Robustness of Bayesian D-optimal design for the logistic mixed model against misspecification of autocorrelation

In medicine and health sciences mixed effects models are often used to study time-structured data. Optimal designs for such studies have been shown useful to improve the precision of the estimators of the parameters. However, optimal designs for such studies are often derived under the assumption of a zero autocorrelation between the errors, especially for binary data. Ignoring or misspecifying the autocorrelation in the design stage can result in loss of efficiency. This paper addresses robustness of Bayesian D-optimal designs for the logistic mixed effects model for longitudinal data with a linear or quadratic time effect against incorrect specification of the autocorrelation. To find the Bayesian D-optimal allocations of time points for different values of the autocorrelation, under different priors for the fixed effects and different covariance structures of the random effects, a scalar function of the approximate variance–covariance matrix of the fixed effects is optimized. Two approximations are compared; one based on a first order penalized quasi likelihood (PQL1) and one based on an extended version of the generalized estimating equations (GEE). The results show that Bayesian D-optimal allocations of time points are robust against misspecification of the autocorrelation and are approximately equally spaced. Moreover, PQL1 and extended GEE give essentially the same Bayesian D-optimal allocation of time points for a given subject-to-measurement cost ratio. Furthermore, Bayesian optimal designs are hardly affected either by the choice of a covariance structure or by the choice of a prior distribution.

[1]  Martijn P. F. Berger,et al.  Maximin D-optimal designs for binary longitudinal responses , 2008, Comput. Stat. Data Anal..

[2]  K. Chaloner,et al.  Optimal Bayesian design applied to logistic regression experiments , 1989 .

[3]  S. Sinha,et al.  Sequential D-optimal designs for generalized linear mixed models , 2011 .

[4]  Kathryn Chaloner,et al.  Bayesian Experimental Design for Nonlinear Mixed‐Effects Models with Application to HIV Dynamics , 2004, Biometrics.

[5]  M. Berger,et al.  Robust designs for linear mixed effects models , 2004 .

[6]  Martijn P. F. Berger,et al.  Optimal experimental designs for multilevel logistic models , 2001 .

[7]  Cora J. M. Maas,et al.  Optimal Experimental Designs for Multilevel Logistic Models with Two Binary Predictors , 2005 .

[8]  Martijn P. F. Berger,et al.  A maximin criterion for the logistic random intercept model with covariates , 2006 .

[9]  K. Chaloner,et al.  Bayesian Experimental Design: A Review , 1995 .

[10]  N. Breslow,et al.  Approximate inference in generalized linear mixed models , 1993 .

[11]  Ilse Mesters,et al.  Short-term effects of a randomized computer-based out-of-school smoking prevention trial aimed at elementary schoolchildren. , 2002, Preventive medicine.

[12]  Martijn P. F. Berger,et al.  A Comparison of Estimation Methods for Multilevel Logistic Models , 2003, Comput. Stat..

[13]  M. Berger,et al.  Bayesian design for dichotomous repeated measurements with autocorrelation , 2015, Statistical methods in medical research.

[14]  Blaza Toman,et al.  Bayesian Experimental Design , 2006 .

[15]  W. Gilks,et al.  Adaptive Rejection Metropolis Sampling Within Gibbs Sampling , 1995 .

[16]  Martijn P. F. Berger,et al.  OPTIMAL ALLOCATION OF TIME POINTS FOR THE RANDOM EFFECTS MODEL , 1999 .

[17]  Weng Kee Wong,et al.  An Introduction to Optimal Designs for Social and Biomedical Research , 2009 .

[18]  Martijn P. F. Berger,et al.  On the Choice of a Prior for Bayesian D-Optimal Designs for the Logistic Regression Model with a Single Predictor , 2014, Commun. Stat. Simul. Comput..

[19]  Woncheol Jang,et al.  A Numerical Study of PQL Estimation Biases in Generalized Linear Mixed Models Under Heterogeneity of Random Effects , 2009, Commun. Stat. Simul. Comput..

[20]  Martijn P. F. Berger,et al.  Bayesian D-optimal designs for the two parameter logistic mixed effects model , 2014, Comput. Stat. Data Anal..

[21]  G. Molenberghs,et al.  Models for Discrete Longitudinal Data , 2005 .

[22]  Mehrdad Niaparast,et al.  On optimal design for a poisson regression model with random intercept , 2009 .

[23]  S. Silvey.  Optimal Design: An Introduction to the Theory for Parameter Estimation , 1980 .

[24]  Peter M. van de Ven,et al.  Blocked Designs for Experiments With Correlated Non-Normal Response , 2011, Technometrics.

[25]  Linda M. Haines,et al.  14 Designs for nonlinear and generalized linear models , 1996, Design and analysis of experiments.

[26]  M. Ghosh,et al.  Design Issues for Generalized Linear Models: A Review , 2006, math/0701088.

[27]  Anthony C. Atkinson,et al.  Optimum Experimental Designs, with SAS , 2007 .

[28]  G. Breart,et al.  Fall-related factors and risk of hip fracture: the EPIDOS prospective study , 1996, The Lancet.

[29]  W. Gilks,et al.  Adaptive Rejection Sampling for Gibbs Sampling , 1992 .

[30]  P. Diggle,et al.  Analysis of Longitudinal Data , 2003 .

[31]  A. Sommer,et al.  Increased risk of respiratory disease and diarrhea in children with preexisting mild vitamin A deficiency. , 1984, The American journal of clinical nutrition.

[32]  P. Albert,et al.  Models for longitudinal data: a generalized estimating equation approach. , 1988, Biometrics.

[33]  Jean Bouyer,et al.  Choosing marginal or random-effects models for longitudinal binary responses: application to self-reported disability among older persons , 2002, BMC medical research methodology.

[34]  Rainer Schwabe,et al.  Optimal design for quasi-likelihood estimation in Poisson regression with random coefficients , 2013 .

[35]  H. Chernoff.  Locally Optimal Designs for Estimating Parameters , 1953 .