Empirically derived criteria cast doubt on the clinical significance of antidepressant-placebo differences

Abstract

Meta-analyses indicate that antidepressants are superior to placebos in statistical terms, but the clinical relevance of the differences has not been established. Previous suggestions of clinically relevant effect sizes have not been supported by empirical evidence. In the current paper we apply an empirical method that consists of comparing scores obtained on the Hamilton rating scale for depression (HAM-D) and scores from the Clinical Global Impressions-Improvement (CGI-I) scale. This method reveals that a HAM-D difference of 3 points is undetectable by clinicians using the CGI-I scale. A difference of 7 points on the HAM-D, or an effect size of 0.875, is required to correspond to a rating of 'minimal improvement' on the CGI-I. By these criteria differences between antidepressants and placebo in randomised controlled trials, including trials conducted with people diagnosed with very severe depression, are not detectable by clinicians and fall far short of levels consistent with clinically observable minimal levels of improvement. Clinical significance should be considered alongside statistical significance when making decisions about the approval and use of medications like antidepressants.

Keywords: Antidepressant efficacy; Clinical relevance of antidepressant effects; Effect size in depression; Hamilton rating scale for depression; Measurement of depression.