검색
검색 팝업 닫기

Ex) Article Title, Author, Keywords

Article

J Vet Clin 2022; 39(5): 199-206

https://doi.org/10.17555/jvc.2022.39.5.199

Published online October 31, 2022

Receiver Operating Characteristic Analysis for Prediction of Postpartum Metabolic Diseases in Dairy Cows in an Organic Farm in Korea

Dohee Kim1 , Woojae Choi1 , Younghye Ro1 , Leegon Hong1 , Seongdae Kim1 , Ilsu Yoon1 , Eunhui Choe2 , Danil Kim1,2

1College of Veterinary Medicine and Research Institute for Veterinary Science, Seoul National University, Seoul 08826, Korea
2Farm Animal Clinical Training and Research Center, Institute of Green-Bio Science and Technology, Seoul National University, Pyeongchang 25354, Korea

Correspondence to:*danilkim@snu.ac.kr

Received: June 20, 2022; Revised: August 8, 2022; Accepted: September 20, 2022

Copyright © The Korean Society of Veterinary Clinics.

Postpartum diseases should be predicted to prevent productivity loss before calving especially in organic dairy farms. This study was aimed to investigate the incidence of postpartum metabolic diseases in an organic dairy farm in Korea, to confirm the association between diseases and prepartum blood biochemical parameters, and to evaluate the accuracy of these parameters with a receiver operating characteristic (ROC) analysis for identifying vulnerable cows. Data were collected from 58 Holstein cows (16 primiparous and 42 multiparous) having calved for 2 years on an organic farm. During a transition period from 4 weeks prepartum to 4 weeks postpartum, blood biochemistry was performed through blood collection every 2 weeks with a physical examination. Thirty-one (53.4%) cows (9 primiparous and 22 multiparous) were diagnosed with at least one postpartum disease. Each incidence was 27.6% for subclinical ketosis, 22.4% for subclinical hypocalcemia, 12.1% for retained placenta, 10.3% for displaced abomasum and 5.2% for clinical ketosis. Between at least one disease and no disease, there were significant differences in the prepartum levels of parameters like body condition score (BCS), non-esterified fatty acid (NEFA), total bilirubin (T-bil), direct bilirubin (D-bil) and NEFA to total cholesterol (T-chol) ratio (p < 0.05). The ROC analysis of each of these prepartum parameters had the area under the curve (AUC) <0.7. However, the ROC analysis with logistic regression including all these parameters revealed a higher AUC (0.769), sensitivity (71.0%), and specificity (77.8%). The ROC analysis with logistic regression including the prepartum BCS, NEFA, T-bil, D-bil, and NEFA to T-chol ratio can be used to identify cows that are vulnerable to postpartum diseases with moderate accuracy.

Keywords: organic dairy farm, periparturient disease, metabolic parameters, ROC analysis, logistic regression.

The transition period from 4 weeks before and after calving is crucial period for dairy cow (12). Because of the increased energy requirements for milk production and decreased dry matter intake (DMI), most transition cows experience a negative energy balance (NEB) (2). The NEB status induces lipid mobilization. Further, adipose mobilization is accompanied by high serum concentrations of non-esterified fatty acids (NEFA) (13,28). NEFA is transported for ketone body synthesis in the liver and is released into the blood to increase the concentration of ketone body (11,22,26). Also, a sudden increase in calcium (Ca) demand at the onset of lactation can lead to hypocalcemia as calcium homeostatic mechanism does not adapt to the change in demand optimally (8). These nutritional and physiological changes during the transition period increase the risk of metabolic and infectious diseases after calving in dairy cows (30).

Periparturient disease has a direct impact on the profitability of a dairy farm, owing to productivity loss as well as treatment costs (3). Therefore, early prediction of the occurrence of periparturient disease and the execution of preventive measures through timely monitoring and intervention can reduce economic losses (30). A metabolic profile refers to the analysis of blood biochemical parameters that are useful for the assessment and prevention of metabolic diseases in dairy cows (26). Using metabolic parameters, many studies have predicted postpartum diseases such as displaced abomasum (20,25), retained placenta (17,25,27), ketosis (1,17,30), and hypocalcemia (4). These parameters include NEFA, total cholesterol (T-chol), and β-hydroxybutyrate (BHB). Therefore, a method for predicting the occurrence of postpartum metabolic diseases by evaluating metabolic profiles is warranted.

