Abstract:
|
Bayesian meta-analysis has been frequently utilized for synthesizing safety and efficacy information to support landmark decision-making due to its flexibility of incorporating prior information and availability of computing software. However, when outcomes are binary and events are rare, that is zero event counts appear in either one or both study arms, conventional Bayesian meta-analysis methods cannot be directly applied unless artificial continuity corrections are made to zero cells, which may cause biases. To better model zero counts without artificial continuity corrections, we propose a novel Bayesian method, Beta prior BInomial model for Risk Differences (B-BIRD). Simulation studies have shown that compared with the exact method (Tian et al. 2009) the credible intervals obtained from B-BIRD have comparable coverage rates but shorter interval lengths when the event rates are within the (0, 1%) range.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.