Hierarchical Modeling for Diagnostic Test Accuracy Using Multivariate Probability Distribution Functions

Models implemented in statistical software for the precision analysis of diagnostic tests include random-effects modeling (bivariate model) and hierarchical regression (hierarchical summary receiver operating characteristic). However, these models do not provide an overall mean, but calculate mean a central study when random effect is equal to zero; hence, it difficult covariance between sensitivity specificity number studies meta-analysis small. Furthermore, estimation correlation affected by included meta-analysis, or variability among analyzed studies. To model relationship test results, binary matrix assumed. Here we used copulas as alternative capture dependence specificity. The posterior values were estimated using methods that consider sampling algorithms from probability distribution (Markov chain Monte Carlo), estimates compared with results bivariate model, which assumes independence results. illustrate applicability their respective comparisons, data 14 published reporting accuracy Alcohol Use Disorder Identification Test used. Using simulations, investigated performance four copula incorporate scenarios designed replicate realistic situations meta-analyses tests. models’ performances evaluated based on p-values Cramér–von Mises goodness-of-fit test. Our indicated are valid assumptions fulfilled.