A0435
Title: New allocation probability tests for improved power in response-adaptive clinical trials
Authors: Stina Zetterstrom - University of Cambridge (United Kingdom) [presenting]
David Robertson - University of Cambridge (United Kingdom)
Sofia Villar - University of Cambridge (United Kingdom)
Abstract: Response-adaptive clinical trials update the allocation probabilities to treatment arms based on the data collected so far in the trial. One objective for response-adaptive clinical trials is to ensure that patients in the trial will have a higher probability of getting the best treatment compared to using equal randomization. However, this imbalance in treatment allocation can lead to low power when testing for a treatment difference. Recent works propose a new testing approach that is based on the allocation probability (AP) instead of the outcome directly, which can increase the power. Alternative versions of the AP test are proposed, where the functional form of the test statistic is changed. The AP tests are evaluated in simulation studies, using the Bayesian response adaptive randomization (BRAR) algorithm for binary outcomes in a two-arm setting. The simulation studies show that the AP test can perform better in terms of power compared to traditional tests, while controlling type I error. Furthermore, changing the functional form of the AP test statistic can result in a higher power than the original AP test. While BRAR is used, the AP test can be used for any response-adaptive randomization, and its performance is studied in that intersection.
