Modified truncated Hochberg procedure for multiple endpoints: An application in a confirmatory trial for pediatric functional constipation.

BACKGROUND

In confirmatory clinical trials, it is critical to have appropriate control of multiplicity for multiple comparisons or endpoints. When multiplicity-related issues arise from different sources (e.g., multiple endpoints, multiple treatment arms, multiple interim data-cuts and other factors), it can become complicated to control the family-wise type I error rate (FWER). Therefore, it is crucial for statisticians to fully understand the multiplicity adjustment methods and the objectives of the analysis regarding study power, sample size and feasibility in order to identify the proper multiplicity adjustment strategy.

METHODS

In the context of multiplicity adjustment of multiple dose levels and multiple endpoints in a confirmatory trial, we proposed a modified truncated Hochberg procedure in combination with a fixed-sequence hierarchical testing procedure to strongly control the FWER. In this paper, we provided a brief review of the mathematical framework of the regular Hochberg procedure, the truncated Hochberg procedure and the proposed modified truncated Hochberg procedure. An ongoing phase 3 confirmatory trial for pediatric functional constipation was used as a real case application to illustrate how the proposed modified truncated Hochberg procedure will be implemented. A simulation study was conducted to demonstrate that the study was adequately powered and the FWER was strongly controlled.

CONCLUSION

This work is expected to facilitate the understanding and selection of adjustment methods for statisticians.