Hierarchical Logistic Regression Modeling with SAS GLIMMIX

Data often have hierarchical or clustered structures, such as patients clustered within hospitals or students nested within schools. Hierarchical models are statistical models that are used to analyze hierarchical or multilevel data. SAS GLIMMIX procedure is a new and highly useful tool for hierarchical modeling with discrete responses. This paper is focused on hierarchical logistic regression modeling with GLIMMIX. We present several applications of these models and show how to use GLIMMIX to fit the models and test hypotheses. We illustrate the applications using a sample data from a multi-institution database on coronary artery bypass grafting surgeries developed by the California Office of Statewide Health Planning and Development.