Introduction to Joint Modeling of Longitudinal Biomarkers and Kidney Disease Outcomes

• 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): Tom Greene and Ken Boucher*+
• Companies: University of Utah and University of Utah
• Keywords: longitudinal analysis ; truncation by death ; joint models ; informative censoring ; ROC analysis

Abstract

Analysis of longitudinal outcomes in cohorts with chronic kidney disease is complicated by high rates of attrition due to death and end stage renal disease (ESRD). Methods that treat death or ESRD as censoring events are problematic, as the analysis then implicitly imputes longitudinal outcomes after these terminating events. We present two complementary approaches that circumvent this difficulty. The first integrates data from the longitudinal and time-to-event outcomes to characterize evolution of cohorts based on changes in the distribution of patients across clinically relevant states over time. The second approach again considers the multivariate outcome defined by the clinical events and the longitudinal outcome, but focuses on patient trajectories over time rather than the distribution of states across patients at specific time points. We consider flexible joint models which incorporate potentially nonlinear trajectories in the longitudinal outcome and terminating clinical events in the same joint model, and examine the use of time-dependent ROC analysis to characterize relationships between the longitudinal outcome and the clinical events.