A comparison of antidepressant trials using active and inert placebos

In placebo-controlled antidepressant trials, there are concerns that the double blind may be breached, which may bias results. Some trials have therefore used placebos containing active substances in order to mimic the side effects of antidepressants. This study set out to examine the impact of 'unblinding' by comparing the results of a meta-analysis of a sample of trials using ordinary inert placebos with a meta-analysis of trials using active placebos. An a priori sub-group analysis of trials conducted with inpatients and outpatients was carried out. Quality was assessed using a quality assessment instrument and meta-regression analysis was conducted to explore heterogeneity. Results of meta-analysis did not differ between trials using active placebos and trials using inert ones. However, results were strongly influenced by two large trials. When these were excluded trials using inert placebos showed greater antidepressant-placebo differences than trials using active placebos. There was very high heterogeneity especially among trials using inert placebos. Meta-regression analysis showed the only significant predictor of outcome was whether trials were conducted with inpatients or outpatients. Sub-group analysis showed significant differences between antidepressants and any type of placebo in trials conducted with outpatients but not in inpatient trials. Lack of data, heterogeneity and the influence of large studies limited the analysis. The integrity of the double blind design, its effect on results and the heterogeneity between studies, require further exploration in antidepressant trials.