Simulation Study of a Joint Modeling Approach to Assess the Potential Effect of Biomarker Variability on Clinical Outcomes

• Activity Number: 64
• Type: Contributed
• Date/Time: Sunday, August 2, 2009 : 4:00 PM to 5:50 PM
• Sponsor: Biometrics Section
• Author(s): Feng Gao*+ and Philip Miller and Mae Gordon and Chengjie Xiong
• Companies: Washington University in St. Louis and Washington University in St. Louis and Washington University in St. Louis and Washington University
• Keywords: linear mixed model with patient-specific variance ; Cox proportional hazard model ; joint-model of longitudinal and survival data

Abstract

In clinical and epidemiology studies, sometimes investigators are interested to see whether biomarker variability is an independent prognostic predictor of a clinical outcome. In many applications, the variability is represented by the standard deviation of several repeated measurements from a subject, and Cox model is often used to assess its effect on the outcome, while treating such a sample-based estimate as a covariate free of measurement error. In Cox model, however, it is well known that measurement error in covariates causes underestimation of the true effects and may mislead one into believing that an actually good predictive biomarker is not related to the outcome. In this study, via simulated data, we evaluated the performance of a joint model that incorporates subject-specific variance and also evaluated the performance of different estimates for biomarker variability.