Use of a weighted random regression test-day model to better relate observed somatic cell score to mastitis infection likelihood

Current methods to analyze somatic cell counts or scores do not differentiate the origins of observed values, they rely on the simple hypothesis that a "higher" value is linked to an intra-mammarian infection. This hypothesis may not be totally valid, but until now proposed alternatives were not usable in large scale genetic evaluations. A simple modification of currently used random regression test-day models is presented that weight observed somatic cell scores by a weight that expresses mastitis infection likelihood. Results showed that despite high correlation with breeding values obtained without the weight, differences that were observed were highly related to mastitis results in Nordic countries. Sire breeding values obtained with this method were also submitted to the March 2003 INTERBULL test-run. They showed the largest average correlations of all populations with mastitis results from Nordic populations. The proposed method is currently implemented and used in routine in the genetic evaluation for somatic cell score in the Walloon Region of Belgium.