Online Program

The Impact of Covariance Priors on Arm-based Bayesian Network Meta-Analyses with Binary Outcomes (306649)

Haitao Chu, Division of Biostatistics, University of Minnesota 
James S. Hodges, Division of Biostatistics, University of Minnesota 
Lifeng Lin, Department of Statistics, Florida State University 
*Zhenxun Wang, Division of Biostatistics, University of Minnesota 

Keywords: Bayesian inference, covariance matrix, network meta-analysis, prior

Performing Bayesian analyses with the arm-based network meta-analysis (NMA) model requires researchers to specify the prior distribution for the covariance matrix of the study-specific underlying treatment effects. The commonly-used conjugate prior for the covariance matrix, the inverse-Wishart (IW) distribution, has several limitations. For example, although the IW distribution is often described as a weakly-informative prior, it may lead to underestimation of correlations between treatments, which are critical for making efficient and less biased estimates of treatment effects by borrowing strength across treatment arms. Alternatively, several separation strategies can be considered for modeling the covariance matrix. To study the impact of the IW prior on network meta-analysis results and compare it with separation strategies, we perform simulation studies under different mechanisms of missing treatments and we find that a separation strategy with appropriate priors for the correlation matrix and variances separately performs better. Finally, we illustrate the importance of sensitivity analyses with different prior specifications when performing NMA by re-analyzing case studies.