Mining adverse events in large frequency tables with ontology, with an application to the vaccine adverse event reporting system

Abstract

Many statistical methods have been applied to VAERS (vaccine adverse event reporting system) database to study the safety of COVID-19 vaccines. However, none of these methods considered the adverse event (AE) ontology. The AE ontology contains important information about biological similarities between AEs. In this paper, we develop a model to estimate vaccine-AE associations while incorporating the AE ontology. We model a group of AEs using the zero-inflated negative binomial model and then estimate the vaccine-AE association using the empirical Bayes approach. This model handles the AE count data with excess zeros and allows borrowing information from related AEs. The proposed approach was evaluated by simulation studies and was further illustrated by an application to the Vaccine Adverse Event Reporting System (VAERS) dataset. The proposed method is implemented in an R package available at https://github.com/umich-biostatistics/zGPS.AO.