Gatekeeping procedures with clinical trial applications

Abstract

The objective of this paper is to give an overview of a relatively new area of multiplicity research that deals with the analysis of hierarchically ordered multiple objectives. Testing procedures for this problem are known as gatekeeping procedures and have found a variety of applications in clinical trials. This paper reviews main classes of these procedures, including serial and parallel gatekeeping procedures, and tree gatekeeping procedures that account for logical restrictions among multiple objectives. We focus on procedures based on marginal p-values; extensions to procedures that exploit the joint distribution of the p-values are also noted. Clinical trial examples are used to illustrate the procedures and their important properties.

Intravenous Dexamethasone and Perineural Dexamethasone Similarly Prolong the Duration of Analgesia After Supraclavicular Brachial Plexus Block

“…We used the sequential testing approach as a serial gatekeeping procedure for this hypothesis testing. 21 As such, the duration of analgesia observed in the Dex IV group was first tested for superiority over the Control group and then tested for superiority over the Dex P group.…”

A Dunnett–Bonferroni‐based parallel gatekeeping procedure for dose–response clinical trials with multiple endpoints

“…In their setting, Liu et al [17] impose two layers of logical restriction: (1) requiring that the significance in the secondary endpoint for a dose level cannot be claimed unless its corresponding primary hypothesis was found [17] procedure with the other two, we modify it as follows: only the first logical restriction will be required or, more specifically, the same form of decision rules (Table AI) as those for our proposed procedure and the Dunnett-based procedure [7] will be used. This modification will be referred to as the independent-T method hereafter and denoted by us as the IT procedure.…”

“…This is achieved by testing multiple families of analyses sequentially and each family serves as a gatekeeper for subsequent families. As pointed out by Dmitrienko and Tamhane [1], multiple families of analyses are often related to multiple endpoints but can also represent dose-to-control comparisons, non-inferiority and superiority tests or inferences at several time points in a longitudinal study. In this paper, we focus on the problem of dose-to-control comparisons with multiple endpoints.…”

“…Two basic types of gatekeeping procedures have been studied in the literature: serial gatekeeping (see [1][2][3][4]) and parallel gatekeeping (see [1,[5][6][7]). More recent tree-structured gatekeeping procedures [8] are a generalization of serial and parallel gatekeeping procedures.…”

Analysis of multiple endpoints in clinical trials: it's time for the designations of primary, secondary and tertiary to go

“…For example, Dmitrienko and Tamhane [4] provide examples of serial gatekeeping procedures, parallel gatekeeping procedures, and tree gatekeeping procedures, that are all designed to provide strong control of the type 1 error rate. It is difficult to imagine how one would classify endpoints as primary, secondary, and tertiary in any coherent way using these procedures.…”