The application of Gatekeeping strategies in dose-response clinical trials with multiple endpoints.

Abstract

Gatekeeping strategies are methods which deal with the analysis of the hierarchically ordered multiple objectives. The methods control the study-wise error rate in the strong sense. The objective of this report is to apply the Dunnett-Bonferroni-based parallel and other gatekeeping procedures in a clinical trial with ordered endpoints and multiples doses for Type II diabetes patients. The original hierarchical testing strategy does not take into account multiplicity in the family of the secondary endpoint. The efficacy of the experimental drug will be evaluated using the primary endpoint: the mean change from baseline in glycohaemoglobin A1C (HbA1C) and the first secondary endpoint: fasting plasma glucose (FPG). Three doses of the active treatment will be compared versus placebo. The results show an increase in pvalues compared to the p-values presented in the final clinical study report (CSR). The comparison of the estimated power is performed between Dunnett-Bonferroni-based parallel gatekeeping procedure and the study design of CSR based on simulation. In addition, the changes in the adjusted p-values are analyzed in details to illustrate the working mechanism of different gatekeeping procedures.