Duloxetine: meta-analyses of suicidal behaviors and ideation in clinical trials for major depressive disorder.

BACKGROUND Uncertainty regarding relationships of antidepressant treatment and suicidality encouraged systematic review of data on suicidal behaviors and ideation from Phase II and III clinical trials of duloxetine for major depressive disorder (MDD). METHODS We evaluated all completed duloxetine trials in MDD with data lock by February 2, 2004. We compared incidence of suicide-related events with duloxetine versus placebo in controlled trials, using Mantel-Haenszel incidence difference (MHID) and exposure time-adjusted rate difference (MHRD) methods, and analyzed changes in Hamilton Depression Scale (HAMD) Item-3 (suicidality) scores. RESULTS There were no significant differences in the incidence of suicide-related events with duloxetine versus placebo in 12 placebo-controlled trials (duloxetine, 1812; placebo, 1184 [corrected] patients). The MHID for suicide-related behaviors was -0.03% (95% confidence interval [CI], -0.48 to 0.42) and MHRD -0.002 (95% CI, -0.02 to 0.02). Changes in HAMD Item-3 suicidality scores showed more improvement with duloxetine (MHID, 9.56%; 95% CI, 4.50 to 14.6; P < 0.001) and less worsening of suicidal ideation with duloxetine (MHID, -4.25%; 95% CI, -6.55 to -1.95; P < 0.001). Other Item-3 findings showed no consistent pattern; a slightly higher proportion of duloxetine-treated patients with a change from 0 (absent) to 3 was balanced against a higher proportion of placebo-treated patients changing from 0 to 2. CONCLUSIONS We found no evidence of an increased risk of suicidal behaviors or ideation during treatment with duloxetine compared with placebo in MDD patients. HAMD Item-3 suicidality scores had more improvement and less worsening of suicidal ideation with duloxetine than placebo.