Send to

Choose Destination
Epidemiol Infect. 2019 Jan;147:e73. doi: 10.1017/S0950268818003503.

Bayesian evaluation of three serological tests for the diagnosis of bovine brucellosis in Bangladesh.

Author information

Department of Medicine,Bangladesh Agricultural University,Mymensingh-2202,Bangladesh.
Department of Biomedical Sciences,Institute of Tropical Medicine,Nationalestraat 155, B-2000 Antwerp,Belgium.
Department of Epidemiology and Public Health, Sciensano, Rue Juliette Wytsmanstraat 14, 1050 Brussels,Belgium.
Laboratory of Epidemiology, Biostatistics and Animal Health Economics,School of Health Sciences, Faculty of Veterinary Science, University of Thessaly,Karditsa,224 Trikalon st. 43100,Greece.
Department of Applied Mathematics, Computer Science and Statistics, Faculty of Sciences,Ghent University,281 Krijgslaan, B-9000, Ghent,Belgium.
Research Unit of Epidemiology and Risk Analysis applied to Veterinary Science (UREAR-ULg), Fundamental and Applied Research for Animals & Health (FARAH) Center, Faculty of Veterinary Medicine, University of Liege,Quartier Vallée 2, Avenue de Cureghem 7A, B42, Sart-Tilman Liege,Belgium.
Department of Surgery and Obstetrics,Bangladesh Agricultural University,Mymensingh-2202,Bangladesh.
Sydney School of Veterinary Science, The University of Sydney,425 Werombi Road, Camden, 2570 NSW,Australia.


We evaluated the performance of three serological tests - an immunoglobulin G indirect enzyme linked immunosorbent assay (iELISA), a Rose Bengal test and a slow agglutination test (SAT) - for the diagnosis of bovine brucellosis in Bangladesh. Cattle sera (n = 1360) sourced from Mymensingh district (MD) and a Government owned dairy farm (GF) were tested in parallel. We used a Bayesian latent class model that adjusted for the conditional dependence among the three tests and assumed constant diagnostic accuracy of the three tests in both populations. The sensitivity and specificity of the three tests varied from 84.6% to 93.7%, respectively. The true prevalences of bovine brucellosis in MD and the GF were 0.6% and 20.4%, respectively. Parallel interpretation of iELISA and SAT yielded the highest negative predictive values: 99.9% in MD and 99.6% in the GF; whereas serial interpretation of both iELISA and SAT produced the highest positive predictive value (PPV): 99.9% in the GF and also high PPV (98.9%) in MD. We recommend the use of both iELISA and SAT together and serial interpretation for culling and parallel interpretation for import decisions. Removal of brucellosis positive cattle will contribute to the control of brucellosis as a public health risk in Bangladesh.


Animal pathogens; Bayesian analysis; brucellosis; infectious disease epidemiology; veterinary epidemiology


Supplemental Content

Full text links

Icon for Cambridge University Press
Loading ...
Support Center