Bayesian Meta-Analysis of Rare Events Incorporating Historical Controls

• Activity Number: 239
• Type: Contributed
• Date/Time: Monday, August 2, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Yan Zheng*+ and Xin Zhi
• Companies: sanofi-aventis and sanofi-aventis
• Keywords: Meta-analysis ; Bayesian approach ; rare event ; historical control

Abstract

In clinical trials, we frequently encounter the problem to compare the event rate between treatment groups (e.g. control vs. active treatment and/or low dose vs. high dose). When the event is rare, the variation associated with the estimated event rate can be quite large compared to the event rate itself and could lead to alarming treatment difference by pure chance. It may be difficult to find a large number of historical studies containing all treatment groups of interest, especially if the active treatment/high dose treatment was recently developed. As a result, commonly used meta-analysis methods which are based on large sample approximation may not be appropriate. We develop a Bayesian approach of meta-analysis and compare the new method to commonly used approaches through simulations.