Latent class analysis of the diagnostic characteristics of PCR and conventional bacteriological culture in diagnosing intramammary infections caused by <em>Staphylococcus aureus</em> in dairy cows at dry off
Cederlöf, Sara Ellinor4; Toft, Nils6; Aalbæk, Bent7; Klaas, Ilka Christine8
1 Population Biology, Department of Large Animal Sciences, Faculty of Life Sciences, Københavns Universitet2 Section of Microbiology, Department of Veterinary Disease Biology, Faculty of Life Sciences, Københavns Universitet3 Production & Health, Department of Large Animal Sciences, Faculty of Life Sciences, Københavns Universitet4 Hushållningssällskapet Västernorrland5 Section for Production and Health, Department of Large Animal Sciences, Faculty of Health and Medical Sciences, Københavns Universitet6 Population Biology, Department of Large Animal Sciences, Faculty of Life Sciences, Københavns Universitet7 Section of Microbiology, Department of Veterinary Disease Biology, Faculty of Life Sciences, Københavns Universitet8 Section for Production and Health, Department of Large Animal Sciences, Faculty of Health and Medical Sciences, Københavns Universitet
ABSTRACT: BACKGROUND: Staphylococcus aureus is one of the most common causes of intramammary infections in dairy cows at dry off. Reliable identification is important for disease management on herd level and for antimicrobial treatment of infected animals. Our objective was to evaluate the test characteristics of PathoProof TM Mastitis PCR Assay and bacteriological culture (BC) in diagnosing bovine intramammary infections caused by S. aureus at dry off at different PCR cycle threshold (Ct)-value cut-offs. METHODS: Sterile quarter samples and non-sterile composite samples from 140 animals in seven herds were collected in connection with the dairy herd improvement (DHI) milk recording. All quarter samples were analyzed using BC whereas all composite samples were analyzed with PathoProof TM Mastitis PCR Assay. Latent class analysis was used to estimate test properties for PCR and BC in the absence of a perfect reference test. The population was divided into two geographically divided subpopulations and the Hui-Walter 2-test 2-populations model applied to estimate Se, Sp for the two tests, and prevalence for the two subpopulations. RESULTS: The Se for PCR increased with increasing Ct-value cut-off, accompanied by a small decrease in Sp. For BC the Se decreased and Sp increased with increasing Ct-value cut-off. Most optimal test estimates for the real-time PCR assay were at a Ct-value cut-off of 37; 0.93 [95% posterior probability interval (PPI) 0.60-0.99] for Se and 0.95 [95% PPI 0.95-0.99] for Sp. At the same Ct-value cut-off, Se and Sp for BC were 0.83 [95% PPI 0.66-0.99] and 0.97 [95% PPI 0.91-0.99] respectively. Depending on the chosen PCR Ct-value cut-off, the prevalence in the subpopulations varied; the prevalence increased with increasing PCR Ct-value cut-offs. CONCLUSION: Neither BC nor real-time PCR is a perfect test in detecting IMI in dairy cows at dry off. The changes in sensitivity and prevalence at different Ct-value cut-offs for both PCR and BC may indicate a change in the underlying disease definition. At low PCR Ct-value cut-offs the underlying disease definition may be a truly/heavily infected cow, whereas at higher PCR Ct-value cut-offs the disease definition may be a S. aureus positive cow.