Use of total and differential somatic cell counts from composite milk samples to detect mastitis in individual cows

The objective of this study was to ascertain the value of variables measured on composite milk samples as predictors of mastitis in individual cows. The standard of comparison was the results obtained from the bacteriological examination of individual quarter foremilk samples. Cows were classified as negative or positive with regard to mastitis on the basis of one quarter sampling only and cows which were impossible to classify in this manner were omitted from subsequent analyses. The variables that were examined were: the presence or absence of specific bacteria, demographic data, and logarithmically transformed total somatic cell counts and percentages of cell volume in channels 7 through 12 of a Coulter Counter. It was found that the inclusion of all variables resulted in correct classification of 95.9% of cows with regard to their mastitis status. Sequential elimination of individual variables or groups of variables in an attempt to simplify the procedure reduced the correct classification to 86.8% when only the log transformation of the total somatic cell count and the demographic data were included. The ability of a function which included the logarithm of the total somatic cell count, the logarithm of the percentage in channel 8 and demographic data, to classify cows was examined in detail and the sensitivity and specificity of the function also discussed. It is also shown that with increasing age the minimum total somatic cell count required to classify a cow as positive increased and possible explanations of this phenomenon are discussed.