Interpretation of an enzyme-linked immunosorbent test using different cutoffs between positive and negative samples for diagnosis of paratuberculosis

The purpose of this study was to compare the serologic responses of cows infected and non-infected with Mycobacterium paratuberculosis in infected and non-infected herds using different cut-offs between positive and negative samples. Infection status of cattle was determined by fecal culture. The serological test was an enzyme-linked immunosorbent assay (ELISA) using an affinity-purified antigen. Cows differed in their level of exposure (high- vs. low-prevalence and negative herds) and disease manifestation (clinical vs. subclinical infection). The ELISA optical density (OD) for cow samples from fecal culture-positive herds (no clinical cases) was higher than that for samples from the fecal culture-negative herd (0.49 vs. 0.26, respectively). The mean OD value (0.61) for clinically-affected cows was not different from that of fecal culture-positive, clinically-normal cows (0.49), nor were differences found between the OD values from fecal culture-negative cows from the herd with no history of paratuberculosis (0.21) and cattle with negative cultures from paratuberculosis-infected herds (0.26). The likelihood-ratio method was used to calculate the optimum ELISA discrimination value between test-positives and -negatives at a given prevalence in a herd, and a receiver-operator characteristic (ROC) curve A as constructed.