Improved Small Sample Inference Methods for a Mixed-Effects Model for Repeated Measures Approach in Incomplete Longitudinal Data Analysis

Yoshifumi Ukyo,Masahiko Gosho,Kazushi Maruo,Hisashi Noma

doi:10.3390/stats2020013

Abstract

The mixed-effects model for repeated measures (MMRM) approach has been widely applied for longitudinal clinical trials. Many of the standard inference methods of MMRM could possibly lead to the inflation of type I error rates for the tests of treatment effect, when the longitudinal dataset is small and involves missing measurements. We propose two improved inference methods for the MMRM analyses, (1) the Bartlett correction with the adjustment term approximated by bootstrap, and (2) the Monte Carlo test using an estimated null distribution by bootstrap. These methods can be implemented regardless of model complexity and missing patterns via a unified computational framework. Through simulation studies, the proposed methods maintain the type I error rate properly, even for small and incomplete longitudinal clinical trial settings. Applications to a postnatal depression clinical trial are also presented.

Highlights

Clinical trials for new drug development are often longitudinal trials in which outcome variables are repeatedly measured
We proposed and investigated two improved inference methods involved in mixed-effects model for repeated measures (MMRM)
We considered the testing problem for individual regression coefficients of MMRM, which corresponded to the primary analysis of longitudinal clinical trials

Summary

Introduction

Clinical trials for new drug development are often longitudinal trials in which outcome variables are repeatedly measured. Regulatory guidelines for preventing and treating the missing data in clinical trials have been issued [1,2,3], and adequate practices have been strongly pursued in recent years. Following these discussions, the mixed-effects model for repeated measures (MMRM) [4,5,6] has been widely applied for primary analyses of clinical trials in drug development. This type of model allows for valid statistical inference under incomplete longitudinal repeated measurements based on the direct likelihood approach

Methods

Results

Conclusion