Evaluation of the Minnesota Easy Culture System II Bi-Plate and Tri-Plate for identification of common mastitis pathogens in milk

The objective of this study was to validate use of the Minnesota Easy Culture System II Bi-Plate and Tri-Plate (University of Minnesota Laboratory for Udder Health, St. Paul) to identify common mastitis pathogens in milk. A total of 283 quarter and composite milk samples submitted to the University of Minnesota Laboratory for Udder Health during the spring of 2010 were cultured simultaneously using 3 methods: standard laboratory culture (reference method) and the Minnesota Easy Culture System II Bi-Plate and Tri-Plate methods. Bi-Plate and Tri-Plate cultures were incubated for 18 to 24 h and interpreted by 2 independent, untrained readers within 5 h of each other. An experienced technician completed the standard laboratory culture. For each sample, all 3 study personnel recorded the culture result (yes/no) for each of the following diagnostic categories: no bacterial growth (NG), mixed (2 organisms), contaminated (3 or more organisms), gram-positive (GP), gram-negative (GN), Staphylococcus spp., Streptococcus spp., Streptococcus agalactiae, Streptococcus dysgalactiae, Streptococcus uberis, Enterococcus spp., Staphylococcus aureus, coagulase-negative staphylococci, Escherichia coli, Klebsiella spp., and other. For each category, the prevalence, sensitivity, specificity, accuracy, and predictive values of a positive and negative test were calculated, and the agreement between readers and between each reader and the laboratory was assessed. Specificity, overall accuracy, and negative predictive values were generally high (>80%) for the Bi-Plate and Tri-Plate for each category. Sensitivity and positive predictive values were intermediate (>60%) or high (>80%) for the broad categories of NG, GP, GN, Staphylococcus spp. and Streptococcus spp., and for Staph. aureus, but were generally lower (<60%) for other more specific categories. Similarly, interreader agreement (kappa value) was moderate to substantial (40-80%) for the broad categories of NG, GP, GN, Staphylococcus spp. and Streptococcus spp., and for Staph. aureus and E. coli, but was lower for other categories. The Tri-Plate had a higher sensitivity, accuracy, and negative predictive value for Streptococcus spp., and higher interreader agreement for some of the more specific categories. Our conclusion was that Bi-Plate and Tri-Plate results will be most reliable when used to classify infections in broad diagnostic categories such NG, GP, or GN. The Bi-Plate and Tri-Plate will have intermediate ability to identify infections as being caused by Staphylococcus spp., Streptococcus spp., or Staph. aureus.