A Variance Shrinkage Method Improves the Arm-Based Bayesian Network Meta-Analysis

• Activity Number: 323 - Estimating Treatment Effects: Applications in Health Policy
• Type: Contributed
• Date/Time: Wednesday, August 5, 2020 : 10:00 AM to 2:00 PM
• Sponsor: Health Policy Statistics Section
• Author(s): Zhenxun Wang* and Lifeng Lin and James S. Hodges and Richard MacLehose and Haitao CHU
• Companies: Division of Biostatistics, University of Minnesota and Florida State University and Division of Biostatistics, University of Minnesota and Division of Epidemiology and Community Health, University of Minnesota and the University of Minnesota Twin Cities
• Keywords: Bayesian inference; variance prior; network meta-analysis; variance shrinkage method

Abstract

Network meta-analysis (NMA) is a useful tool to combine direct and indirect evidence in systematic reviews of multiple treatments to improve estimation compared to traditional pairwise meta-analysis. Unlike the contrast-based approach, the arm-based NMA approach can estimate absolute risks, which are more useful in public health. However, the number of clinical studies involving each treatment is often small in an NMA, leading to unstable treatment-specific variance estimates in the AB-NMA when using non-informative priors under unequal variance assumption. Additional assumptions, such as equal variances for all treatments, may be used to remedy this problem but such assumption may be inappropriately strong. This article introduces a variance shrinkage method for an AB-NMA. Specifically, we assume different treatment variances share a common prior with unknown hyper-parameters. This assumption is weaker than the homogeneous-variance assumption and improves estimation by shrinking the variances in a data-dependent way. We illustrate the advantages of our method by re-analyzing an NMA of different organised inpatient care interventions for stroke and comprehensive simulation studies.