Online Program

A Bayesian Hierarchical Summary Receiver Operating Characteristic Model for Network Meta-Analysis of Diagnostic Tests

*Qinshu Lian, University of Minnesota 
Xiaoye Ma, University of Minnesota 
Haitao Chu, University of Minnesota 

Keywords: multiple diagnostic tests, Bayesian hierarchical models, missing data, network meta-analysis, hierarchical summary receiver operating characteristic models

In the studies of the accuracy of multiple diagnostic tests, three designs are commonly used: (1) the multiple test comparison design; (2) the randomized design; and (3) the noncomparative design. Yet the handful methods, which can compare multiple tests simultaneously in a meta-analysis setting, only consider the simple case when all or none of the reference tests can be considered as a gold standard test. In this paper, we extend the Bayesian hierarchical summary receiver operating characteristic model for network meta-analysis of diagnostic tests (HSROC-NMADT) to compare multiple tests simultaneously. A missing data framework is employed to combine studies with and without a gold standard test, as well as studies with different designs. Our method accounts for the potential correlation between multiple tests within a study and heterogeneity across studies. In addition, it allows studies to perform different subsets of the diagnostic tests and provides flexibility on the choice of the summary statistics. The performance of HSROC-NMADT is evaluated through simulation studies and illustrated using real data on network meta-analysis of deep vein thrombosis tests.