Comparison of Meta-Analytic Methods That Combine Direct and Indirect Evidence to Assess Dose Relationship

• Activity Number: 514
• Type: Contributed
• Date/Time: Wednesday, August 4, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Qi Zhang and Mary Nilsson and Yoko Tanaka and Brenda Crowe and Xiaobi Huang*+
• Companies: Eli Lilly and Company and Eli Lilly and Company and Eli Lilly and Company and Eli Lilly and Company and University of Michigan
• Keywords: dose response ; direct and indirect comparison ; meta analysis ; simulation

Abstract

In clinical trials, it is often of interest to assess dose response relationship for certain adverse events through combination of all the evidence from the clinical trials with various dose designs. Direct comparison can lead to the exclusion of a lot of the trials; indirect comparison is prone to bias due to loss of randomization. This talk will present the results of a simulation study that compared Gleser's regression method, a Bayesian mixed treatment comparison model and standard logistic regression for comparison of log odds ratios by combing both direct and indirect evidence. We found that the usual logistic regression method had slightly larger bias compared with the other 2 methods. The Bayesian model had the smallest bias and is flexible to implement. In order to account for duration of exposure, we then further extend both methods to survival models.