Bayesian inference for multivariate longitudinal data analysis using robust distributions
Many clinical trials collect information on multiple longitudinal outcomes such as Parkinson's disease. As a method that account for all the information from multiple longitudinal outcomes, the multilevel item response theory (MLIRT) models have been increasingly used in clinical studies. The MLIRT models account for all the information from multiple longitudinal outcomes of mixed types (e.g. continuous, binary and ordinal) and can provide valid inference for the overall treatment effects. However, the continuous outcomes and the random effects in the MLIRT models are often assumed to be normally distributed. The normality assumption can be violated due to skewness or outliers and thus may produce misleading results. In addition, patients' follow-up in the longitudinal studies may be stopped by terminal events such as death or dropout due to disease progression. The normal/independent (NI) distribution and the skew-normal/independent (SNI) distribution has been increasingly used to handle the outlier and skewness problem to produce robust inference. In the first paper, we developed approaches that implement the NI distributions on both continuous outcomes and random effects in the MLIRT models and discussed the model performance on different strategies of implementing the NI distribution. In the second paper, we developed approaches that implement the SNI distributions to the joint MLIRT models framework and evaluated the performance of the models on all the three data features skewness, outliers and dependent censoring. Extensive simulation studies were conducted to evaluate the performance of various models. Specifically, we considered two continuous outcomes and two ordinal outcomes with the first outcome have outliers in the first paper; and we considered one continuous outcome and two ordinal outcomes with the continuous outcome has outliers and skewness in the second paper. Our proposed model Indep-CN-MLIRT in paper 1 and JMST model in paper 2 performed significantly better than their corresponding competing models. Our proposed methods were applied to a motivating Parkinson's disease study, the DATATOP study, to investigate the effect of deprenyl in slowing down the disease progression. ^
Chen, Geng, "Bayesian inference for multivariate longitudinal data analysis using robust distributions" (2014). Texas Medical Center Dissertations (via ProQuest). AAI3689749.