Study duration in antidepressant research: advantages of a 12-week trial.

There has been an increasing interest in studying the effectiveness of antidepressants in non-melancholic depressives. For non-melancholic patients who are characterized by transient mood improvement, a single cross-sectional evaluation may reflect temporary change. Transient improvement, like any source of instability in an outcome measure, reduces validity and consequently power. To minimize the effects of transient mood change in non-melancholic depressives, we designed a two-phase drug trial. The first phase was a standard 6-week trial. Those judged responders at the end of the first phase entered the second 6-week phase. Between weeks 6 and 12 it was anticipated that a smaller proportion of six-week responders would relapse on drug than placebo, thus sharpening the contrast between treatments. The purpose of this paper is to demonstrate the power advantages of the 12-week design. Data is presented which suggests that a 12-week design reduces transient improvement, increases treatment effect size, and requires a smaller number of patients for equivalent statistical power. It also offers a better estimate than a 6-week study of the proportion of patients who will have persistent clinical benefit.