Longitudinal Data Analysis

In this chapter, we describe statistical methods for analyzing data from longitudinal studies in which the response from each individual is a sequence of repeated measurements on an outcome variable of scientific interest. We use the setting of the linear model to show how to handle two of the most important ways in which longitudinal studies differ from cross-sectional studies: the non-independence of repeated measurements on the same individual; and the treatment of missing values. We illustrate the methods with an analysis of repeated measurement data from a clinical trial of drug therapies for schizophrenia patients and with an observational study of factors affecting the long-term progression of lung function in cystic fibrosis patients. We describe generalized linear models for repeated count or binary responses and joint models for repeated measurements and a time-to-event outcome.