Comparison of some screening tests for detecting mastitis

Quarter milk samples were analyzed for somatic cell counts, ATP, NAGase, bovine serum albumin, antitrypsin, conductivity, and bacteriological findings. The diagnostic tests were compared with respect to their ability to predict the infection status of a quarter on test day. Predictive ability was evaluated by logistic regression. All components measured increased in the presence of infection. Correlations between components were higher in quarters with pathogens than in quarters free from infection. Predictive ability was highest for ATP, somatic cell counts, and NAGase. Comparisons within udder using ratios from different quarters gave no better prediction than absolute values. For all diagnostic tests except conductivity, combinations of absolute values and interquarter ratios were not better than predictions based on absolute values only. If conductivity is to be used to predict infection status, a combination of absolute value and interquarter evaluation is recommended. Combinations of two diagnostic tests increased the predictive ability in most cases, although the increase was only minor.