RUMINATION, ACTIVITY, MILK YIELD AND MILK COMPONENTS ANALYSIS FOR DISEASE DETECTION DURING THE TRANSITION PERIOD OF DAIRY COWS
Early detection of disease is the key to successful management of the transition dairy cows, leading to timely treatment and prevention of costs associated with prolonged treatment and prolonged milk yield reduction. Electronic systems that allow for monitoring rumination, activity and milk yield of individual cows are now available. Our objective was to determine the association between changes in rumination behavior, activity milk yield and milk components with health disorders in the peripartum and early lactation cows. Three weeks before the estimated calving date, 198 multiparous Holstein cows housed at the University of Florida (UF) dairy unit were affixed with neck collars containing rumination loggers and activity (Hr-Tag rumination monitoring system, SCR Engineers Ltd., Netanya, Israel) providing rumination time (RT) and activity in 2-h intervals. Blood was collected 12-72 h after calving for non-esterified fatty acid (NEFA), beta-hydroxy butyraic acid (BHBA), and calcium(Ca) determination (n=136). Occurrence of clinical health disorders [mastitis (MAS), metritis (MET), clinical hypocalcemia (HYC), depressed and dehydrated (DEP), digestive disorders (DIG), lameness (LAM), and ketosis (KET)] was assessed until 60 DIM by UF veterinarians and farm personnel and this was regarded as gold standard for the experiment. For the blood analysis cows were considered in negative energy balance (NEB) if serum NEFA concentration were > 0.5 mmol/L; positive for subclinical ketosis (SCK) if serum BHBA cocentrations were ≥ 1.4 mmol/L; and hypocalcemic (HYC) if serum Ca concentrations were < 8.0 mg/dl. Two indexes were developed to explore the potential associations between the proposed parameters and health disorders. i. CowIndex (CIx) that considered the difference in RT between the day of diagnosis (d0) and the daily average RT from d -3 to -5 relative to diagnosis (0vs-3to- 5RT) divided by the daily average RT from d -3 to -5 (-3to-5RT) in the affected cow. ii. MatesIndex (MIx) that considered the difference between the affected cow 0vs- 3to-5RT and the pen mates 0vs-3to-5RT relative to the average d0 RT for healthy pen mates. Using a CIx index value -0.1 as the cut off value for the change in rumination, we set a cow alarm (CAL) and marked the cow as flagged for a health disorder. Similarly, whenever the MIx was less than -0.1, we set a herd alarm (HAL) and marked the cow as flagged for a health disorder. A combined alarm (CombA) was created using the parallel combination of the HAL and CAL.A separate analysis was done to detect disorders one day prior to disease diagnosis. The same procedure for calculation of CIx and MIx and the corresponding alarms were applied for activity, milk yield, and milk components (fat %, protein%, lactose%, and fat/lactose ratio) analyses. The average rumination CIx in healthy cows was 0.049 while CIx in sick cows were -0.165, -0.029, -0.513,-0.048, -0.022, -0.098, -and 0.081 for MAS, MET, HYC, DEP, DIG, LAM, and KET, respectively. Average rumination MIx in healthy cows ranged from 0.0001 to 0.001 and MIx in sick cows were -0.183, -0.101, -0.424,-0.114 -0.101, -0.148, -0.147 for MAS, MET, HYC, DEP, DIG, LAM, and KET, respectively. Sensitivity/specificity (%) of rumination CAL were 56/77, 39/77, 100/77, 47/77, 44/77, 67/77, and 61/77, for MAS, MET, HYC, DEP, DIG, LAM, and KET, respectively. Sensitivity/specificity (%) of rumination HAL were 63/77, 42/77, 100/77, 50/77, 48/77, 56/77, and 67/77 for MAS, MET, HYC, DEP, DIG, LAM, and KET, respectively. The AUC values for rumination CAL ranged from 0.55 to 0.86 whereas the AUC values for rumination HAL ranged from 0.57 to 0.86. Sensitivity/specificity of activity CAL were 34/81, 50/81, 71/81, 31/81, 48/81, 44/81, 48/81 % to detect each disease on the day of diagnosis and area under the curve (AUC) resulting from receiver operating characteristic (ROC) curve analyses for our proposed cut-off value were 0.58, 0.65, 0.76, 0.56, 0.64, 0.70 and 0.65 for MAS, MET, HYC, DEP, DIG, LAM, and KET, respectively. Sensitivity/specificity (%) of milk CAL was 59/77, 67/77, 100/77, 59/77, 61/77, 27/77, and 61/77 for MAS, MET, HYC, DEP, DIG, LAM, and KET, respectively. Area under the curve for milk CAL on our proposed cut-off value (-0.1) ranged from 0.52 to 0.88. Sensitivity/specificity (%) of milk MAV was 53/82, 46/82, 75/83, 52/83, 49/83, 20/83, and 49/83 for MAS, MET, HYC, DEP, DIG, LAM, and KET, respectively. Area under the curve for milk MAL on our proposed cut-off value ranged from 0.51 to 0.78. Sensitivity/specificity (%) for fat CAL were 61/73, 48/73, 67/73, 35/73, 47/73, 10/73, and 46/73 and for fat MAL were 70/75, 57/75, 67/65, 30/75, 45/75, 19/75, 54/75 for MAS, MET, HYC, DEP, DIG, LAM, and KET, respectively. The AUC values for fat CAL ranged from 0.53 to 0.69 and that for fat HAL ranged from 0.53 to 0.72. The sensitivity/specificity (%) of protein CAL were 42/89, 44/89, 100/89, 22/89, 20/89,10/89, and 19/89 and the protein HAL were 42/90, 43/90, 100/90, 22/90,22/90, 10/100, and 22/90 for MAS, MET, HYC, DEP, DIG, LAM, and KET respectively. The AUC values for protein CAL ranged from 0.50-0.95 and the AUC values for protein HAL ranged from 0.51-0.95. The sensitivity/specificity (%) of lactose CAL were 35/95, 17/95, 100/95, 13/95, 8/95, 5/95, and 5/95 and the sensitivity/ specificity (%) for lactose HAL were 35/96, 4/96, 100/96, 4/96, 8/95, 5/100 and 5/96 for MAS, MET, HYC, DEP, DIG, LAM, and KET, respectively. The AUC values for lactose CAL ranged from 0.50 to 0.98 and the lactose HAL ranged from 0.50 to 0.98 for different disorders. The sensitivity/specificity (%) of fatbylactose CAL were 78/59, 57/59, 80/59, 52/59, 58/59, 58/59, and 61/59 and the sensitivity/specificity (%) for fatbylactose HAL were 80/60, 77/60, 80/60, 48/60, 63/60, 58/60 and 71/60 for MAS, MET, HYC, DEP, DIG, LAM, and KET, respectively. The AUC values for fatbylactose CAL ranged from 0.55 to 0.69 and the lactose HAL ranged from 0.54 to 0.72 for different disorders. In conclusion consistent negative changes in rumination time, activity, and milk yield were observed on the day of clinical diagnosis of disease and could be used to assist in the early detection of periparturient cow disorders.