GEECORR: A SAS macro for regression models of correlated binary responses and within-cluster correlation using generalized estimating equations

TL;DR

The paper introduces GEECORR, a SAS macro that extends GEE methods to model correlated binary data by including within-cluster correlation, enabling flexible analysis of marginal means and correlations.

Contribution

It presents a new SAS macro, GEECORR, that implements an extended GEE method for correlated binary responses, including correlation modeling.

Findings

Successfully applied to three datasets

Demonstrated efficiency through simulation studies

Allows flexible modeling of mean and correlation

Abstract

A SAS macro, GEECORR, has been developed for the analysis of correlated binary data based on the Prentice (1988) estimating equations method that extends the Liang and Zeger (1986) generalized estimating equations (GEE) method to include additional estimating equations for the pairwise correlation between binary variates. This extension allows for flexible modeling of both the marginal mean and within-cluster correlation as a function of their respective covariate risk factors. This paper provides an overview of the extended estimating equations method, describes the features and capabilities of the GEECORR macro, and applies the GEECORR macro to three different datasets. In addition, this paper describes the more detailed fitting algorithm proposed by Prentice (1988), of which a variation has been implemented in the GEECORR macro. We provide a small simulation study to demonstrate the efficiency of the detailed method for estimating correlation parameters.