Reliability of three bulk-tank antimicrobial residue detection assays used to test individual milk samples from cows with mild clinical mastitis.

Objective: To determine the likelihood of false-positive results when testing milk samples from individual cows by use of 3 commercially available assays (Penzyme MilkTest and the SNAP beta-lactam and Delvo-SP assays) labeled for use with commingled milk.
Sample Population: Milk samples from 111 cows with mild clinical mastitis.
Procedure: Cows were randomly assigned to the control (no antimicrobials) or intramammary treatment group. Posttreatment milk samples were collected at the first milking after the labeled withholding period or an equivalent time for controls, randomly ordered, and tested twice by use of each assay and once by use of high-performance liquid chromatography. Sensitivity, specificity, and positive and negative predictive values were determined for each assay. Concordance of results for the same sample was assessed for each assay by calculating kappa.
Results: Sensitivities of the Delvo-SP and SNAP lactam assays were > 90%, whereas the sensitivity of the Penzyme Milk Test was 60%. Positive predictive values (range, 39.29 to 73.68%) were poor for all 3 assays. Concordance of test results was excellent for the SNAP beta-lactam and Delvo-SP assays (kappa = 0.846 and 0.813, respectively) but was less for the Penzyme MilkTest (kappa = 0.545).
Conclusions and Clinical Relevance: Because of the low positive predictive values, these 3 assays may not be useful for detecting violative antimicrobial residues in individual milk samples from cows treated for mild clinical mastitis. However, repeatability of each assay was considered good to excellent.