Participants included 20,744 men aged 30-64 years who first visited the general hospital in Tokyo for routine health assessments during 2003-2007. Untreated diabetes was defined as not being under diabetes treatment and fasting blood glucose of 猢?26 mg/dL or hemoglobin A1c of 猢?.5%. Sleep duration (猢?, 6, 7, or 猢? h) and confounders were assessed using a questionnaire. Logistic regression was used to calculate odds ratios (ORs) for diabetes after adjustment for confounders.
The prevalence of untreated diabetes was 3.4%in all men. The prevalence of untreated diabetes by sleep duration was as follows: 猢? h: 3.5%; 6 h: 3.3%; 7 h: 3.2%; and 猢? h: 5.1%. Men who reported sleeping for 猢? h (OR: 1.52, 95%confidential interval [CI]: 1.22-1.90) and 猢? h (OR: 1.39, 95%CI: 1.05-1.85) were significantly more likely to have untreated diabetes compared with those who reported sleeping for 7 h. In stratified analyses by obesity (body mass index 猢?5.0 kg/m2), the association between long sleep duration (猢? h) and untreated diabetes remained significant only in obese men.
Short sleep duration was significantly associated with untreated diabetes in both nonobese and obese men. Regarding long sleep duration, a significant association was observed in obese men only.