PS09: Battle of the Flexible Sample Sizes: Group Sequential Versus Promising Zone

Flexible sample size designs, dating back to at least the 1970s, are now a fundamental component of many clinical trials. Accounting for uncertainty in treatment effect as well as decreasing sample size, modern techniques can preserve power, save sample size, and provide robustness when the treatment effect is highly uncertain prior to a trial. The two most common paradigms for flexible sample sizes today are group sequential designs (including "goldilocks" trials when outcomes are delayed) and sample size re-estimation, most commonly implemented by the promising zone design and its variants. These methods continue to generate debate among practitioners over which is superior, including a debate over which is more accepted from a regulatory perspective; a famous regulatory "endorsement" of promising zone, combined with no similar endorsement of group sequential designs, contributes to this debate. We propose a session that will review current methodology for flexible sample sizes, discuss regulatory aspects of these trials, and discuss pros and cons of these methods under different conditions (for example with different accrual profiles with delayed outcomes).