Predictors of relapse or recurrence in bipolar I disorder

Abstract

Objectives: Bipolar disorder represents a major public health concern and, despite treatment, is characterized by recurring episodes of mania, depression, or mixed states. Prevention of relapse or recurrence is a primary treatment objective in the management of the disorder. The objective of the current study was to identify predictors of relapse/recurrence in patients with bipolar I disorder treated with olanzapine, lithium, divalproex, or olanzapine plus divalproex/lithium.

Methods: Data from four clinical trials studying the efficacy of olanzapine compared to placebo and active comparators (lithium, divalproex, olanzapine plus divalproex/lithium) for bipolar I disorder were pooled for this analysis. Patients achieving remission after pharmacological treatment and entering randomized double-blind maintenance phase for 44 to 72 weeks were included. Cox Proportional Hazard models and Kaplan-Meier analyses were used to determine predictors of relapse/recurrence for the pooled data and within each treatment group.

Results: A total of 929 patients meeting the criteria for remission and followed by maintenance treatment were included in this analysis, and 427 patients (46.0%) experienced symptomatic relapse/recurrence during the follow-up period. A 21-item Hamilton Depression Rating Scale (HAMD-21) total score<4, gender, rapid cycling and treatment emerged as significant predictors of relapse/recurrence and may be generalized to treatment with olanzapine and to some extent to treatment with lithium and divalproex. The results on treatment-specific predictors of relapse/recurrence are considered to be exploratory and no adjustments were made for multiple comparisons.

Conclusion: The major findings from this study suggest that a HAMD-21 total score<4 may be a better predictor of maintenance of remission in bipolar I patients than HAMD-21 total score<8. The prophylactic effect of olanzapine, lithium, divalproex, olanzapine plus divalproex or lithium, and placebo was assessed and baseline predictors of relapse/recurrence were identified.