ABSTRACT
In group-sequential designs, multiplicity arises from repeated analyses of data that accrues in a clinical trial. At interim times, data is analyzed to decide about stopping the trial. The treatment effect is judged by statistical tests with control of the family-wise error rate (FWER) of false rejection under the null hypothesis of no treatment differences. The only permitted decisions at interim are to stop the study for efficacy or futility. In adaptive designs, additional design modifications are permitted. In practice, sample size re-estimation, dropping of treatment arms or restriction of subsequent randomization to subpopulations of patients are the most common modifications. To guarantee FWER control, combination test approaches such as p-value combination approaches and conditional error function approaches are applied. In recent years, trial designs are increasingly using combinations of these techniques with other multiplicity adjustment methods such as the closed test principle. For example, some recent trials investigated several treatment arms with several interim analyses and the possibility to drop treatment arms. We present examples of such trials and illustrate how the closed test principle, combination test approaches, and group-sequential error spending can be combined to retain FWER control and high power for relevant treatment effect sizes.
