The Use of Prepartal Standing Behavior As a Parameter for Early Detection of Subclinical Ketosis in Postpartal Dairy Cows

Subclinical ketosis is a common disease that typically occurs during early lactation in dairy cows, and consequently has a great economic impact on the dairy industry. Twenty-four multiparous Holstein cows were enrolled at 30 d prior to calving and received a common prepartal (1.5 Mcal/kg DM, 15% CP) and postpartal (1.76 Mcal/kg DM, 18% CP) diet. From -30 d until 15 d postpartum, Hobo Pedant G data loggers (Pocasset, MA) were used to record 3 dimensional acceleration at 60-s intervals. The acceleration data were used to analyze cow behavior such as standing and lying time, bouts (n/24 h), and bout duration. Blood samples were taken from the coccygeal vein or artery at 1, 3, 5, 7, 9, 11, 13, and 15 d postpartum and used to measure the blood concentration of ketones (BHB) with the handheld Precision Xtra. A retrospective analysis of BHB data allowed the classification of cows as subclinically-ketotic (KET; n = 13; ketones > 1.4 mmol/L) or non-ketotic (NONKET; n = 11; ketones < 1.4 mmol/L) during early postpartum. The behavior and BHB data were analyzed using the MIXED procedure of SAS. The CORR procedure of SAS was used to analyze correlations between behavior data and BHB. The REG procedure of SAS was used to build a model to predict postpartal BHB using prepartal data. Prepartal standing time was greater (P = 0.02) in NONKET cows than KET cows, whereas lying time was greater (P = 0.02) in KET cows. As expected, BHB were greater (P < 0.01) in KET cows than NONKET cows. Overall BHB reached 1.2 mmol/L by 3 d postpartum and continued to increase over time. Therefore, correlation analysis was performed between behavior data from -30 d to 3 d relative to parturition and BHB at 3 d postpartum. The greatest significant correlation between BHB and behavior data was observed with standing time at 6 d prepartum (P < 0.01; r = -0.84). The latter meant that cows least-susceptible to postpartal subclinical ketosis remained standing for a longer time in the prepartum period. A model containing standing time, BW, BCS, and DMI at 6 d prepartum accounted for most of the variation (P < 0.01; R²= 0.90) in the BHB data at 3 d postpartum. These preliminary data suggest that behavioral data such as standing and lying time can be a powerful tool to build prediction models for early detection of postpartal subclinical ketosis.