Title: Likelihood based analysis of longitudinal discrete data with overdispersion
Authors: Justine Shults - Perelman School of Medicine at University of Pennsylvania and Children's Hospital of Philadelphia (United States) [presenting]
Abstract: The focus is on longitudinal discrete data that may be unequally spaced in time and may exhibit overdispersion, so that the variance of the outcome variable is inflated relative to its assumed distribution. We propose an approach that extends generalized linear models for analysis of longitudinal data and is likelihood based, in contrast to generalized estimating equations (GEE) that are semi-parametric. We demonstrate application of the method in an analysis of seizure counts and kidney transplant centers. Simulations for both studies show that the likelihood based approach outperforms GEE, especially as the degree of intra-subject correlation and over-dispersion in the outcome distribution increases.