Introduction It has been argued that the relationship between depression and diabetes is bi-directional, but this hypothesis has not been explicitly tested. We conducted two separate cohort studies, based on Taiwan's National Health Insurance claims, to determine the bi-directional prospective relationships between depression and type 2 diabetes.

Methods The first cohort analysis identified all 390 011 diabetic patients in 2000 and the same number of randomly selected non-diabetic beneficiaries. The second cohort analysis identified 5847 depressive patients and a random sample of non-depressive beneficiaries of the same number in 2000. The subsequent information on incident depression and diabetes was retrieved from ambulatory cares from 2000 to 2006. We evaluated the age-and sex-specific relative hazards of depression/diabetes in relation to diabetes/depression with Cox proportional hazard regression model adjusted for potential confounders.

Results The first cohort analysis noted a covariate adjusted HR of 1.43 (95% CI 1.38 to 1.48) for incident depression among diabetes. The second cohort analysis noted that the depressive patients experienced significantly elevated HR (2.02, 95% CI 1.80 to 2.27) for incident diabetes.

Conclusion The two cohort studies provided support for the bi-directional prospective relationships between diabetes and depression, with a stronger association noted for the depression predicting onset of diabetes. We also noted that the bi-directional relationships were most obvious in younger (<35 years) patients, regardless of gender.