Some Methodological Issues on Glomerular Filtration Rate Progression Trajectories Among Patients with Chronic Kidney Disease

• 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): Liang Li*+
• Companies: Cleveland Clinic
• Keywords: Bayesian ; joint modeling ; longitudinal ; multistate ; survival ; chronic kidney disease

Abstract

Patients with chronic kidney disease (CKD) often have their glomerular filtration rate (GFR) monitored till the occurrence of a terminal event such as the end stage renal disease or death. In the traditional paradigm, it is often assumed that GFR declines relentlessly over time in a near linear fashion. Our recent research suggests that a substantial proportion of CKD patients may have nonlinear GFR trajectories with both periods of stable and declining GFR. In this talk, we discuss four methodological development related to this finding: (1) a Bayesian approach to measure and detect nonlinear GFR trajectory; (2) a within-patient case-control design to compare the stable and decline periods of GFR trajectory and identify time-dependent risk factors of CKD progression, removing both measured and unmeasured confounding at baseline; (3) a class of joint models of the nonlinear GFR trajectory and terminal clinical events, accommodating possibly high dimensional random effects; (4) a multistate representation of both longitudinal biomarker data and clinical events, as both a descriptive and inferential tool to summarize a longitudinal cohort.