A Bayesian Exposure-Time Method for Clinical Trial Safety Monitoring With Blinded Data

Abstract

The FDA safety reporting Final Rule requires an expedited safety report whenever aggregate analysis indicates a clinically meaningful imbalance with an adverse event occurring more frequently in the drug treatment group than in a concurrent or historic control group. We introduce a safety monitoring procedure for two-arm blinded clinical trials that can be used to help address new requirements from the recent FDA safety reporting Final Rule. This procedure incorporates a Bayesian hierarchical exposure-time model for using prior information and blinded event data to make inferences on the rate of adverse events of special interest in the test treatment arm. We describe a collaborative process and provide free software for eliciting the required prior information and calibrating operating characteristics through simulation. We illustrate the use of our procedure with a case study composed of a combination of real and simulated data. Our procedure provides good operating characteristics for detecting higher than expected rates of adverse events in the drug treatment group, and is appropriate for inferring the rate of adverse events in multi-armed clinical trials with blinded data.