Rapid-cycling bipolar disorder: effects of long-term treatments

To compare responses to long-term treatment of rapid-cycling (RC) vs. non-RC bipolar disorder patients and assess relative effectiveness of specific agents in RC patients.

Studies identified by literature searching were analyzed for effects of RC status and treatment-type on clinical outcome (recurrence or non-improvement per exposure-time), using random-effects methods to estimate pooled rates and their 95% CI for quantitative meta-analytic modeling.

Data were obtained from 16 reports with 25 trial-arms involving 1856 (905 RC and 951 non-RC) patients treated with carbamazepine, lamotrigine, lithium, topiramate, or valproate, alone or with other agents over an average of 47.5 months (7347 total patient-years). Estimated RC prevalence was 15.4%. Crude rates (%/month) of recurrence (2.31/1.20) and clinical non-improvement (1.93/0.49) averaged 2.9-fold greater in RC vs. non-RC subjects. The pooled RC/non-RC risk ratio (RR) for inferior treatment-response (in 13 direct comparisons) was 1.40 (CI 1.26-1.56; P < 0.0001). Pooled crude recurrence and non-improvement rates suggested no clear advantage for any treatment, nor superiority for anticonvulsants over lithium. However, only lithium vs. carbamazepine could be directly compared (in four treatment-arms) meta-analytically in RC patients (RR = 0.93, CI 0.74-1.18, indicating no difference in effectiveness).

As expected, RC was associated with lower effectiveness of all treatments evaluated. Direct comparisons of specific treatment alternatives for RC patients were rare, and provided no secure evidence of superiority of any treatment. Additional long-term studies comparing RC/non-RC patients randomized to specific treatments are required.