GEE with Gaussian estimation of the correlations when data are incomplete

This paper considers a modification of generalized estimating equations (GE E) for handling missing binary response data. The proposed method uses Gaus sian estimation of the correlation parameters. i.e., the estimating functio n that yields an estimate of the correlation parameters is obtained from th e multivariate normal likelihood. The proposed method yields consistent est imates of the regression parameters when data are missing completely at ran dom (MCAR). However, when data are missing at random (MAR), consistency may not hold. In a simulation study with repeated binary outcomes that are mis sing at random, the magnitude of the potential bias that can arise is exami ned. The results of the simulation study indicate that, when the working co rrelation matrix is correctly specified, the bias is almost negligible for the modified GEE. In the simulation study, the proposed modification of GEE is also compared to the standard GEE. multiple imputation, and weighted es timating equations approaches. Finally, the proposed method is illustrated using data from a longitudinal clinical trial comparing two therapeutic tre atments, zidovudine (AZT) and didanosine (ddI), in patients with HIV.