Predictive model of mastitis occurrence in the dairy cow

Mastitis occurrence within a lactation and times of mastitis onset were studied for about 1000 cows. The number of mastitis cases within a lactation was modeled through overdispersed Poisson regression with individual and herd covariants. The results emphasized the role of the herd variable. Increased production potential increased the number of cases per lactation at a rate of 1.4/10 kg. Calving month also played an important role. The incidence of mastitis was greater when calving took place in early autumn or winter, which led to an expanded housing period. The interval from calving to the first case of mastitis and the intervals between successive cases were modeled for cases occurring during lactation through random selections from fitted gamma distributions, these distributions being truncated to consider the lactation length. The results of both steps can be used to simulate mastitis occurrence in different conditions.