Estimation of a generalized linear mixed model for response‐adaptive designs in multi‐centre clinical trials. Issue 3 (23rd May 2017)

• Record Type: Journal Article
• Authors: Selvaratnam, Selvakkadunko; Oyet, Alwell J.; Yi, Yanqing; Gadag, Veeresh
• Is Part Of: Canadian journal of statistics. Volume 45:Issue 3(2017)
• Page Start: 310
• Page End: 325
• Publication Date: 2017-05-23
• Subjects: Asymptotic properties -- Clinical trials -- generalized linear mixed model -- influence function -- maximum likelihood -- response‐adaptive designs -- MSC 2010: Primary 62F12 -- secondary 62L05; Mathematical statistics -- Periodicals; 519.5
• DOI: 10.1002/cjs.11324
• Languages: English
• ISSNs: 0319-5724

Abstract

Abstract: Response‐adaptive designs are important alternatives to equal allocation in clinical trials because equal treatment allocation has been found to have ethical issues. In this article we discuss the implementation of response‐adaptive designs in multi‐centre clinical trials. We develop a generalized linear mixed model (GLMM) for analyzing data obtained from multi‐centre clinical trials and use the maximum likelihood (ML) approach to estimate the model parameters. We apply influence function techniques to derive the asymptotic properties of our estimators. The advantage of using the influence function approach is that it leads to a closed form expression for the asymptotic covariance of the estimated parameters. To our knowledge such a closed form expression does not currently exist in the literature. The performance of the ML estimator under various response‐adaptive designs is examined through simulation studies. We use our simulation studies to compare the asymptotic covariance matrix, based on the influence function to that based on the inverse of the Hessian matrix obtained from the likelihood function of the observations. The techniques are applied to real data obtained from a multi‐centre clinical trial designed to compare two cream preparations (active drug/control) for treating an infection. The Canadian Journal of Statistics 45: 310–325; 2017 © 2017 Statistical Society of Canada