Bivariate hierarchical model for the Meta-analysis of diagnostic tests in studies with binary responses: its application from SAS and R

Studies on the precision of diagnostic tests usually report the number of true positives, false positives, true negatives, and false negatives. There is generally a negative association between the number of true positives and true negatives, as studies that adopt less strict criteria to declare a test as positive need higher sensitivities and lower specificities. Given this particularity, the bivariate nature of the data must be preserved, by modeling sensitivity and specificity together. In this paper, we will use the bivariate hierarchical model applied to a meta-analysis data set which was an update to a previous systematic review of diagnostic tests for chronic Chagas disease. Our modeling framework was implemented with SAS NLMIXED procedure, making it possible to obtain summary measures for sensitivity and specificity, with values of 0.725 and 0.995, respectively, out of a total of 35 studies with 6057 patients.