gms | German Medical Science

Background: In the planning stage of a clinical study, several challenges need to be considered in the statistical design apart from the classical tasks (for example, the definition of the primary and secondary analysis models and the according sample size). We designed two phase III twin studies to assess the efficacy and safety of our treatment strategy. In these trials, we decided to use one primary and two secondary endpoints to assess efficacy and to compare two dosing regimens of the active treatment to a placebo arm. During the planning stage, we came across two main challenges: 1) the phase III was initiated right after the completion of a successful proof-of-concept study on another molecule with a similar mechanism of action Thus, very limited information on the effect of our active treatment was available; 2) one of the secondary endpoints would have required the inclusion of a very high number of patients to reach a reasonable power value.

Methods: We implemented a statistical study design tackling both issues mentioned above. Firstly, we included a futility analysis to be performed as soon as 40% of the patients would reach the assessment of the primary endpoint. This allows to stop the study early, either for efficacy or futility, in case our assumptions do not match reality. Secondly, we implemented a hierarchical testing procedure starting with the primary endpoint and succeeding to both secondary endpoints. Furthermore, we implemented a pooling strategy for one of the second secondary endpoints which doubled the number of subjects our statistical test was based on. This leads to a much higher power level and enables us to evaluate this endpoint in a sensible way.

Results: The combination of the futility analysis, the hierarchical testing procedure, and the pooling strategy (only for one secondary endpoint while the primary and the other secondary endpoint are still tested on a study level) leads to a rather complex study design. However, there are big advantages, such as the flexibility and the increase in power for the secondary endpoint. In our contribution, we will present the details of our design and give an overview of its advantages and disadvantages.

Conclusion: The statistical study design is the result of several internal and external discussions and tackles a situation that may occur not infrequently in submission studies. Thus, we are convinced that sharing technical aspects of our design might be helpful to other researchers who are involved in the planning of clinical studies implemented for submission purposes.

The authors declare that they have no competing interests.

The authors declare that a positive ethics committee vote has been obtained.