Keywords: classification models; multivariate longitudinal data; health-related quality of life
Multivariate longitudinal data, in which multiple outcomes are repeatedly measured at two or more occasions, are commonly collected in health-related quality of life (HRQOL) studies. Researchers are often interested in evaluating differences in longitudinal trajectories of multiple outcomes. This study investigates the accuracy of classification models based on mixed-effects models and generalized estimating equations for discriminating between groups based on longitudinal trajectories of multiple domains of HRQOL. Monte Carlo methods were used to assess the effects of sample size, the number of outcomes, covariance heterogeneity, data distribution on the accuracy of longitudinal discriminant analysis and quadratic inference classifiers in discriminating between population groups over time. We present an example using data from a longitudinal study of individuals with coronary artery disease where the interest is in predicting mortality. Recommendations for guiding researchers’ choice of an appropriate classifier under a variety of data analytic conditions will be provided.