A Novel Bayesian Latent Class Model (BLCM) Evaluates Multiple Continuous and Binary Tests: A Case Study for Brucella Abortus in Dairy Cattle

Bovine brucellosis, primarily caused by Brucella abortus, severely affects both animal health and human well-being. Accurate diagnosis is crucial for designing informed control and prevention measures. Lacking a gold standard test makes it challenging to determine optimal cut-off values and evaluate the diagnostic performance of tests. In this study, we developed a novel Bayesian Latent Class Model that integrates both binary and continuous testing outcomes, incorporating additional fixed (parity) and random (farm) effects, to calibrate optimal cut-off values by maximizing Youden Index. We tested 651 serum samples collected from six dairy farms in two regions of Henan Province, China with four serological tests: Rose Bengal Test, Serum Agglutination Test, Fluorescence Polarization Assay, and Competitive Enzyme-Linked Immunosorbent Assay. Our analysis revealed that the optimal cut-off values for FPA and C-ELISA were 94.2 mP and 0.403 PI, respectively. Sensitivity estimates for the four tests ranged from 69.7% to 89.9%, while specificity estimates varied between 97.1% and 99.6%. The true prevalences in the two study regions in Henan province were 4.7% and 30.3%. Parity-specific odds ratios for positive serological status ranged from 1.2 to 2.2 for different parity groups compared to primiparous cows. This approach provides a robust framework for validating diagnostic tests for both continuous and discrete tests in the absence of a gold standard test. Our findings can enhance our ability to design targeted disease detection strategies and implement effective control measures for brucellosis in Chinese dairy farms.