Bayesian Estimation of Sensitivity and Specificity of Rose Bengal, Complement Fixation, and Indirect ELISA Tests for the Diagnosis of Bovine Brucellosis in Ethiopia

Abstract

Test evaluation in the absence of a gold standard test was conducted for the diagnosis and screening of bovine brucellosis using three commercially available tests including RBPT, CFT, and I-ELISA in National Animal Health Diagnostic and Investigation Center (NAHDIC) Ethiopia. A total of 278 sera samples from five dairy herds were collected and tested. Each serum sample was subjected to the three tests and the results obtained were recorded and the test outcomes were cross-classified to estimate the sensitivity and specificity of the tests using Bayesian model. Prior information generated on the sensitivity and specificity of bovine brucellosis from published data was used in the model. The three test-one population Bayesian model was modified and applied using WinBug software with the assumption that the dairy herds have similar management system and unknown disease status. The Bayesian posterior estimate for sensitivity was 89.6 (95% PI: 79.9–95.8), 96.8 (95% PI: 92.3–99.1), and 94 (95% PI: 87.8–97.5) and for specificity was 84.5 (95% PI: 68–94.98), 96.3 (95% PI: 91.7–98.8), and 88.5 (95% PI: 81–93.8) for RBT, I-ELISA, and CFT, respectively. In this study I-ELISA was found with the best sensitivity and specificity estimates 96.8 (95% PI: 92.3–99.1) and 96.3 (95% PI: 91.7–98.8), compared to both CFT and RBPT.