The Randomization Effect in a Double-Blind Placebo-Controlled Depression Study

Abstract

BACKGROUND: Double-blind, placebo-controlled trials are often affected by higher-than-anticipated placebo responses that may contribute to study failure. It is common to observe symptomatic improvement immediately after randomization regardless of treatment allocation. METHODS: This post hoc analysis examined the early randomization effect in a recent 4-week double-blind, placebo-controlled study that examined a novel presynaptic mGluR2 inhibitor, MK-1942, in a daily or twice-weekly dosing strategy versus placebo as an adjunctive treatment in acutely depressed participants with major depressive disorder who had not adequately responded to antidepressant treatment. RESULTS: The original study failed to meet its primary endpoints, but a post hoc analysis revealed an early randomization effect that may have impeded signal detection. Over 60% of the overall Montgomery Asberg Depression Rating Scale (MADRS) score improvement observed in the placebo group occurred within the first week of this 4-week study, and >20% of placebo-assigned participants met criteria as treatment responders within 1 week after randomization. Using week 1 as an alternative ("revised") baseline, the percentage of placebo responders at week 4 was reduced from 48.6% to 31.4%. Cohen standardized effect size favoring the twice-weekly MK-1942 dosing over placebo improved from -0.33 (original baseline) to -0.54 at week 4 by applying the revised baseline. CONCLUSIONS: These post hoc results suggest an early randomization effect may complicate the interpretation of studies that examine treatment effects of drugs in development. Masking the timing of randomization with blinded, placebo-lead-in designs may warrant consideration for depression studies, particularly for rapid-acting antidepressants.