Bayesian Safety Monitoring and Stopping Rules with Incorporation of Historical Data in Medical Device Clinical Trials

• Activity Number: 37
• Type: Contributed
• Date/Time: Sunday, July 29, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Zengri Wang*+
• Keywords: Bayesian Safety Monitoring ; Clinical Trials ; Historical Data ; Medical Devices ; Operating Characteristics ; Stopping Rule

Abstract

There is increasing interest in applying Bayesian statistical methodology in monitoring and evaluating safety outcomes in medical device clinical trials. The flexibility of incorporating previous and existing knowledge of a study device into risk quantification is well suited in safety monitoring for an ongoing clinical trial, especially when safety data on a device is limited. With appropriate integration of historical data, estimation precision of a target risk quantity may be improved and robust conclusions may be reached early. In this presentation, a Bayesian safety monitoring and decision-analysis framework is proposed for a neurostimulation clinical trial allowing for incorporation of historical data into explicit safety stopping rules. The prior distribution is discussed and critical safety threshold values are justified based on the balance between unnecessarily stopping early and mistakenly stopping late. Additionally, safety decision operating characteristics with limited or no previous data integration is investigated and compared.