Maintenance efficacy designs in psychiatry: Randomized discontinuation trials – enriched but not better

INTRODUCTION Although classical randomized clinical trials (RCTs) are the gold standard for proof of drug efficacy, randomized discontinuation trials (RDTs), sometimes called "enriched" trials, are used increasingly, especially in psychiatric maintenance studies. METHODS A narrative review of two decades of experience with RDTs. RESULTS RDTs in psychiatric maintenance trials tend to use a dependent variable as a predictor: treatment response. Treatment responders are assessed for treatment response. This tautology in the logic of RDTs renders them invalid, since the predictor and the outcome are the same variable. Although RDTs can be designed to avoid this tautologous state of affairs, like using independent predictors of outcomes, such is not the case with psychiatric maintenance studies Further, purported benefits of RDTs regarding feasibility were found to be questionable. Specifically, RDTs do not enhance statistical power in many settings, and, because of high dropout rates, produce results of questionable validity. Any claimed benefits come with notably reduced generalizability. CONCLUSIONS RDTs appear to be scientifically invalid as used in psychiatric maintenance designs. Their purported feasibility benefits are not seen in actual trials for psychotropic drugs. There is warrant for changes in federal policy regarding marketing indications for maintenance efficacy using the RDT design.