Relationships of Bulk Tank Somatic Cell Counts to Prevalence of Intramammary Infection and to Indices of Herd Production

Abstract

Linear regressions and correlations between bulk tank somatic cell counts and (a) prevalence of intrammary infection (n 80), (b) test day average daily milk and fat production per cow (n = 85), and (c) rolling herd average milk and fat production per cow (n = 83) were determined. The correlation coefficient between bulk tank somatic cell count and prevalence of infection was +. 77 and the coefficient of determination was .59; prevalence of infection was thus the major determinant of bulk tank somatic cell count. Highly significant negative correlations were found between bulk tank somatic cell counts and indices of herd production. Somatic cell counts, presumably as an index of the mas-titis situation in a herd, are routinely determined on bulk tank samples. Several previous workers have concluded that bulk tank somatic cell counts (BTSCC) only poorly reflect the extent of intramammary infection (percent of cows or quarters infected) in the herd (9,13,15,17,18). In another study, however, relatively high correlations between BTSCC and prevalence of infection were found (11). Natzke (9), noting the day-today variation in BTSCC from the same herd, questioned the reliability and fairness of BTSCC as a basis for payment or regulatory programs. Furthermore, many dairy farmers seem unaware of the possible significance of BTSCC. The objectives of this study were to examine relationships between BTSCC and prevalence of infection and between BTSCC and various indices of herd production. and Clinton Counties, Pennsylvania. During an 18-month period, 27 herds were studied 3 times each and 2 herds were studied twice at approximately 6-month intervals, resulting in a total of 85 herd studies. For each herd study, BTSCC was determined, prevalence of infection was determined by culture of individual quarter samples, and production data were obtained from DHIA records. Although each herd was studied more than once, results from each herd study were considered as independent observations. For each herd study the herd was visited twice, usually within 2 d and always within 5 d after a monthly DHIA test. Foremilk samples were collected aseptically from each quarter of all lactating cows just before milking at each of the 2 herd visits which were usually I d apart. Transport and culture of samples and identification of bacterial isolates were done according to standard methods (7). California Mastitis Tests (CMT) (14) were done in the laboratory on samples collected at the first visit. Bacteriologic diagnosis for each quarter was based on combined results from the 2 samples. When bacteriologic results of the 2 samples did not agree, the CMT score was considered in making the diagnosis. When results from an individual quarter were so conflicting as to prevent a reliable diagnosis, a classification of "no diagnosis" was made. This classification was assigned to 2.2% of all quarters tested. Quarters infected with streptococci, Staphylococcus au-reus (coagulase-positive), coliform bacteria or, rarely, yeasts or Corynebac-terium pyogenes, were considered to have major infections. Quarters infected with coagulase-negative staphylococci or Corynebacterium bovis were considered to have minor infections. BTSCCs were the average of I count done on each of 2 well mixed samples collected from the bulk tank at the completion of the two milkings during which quarter samples were collected. Somatic cell counts were done by the Direct Microscopic Somatic Cell Count Method (8). Test day average daily milk and fat yield per cow and 12-month rolling herd average milk and fat production data were obtained from DHIA records. At the conclusion of the 85 herd studies, data available for analysis were as follows: BTSCC from 85 herd studies, test day average milk and fat production per cow from 85 DHIA monthly samplings, rolling herd average milk and fat production data for the 12-month period preceding 83 herd studies (in two instances the herds had been on test for less than 12 months and did not have rolling herd averages), and prevalence of quarter infection data from 80 herd studies (data from 5 herd studies were removed from the analysis either because only I bacteriological survey was conducted or because of excessive numbers of contaminated samples). Relationships between various observations were determined by correlation and regression analylsis. 4