gms | German Medical Science

Background: In order to save time and cost for the development of a new compound, pivotal trials tend to become more complex by testing several hypothesis within the same trial. Here, we focus on a trial testing two dose groups in a full population as well as a subgroup based on two binary co-primary endpoints. The trial is designs as a two-stage procedure allowing stopping for futility at interim based on the conditional power. In order to adjust for multiplicity a truncated Hochberg procedure is implemented.

Objectives: To derive analytical solutions allowing the calculation of the conditional power at interim as well as the overall power and the sample size.

Methods: Based on group sequential trial methodology, we derive approximate analytical solutions for the joint test statistics allowing calculation of the (conditional) power to reject the individual dose levels for each population as well as the joint power to reject any dose.

Results: We show how to derive the approximate analytical solutions for this complex trial design. Our results are checked against simulation studies. We also show the impact of the choice of the truncation parameter on the power for the various hypotheses being tested. The findings are illustrated using a real trial example.

Conclusion: Analytical solutions can be derived even in complex trial design situations allowing a more precise planning of the trial than could be achieved via simulation studies alone. However, we also found that in some cases, the calculations involved in the analytical solutions can also be time consuming and might occasionally be longer than for the simulation study.

The authors declare that they have no competing interests.

The authors declare that an ethics committee vote is not required.