Recently, with an increase in popularity of eco-friendly foods, interest in organic milk has grown as well. Organic milk refers to the milk that is produced in a farm that is certified as an organic product from the National Agricultural Products Quality Management Service in Korea. To obtain organic farm certification, organic feed and supplements are consumed by the cattle. Feed materials of organic farm may have narrower types and specifications of feed formulations, raw materials of feedstuff, and supplements compared to those of conventional farms. For example, with the restriction of fat protection-oriented organic supplementary feed that can supplement insufficient energy due to peak of milk yields in the early stage of lactation, farmer set the goal of milk yield lower than conventional farm and set feed mixing ratio based on nutritional balance. Moreover, various regulations are followed, such as prohibiting the use of antibiotics as well as other drugs. Due to a limited treatment regimen, parturient cows in organic farms might not receive active supportive care for adaptation to abrupt physiological changes as compared to cows in conventional farms. Therefore, it is important to predict the occurrence of postpartum diseases prior to parturition and to prepare preventive measures, especially for organic farms.

The objectives of this study were to investigate the incidence of postpartum metabolic diseases on an organic dairy farm in Korea and to confirm the association between the occurrence of metabolic diseases and prepartum blood biochemical parameters. Finally, the availability of prepartum blood biochemical parameters was evaluated to identify cows that are vulnerable to postpartum diseases, using a receiver operating characteristic (ROC) analysis.

In this study, data from 58 Holstein parturient cows (16 primiparous and 42 multiparous) were collected from November 2019 to November 2021. The data were collected in an organic farm located in Pyeongchang, Gangwon-do. Every 2 weeks (from 4 weeks prepartum to 4 weeks postpartum) periparturient cows were physically examined and their body condition score was determined by two experienced veterinarians (7). Also, blood biochemical analysis was also performed. The average milk production recorded by the robotic system (DeLaval, Sweden) was 24.9 L. The robot milking machine (automatic milking system) was allowed milking from 1.5 to 3 times a day on average. The cows were fed as shown in Table 1. Although organic farms may have a narrow range of feed formulations and specifications, efforts have been made to feed total mixed ration (TMR) to meet the nutritional requirements of each period (pre- and postpartum) of cows according to NRC (2001).

Table 1 Ingredient and nutrient composition of the prepartum and postpartum cow diet

CompositionDry cow dietLactating cow diet
Ingredient (%)
Formulated feed mixture26.245.2
Silage-16.14
Alfalfa-10.90
Oat hay36.98.72
Meadow hay36.917.43
Kapok-1.61
Nutrient (%)
DM (%)88.078.61
CP (% of DM)10.4615.54
EE (% of DM)1.923.64
CF (% of DM)24.2920.08
Ash (% of DM)7.357.93
Ca (% of DM)0.520.92
P (% of DM)0.310.39
ADF (% of DM)31.2521.59
NDF (% of DM)48.5035.31
NFE (% of DM)44.5431.42
DM Intake (kg/day)12-1321.5


Samples of heparinized blood from the coccygeal vessel were sent to a laboratory. Plasma was separated and used for metabolic profile analyses, which included measuring the levels of total protein (TP), albumin (Alb), triglyceride (TG), T-chol, BHB, NEFA, glucose (Glu), total bilirubin (T-bil), direct bilirubin (D-bil), aspartate aminotransferase (AST), γ-glutamyl transferase (GGT), calcium (Ca), and inorganic phosphate (iP). An automatic analyzer BS-400 (Mindray, Shenzhen, China) was used. The data were divided into four time periods: 15-30 days before calving (BC-2), 1-14 days before calving (BC-1), 0-14 days after calving (AC+1), and 15-30 days after calving (AC+2).

Postpartum metabolic diseases were based on the occurrence from calving to 30 days in milk. Retained placenta was defined as the failure to excrete the placenta by 24 hours after calving (21,25,27). Subclinical hypocalcemia was characterized as the plasma concentration of total Ca < 2.0 mmol/L (29). Cows were considered to have subclinical ketosis when the BHB concentration was 1.2 to 2.9 mmol/L (22), and clinical ketosis if the concentration was ≥ 3.0 mmol/L (25). Displaced abomasum was diagnosed by a veterinarian, based on the auscultation of a characteristic tympanic resonance during percussion (21). Subsequently, the incidence of postpartum metabolic disease was calculated. Appropriate veterinary treatment was provided for each disease immediately after a definite diagnosis.

