Re-analysis of the earliest controlled trials of imipramine

AimsRe-analysis of outcomes of the earliest controlled trials of imipramine for comparison to more recent findings in antidepressant trials.MethodsControlled trial-reports of imipramine reviewed by Klerman and Cole (1965) were re-analyzed for their methods and response rates, using random-effects meta-analytic modeling.FindingsIn 18 early trials (1959-1965), imipramine yielded a large and highly significant, pooled drug/placebo response rate-ratio (RR) of 2.17 (CI: 1.87-2.51), with an estimated number-needed-to-treat (NNT) of 3.1 (CI: 2.1-5.8), even though only 9/18 (50%) trials, individually yielded statistically significant drug-placebo differences.ConclusionsResponses to imipramine in its earliest controlled trials were much larger than in recent antidepressant trials. Drug-placebo differences declined significantly between 1959 and 1965, with rising placebo-associated responses. Frequent failure to find superior drug-over-placebo outcomes may reflect patient characteristics and limited statistical power. Antidepressant-trial methods have become much more standardized, samples larger and more complex, and effect-sizes much smaller since the 1960s.LimitationsSome reports did not include relevant information, diagnostic and outcome criteria varied, and only 18/30 trials included responder-rates.