Doubly Robust-Based Generalized Estimating Equations for the Analysis of Longitudinal Ordinal Missing Data

TL;DR

This paper introduces a new doubly robust estimator for analyzing longitudinal ordinal data with missing responses, combining inverse probability weighting and multiple imputation to improve accuracy under the MAR mechanism.

Contribution

It proposes a novel doubly robust GEE method specifically designed for ordinal longitudinal data with missing values, enhancing robustness and performance.

Findings

Simulation shows improved accuracy over existing methods

Method performs well with intermittent missing data

Applied successfully to childbirth analgesia data

Abstract

Generalized Estimation Equations (GEE) are a well-known method for the analysis of non-Gaussian longitudinal data. This method has computational simplicity and marginal parameter interpretation. However, in the presence of missing data, it is only valid under the strong assumption of missing completely at random (MCAR). Some corrections can be done when the missing data mechanism is missing at random (MAR): inverse probability weighting (WGEE) and multiple imputation (MIGEE). In order to obtain consistent estimates, it is necessary the correct specification of the weight model for WGEE or the imputation model for the MIGEE. A recent method combining ideas of these two approaches has doubly robust property. For consistency, it requires only the weight or the imputation model to be correct. In this work it is assumed a proportional odds model and it is proposed a doubly robust estimator for the analysis of ordinal longitudinal data with intermittently missing response and covariate under the MAR mechanism. Simulation results revealed better performance of the proposed method compared to WGEE and MIGEE. The method is applied to a data set related to Analgesia Pain in Childbirth study.