For statistical analysis, parturient cows were divided into two groups: cows with at least one disease and cows with no disease. Further statistical analysis was performed with 1) cows with subclinical disease vs. clinical disease vs. no disease 2) cows with one disease vs multiple diseases. Subclinical hypocalcemia and subclinical ketosis were included in the subclinical disease, and retained placenta, displaced abomasum, and clinical ketosis were included in the clinical disease.

Data were expressed as mean ± standard deviation (SD). According to the normality test results, parametric data were analyzed with the Student’s t-test, and non-parametric data were analyzed with the Mann-Whitney U test to compare the significant differences in the parameters between diseased and non-diseased cows. To determine the cut-off values for predicting the occurrence of postpartum metabolic diseases, ROC analysis was performed on blood parameters from the two periods before calving. The point on the ROC curve with the highest combined sensitivity and specificity was regarded as the critical threshold (10). Interpretation of this cut-off point was based on the area under the curve (AUC) such that if AUC = 0.5 it was noninformative; if 0.5 < AUC ≤ 0.7, it was less accurate; if 0.7 < AUC ≤ 0.9, it was moderately accurate; if 0.9 < AUC < 1, it was highly accurate; and if AUC = 1, it was considered perfect (10). With all the parameters showing significant differences, the predicted probability model was obtained using logistic regression analysis. And, ROC curves were calculated to determine the sensitivity, specificity, and AUC of predictive models using logistic regression and presence or absence of postpartum disease. The BCS and blood parameters for subclinical vs. clinical vs. no disease cows were analyzed using one-way ANOVA followed by Tukey test at p < 0.05. All statistical analyses were performed using SPSS version 25.0 (IBM SPSS Inc., Chicago, USA).

A total of 31 parturient cows among the 58 cows (53.4%) were diagnosed with at least one disease after calving. The incidence of subclinical, clinical and no disease was 26, 13, and 27 cases, respectively. Nineteen of the 31 cows with periparturient disease had one disease, and 12 of the 31 cows with periparturient disease had two or more diseases. The incidence of disease was 16 cows with subclinical ketosis (27.6%), 13 with subclinical hypocalcemia (22.4%), seven with retained placenta (12.1%), six with displaced abomasum (10.3%) and three with clinical ketosis (5.2%). There was no significant difference in incidence rate according to the parity through chi-test (x2 = 0.07 and p = 0.792): nine of the 16 primiparous cows (56.3%) and 22 of the 42 multiparous cows (52.4%) had postpartum disease.

At BC-2, there was no significant difference in blood biochemistry parameters between diseased and non-diseased cows except for BCS (p = 0.033) in Fig. 1. At BC-1, plasma concentrations of NEFA, T-bil, D-bil, and the NEFA to T-chol ratio were greater for cows in the periparturient disease group than those for cows in the non-periparturient disease group (p = 0.011, p = 0.017, p = 0.026, and p = 0.010, respectively). Interestingly, the aforementioned parameters of BC-1 were also significantly different between the groups at AC-1 and AC-2 (p < 0.01).

Figure 1.Comparison of body condition score (BCS), plasma total protein (TP), albumin (Alb), triglycerides (TG), total cholesterol (T-chol), β-hydroxybutyrate (BHB), non-esterified fatty acid (NEFA), NEFA to T-chol ratio, glucose (Glu), total bilirubin (T-bil), direct bilirubin (D-bil), aspartate aminotransferase (AST), γ-glutamyl transferase (GGT), calcium (Ca), and inorganic phosphate (iP) in cows with periparturient disease and non-periparturient disease at for four time periods. Data are expressed as means ± SD. Significant difference between differences between periparturient disease cows and non-periparturient disease cows is indicated by *p < 0.05 and **p < 0.01.

On the other hands, a significant difference between subclinical vs clinical vs non-disease with prepartum blood biochemical parameters were only in the case of NEFA, in which subclinical cases had significantly higher NEFA concentrations at BC-1 than non-disease cases (p = 0.038). Moreover, the levels of BCS and Ca of multiple disease cases was significantly lower than those of one disease cases at BC-1 (p = 0.019, p = 0.007, respectively).

