Substance use disorder and other predictors of antidepressant-induced mania: a retrospective chart review.

Objective: To determine if substance use disorder (SUD) is a predictor of antidepressant-induced mania (ADM) in bipolar disorder, correcting for confounding factors in a regression model.
Method: 335 antidepressant trials were identified in 98 patients treated in an academic bipolar specialty clinic from 2000 to 2004. Patient charts were reviewed, and histories of SUD and ADM (primary outcome; defined as a hypomanic or manic episode within 12 weeks of beginning an antidepressant trial) were identified. Mood disorder diagnoses were made using the Structured Clinical Interview for DSM-IV mood module, and SUD diagnoses were defined using DSM-IV criteria. Potential confounding variables were also examined and included in a multivariable regression model. Concomitant mood stabilizer, antimanic, and antidepressant use was adjusted for in the regression model.
Results: In univariate analyses, there was no evidence of an association between ADM and past SUD. However, after adjustment for confounding variables in a multivariable regression model, there was a strong relationship (OR = 5.06, 95% CI = 1.31 to 19.64, p < .05). Other statistically significant predictors of ADM in the regression model were type II subtype of bipolar illness, female gender, and tricyclic antidepressant (TCA) use (vs. bupropion).
Conclusions: Along with other factors, a history of SUD was a strong predictor of ADM. Possible underestimation of ADM in randomized clinical trials may occur due to the exclusion of subjects with SUD. Type II illness, female gender, and TCA use also appeared to be predictors of ADM, while bupropion use appeared to predict lower likelihood of ADM.