Adaptive trial design discussions tend to focus on early stopping for efficacy or futility, but sample size re-estimation, adjusting the planned enrollment target based on an interim look at accumulating data, is a distinct and increasingly common adaptive element that addresses a different problem entirely: getting the original sample size assumptions wrong.
Blinded sample size re-estimation, which adjusts enrollment targets based on nuisance parameters like observed variance or event rates without unblinding treatment assignment, is the more widely accepted approach among regulators, since it preserves the statistical integrity of the trial while correcting for design assumptions that turn out to differ from what interim data actually shows.
Unblinded re-estimation, which looks at the actual treatment effect size partway through the trial to adjust enrollment, offers more statistical power but requires considerably more rigorous firewalling between the data monitoring committee and the trial’s operational team, since any leakage of interim effect size information could compromise the blinding for everyone else involved in running the trial.
The practical value shows up most clearly in trials where the initial sample size calculation relied on assumptions from prior literature or a different patient population, a common scenario when a drug moves into a new indication or a new region where baseline event rates or variance may differ meaningfully from the historical data the original calculation was based on.
Regulatory acceptance of sample size re-estimation has grown considerably, but agencies generally expect this adaptation to be fully pre-specified in the statistical analysis plan before the trial begins, including exactly what triggers a re-estimation and what range of adjustment is permitted, rather than treated as an ad hoc decision made in response to disappointing interim results.
Deep Dive coverage examining how specific trials have implemented sample size re-estimation in practice, including what triggered the adjustment and how it affected overall trial timeline and cost, such as the analysis published by The Clinical Trial Vanguard, gives statisticians and trial designers a more concrete reference than the general methodology literature alone provides.
