Diagnosis of pregnancy in dairy cows based on the progesterone content of milk. Part 1. Discriminating between pregnant and non-pregnant dairy cows on the basis of milk progesterone levels

A linear discriminant function (LDF) was used to estimate the level of milk progesterone which allowed the best overall classification of dairy cows into pregnant and non-pregnant groups (confirmed by rectal palpation). Progesterone levels were measured in milk samples drawn between 20 and 24 days after insemination. This function, with milk progesterone as the only variable, was tested on a population of 188 pregnant and 103 nonpregnant cows, correctly classifying 98,0% of all cases (98,9% of pregnant cows and 97,1% of non-pregnant cows), when the discriminating level for progesterone was set at 5 ng/ml milk. At the same level of progesterone this function correctly classified 94,4% of all cases belonging to a different sample of dairy cows. This technique appears to allow selection of a milk progesterone level which maximizes accuracy of classification into both pregnant and non-pregnant categories.