Reanalysis of The National Institute of Mental Health Treatment of Depression Collaborative Research Program General Effectiveness Report

Abstract

The National Institute of Mental Health (NIMH) Treatment of Depression Collaborative Research Program General Effectiveness Report statistical analyses are criticized. Their analyses, which fostered the belief that the active treatments were indistinguishable, were compromised by an inappropriately stringent level of significance with regard to both heterogeneity of slope and pairwise group differences. Once slope heterogeneity is detected, the Johnson-Neyman technique is more appropriate than arbitrary sample subdivision. All of these tactics lowered power substantially. Our reanalysis indicates a reasonable ordering for the treatments with medication superior to the psychotherapies and the psychotherapies somewhat superior to placebo. These effects are particularly marked among the more symptomatic and impaired patients. The lack of dosage by severity analyses renders the severity findings ambiguous. Scientific and public health implications are discussed.