Use of antidepressant agents and the incidence of type 2 diabetes mellitus : a methodological comparison
The main study purposes were to determine: whether antidepressant (AD) use increases the risk of type 2 diabetes mellitus; and whether results differ when using different methodological designs: retrospective cohort design and nested case-control design. A retrospective Texas Medicaid database analysis of new AD (exposed cohort) and benzodiazepine (unexposed cohort [BZ]) users from January 1, 2002 to December 31, 2009 was conducted. Patients aged 18-64 years without diabetes at cohort entry were included. The primary outcome was incident diabetes and the main independent variable was AD vs. BZ use. Covariates included age, gender, race/ethnicity, medication adherence, persistence, number of concomitant diabetogenic medications, Chronic Disease Score, treatment duration, year of cohort entry, and use of both AD and BZ at index. Regression analyses (adjusted) were used to address the study purposes. Of the study cohort (N=44,715), 35,552 (79.5%) were AD users and 9,163 (20.5%) were BZ users. Patients were followed for an average of 2.3±1.9 years (Median=1.8 years), were on average 38.6±14.2 years old, and 69.3% were female. Using the retrospective cohort design, AD use was associated with a 48.9% increase (logistic regression) and 60.0% increase (Cox regression) in the risk of diabetes compared to BZ use (logistic regression analysis: RR[subscript adj]. =1.489; 95% CI: 1.331-1.667; Cox regression analysis: HR[subscript adj]. =1.600; 95% CI: 1.437 - 1.783). Using a nested case-control design within the entire study cohort, AD use was associated with a 54.1% increase in the risk of diabetes compared to BZ use (OR[subscript adj]. =1.541; 95% CI: 1.368 - 1.735). Using a nested case-control design within the exposed cohort, current AD use was associated with a two-fold higher risk of diabetes compared to former AD use (OR[subscript adj]. =1.995; 95% CI: 1.759 - 2.264). Among antidepressant classes, TCAs, SSRIs, SNRIs, and Other ADs were associated with a higher diabetes risk compared with BZs. The results from the present study suggest that AD use is associated with an increased risk of diabetes. Clinicians may need to take this into account when choosing treatment for depression in patients at high risk of diabetes.