Hierarchical Bayesian Methods for Combining Efficacy and Safety in Multiple Treatment Comparisons
View Presentation View Presentation
*Bradley P. Carlin, University of Minnesota  Haitao Chu, University of Minnesota  Hwanhee Hong, University of Minnesota 

Keywords: Bayesian methods, Indirect comparisons, Missing data,

Biomedical decision makers confronted with questions about the comparative effectiveness and safety of interventions often wish to combine all sources of data. Such multiple treatment comparisons (MTCs) often rely largely on indirect comparisons. In such settings, hierarchical Bayesian meta-analytic methods offer a natural approach (e.g., by enabling full posterior inference on the probability that a given treatment is best). We summarize the current state of such methods for binary and continuous responses, and consider extension to multiple outcomes in a missing data framework. We also propose a new arm-based model that is less constrained than existing models. We illustrate our methods with data from two recent MTCs, one comparing pharmacological treatments for female urinary incontinence, and another on physical therapy interventions for knee pain secondary to osteoarthritis.