There were significant group differences in the levels of BCS, NEFA, T-bil, D-bil, and NEFA to T-chol ratio before parturition. These differences were evaluated using ROC analysis to determine a cut-off value for association with the occurrence of periparturient diseases (Table 2). Each parameter included in the ROC curve analysis was less accurate, as AUC was less than 0.7.

Table 2 Receiver operator characteristic (ROC) curve for the prepartum body condition score (BCS) at 15 to 30 days before calving (BC-2). The prepartum plasma non-esterified fatty acid (NEFA), total bilirubin (T-bil), and direct bilirubin (D-bil) concentrations and NEFA to T-chol ratio at 1 to 14 days before calving (BC-1) were measured as predictors of periparturient disease

PeriodParameterCut-off valueSenSpeStandard errorAUC+LR95% CI
BC-2BCS≥ 3.4480.648.10.0740.643
p = 0.061
1.560.498-0.789
BC-1NEFA (mmol/L)≥ 0.3945.285.20.0680.697
p = 0.010
3.050.563-0.831
BC-1T-bil (μmol/L)≥ 15061.374.10.0720.683
p = 0.017
2.360.542-0.823
BC-1D-bil (μmol/L)≥ 85.564.570.40.0720.670
p = 0.026
2.180.529-0.812
BC-1NEFA to T-chol ratio≥ 0.01258.177.80.0690.697
p = 0.010
2.610.563-0.831

Sen, sensitivity; Spe, specificity; AUC, area under the ROC curve; +LR, positive likelihood ratio; CI, confidence interval.



With BCS at BC-2 and plasma T-bil, D-bil concentrations, and NEFA to T-chol ratio at BC-1 were calculated for the predictive models using logistic regression and the presence or absence of postpartum diseases (Fig. 2). The sensitivity, specificity, and AUC were determined as 71.0%, 77.8%, and 0.769, respectively (Table 3).

Table 3 Receiver operator characteristic (ROC) curve for to the predicted probability model using logistic regression analysis, that was measured as a predictor of periparturient disease

Predicted modelCut-off valueSenSpeStandard errorAUC+LR95% CI
p = 11+ez≥ 0.52371.077.80.0630.769
p = 0.000
3.190.646-0.893

Sen, sensitivity; Spe, specificity; AUC, area under the ROC curve; +LR, positive likelihood ratio; CI, confidence interval.

z, –12.874 + 3.376x1 – 0.013x2 + 0.021x3 + 103.893x4.

x1, Body condition score (BCS) 15 to 30 days before calving (BC-2).

x2, Total bilirubin (T-bil) 1 to 14 days before calving (BC-1).

x3, Direct bilirubin (D-bil) BC-1.

x4, Non-esterified fatty acid (NEFA): total cholesterol (T-chol) ratio BC-1.



Figure 2.Comparison of diseased and non-diseased cows with regard to the predicted probability model using logistic regression analysis. Data are presented as mean ± SD and minimum and maximum values to non-periparturient disease and periparturient disease groups.

In this study, 53.4% of cows experienced one or more diseases after calving. This is consistent with several studies showing that more than 50% of cows experience one or more periparturient diseases in conventional farms (21,32,33). The incidence of ketosis (27.6% for subclinical ketosis and 5.2% for clinical ketosis) and displaced abomasum (10.3%) in this study were higher than in those from two conventional farm studies in Korea (15,16). However, the incidence of placental retention (12.1%) was lower than that reported in the aforementioned two studies (16.7% and 30.4%) (15,16). In view of these results, by increasing the feeding amount from 1 week before parturition on this organic farm, retained placenta that checked within 24 h after calving may be less. However, as it progresses to the NEB state, it may be judged that the incidence of ketosis and displaced abomasum is high due to the limitation of organic supplementary feed to supplement the insufficient energy for peak of milk yield in the early lactation period.

Prepartum blood data between cows with one disease and multiple diseases were significantly different for Ca in this study. Periparturient diseases are not single entities but are interconnected (33). Indeed, cows with hypocalcemia are more likely to develop displaced abomasum, ketosis, and retained placenta (5,8,9). Eleven of 12 cows with multiple diseases showed subclinical hypocalcemia in the present study. Subclinical hypocalcemia may result in decreased intestinal motility due to decreased muscle contractility, which may lead to decreased appetite. In other words, it increases the simultaneous risk of abomasal displacement, ketosis and even hepatic lipidosis. Thus, subclinical hypocalcemia is a significant health risk for dairy cows and should be taken preventive measures to reduce potential economic losses to farms.

