Abstract:
|
For randomized time-to-event studies targeting a fixed number of events, it is often thought that randomizing patients 1:1 maximizes power. However, Yung and Liu (2019) recently showed that this is not true; power is not maximized by balancing number of patients, but rather by balancing the number of events (the effective sample size) across arms. This may encourage trials involving effective therapies (i.e. hazard ratio < 1) to randomize more patients to the active arm (e.g. 3:2 or 2:1). Of course, in practice other factors besides power should be taken into consideration when deciding the randomization ratio: study duration, costs, anticipated benefit-risk profiles of the two involved drugs, and ethics, to name a few. In this talk, we will share findings from our additional simulation and research which focuses on the potential interplay between power and study duration. We will also provide general guidance as to when unequal allocation might be attractive to both patients and sponsors.
|