Factors that influence the outcome of placebo-controlled antidepressant clinical trials.Psychopharmacol Bull 1997; 33(1):41-51PB
An important issue in judging the therapeutic potential of a new antidepressant drug is the effect size it generates in placebo-controlled trials which has to be compared with the effect of an active control. As this effect size tends to vary substantially it is not easy to predict the sample size in a clinical trial. We carried out two dose finding placebo- and imipramine-controlled, double-blind, multicenter and multinational trials (northern part of Europe) with a new psychotropic compound (NPC) currently being investigated in Phase II/III as a potential antidepressant in the indication major depressive disorder (MDD). Statistical analysis showed that the effect size of 150 mg per day imipramine was meager (1.04 points difference from placebo after 6 weeks according to total scores on the 17-item Hamilton Rating Scale for Depression [HAM-D-17] based on the Intent-to-Treat group using last observation carried forward [LOCF] approach). To increase the discriminative power of the analysis we selected centers that were able to detect a difference of at least two points between imipramine and placebo on the HAM-D-17 total score at 6 weeks in the LOCF analysis (discriminative centers, DC). The other group of centers will be called non-discriminative centers (NDC). We found that 36 percent of centers were DC and recruited about 45 percent of patients. Further analysis revealed no statistically significant differences between the groups of centers concerning the important baseline characteristics of the patients. Also the pattern of reported adverse events did not differ between DC and NDC. We found a tendency for the DC to select more patients with recurrent illness, in particular with a previous good response to antidepressant therapy. The groups of centers differed in dropout rates for both active treatments (DC 19.3-20.4% vs. NDC 35.1-37.7%) but not for the placebo (DC 38.2% vs. NDC 30.5%) that could suggest that different treatment strategies were employed by different centers regarding the occurrence of adverse events with and without therapeutic effects.