One study found that a lower BCS in primiparous cows during peripartum influences the occurrence of postpartum health problems (34). However, other studies have shown that dry cows with a better body condition had a greater depression of DMI and a deeper NEB than cows with a lower body condition in the peripartum period (31). According to the results of this study, cows with at least one disease had a higher BCS at BC-2, and when cows with one disease and multiple diseases were compared, cows with multiple diseases had significantly lower BCS in BC-1. Through these studies, prepartum BCS screening itself is vital in predicting the disease, but the degree of prepartum decrease in BCS also seems valuable.

Several studies have indicated that monitoring prepartum NEFA concentration is crucial for predicting postpartum diseases such as displaced abomasum (21,25), retained placenta (27), and ketosis (25). In this study, there was a significant difference between no disease and subclinical disease cows in the case NEFA in BC-1. Elevated NEFA concentrations convert to TG, which must bind to very low density lipoprotein (VLDL) to be transported to the liver (26,28). Low cholesterol levels limit VLDL production and TG accumulates in the liver, increasing the potential for fat infiltration (13,18). Hence, a decrease in plasma T-chol concentration may indicate a decreased DMI and lead to fatty liver due to a decrease in TG transport. Some investigators have suggested the need to assess the NEFA to T-chol ratio for the association between NEFA and T-chol (14,17). The NEFA to T-chol ratio of postpartum diseased cows was significantly higher than that of non-postpartum diseased cows at BC-1 in the present study (p < 0.05).

A previous study reported that the accumulation of fat and changes in cytoplasmic organelles in the liver of fasted cows was accompanied by impaired hepatocyte function, as indicated by the elevated serum T-bil concentration and AST activity (28). Another study also showed that in cows with hyperbilirubinemia, the frequently described signs were anorexia and rumen stasis (23). Hyperbilirubinemia has also been observed in cattle with fatty liver associated with ketosis (26). Bilirubin concentrations were significantly higher in cattle with hepatic lipidosis than in those with non-fatty liver (24). It is believed that increase in T-bil and D-bil is related to a decrease in the hepatic absorption of bilirubin due to competition with other anions such as free fatty acids and an impaired hepatocyte function (23). Our findings regarding the association of high NEFA and T-bil concentrations with the postpartum metabolic disease before calving are consistent with those of previous studies (26,28). Starting with the increase in NEFA, which reflects the status of NEB, which showed a significant difference between subclinical disease and the absence of disease, changes in various blood levels can be viewed as related rather than independently increasing or decreasing levels.

To predict the risk of developing postpartum disease, the cutoff values of prepartum BCS, NEFA, NEFA to T-chol ratio, T-bil and D-bil of dairy cows were determined using ROC curve analysis. However, there no parameter demonstrated an AUC greater than 0.7. ROC analysis based on logistic regression with all parameters with significant differences revealed a higher AUC (0.769), sensitivity (71.0%), and specificity (77.8%). One study confirmed that the results of an ROC analysis with a logistic regression model that combined milk yield and composition for the presence or absence of ketosis had a higher AUC than that of one composition (19). In another study, a high level of accuracy was achieved using a predictive model that combined multiple variables to predict calving time (6). It is suggested that better results that disease can be prevented can be obtained by analyzing each parameter and developing a predictive model by multiplying the logistic regression coefficients according to the negative and positive correlations of the parameters for the presence or absence of postpartum metabolic disease, which is a better direction than predicting postpartum disease through each variable.

In conclusion, this study investigated the incidence rate of postpartum metabolic diseases in dairy cows in an organic farm. We confirmed the presence of significant differences between the diseased and non-diseased group in prepartum BCS, NEFA, T-bil, D-bil, and the NEFA to T-chol ratio. ROC analysis using these parameters could be used to predict disease occurrence with moderate accuracy. However, this study was conducted on only one organic farm, and since the number of individuals used in the study is small, additional research is needed to increase the number of individuals and farms to create an index that can be applied to various farms.

This study was partially supported by the Research Institute for Veterinary Science, Seoul National University.

