Background: Network meta-analyses have extensively been used to compare the effectiveness of multiple interventions for healthcare policy and decision making. However, methods for evaluating the performance of multiple diagnostic tests are less established. In a decision-making context, we are often interested in comparing the performance of multiple diagnostic tests, at varying levels of test thresholds, in one simultaneous analysis.

Objectives: To develop a network meta-analysis framework in which diagnostic test accuracy data from multiple tests and thresholds can be synthesised and ranked in a single coherent analysis.

Methods: Motivated by an example of cognitive impairment diagnosis following stroke, we synthesised data from 13 studies assessing the efficiency of two diagnostic tests: Mini-Mental State Examination (MMSE) and Montreal Cognitive Assessment (MoCA), at two test thresholds: MMSE <25/30 and <27/30, and MoCA <22/30 and <26/30. We fitted a bivariate network meta-analysis model to account for the correlation between paired measures of test accuracy, i.e. sensitivity and specificity. Building on this model, we further incorporated constraints on increasing test thresholds, assuming that higher-test thresholds had an increased sensitivity but decreased specificity. All models were fitted in WinBUGS using Bayesian Markov Chain Monte Carlo (MCMC) methods.

Results: MoCA at threshold <26/30 appeared to have the optimal true positive rate (estimated sensitivity: 0.98; 95% credible interval (CrI): 0.93, 0.99), whilst MMSE at threshold <25/30 appeared to have the optimal true negative rate (estimated specificity: 0.82, 95%CrI: 0.73, 0.89). Both of which ranked in first place for 99% of MCMC iterations. Applying constraints on increasing test thresholds reduced between-study heterogeneity and increased the precision in estimates of sensitivity and specificity.

Conclusions: In a health-technology assessment setting, there is an increasing need to compare the efficiency of multiple diagnostic tests. Use of a bivariate network meta-analysis allows us to compare and rank all tests and thresholds of interest for healthcare policy and decision making.