Does placebo help establish equivalence in trials of new antidepressants?

Clinical trials of new antidepressants usually compare a new drug to a refe rence antidepressant and to a placebo. The placebo is intended to validate the trial in the case of a no-difference outcome, i.e., it helps in assessi ng equivalence. The aim of the present paper is to test whether placebo has indeed helped establish equivalence of effect in comparative trials of new antidepressants. We carried out an example of sample size determination fi rst in a trial to show a difference between the new and control drug, and s econd in a trial to assess equivalence between two competing drugs. Finally , we retrospectively calculated the maximum difference accepted as equivale nce of effect in published trials of new antidepressants. Assuming a respon se rate to antidepressants of 70%, 294 subjects for each treatment group ar e needed to show a 10% difference between two antidepressant drugs and more than 1,300 to assess equivalence at a 5% level of delta, the maximum diffe rence acceptable as equivalence of effect. The level of delta in published trials of new antidepressants ranges between 12 and 43%, suggesting they ca nnot claim to demonstrate equivalence of effect. Therefore, the presence of a placebo arm for comparison didn't help establish whether both drugs real ly worked the same way. Comparative trials of new antidepressants should ad opt a two-arm design, a suitable number of patients and a high standard in the experimental design in order to minimise possible control-event rate va riation. (C) 2000 Editions scientifiques et medicales Elsevier SAS.