A GEE Approach for Estimating Correlation Coefficients Involving Left-censored Variables

Jingli Song,Huiman X Barnhart,Robert H Lyles

doi:10.6339/jds.2004.02(3).166

Abstract

HIV (Human Immunodeficiency Virus) researchers are often con cerned with the correlation between HIV viral load measurements and CD4+ lymphocyte counts. Due to the lower limits of detection (LOD) of the avail able assays, HIV viral load measurements are subject to left-censoring. Mo tivated by these considerations, the maximum likelihood (ML) method under normality assumptions was recently proposed for estimating the correlation between two continuous variables that are subject to left-censoring. In this paper, we propose a generalized estimating equations (GEE) approach as an alternative to estimate such a correlation coefficient. We investigate the robustness to the normality assumption of the ML and the GEE approaches via simulations. An actual HIV data example is used for illustration.

Highlights

Viral load assessment via quantification of plasma viral RNA (Ribonucleic Acid) plays an important role in current HIV (Human Immunodeficiency Virus) research
We have proposed a generalized estimating equations (GEE) approach for estimating the correlation between two continuous variables, where one variable is subject to left censoring
We investigated the robustness to the normality assumption for both the maximum likelihood (ML) approach and the GEE approach via simulations

Summary

Introduction

Viral load assessment via quantification of plasma viral RNA (Ribonucleic Acid) plays an important role in current HIV (Human Immunodeficiency Virus) research. It has provided valuable insights into the pathogenesis of HIV disease and the activity of anti-viral drugs. Inherent limits of detection (LOD) in existing HIV RNA assays lead to the possibility of left-censored ( termed missing) data. Such left-censored data is characteristic of many other types of bioassay studies (Lynn, 2001). There is a need to estimate the correlation between two variables, where one of them may be left-censored

Methods

Results

Conclusion