Predictors of relapse or recurrence in bipolar I disorder

ObjectivesBipolar 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.MethodsData 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.ResultsA 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.ConclusionThe 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.