendency to develop DM due to decreased insulin secretory function.