Agreement between four commercial diagnostic tests and routine bacteriological culture of milk to determine the udder infection status of dairy cows

Mastitis is usually treated based on clinical signs or somatic cell count information rather than on results of bacteriological culture of milk. In many countries an optimal mastitis treatment is considered important from the perspective of therapy efficacy, prudent antimicrobial use and farm economics. Farmers can optimize their mastitis treatment decisions if they know whether and which mastitis pathogen is involved. Information on the mastitis pathogen involved can be acquired from diagnostic mastitis tests such as culture-based tests. This study aimed to determine the agreement of four commercial culture-based mastitis tests with routine bacteriological culture of milk to determine the intramammary infection status of a quarter or cow. The commercial culture-based tests evaluated in this study were CHROMagar Mastitis (CHROMagar, France), Hardy Diagnostics Mastitis Triplate (Hardy Diagnostics, USA), Minnesota Easy Culture System II Tri-plate (University of Minnesota, USA), and VétoRapid (Vetoquinol, the Netherlands). We used 866 prospectively collected milk samples, routinely submitted to the bacteriological laboratory of GD Animal Health for routine bacteriological culture of milk from April to June 2016. Samples were cultured on routine bacteriological culture of milk and on the commercial culture-based tests. We calculated the agreement beyond chance of each commercial culture-based test result with the result of routine bacteriological culture using 2x2 contingency tables. Furthermore, inter-reader agreement was determined for 597 samples read by two masked readers. The agreement of the four commercial culture-based mastitis tests with routine bacteriological culture of milk for Gram-positive bacteria ranged from 0.14 (95% CI 0.11-0.16) using Hardy Diagnostics Mastitis Triplate to 0.25 (95% CI 0.22-0.28) using Minnesota Easy Culture System II Tri-plate. The agreement for Gram-negative bacteria was approximately 0.70 (95% CI 0.66-0.74) for all four commercial culture-based tests. The agreement for no growth ranged from 0.22 (95% CI 0.19-0.25) using Hardy Diagnostics Mastitis Triplate to 0.34 (95% CI 0.31-0.38) using VétoRapid. This category was affected by prevalence and bias as the prevalence adjusted and bias adjusted kappa ranged from 0.63 (95% CI 0.56-0.69) using CHROMagar Mastitis to 0.68 (95% CI 0.62-0.74) using Hardy Diagnostic Mastitis Triplate. Agreement between readers was almost perfect. Although only for Gram-negative bacteria a good agreement was found between commercial culture-based tests and routine bacteriological culture of milk, and further on-farm evaluations are needed to determine the effect of these findings on udder health, commercial culture-based tests are of added value to support decisions whether and how to treat cows with mastitis.