The 10-year follow-up study involved 4,681 men and women aged 65 and over who were free from diabetes at baseline.
Each year, the participants were evaluated for the presence of 10 symptoms of depression, including mood, irritability, calorie intake and concentration. Symptoms were scored on a scale of zero to 30, with scores of eight or higher indicating depression.
At the beginning of the study, the average depressive symptoms score was 4.5. A fifth of participants had a score of eight or higher. During the follow-up period, scores increased by at least five points in nearly half of participants, and 234 individuals developed diabetes.
The researchers found that rates of diabetes were higher in those with a depressive symptom score of eight or higher, who had a low score at baseline.
Lead researcher Dr Mercedes Carnethon, assistant professor of preventative medicine at Northwestern's Feinberg School of Medicine, Chicago, said that the mechanism for the association was unclear.
One possibility was that depressed people may have high levels of cortisol, which may decrease insulin sensitivity and increase fat deposits, said Dr Carnethon.
Arch Intern Med 2007; 167: 802-7