Should Clinical Trials Be Terminated Early?

In many trials, interim analyses are often performed to decide whether the trial should be terminated early, for a variety of reasons. These reasons may include an imbalance in adverse events in one group versus the other(s), a situation in which, given the sample size, it would be impossible to demonstrate superiority of one treatment (futility) or a significant difference between groups. This commentary argues that ending a trial prematurely for the latter reason is fraught with problems and often results in an overestimation of the effect that would have been obtained were the trial allowed to continue. It concludes that stopping a trial early for apparent superiority of a treatment should be avoided.