ABSTRACT
To perform a comparative evaluation of the indirect enzyme-linked immunosorbent assay (iELISA) in milk, for the detection of Brucella abortus in cattle, 1,523 milk samples were taken from individual animals and bulk milk belonging to 200 herds in the province of Cundinamarca, Colombia. All these herds were part of the official programme of monitoring free herds and determining herd prevalence in the province. The samples were submitted simultaneously to the milk ring test and the iELISA. A total of 584 individual and 497 bulk milk samples taken from free herds were considered negative, and served to determine the initial ELISA cut-off point. An optimised determination of the cut-off point involved an operational analysis, using a statistical programme. In the analysis, the total population was considered: 756 individual positive and negative samples, as defined by herd history and previously established serologies. The cut-off point was set at > 0.20 optical density units, which, expressed in percentage positivity, corresponded to 20.5% against the controls, with a sensitivity of 95.3%, a specificity of 95.1%, and a confidence interval of 95%. Indirect confirmation of the state of infection was made with competitive ELISA in the serum of the individual animals considered positive, and an attempt was made to determine bacterial presence by isolation in culture and by reverse transcription-polymerase chain reaction. The results confirm that the indirect ELISA is highly sensitive and specific, provides confirmed results in strict quality-control conditions, and may be used to test a large number of herd or individual samples, thus enhancing the efficiency of surveillance programmes and control campaigns.