Presented by: by Dr. Shouhao Zhou, PhD, Assistant Professor, Division of Biostatistics and Bioinformatics, Public Health Sciences, and Zizhong Tian, MS, PhD Candidate of Biostatistics and Bioinformatics

Information on drug safety is critical to health care and policy decisions. For evidence synthesis, meta-analysis offers a unique opportunity to assess adverse event (AE) risks in a large sample size for a broad population, but poses significant challenges and requires careful consideration. In particular, many AEs are rare, and AEs which occur infrequently may be omitted from official trial reports. The observed data therefore correspond to low counts which may be subject to informative censoring.

We propose a novel Bayesian modeling framework, named Meta-analysis of Adverse Drug Effects with data Censoring (MAGEC), to synthesize clinical toxicity data. It can allow for incomplete reporting and handle hierarchical multivariate outcome data. Additionally, it is capable of incorporating AE data reported under different severity grades, and conducting comprehensive drug-wise comparison over multiple AE categories while controlling for false discovery. We illustrate our method by applying it to a meta-analysis of adverse events in cancer immunotherapy and targeted therapy.