Online Program Home
My Program

Abstract Details

Activity Number: 196 - SPEED: Biometrics and Biostatistics Part 2
Type: Contributed
Date/Time: Monday, July 29, 2019 : 11:35 AM to 12:20 PM
Sponsor: Section on Bayesian Statistical Science
Abstract #307590
Title: Bayesian Modeling of Rare Events with Informative Censoring in Meta-Analysis
Author(s): Xinyue Qi* and Yucai Wang and Chan Shen and Michael Wang and Shouhao Zhou
Companies: UT MD Anderson Cancer Center and Mayo Clinic and College of Medicine, Penn State University and The University of Texas MD Anderson Cancer Center and PennState College of Medicine
Keywords: Bayesian hierarchical model; censoring; rare events; meta-analysis

This study is motivated by a meta-analysis for drug safety in clinical trials, when a large number of rare adverse events (AEs) are not reported if they are less frequently observed. As a typical missing not at random problem, is the censored information ignored, the inference on incidence rate of AEs would be overestimated by nearly 40%. We propose a modified Bayesian multilevel logistic regression model to accommodate the censored sparse binomial event data, and implement in JAGS based on a tailored modeling strategy. We conduct simulation studies to examine the performance of our proposed Bayesian model compared to other popular methods in finite samples under four scenarios. The proposed approach is illustrated using data from a recent meta-analysis of 87 clinical trials involving PD-1/PD-L1 inhibitors with respect to their toxicity profiles.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2019 program