Short communication: Comparative performance of 3 on-farm culture systems for detection of mastitis pathogens interpreted by trained and untrained observers

• Published in: Journal of dairy science 2021-04, Vol.104 (4), p.4936-4941
• Main Authors: Sipka, A., Wieland, M., Biscarini, F., Rossi, R.M., Roman, N., Santisteban, C., Moroni, P., Nydam, D.V.
• Format: Article
• Language: English
• Subjects: Animals; Cattle; Cattle Diseases; Farms; Female; mastitis; Mastitis - veterinary; Mastitis, Bovine - diagnosis; Milk; Minnesota; on-farm culture; Sensitivity and Specificity; test evaluation; training
• ISSN: 0022-0302; 1525-3198
• DOI: 10.3168/jds.2020-19166

On-farm culture (OFC) systems facilitate pathogen-based mastitis management and can facilitate antimicrobial stewardship on dairy farms. Interpretation of the results, however, may present a challenge for those with limited microbiology experience. Here, we compared results of 3 OFC systems interpreted by trained and untrained observers against results of a standard laboratory reference method (aerobic culture and mass spectrometry). Milk samples (280 quarter and 60 composite) were selected from submissions for routine diagnostic testing to Quality Milk Production Services (Cornell University, Ithaca, NY) between August 2017 and January 2018. Samples were cultured simultaneously using the standard laboratory reference method and 3 commercially available OFC systems that varied in detail of pathogen identification (provided in parentheses) as follows: (1) Minnesota Easy Culture System II Bi-plate (University of Minnesota Laboratory for Udder Health, St. Paul; gram-positive, gram-negative), (2) Minnesota Easy Culture System II Tri-plate (gram-positive, gram-negative, some genus level), and (3) FERA Diagnostics and Biologicals AccuMast plate (Ithaca, NY; genus level, some species level). After 18 to 24 h of incubation, OFC plates were interpreted by 1 trained observer (>10 yr of experience in milk microbiology) and 6 untrained observers with no previous milk microbiology training, using only the manufacturers' instructions for guidance. Strength of agreement (κ) between observer groups and the reference method was determined for the available outcomes of each system. Interpreted by the trained observer, agreement was moderate for identifying gram-positive organisms (Bi-plate, κ = 0.56) and substantial for Streptococcus spp. (Tri-plate, κ = 0.64, AccuMast κ = 0.61). Interpretation by untrained observers resulted in fair agreement (κ = 0.29–0.37) for these organisms. Moderate agreement (κ = 0.43–0.59) was found across all 3 OFC for the identification of gram-negative organisms (Bi-plate), non-aureus staphylococci (Tri-plate and AccuMast), Lactococcus spp., and Enterococcus spp. (AccuMast) when interpreted by the trained observer, and fair to moderate agreement was found (κ = 0.31–0.53) among untrained observers. Across all 3 OFC, agreement was almost perfect (κ = 0.80–0.89) for Staphylococcus aureus for the trained observer, and moderate to substantial (κ = 0.56–0.61) for untrained observers. We concluded that all 3 OFC appeared suitable to support pathogen-bas