Evaluation of three serological tests for diagnosis of bovine brucellosis in smallholder farms in Pakistan by estimating sensitivity and specificity using Bayesian latent class analysis

The present study aimed to estimate the herd-level sensitivity (Se) and specificity (Sp) of three commonly used serological tests in naturally-infected cattle and buffalo in smallholder farms in Pakistan. Between February and June 2015, a cross-sectional study was carried out in five districts of Punjab (Kasur, Okara, Pakpattan, Jhelum, and Bhakkar) and two districts of Sindh (Badin and Thatta). Serum samples from mixed farms of cattle (n = 441) and buffalo (n = 621) were collected and tested using the Rose Bengal test (RBT), indirect ELISA (I-ELISA) and competitive ELISA (C-ELISA). A Bayesian latent class analysis (LCA) approach was used to estimate the Se and Sp of these three serological tests and the true herd-level prevalence in each district. The model was fitted under the assumption of conditional independence between three tests and also conditional dependence by including covariances between the two ELISAs. In addition, the model was fitted using three different shapes of beta distributions to incorporate prior information in the model. The test with the highest Se was the C-ELISA, with a range from 76.3% (95% PCI (Posterior Credibility Interval), 62.6-88.2%) to 81.4% (95% PCI, 68.2-92.8%). The RBT was found to have the highest Sp (99.1-99.4%) of the tests. The highest estimated herd-level prevalence, 45% (95% PCI, 32-59%), was found in Jhelum district and the lowest in Thatta district, 1.1% (95% PCI 0.04-6.0%). The results of this study identified some discrepancy between the published literature on the level of Se of these tests, especially for RBT. It appears that RBT has lower Se and higher Sp when used in the field conditions of the present study. Consequently, it is recommended that none of the evaluated tests should be performed in isolation for the diagnosis of bovine brucellosis in the field conditions of Pakistan, but the use of tests in combination, with RBT and C-ELISA used in parallel returning optimal Se and Sp, is warranted.