Latent class evaluation of three serological tests for the diagnosis of human brucellosis in Bangladesh

BACKGROUND A Bayesian latent class evaluation was used to estimate the true prevalence of brucellosis in livestock farmers and patients with prolonged pyrexia (PP) and to validate three conditionally dependent serological tests: indirect ELISA (iELISA), Rose Bengal Test (RBT), and standard tube agglutination (STAT). A total of 335 sera from livestock farmers and 300 sera from PP patients were investigated. RESULTS The true prevalence of brucellosis in livestock farmers and PP patients was estimated to be 1.1 % (95 % credibility interval (CrI) 0.1-2.8) and 1.7 % (95 % CrI 0.2-4.1), respectively. Specificities of all tests investigated were higher than 97.8 % (95 % CrI 96.1-99.9). The sensitivities varied from 68.1 % (95 % CrI 54.5-80.7) to 80.6 % (95 % CrI 63.6-93.8). The negative predictive value of all the three tests in both populations was very high and more than 99.5 % (95 % CrI 98.6-99.9). The positive predictive value (PPV) of all three tests varied from 27.9 % (95 % CrI 3.6-62.0) to 36.3 % (95 % CrI 5.6-70.5) in livestock farmers and 39.8 % (95 % CrI 6.0-75.2) to 42.7 % (95 % CrI 6.4-83.2) in patients with PP. The highest PPV were 36.3 % for iELISA and 42.7 % for RBT in livestock farmers and pyrexic patients, respectively. CONCLUSIONS In such a low prevalence scenario, serology alone does not help in diagnosis and thereby therapeutic decision-making. Applying a second test with high specificity and/or testing patients having history of exposure with known risk factors and/or testing patients having some clinical signs and symptoms of brucellosis may increase the positive predictive value of the serologic tests.