  1. Asl AN, Nazifi S, Ghasrodashti AR, Olyaee A. Prevalence of subclinical ketosis in dairy cattle in the Southwestern Iran and detection of cutoff point for NEFA and glucose concentrations for diagnosis of subclinical ketosis. Prev Vet Med 2011; 100: 38-43.
    Pubmed CrossRef
  2. Bell AW. Regulation of organic nutrient metabolism during transition from late pregnancy to early lactation. J Anim Sci 1995; 73: 2804-2819.
    Pubmed CrossRef
  3. Caixeta LS, Omontese BO. Monitoring and improving the metabolic health of dairy cows during the transition period. Animals (Basel) 2021; 11: 352.
    Pubmed KoreaMed CrossRef
  4. Chamberlin WG, Middleton JR, Spain JN, Johnson GC, Ellersieck MR, Pithua P. Subclinical hypocalcemia, plasma biochemical parameters, lipid metabolism, postpartum disease, and fertility in postparturient dairy cows. J Dairy Sci 2013; 96: 7001-7013.
    Pubmed CrossRef
  5. Curtis CR, Erb HN, Sniffen CJ, Smith RD, Powers PA, Smith MC, et al. Association of parturient hypocalcemia with eight periparturient disorders in Holstein cows. J Am Vet Med Assoc 1983; 183: 559-561.
  6. Fadul M, Bogdahn C, Alsaaod M, Hüsler J, Starke A, Steiner A, et al. Prediction of calving time in dairy cattle. Anim Reprod Sci 2017; 187: 37-46.
    Pubmed CrossRef
  7. Ferguson JD, Galligan DT, Thomsen N. Principal descriptors of body condition score in Holstein cows. J Dairy Sci 1994; 77: 2695-2703.
    CrossRef
  8. Goff JP. The monitoring, prevention, and treatment of milk fever and subclinical hypocalcemia in dairy cows. Vet J 2008; 176: 50-57.
    Pubmed CrossRef
  9. Goff JP, Horst RL. Physiological changes at parturition and their relationship to metabolic disorders. J Dairy Sci 1997; 80: 1260-1268.
    CrossRef
  10. Greiner M, Pfeiffer D, Smith RD. Principles and practical application of the receiver-operating characteristic analysis for diagnostic tests. Prev Vet Med 2000; 45: 23-41.
    CrossRef
  11. Grummer RR. Etiology of lipid-related metabolic disorders in periparturient dairy cows. J Dairy Sci 1993; 76: 3882-3896.
    CrossRef
  12. Grummer RR. Impact of changes in organic nutrient metabolism on feeding the transition dairy cow. J Anim Sci 1995; 73: 2820-2833.
    Pubmed CrossRef
  13. Herdt TH. Ruminant adaptation to negative energy balance. Influences on the etiology of ketosis and fatty liver. Vet Clin North Am Food Anim Pract 2000; 16: 215-230, v.
    CrossRef
  14. Holtenius P, Hjort M. Studies on the pathogenesis of fatty liver in cows. Bov pract 1990; 25: 91-94.
  15. Jeong JK, Choi IS, Kang HG, Jung YH, Hur TY, Kim IH. Postpartum reproductive tract recovery and prevalence of health problems in dairy cows. J Vet Clin 2015; 32: 168-173.
    CrossRef
  16. Jeong JK, Hur TY, Jung YH, Kang HG, Kim IH. Identification of predictive biomarkers of peri- and postpartum disorders in dairy cows. Korean J Vet Res 2019; 59: 1-8.
    CrossRef
  17. Kaneene JB, Miller R, Herdt TH, Gardiner JC. The association of serum nonesterified fatty acids and cholesterol, management and feeding practices with peripartum disease in dairy cows. Prev Vet Med 1997; 31: 59-72.
    CrossRef
