Many studies on optimal designs for linear mixed model analysis of repeated measures data have focussed on estimating the fixed effects. The present study investigates the optimal number of time points and subjects in case random effects have to be estimated. Linear mixed models with a linear or quadratic trend across equidistant time points are studied. Given a particular cost function, we examine which designs minimise the expected average squared prediction error. Robustness of the optimal design, important when one does not know the underlying model, is also treated.
Get full access to this article
View all access options for this article.
References
1.
Albert PSTutorial in biostatistics: longitudinal data analysis (repeated measures) in clinical trials. Statistics in Medicine1999; 18: 1707-32.
2.
Bours GJJW , Halfens RJG , Candel MJJM , Grol RTPM , Huijer Abu-Saad H.A pressure ulcer audit and feedback project across multi-hospital settings in the Netherlands. The International Journal for Quality in Health Care2004; 16: 211-18.
3.
Hedeker D., Gibbons RDMIXREG: A computer program for mixed-effects regression analysis with autocorrelated errors. Computer Methods and Programs in Biomedicine1996; 49: 229-52.
4.
Reisby N., Gram LF, Bech P. et al. Imipramine: clinical effects and pharmacokinetic variability. Psychopharmacology1977; 54: 263-72.
5.
Raudenbush SW , Xiao-Feng L.Effects of study duration, frequency of observation, and sample size on power in studies of group differences in polynomial change . Psychological Methods2001; 6: 387-401.
6.
Bock RDMultivariate statistical methods in behavioral research. McGraw-Hill, New York; 1975.
7.
Heo M., Faith MS, Mott JW, Gorman BS, Redden DT, Allison DBTutorial in biostatistics.Hierarchical linear models for the development of growth curves: an example with body mass index in overweight/obese adults. Statistics in Medicine2003; 22: 1911-42.
8.
Van Dommelen P., Van Buuren S., Zandwijken GRJ, Verkerk PHIndividual growth curve models for assessing evidence-based referral criteria in growth monitoring . Statistics in Medicine2005; 24: 3663-74.
9.
Lyles RH, Xu, J.Classifying individuals based on predictors of random effects. Statistics in Medicine1999; 18: 35-52.
10.
Kriström B. , Wikland KAGrowth prediction models: concepts and use . Hormone research2002; 57(suppl): 66-70.
11.
Chvatal BA, Behrents RG, Ceen RF, Buschang PHDevelopment and testing of multilevel models for longitudinal craniofacial growth prediction. American Journal of Orthodontics and Dentofacial Orthopedics2005; 128: 45-56.
12.
Cook NREstimating predictive values for blood pressure measurements from multivariate regression models with covariates. Statistics in Medicine1996; 15: 2013-28.
13.
Ware JH, Wu MCTracking: prediction of future values from serial measurements. Biometrics1981; 37: 427-37.
14.
Darby SC, Fearn T.The Chatham blood pressure study.An application of Bayesian growth curve models to a longitudinal study of blood pressure in children. International Journal of Epidemiology1979; 8: 15-21.
15.
Verbeke G., Molenberghs G.Linear mixed models for longitudinal data. Springer ; New York: 2000.
Candel Mjjm , Winkens B.Performance of empirical Bayes estimators of level-2 random parameters in multilevel analysis: a Monte Carlo study for longitudinal designs. Journal of Educational and Behavioral Statistics2003; 28: 169-94.
20.
Galbraith S. , Marschner ICGuidelines for the design of clinical trials with longitudinal outcomes. Controlled Clinical Trials2002; 23: 257-73.
21.
Winkens B., Schouten HJA, Van Breukelen GJP, Berger MPF.Optimal time-points in clinical trials with linearly divergent treatment effects . Statistics in Medicine2005; 24: 3743-56.
22.
Winkens B., Schouten HJA, Van Breukelen GJP, Berger MPF.Optimal number of repeated measures and group sizes in clinical trials with linearly divergent treatment effects. Contemporary Clinical Trials2006; 27: 57-69.
23.
Bryk AS, Raudenbush SWHierarchical linear models: applications and data analysis methods . Sage; Newbury Park, 1992.
24.
Rao CRSimultaneous estimation of parameters in different linear models and applications to biometric problems. Biometrics1975; 31: 545-54.
25.
Taylor JMG , Lay N.Does the covariance structure matter in longitudinal modelling for the prediction of future CD4 counts?Statistics in Medicine1998; 17: 2381-94.
26.
Chaloner K. , Verdinelli I.Bayesian experimental design: A review. Statistical Science1995; 10: 273-304.
27.
Verdinelli I.A note on Bayesian design for the normal linear model with unknown error variance. Biometrika2000; 87: 222-27.
28.
Box GEP.Problems in the analysis of growth and wear curves. Biometrics1950; 6: 362-89.
29.
Singer JDUsing SAS PROC MIXED to fit multilevel models, hierarchical models, and individual growth models. Journal of Educational and Behavioral Statistics1998; 24: 323-55.
30.
Moerbeek M. , Van Breukelen GJP, Berger MPF.Design issues for experiments in multilevel populations. Journal of Educational and Behavioral Statistics2000; 25: 271-84.
31.
Raudenbush SW, Xiao-Feng L.Statistical power and optimal design for multisite randomized trials. Psychological Methods2000 ; 5: 199-213.
32.
Tan ES, Berger MPF.Optimal allocation of time points for the random effects model. Communications in Statistics: Simulation and Computation1999 ; 28: 517-40.
33.
Rao CRPrediction of future observations with special reference to linear models . In: Krishnaiah PR ed. Multivariate analysis - IV. North-Holland Publishing Company , 1997: 193-208.
