We propose a class of adaptive randomized trial designs for comparing two treatments to a common control in two disjoint subpopulations. The type of adaptation, called adaptive enrichment, involves a preplanned rule for modifying enrollment and arm assignment based on accruing data in an ongoing trial. The motivation for this adaptive feature is that interim data may indicate that a subpopulation, such as those with lower disease severity at baseline, are unlikely to benefit from a particular treatment, while uncertainty remains for the other treatment and/or subpopulation. We developed a new multiple testing procedure tailored to this design problem. The procedure improves power by: leveraging the correlation between the test statistics arising from the two treatments being compared to a common control; reallocating alpha across subpopulations, and using the data only through minimally sufficient statistics. We optimize expected sample size over this class of designs, focusing on designs with 2 stages. Our approach is demonstrated in simulation studies that mimic features of a completed trial of a medical device for treating heart failure. User-friendly, open-source software that implements the trial design optimization is provided.
Biostatistics | Statistical Methodology
Steingrimsson, Jon Arni; Betz, Joshua; Qian, Tiachen; and Rosenblum, Michael, "OPTIMIZED ADAPTIVE ENRICHMENT DESIGNS FOR MULTI-ARM TRIALS: LEARNING WHICH SUBPOPULATIONS BENEFIT FROM DIFFERENT TREATMENTS" (September 2017). Johns Hopkins University, Dept. of Biostatistics Working Papers. Working Paper 288.