Operational validation of a commercial ELISA for Brucella ovis, Chile

Abstract

Sensitivity and specificity are usually estimated for a reference population, and these differ from the operational interpretation. The aim of this study was to carry out an operational validation of a commercial Enzyme-Linked Immunosorbent Assay (ELISA) for Brucella ovis in the sheep population of Magallanes Region. The assay was validated using sera from 82 rams with bacterial isolation and 421 sera from rams with culture-negative semen and not reacting to agar gel immunodiffusion test. The cut-off point was 58.15% optical density, estimated by using the receiver-operating characteristic method; with 82.9% of sensibility and 91.4% of specificity; the low sensibility indicates the failure of the ELISA to detect a great number of positives, therefore an additional diagnostic test must be used in eradication programmes; the low positive predictive value was mainly influenced by the low prevalence. The estimated parameters should be used in the interpretation of future diagnostic tests and an operational validation must be carried out when a new diagnostic test is adopted, establishing the parameters under which the laboratory will work, as a way to minimize the misdiagnoses.