Comparison between a Calving Predictive System and a Routine Prepartal Examination in German Holstein Heifers and Cows

The objective was to validate the efficacy of Moocall^® comparing it to a routine clinical examination. Altogether 38 Holstein cows were enrolled in this study (Moocall^® group: 16 heifers and 8 cows; control group: 9 heifers and 5 cows). Clinical examinations were performed every 6 h over the 7 days period before the predicted calving date. The examined traits were changes in pelvic ligament relaxation, edema of the vulva, teat filling, vaginal secretion, tail tip flexibility, tail raising and behavior. There were no significant differences in Moocall^® alerts between heifers and cows. The time lag between the first warning of Moocall^® and the onset of labor was 21.2 ± 20.2 h (max: 95.4 h; min: 0.1 h; p = 0.87) for heifers and 29.6 ± 29.6 h (max: 177.8 h; min: 0 h; p = 0.97) for cows. Linear models including Moocall^® alerts showed a significantly better fit to the time until calving than models without Moocall^® information (without variable selection: p = 0.030, with variable selection: p < 0.01). In the best-fitting model, class 2 alerts (enhanced tail activity over 2 h) contributed with a higher significance (p < 0.01). Vice versa, models including additional traits were outperformed the use of Moocall^® alerts alone. In the best fitting model, class 2 alerts (enhanced tail activity during 2 h) contributed with a higher significance (p < 0.01) than any of the best clinical predictive parameters, such as pelvic ligament relaxation (p = 0.01), tail tip flexibility (p = 0.01) or behavior (p = 0.01).