Depression as a risk factor for diabetes: a meta-analysis of longitudinal studies

The present meta-analysis aimed to assess the risk of incident diabetes associated with clinical depression, depressive symptoms, or both in nondiabetic subjects.

We performed a MEDLINE search for studies published in the English language using the search string diabetes AND (depression OR antidepressant). The search included studies from any date to December 30, 2011.

1,898 studies were independently assessed for eligibility, and longitudinal studies that assessed the risk of incident diabetes in subjects with or without clinical depression were selected.

Study design and characteristics were verified for each study. A meta-analysis was performed for unadjusted and adjusted risk ratios of incident diabetes in subjects with depression by using a random-effects model. Additional analyses were performed to assess heterogeneity, publication bias, and specific hazard ratios for diabetes associated with antidepressant drug use.

The 23 studies included in the meta-analysis enrolled 424,557 subjects, with a mean follow-up of 8.3 years and 19,977 cases of incident diabetes. A higher incidence of diabetes was found in depressed versus nondepressed subjects (0.72% vs 0.47% yearly), with unadjusted and adjusted risk (95% CI) of 1.56 (1.37-1.77) and 1.38 (1.23-1.55), respectively (both P values < .001). The use of antidepressant drugs and untreated depression were associated with an adjusted risk of diabetes of 1.68 (1.17-2.40) (P = .005) and 1.56 (0.92-2.65) (P = .09).

Depressive symptoms are associated with a significantly increased risk for incident diabetes. This association cannot be entirely explained by the use of antidepressant drugs or being overweight. Pathogenetic mechanisms connecting depression with diabetes deserve further exploration. Depression should be included among risk factors that indicate intensified screening for diabetes.