A comparison of several approaches for choosing between working correlation structures in generalized estimating equation analysis of longitudinal binary data

• Published in: Statistics in medicine Vol. 28; no. 18; pp. 2338 - 2355
• Main Authors: Shults, Justine, Sun, Wenguang, Tu, Xin, Kim, Hanjoo, Amsterdam, Jay, Hilbe, Joseph M., Ten-Have, Thomas
• Format: Journal Article
• Language: English
• Published: Chichester, UK John Wiley & Sons, Ltd 15.08.2009
• Subjects: Antidepressive Agents, Second-Generation - therapeutic use; Biometry; Computer Simulation; correlated binary data; Cyclohexanols - therapeutic use; Depressive Disorder, Major - drug therapy; first-order autoregressive correlation structure; generalized estimating equations; Humans; Linear Models; Lithium Compounds - therapeutic use; longitudinal data; Longitudinal Studies; longitudinal study; Prospective Studies; Randomized Controlled Trials as Topic - statistics & numerical data; Regression Analysis; Venlafaxine Hydrochloride
• ISSN: 0277-6715; 1097-0258; 1097-0258
• DOI: 10.1002/sim.3622

The method of generalized estimating equations (GEE) models the association between the repeated observations on a subject with a patterned correlation matrix. Correct specification of the underlying structure is a potentially beneficial goal, in terms of improving efficiency and enhancing scientific understanding. We consider two sets of criteria that have previously been suggested, respectively, for selecting an appropriate working correlation structure, and for ruling out a particular structure(s), in the GEE analysis of longitudinal studies with binary outcomes. The first selection criterion chooses the structure for which the model‐based and the sandwich‐based estimator of the covariance matrix of the regression parameter estimator are closest, while the second selection criterion chooses the structure that minimizes the weighted error sum of squares. The rule out criterion deselects structures for which the estimated correlation parameter violates standard constraints for binary data that depend on the marginal means. In addition, we remove structures from consideration if their estimated parameter values yield an estimated correlation structure that is not positive definite. We investigate the performance of the two sets of criteria using both simulated and real data, in the context of a longitudinal trial that compares two treatments for major depressive episode. Practical recommendations are also given on using these criteria to aid in the efficient selection of a working correlation structure in GEE analysis of longitudinal binary data. Copyright © 2009 John Wiley & Sons, Ltd.