Bayesian models for survival data of clinical trials: Comparison of implementations using R software

Lucie Biard, Anne Bergeron, Sylvie Chevret

Published 2019-08-19Version 1

Objective: To provide guidance for the use of the main functions available in R for performing \emph{post hoc} Bayesian analysis of a randomized clinical trial with a survival endpoint using proportional hazard models. Study Design and Setting: Data derived from the ALLOZITHRO trial, conducted with 465 patients after allograft to prevent pulmonary complications and allocated between azithromycin and placebo; airflow decline--free survival at 2 years after randomization was the main endpoint. Results: Despite heterogeneity in modeling assumptions, in particular for the baseline hazard (parametric or nonparametric), and in estimation methods, Bayesian posterior mean hazard ratio (HR) estimates of azithromycin effect were close to those obtained by the maximum likelihood approach. Conclusion: Bayesian models can be implemented using various R packages, providing results in close agreement with the maximum likelihood estimates. These models provide probabilistic statements that could not be obtained otherwise.