B1776
Title: Studying heart failure progression through Bayesian multi-state survival models
Authors: Jesus Gutierrez-Botella - Universidade de Santiago de Compostela (Spain) [presenting]
Maria Pata - Biostatech Advice Training and Innovation in Biostatistics SL (Spain)
Carmen Armero - Universitat de Valencia (Spain)
Thomas Kneib - University of Goettingen (Germany)
Francisco Gude - Complexo Hospitalario Universitario de Santiago de Compostela (Spain)
Abstract: Heart failure (HF) is a progressive disease caused by the inability of the heart to give the body an adequate oxygen supply. Although it is a chronic condition, cardiac resynchronization therapy (CRT) has been shown to have benefits on the short-term prognosis of HF patients. The objective is to discuss the temporal evolution of HF patients treated with CRT in relation to their demographic and clinical variables. That progression includes transient states such as congestive heart failure or atrial fibrillation as intermediate states of the disease, and an absorbing state associated with death. The different survival times (times between consecutive transitions) have been modelled through Cox regression with Weibull and piecewise constant baseline hazard models, and covariates selected by the medical team. Bayesian methods have been used to estimate the parameters of the full multi-state model and Markov chain Monte Carlo (MCMC) methods have been employed to approximate the relevant posterior distribution through JAGS Software.