Bias Correction in the Hierarchical Likelihood Approach to the Analysis of Multivariate Survival Data

• Activity Number: 588
• Type: Contributed
• Date/Time: Wednesday, August 4, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Section on Risk Analysis
• Author(s): Jihyoun Jeon*+ and Li Hsu and Malka Gorfine
• Companies: Fred Hutchinson Cancer Research Center and Fred Hutchinson Cancer Research Center and Technion - Israel Institute of Technology
• Keywords: Frailty model ; Hierarchical likelihood ; Semiparametric ; NPMLE ; Heterogeneity ; Multivariate survival

Abstract

Frailty models in survival analysis are useful for measuring unobserved heterogeneity in risk of failures across clusters, providing cluster-specific risk prediction. In a frailty model, the latent frailty variable shared by members within a cluster often acts multiplicatively to the hazard function, which results in an extended Cox proportional hazards model. In order to obtain parameter estimates as well as frailty variate estimates, we consider the hierarchical likelihood approach (H-likelihood), in which the latent frailties are treated as ``parameters'' and estimated jointly with other parameters of interest. We propose a bias correction method for the H-likelihood estimators under a shared frailty model, and also extend the method to a multivariate frailty model. Our simulation studies show that the proposed approach performs very well for censoring rates as high as 80\%.