A comparison of indirect methods for diagnosis of subclinical intramammary infection in lactating dairy cows .3. Prediction of the incidence of infections and the critical threshold value for somatic cell count from the herd's bulk milk somatic cell count.

Bacterial status and somatic cell count were measured for 30 cows in each of 15 herds, and data for the three herds described in a previous paper were also included. The relationships between prevalence of infection, estimated bulk milk somatic cell count (EBMSCC) and the critical threshold for the somatic cell count were examined. Log(10) EBMSCC was closely correlated to the percentage of quarters infected by major pathogens (r(2) = 0.71) and by major and minor pathogens (r(2) = 0.85). The corresponding correlations with percentage of cows infected were lower (r(2) < 0.56). The critical threshold for the somatic cell count was correlated to the percentage of quarters infected by minor pathogens and by streptococci (r(2) = 0.52), but correlation with other measures of infection prevalence were lower. However, when one herd with a high incidence of Streptococcus agalactiae was excluded, the critical threshold value was closely correlated with EBMSCC (r(2) = 0.76) and the critical threshold could be predicted reliably by a model which included EBMSCC, cow age and stage of lactation, (r(2) = 0.84). Use of the herd's critical threshold value resulted in a large decrease (x 0.5) in the number of false positive diagnoses, but a large increase (x 2.0) in the number of false negative diagnoses, compared with the use of 250,000 cells/mL as a set value for the threshold. The data suggest that for the majority of herds, critical threshold for SCC can be predicted reliably from the herds EBMSCC. It can then be used to estimate the economic threshold value which is best-suited for each particular application within the mastitis control program.