Categorical Data Analysis Using The SAS® System, 2nd Edition

15.9. Using GEE for Count Data

Sometimes, categorical data come in the form of count data. For example, you may record the number of acute pain episodes in a time interval in a clinical trial evaluating treatments. Other examples might be the number of insurance claims registered during the year or the number of unscheduled medical visits made during a study of a new protocol for asthma medication. Often, Poisson regression is the appropriate strategy for analyzing such data. See Chapter 12, “Poisson Regression,” for further discussion of Poisson regression.

Since Poisson regression is an application of the generalized linear model with the Poisson distribution and the log link function, you can fit models for clustered or repeated data with ...