Linear Mixed-Effects Model to Quantify the Association between Somatic Cell Count and Milk Production in Italian Dairy Herds

Abstract

Milk production loss due to mastitis in dairy herds is economically important. Before estimating the economic impacts of mastitis, it is crucial to quantify the association between mastitis and milk production. The objective of this study was to estimate the association between somatic cell count (SCC, as an indicator of intramammary infection due to mastitis) and milk production for dairy cows in Lombardy, Italy. The test-day (TD) records data of 3816 dairy herds located in three different geographical areas of Lombardy from January 2016 to December 2018 were used. After data editing, the final dataset comprised 10,445,464 TD records from 2970 farms and 826,831 cows. The analysis was carried out by using a mixed-effects model with six fixed effects (geographical Area, Breed, Days in Milk, Parity, Season and Year) and nested random effects for each cow and herd. The results confirmed that the SCC had a negative association with milk production. On average, this study found that any two-fold increase of SCC resulted in a milk production loss of 0.830 (95% CI: -0.832, -0.828) kg/cow/day in the whole of Lombardy. These results can be used for economic calculations on the costs of mastitis.

Keywords: costs; mastitis; milk yield; somatic cell count.

Figure 1

Editing criteria and number of herds, cows and test day records retained and excluded. ¹ Somatic cell count. ² Days in Milk.

Figure 2

Milk production at test-day records level in different breed groups (10,445,464 test-day records from 826,831 cows in 2970 herds). The violin plot combines the probability density (area under the curve) and box plots.

Figure 3

Somatic cell score (SCS) at test-day records level in different milk production quantile groups (10,445,464 test-day records from 826,831 cows on 2970 herds). The violin plot combines the probability density (area under the curve) and box plots.

Figure 4

Somatic cell score (SCS) at test-day records level in different breed groups (10,445,464 test-day records from 826,831 cows in 2970 herds). The violin plot combines the probability density (area under the curve) and box plots.

Figure 5

Relationship between SCC and average milk production (10,445,464 test-day records from 826,831 cows in 2970 herds) classified by the three geographical areas (Po Valley, sub-Alps and Alps).

Figure 6

Relationship between SCC and average milk production (10,445,464 test-day records from 826,831 cows in 2970 herds) classified by different breeds (Holstein, Brown Swiss and other breeds).