Cost-Effectiveness Comparison of Adaptive vs. Equal Randomization for a 'Pick the Winner' Clinical Trial Design

• Activity Number: 30
• Type: Contributed
• Date/Time: Sunday, August 3, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Lingyun Ji*+ and Richard Sposto and Susan Groshen
• Companies: University of Southern California and University of Southern California and University of Southern California
• Keywords: Bayesian Adaptive Randomization ; Phase II Clinical Trial Design ; Cost-Effectiveness ; Pick the Winner

Abstract

Recent research has shown benefits of Bayesian adaptive randomization (AR) compared to equal randomization (ER). However, AR might not be adopted due to logistical and practical concerns, including the need for frequent endpoint evaluation and for a higher level of statistical and computational sophistication in trial design and execution. Using a simulation approach, we evaluated the properties of AR compared to ER designs in a three-arm "pick the winner" Phase II study to examine when the benefits of AR in terms of statistical efficiency and patient safety might be outweighed by the cost of adopting AR. A binary response endpoint was used in the simulated trial. We compared AR and ER designs over a wide range of scenarios using metrics that included the correct arm selection rate, average response rate of each arm, probability of early dropping of an ineffective or effective arm, number of patients in each arm, the total number of responders, as well as estimated relative costs for AR vs. ER. We discuss the magnitude of these differences in different scenarios to illustrate in which situations AR designs are clearly beneficial compared to ER designs.