Background.

ELISA optical density (OD) values are a measure of antibody concentration. Transforming OD values to positive or negative interpretations based on a single cut-off value ignores the clinical value of the magnitude of the ELISA result. Likelihood ratios (LHRs = [sensitivity/(1-specificity)]) calculated on sera from well characterized cases and controls allows the magnitude of ELISA results to be used to estimate the probability the tested animal has or does not have paratuberculosis.

Purpose.

Calculate LHRs for a wide range of ELISA S/P values.

Methods.

LHRs were determined using the a serum antibody ELISA for Johne's disease (IDEXX) on 143 bovine sera from culture- or histopathology-confirmed cases of paratuberculosis and 2,973 bovine sera from animals in herds proven free of M. paratuberculosis infection (herds negative on whole-herd fecal culture a minimum of eight times at intervals of 6 months or more). ELISA ODs were transformed to S/P values as per the kit manufacturer's directions. Sensitivity and specificity were calculated for S/P values 0.0 to 2.0 in intervals of 0.05 S/P units.

Results.

LHRs ranged from 1 to 583. LHR (y-axis) was plotted against ELISA S/P value cut-off (x-axis). These data pairs fitted a line defined by a power function with an r2 value = 0.94.

Conclusions.

The LHR algorithm can be used to estimate the probability of M. paratuberculosis infection.

Example

A cow with an ELISA S/P of 0.40 has an LHR of 41:1 (odds the cow is infected); converted to probability this is 97.6%. Infection probability estimates are more accurate if the pre-test probability of infection is included in the calculation. Comparison of LHR values on the 143 cases of paratuberculosis to other test results catalogued in a repository database showed that LHR values are directly related to the percentage of cattle positive on other tests for paratuberculosis.