  18. Katoh N. Relevance of apolipoproteins in the development of fatty liver and fatty liver-related peripartum diseases in dairy cows. J Vet Med Sci 2002; 64: 293-307.
    Pubmed CrossRef
  19. Kayano M, Kataoka T. Screening for ketosis using multiple logistic regression based on milk yield and composition. J Vet Med Sci 2015; 77: 1473-1478.
    Pubmed KoreaMed CrossRef
  20. LeBlanc S. Monitoring metabolic health of dairy cattle in the transition period. J Reprod Dev 2010; 56 Suppl: S29-S35.
    Pubmed CrossRef
  21. LeBlanc SJ, Leslie KE, Duffield TF. Metabolic predictors of displaced abomasum in dairy cattle. J Dairy Sci 2005; 88: 159-170.
    CrossRef
  22. McArt JAA, Nydam DV, Oetzel GR. Epidemiology of subclinical ketosis in early lactation dairy cattle. J Dairy Sci 2012; 95: 5056-5066.
    Pubmed CrossRef
  23. McSherry BJ, Lumsden JH, Valli VE, Baird JD. Hyperbilirubinemia in sick cattle. Can J Comp Med 1984; 48: 237-240.
  24. Mostafavi M, Seifi HA, Mohri M, Jamshidi A. Optimal thresholds of metabolic indicators of hepatic lipidosis in dairy cows. Revue Med Vet 2013; 164: 564-571.
  25. Ospina PA, Nydam DV, Stokol T, Overton TR. Evaluation of nonesterified fatty acids and beta-hydroxybutyrate in transition dairy cattle in the northeastern United States: critical thresholds for prediction of clinical diseases. J Dairy Sci 2010; 93: 546-554.
    Pubmed CrossRef
  26. Puppel K, Kuczyńska B. Metabolic profiles of cow’s blood; a review. J Sci Food Agric 2016; 96: 4321-4328.
    Pubmed CrossRef
  27. Quiroz-Rocha GF, LeBlanc S, Duffield T, Wood D, Leslie KE, Jacobs RM. Evaluation of prepartum serum cholesterol and fatty acids concentrations as predictors of postpartum retention of the placenta in dairy cows. J Am Vet Med Assoc 2009; 234: 790-793.
    Pubmed CrossRef
  28. Reid IM, Harrison RD, Collins RA. Fasting and refeeding in the lactating dairy cow. 2. The recovery of liver cell structure and function following a six-day fast. J Comp Pathol 1977; 87: 253-265.
    CrossRef
  29. Reinhardt TA, Lippolis JD, McCluskey BJ, Goff JP, Horst RL. Prevalence of subclinical hypocalcemia in dairy herds. Vet J 2011; 188: 122-124.
    Pubmed CrossRef
  30. Ro Y, Choi W, Kim H, Kim D. Prepartal decrease in plasma total cholesterol concentration in dairy cows developed subclinical ketosis. J Vet Clin 2017; 34: 222-224.
    CrossRef
  31. Rukkwamsuk T, Kruip TA, Wensing T. Relationship between overfeeding and overconditioning in the dry period and the problems of high producing dairy cows during the postparturient period. Vet Q 1999; 21: 71-77.
    Pubmed CrossRef
  32. Sepúlveda-Varas P, Weary DM, Noro M, von Keyserlingk MA. Transition diseases in grazing dairy cows are related to serum cholesterol and other analytes. PLoS One 2015; 10: e0122317.
    Pubmed KoreaMed CrossRef
  33. Van Saun RJ. Metabolic profiles for evaluation of the transition period. Am Assoc Bov Pract Conf Proc 2006; 39: 130-138.
  34. Wang Y, Huo P, Sun Y, Zhang Y. Effects of body condition score changes during peripartum on the postpartum health and production performance of primiparous dairy cows. Animals (Basel) 2019; 9: 1159.
    Pubmed KoreaMed CrossRef

