This study aimed to clarify the relationship between serum triglyceride levels (fasting or non-fasting) and incidence of diabetes in a general Japanese population.

We conducted a population-based retrospective cohort study using annual health check-up data of residents of Iki City, Nagasaki Prefecture, Japan. A total of 5632 participants without diabetes at baseline were included in the present analysis. Serum triglyceride levels were classified as tertile1 (≤0.94mmol/L), tertile 2 (0.94–1.51 mmol/L) and tertile 3 (≥1.51 mmol/L) for men, and tertile1 (≤0.84 mmol/L), tertile 2 (0.84–1.28 mmol/L) and tertile 3 (≥1.28 mmol/L) for women. The outcome was incidence of diabetes.

During an average follow-up of 5.73 years, 342 participants (194 men and 148 women) developed diabetes. The incidence (per 1000 person-years) of diabetes increased with an elevation in serum triglyceride levels for men (9.70% in tertile1, 12.0% in tertile2, and 20.3% in tertile 3). This association was significant, even after adjustment for other risk factors: hazard ratio 1.06 [95% CI 0.68–1.65] for tertile2, and 1.56 [1.01–2.42] for tertile3 compared with tertile1 (p = 0.029 for trend). Comparable effects were observed for fasting and non-fasting serum triglyceride levels (p = 0.406 for interaction). In contrast, there was no clear association between serum triglyceride levels and incidence of diabetes after adjustment for confounding factors among women (p = 0.111 for trend) although interaction between men and women was not statistically significant (p = 0.631 for interaction).

High levels of serum triglyceride levels (fasting or non-fasting) were associated with future incidence of diabetes mellitus in general Japanese men but not in women.