Few prospective cohort studies describe the risk of type 2 diabetes mellitus associated with depression or anxiety. We aimed to determine the 2-year diabetes incidence and pattern of explanatory factors associated with depressive and/or anxiety disorders.

A prospective cohort of 2,981 participants (aged 18–65 years, 66% women) recruited in the NESDA from community, primary care and outpatient psychiatric clinics were followed-up for two years. Complete data were analyzed from 2,460 participants without baseline diabetes. Lifetime or current (past 6-month) depressive and/or anxiety disorders at baseline were assessed using the CIDI and classified by the DSM-IV. Diabetes was classified by either self-report, medications, or fasting plasma glucose ≥7.0mmol/L. Baseline covariates included age, gender, lifestyle factors, and medical conditions. Odds ratios (OR[95% confidence intervals]) for diabetes were determined using exact logistic regression.

The unadjusted 2-year diabetes incidence was 0.2% (1/571), 1.1% (6/548), and 1.8% (24/1,340) for no, remitted, and current depressive and/or anxiety disorders, respectively. In comparison to controls, current depressive and/or anxiety disorders was associated with diabetes incidence in unadjusted (OR 10.4[1.7,429.0]) and age-adjusted (OR 11.9[1.9,423.0]) analyses. The strength of this association (β) was slightly changed after further adjustments for impaired fasting glucose (11.4%), high triglycerides (−7.8%), and lifestyle cumulative risk score (−5.0%), in contrast to other covariates when assessed in separate models.

The relative odds of developing diabetes within two years was increased for persons with current depressive and/or anxiety disorders, which was partially explained by, but remained independent of, lifestyle cumulative risk factors.