Oquendo MA, Galfalvy HC, Currier D, Grunebaum MF, Sher L, Sullivan GM, Burke AK, Harkavy-Friedman J, Sublette ME, Parsey RV, Mann JJ.
Am J Psychiatry 2011 Oct;168(10):1050-6.

OBJECTIVE: Bipolar disorder is associated with high risk for suicidal acts. Observational studies suggest a protective effect of lithium against suicidal behavior. However, testing this effect in randomized clinical trials is logistically and ethically challenging. The authors tested the hypothesis that lithium offers bipolar patients with a history of suicide attempt greater protection against suicidal behavior compared to valproate. METHOD: Patients with bipolar disorder and past suicide attempts (N=98) were randomly assigned to treatment with lithium or valproate, plus adjunctive medications as indicated, in a double-blind 2.5-year trial. An intent-to-treat analysis was performed using the log-rank test for survival data. Two models were fitted: time to suicide attempt and time to suicide event (attempt or hospitalization or change in medication in response to suicide plans).RESULTS: There were 45 suicide events in 35 participants, including 18 suicide attempts made by 14 participants, six from the lithium group and eight from the valproate group. There were no suicides. Intent-to-treat analysis using the log-rank test showed no differences between treatment groups in time to suicide attempt or to suicide event. Post hoc power calculations revealed that the modest sample size, reflective of challenges in recruitment, only permits detection of a relative risk of 5 or greater. CONCLUSIONS: Despite the high frequency of suicide events during the study, this randomized controlled trial detected no difference between lithium and valproate in time to suicide attempt or suicide event in a sample of suicide attempters with bipolar disorder. However, smaller clinically significant differences between the two drugs were not ruled out.