Time-Dependent ROC Analysis for Early Detection of End-Stage Renal Disease (ESRD) Using Baseline Glomerular Filtration Rate

• Activity Number: 581
• Type: Topic Contributed
• Date/Time: Wednesday, August 7, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistical Consulting
• Author(s): Nan Hu*+
• Companies: University of Utah
• Keywords: receiver operating characteristic ; chronical kidney disease ; biomarkers ; logitudinal data ; time-varying ; covariate adjusted

Abstract

In clinical trials of chronical kidney diseases (CKDs), biomarkers for the end stage renal disease (ESRD) can often be able to measured at baeline and measured longitudinally up until the time to either ESRD or death. Thus, the prognistic ability of the biomarkers can be evaluated under a time-dependent framework. In this talk, I will present the statistical methodology of extending the traditional ROC analysis in a time varying fashin to accommodate the prognosis of ESRD as a complex end disease status for patients with CKD. The presenter will also illustrate the methods developed to evaluate the biomarker performance for prognosing disease outcomes in African American Study of Kidney disease (AASK), a clinical trial with follow-up to ESRD or death. In addition, methods of covariate adjustment in time-dependent ROC analyses will also be given in the talk.