Trial Simulation to estimate Type I error when a population window enrichment strategy is used to improve efficiency of clinical trials in depression

Performance of recruitment centers is a critical feature for the success of multicentric randomized clinical trials (RCT). An enrichment window strategy (EWS) was recently proposed to enhance signal detection in RCTs based on the identification of centers with non-plausible placebo response, leading to a Per Protocol exclusion of all data from these non-informative centers before statistical analysis. The risk of an inflated Type I error associated with EWS was assessed using Clinical Trial Simulation. Randomized two-arm placebo controlled trials were simulated under the assumption that the drug treatment was ineffective.The results obtained in the absence of population enrichment were compared to the results obtained after EWS application. The results indicated that EWS preserved the Type I error showing no bias, at variance with other procedures. These results were confirmed by simulating 13 three-arm placebo controlled RCTs from the GSK Clinical Database.