Online Program

Risk Assessment with Near Zero Informed Bayesian (NZIB) for Idiosyncratic Adverse Drug Reaction (IADR) Events: Uncertain and Rare AEs Beyond DLT (308498)

Keywords: Near Zero Informed Bayesian (NZIB), Idiosyncratic Adverse Drug Reaction (IADR), biostatistics, Bayesian, clinical trial safety, risk assessment

In clinical study, idiosyncratic adverse drug reaction (IADR) is a type of drug reactions that is uncommon, dose-independent and unpredictable with high mortality and morbidity. Due to this uncertainty, IADRs become a major factor for the failure of investigational new drug development. Before and during a clinical study, new drug investigators and sponsors rely on the stopping rule based on empirical opinion from domain experts without appropriate statistical consideration. Consequently, a risk of having certain number of IADRs in a clinical trial is unknown and not properly quantified. In this project, we compared available methods and proposed a Near Zero Informed Bayesian (NZIB) model to account for the rare property of IADRs and integrate experimental data with historical knowledge. The proposed NZIB model outperforms the other available models with better risk estimates and well controlled type I and type II errors.