Title: Joint modelling of longitudinal and survival data
Authors: Eleni-Rosalina Andrinopoulou - Erasmus Medical Center (Netherlands) [presenting]
Abstract: In epidemiological follow-up studies different types of outcomes are typically collected for each individual. These include longitudinally measured responses (e.g., biomarkers), and the time until an event of interest occurs (e.g., death, intervention). Often these outcomes are separately analysed, but in many occasions, it is of scientific interest to study their association. This type of interest has given rise in the class of joint models for longitudinal and time-to-event data. Joint models can be used when focusing either on the survival outcome when we wish to account for the effect of an endogenous time-dependent covariate or on the longitudinal outcome, and we wish to correct for non-random dropout. The idea behind these models is to couple a survival model for the time-to-event process with a mixed-effects model for the longitudinal outcome. Several extensions of the standard joint model that consists of one longitudinal and one survival outcome have been proposed including among others the use of multiple longitudinal outcomes and the investigation of different manners to associate the longitudinal and the survival process. Several applications of these type of models will be discussed.