Monitoring herd incidence of intramammary infection in lactating cows using repeated longitudinal somatic cell count measurements

• Published in: Journal of dairy science 2013-03, Vol.96 (3), p.1568-1580
• Main Authors: Dufour, S., Dohoo, I.R.
• Format: Article
• Language: English
• Subjects: Animals; Cattle; Cell Count - veterinary; cows; dairy herds; Female; herd improvement; Incidence; intramammary infection; lactating period; lactation; Lactation - physiology; linear models; Longitudinal Studies; mastitis; Mastitis, Bovine - diagnosis; Mastitis, Bovine - epidemiology; Mastitis, Bovine - microbiology; Mastitis, Bovine - physiopathology; milk; Milk - cytology; milking; monitoring; pathogens; regression analysis; somatic cell count
• ISSN: 0022-0302; 1525-3198
• DOI: 10.3168/jds.2012-5902

The objective of the study was to evaluate the ability of an estimate of the herd intramammary infection (IMI) incidence rate computed using repeated somatic cell count (SCC) measurements (quarter- and composite-SCC; hereafter, the SCC-derived herd IMI incidence, SCCI)to predict the incidence rate computed using repeated quarter-milk bacteriological culture (hereafter, bacteriological culture incidence, BCI) during the lactating period. A cohort of 91 Canadian dairy herds was followed in 2007 and 2008. In each herd and at each of 4 sampling periods, a series of 3 to 7 quarter-milk samples was collected from a sample of 15 cows. Routine milk bacteriological culture was conducted to identify IMI, SCC was measured on the quarter-milk samples, and composite-SCC of the preceding and following dairy herd improvement (DHI) tests were obtained. Mastitis pathogens were grouped in 3 categories: major, minor, and any pathogens. For each herd and for each period, BCI was computed for each group of organisms. Similarly, SCCI were computed using quarter- and DHI composite-SCC and using a threshold of 200,000 cells/mL to define infected quarters or cows. A linear regression model taking into account the structure of the data was used to compare the SCCI to the BCI. A similar model was used to compare fluctuations (i.e., changes from one sampling period to the next) over time of the SCCI and BCI. Measures of correlation between observed and predicted rates were computed and limits of agreement plots sketched to better explore the predictive ability of the SCCI. The quarter-milk SCC measurements that could be obtained—for instance, using on-line milking system measurements—appeared to be particularly valuable. Quarter-SCCI showed a positive and significant association with the BCI. However, limits of agreement plots indicated important disagreement for the small proportion of observations with very high BCI. Quarter-level SCCI and BCI fluctuations were also significantly associated, and a substantial correlation (Spearman rho ranging from 0.54 to 0.58) could be seen between observed and predicted rates. Conversely, the predictive value of composite-DHI SCC for monitoring IMI incidence during the lactation seemed to be quite limited. Composite SCCI was strictly associated with major IMI BCI, showed a relatively low correlation with the observed rate (Spearman rho: 0.14), and was of little help for longitudinal monitoring of the IMI incidence.