Determination of blood variables tresholds in prediction of productive and reproductive performances in dairy cows: use of receiver operating characteristic (roc) analysis

Abstract

Decreased milk production and impaired reproductive performances of dairy cows provoke huge economic losses in dairy farm industry. Those disturbances are mostly a consequence of productive diseases (PD) that occurred, commonly, during peripartal period, defined as period from 21 days before until 21 days after calving. The factors that predispose cows to PD mainly include low feed intake combined with increased energy demand and/or endocrine disturbances that disable adequate adaptation of cows to increased milk production. Those factors are manifested during early postpartal period but originated from disturbances that occur during antepartal and/or puerperal period. Therefore prediction of PD could be obtained during periods when there is no visible signs of upcoming diseases. Implementing a model of the PD prediction could be a key factor that ensure profitable dairy farming. Concentrations of different blood variables during peripartal period might be a key factors that determine the success in metabolic adaptation to increased milk production. Several studies have determined blood variable thresholds, using receiver operating characteristic (ROC) analysis as a statistical tool, for predicting different PD in transition dairy cattle. Optimal thresholds of metabolic indicators of hepatic lipidosis, ketosis and abomasal displacement were suggested by different authors. Our research team, including researchers from Faculty of Veterinary Medicine University of Belgrade and Veterinary Faculty University of Ljubljana, preliminary established optimal thresholds of blood variables that were associated with an increased lipid and decreased glycogen contents in postpartum liver of dairy cows. Those results were obtained as a part of bilateral research project supported by Serbia and Slovenia. Nevertheless, there are only few studies that examined thresholds of blood variable concentrations for predicting productive and reproductive performances in dairy cows using ROC analysis. Since this statistical approach demand definition of beneficial (positive) and non-beneficial (negative) outcomes, we have introduce beneficial and non-beneficial productive and reproductive outcomes in Holstein cows on commercial dairy farms in Serbia. Considering genetic potentials of Holstain cows, we have established daily milk production higher than 30 L during early lactation and calving-conception interval shorter than 120 days as profitable for dairy cow industry in Serbia. In our study, blood samples were collected from the jugular vein at day 14 before (antepartum) and day 7 after calving (puerperium). Concentrations of glucose, BHBA, total protein, albumin, total bilirubin, urea, Ca and P were determined in blood serum. Body condition scores were also determined. Our results showed that antepartum, body condition score > 3.87, glucose > 3.45 mmol/l, BHBA > 0.65 mmol/l, total protein < 71.25 g/l and albumin < 36.75 g/l were associated with decreased milk production, while there was no reliable blood variable for prediction of calving-conception interval. During puerperium, glucose > 2.85 mmol/l, BHBA concentration > 1.35 mmol/l and Ca < 2.05 mmol/l were associated with decreased milk production and glucose < 2,75 mmol/l was associated with longer calving-conception interval. In conclusion, ROC analyses may be useful statistic tool for prediction of cow`s productive and reproductive performances on dairy farms.