Background. The role of antidepressant medications in bipolar depression remains controversial, mainly due to a lack of research in this area. In this study the authors examined the episode length in bipolar depression and the relationship between antidepressant therapy and episode length.

Method. A retrospective chart review of 165 subjects identified 50 (30%) with bipolar illness who experienced a major depressive episode between 1 January 1998 and 15 December 2000. Data gathered utilized a structured instrument completed by the clinician at each visit. This instrument includes modified SCID mood modules as well as continuous ratings for each associated symptom of depression and mood elevation. Survival analysis was employed to calculate the median length of the depressive episodes for the entire group. Further survival analysis compared the episode length for subjects treated with antidepressants during the depression (N = 33) with those who did not receive antidepressants (N = 17). The rate of switch into elevated mood states was compared for the two groups.

Results. The survival analysis for the entire sample demonstrated 25%, 50% and 75% probability of recovery at 33 (S.E. 8.7), 66 (S.E. 17.9) and 215 (S.E. 109.9) days, respectively. Comparing those who received (N = 33) and those who did not receive (N = 17) antidepressants during the episode did not reveal any difference in the length of the depressive episode. Switch rates were not significantly different between those receiving antidepressants and those not taking these medications (15.2% v. 17.6%, respectively).

Conclusions. Over the past 20 years little progress has been made in reducing the length of depressive episodes in those with bipolar illness. This is despite increasing pharmacological options available for treating depression. Clinicians treating bipolar depression should discuss with their patients the likelihood that the episode will last between 2–3 months. Our results also suggest that antidepressant treatment may not reduce the length of depressive episodes, neither did it appear to contribute to affective switch in our sample.