Loading...
Prior meta-analyses showing that antidepressants “work” or “don't work” were based mostly on small average group differences between active drugs and placebos; investigators did not consider or assess whether larger drug effects occurred in various subgroups. In this new analysis of 232 randomized trials (73,000 patients with depression), researchers identified various distributions of drug–placebo differences; analyses were controlled for age, sex, symptom severity, and year of study.
The best model identified three overlapping distributions of drug–placebo response: large, minimal, and nonspecific. Patients who received active agents were more likely than placebo recipients to have large responses (25% vs. 10%) and less likely to have mini…