Article

Original Article

J Vet Clin 2022; 39(5): 199-206

Published online October 31, 2022 https://doi.org/10.17555/jvc.2022.39.5.199

Copyright © The Korean Society of Veterinary Clinics.

Receiver Operating Characteristic Analysis for Prediction of Postpartum Metabolic Diseases in Dairy Cows in an Organic Farm in Korea

Dohee Kim1 , Woojae Choi1 , Younghye Ro1 , Leegon Hong1 , Seongdae Kim1 , Ilsu Yoon1 , Eunhui Choe2 , Danil Kim1,2

1College of Veterinary Medicine and Research Institute for Veterinary Science, Seoul National University, Seoul 08826, Korea
2Farm Animal Clinical Training and Research Center, Institute of Green-Bio Science and Technology, Seoul National University, Pyeongchang 25354, Korea

Correspondence to:*danilkim@snu.ac.kr

Received: June 20, 2022; Revised: August 8, 2022; Accepted: September 20, 2022

This is an open access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

Postpartum diseases should be predicted to prevent productivity loss before calving especially in organic dairy farms. This study was aimed to investigate the incidence of postpartum metabolic diseases in an organic dairy farm in Korea, to confirm the association between diseases and prepartum blood biochemical parameters, and to evaluate the accuracy of these parameters with a receiver operating characteristic (ROC) analysis for identifying vulnerable cows. Data were collected from 58 Holstein cows (16 primiparous and 42 multiparous) having calved for 2 years on an organic farm. During a transition period from 4 weeks prepartum to 4 weeks postpartum, blood biochemistry was performed through blood collection every 2 weeks with a physical examination. Thirty-one (53.4%) cows (9 primiparous and 22 multiparous) were diagnosed with at least one postpartum disease. Each incidence was 27.6% for subclinical ketosis, 22.4% for subclinical hypocalcemia, 12.1% for retained placenta, 10.3% for displaced abomasum and 5.2% for clinical ketosis. Between at least one disease and no disease, there were significant differences in the prepartum levels of parameters like body condition score (BCS), non-esterified fatty acid (NEFA), total bilirubin (T-bil), direct bilirubin (D-bil) and NEFA to total cholesterol (T-chol) ratio (p < 0.05). The ROC analysis of each of these prepartum parameters had the area under the curve (AUC) <0.7. However, the ROC analysis with logistic regression including all these parameters revealed a higher AUC (0.769), sensitivity (71.0%), and specificity (77.8%). The ROC analysis with logistic regression including the prepartum BCS, NEFA, T-bil, D-bil, and NEFA to T-chol ratio can be used to identify cows that are vulnerable to postpartum diseases with moderate accuracy.

Keywords: organic dairy farm, periparturient disease, metabolic parameters, ROC analysis, logistic regression.

Introduction

The transition period from 4 weeks before and after calving is crucial period for dairy cow (12). Because of the increased energy requirements for milk production and decreased dry matter intake (DMI), most transition cows experience a negative energy balance (NEB) (2). The NEB status induces lipid mobilization. Further, adipose mobilization is accompanied by high serum concentrations of non-esterified fatty acids (NEFA) (13,28). NEFA is transported for ketone body synthesis in the liver and is released into the blood to increase the concentration of ketone body (11,22,26). Also, a sudden increase in calcium (Ca) demand at the onset of lactation can lead to hypocalcemia as calcium homeostatic mechanism does not adapt to the change in demand optimally (8). These nutritional and physiological changes during the transition period increase the risk of metabolic and infectious diseases after calving in dairy cows (30).

Periparturient disease has a direct impact on the profitability of a dairy farm, owing to productivity loss as well as treatment costs (3). Therefore, early prediction of the occurrence of periparturient disease and the execution of preventive measures through timely monitoring and intervention can reduce economic losses (30). A metabolic profile refers to the analysis of blood biochemical parameters that are useful for the assessment and prevention of metabolic diseases in dairy cows (26). Using metabolic parameters, many studies have predicted postpartum diseases such as displaced abomasum (20,25), retained placenta (17,25,27), ketosis (1,17,30), and hypocalcemia (4). These parameters include NEFA, total cholesterol (T-chol), and β-hydroxybutyrate (BHB). Therefore, a method for predicting the occurrence of postpartum metabolic diseases by evaluating metabolic profiles is warranted.

Recently, with an increase in popularity of eco-friendly foods, interest in organic milk has grown as well. Organic milk refers to the milk that is produced in a farm that is certified as an organic product from the National Agricultural Products Quality Management Service in Korea. To obtain organic farm certification, organic feed and supplements are consumed by the cattle. Feed materials of organic farm may have narrower types and specifications of feed formulations, raw materials of feedstuff, and supplements compared to those of conventional farms. For example, with the restriction of fat protection-oriented organic supplementary feed that can supplement insufficient energy due to peak of milk yields in the early stage of lactation, farmer set the goal of milk yield lower than conventional farm and set feed mixing ratio based on nutritional balance. Moreover, various regulations are followed, such as prohibiting the use of antibiotics as well as other drugs. Due to a limited treatment regimen, parturient cows in organic farms might not receive active supportive care for adaptation to abrupt physiological changes as compared to cows in conventional farms. Therefore, it is important to predict the occurrence of postpartum diseases prior to parturition and to prepare preventive measures, especially for organic farms.

The objectives of this study were to investigate the incidence of postpartum metabolic diseases on an organic dairy farm in Korea and to confirm the association between the occurrence of metabolic diseases and prepartum blood biochemical parameters. Finally, the availability of prepartum blood biochemical parameters was evaluated to identify cows that are vulnerable to postpartum diseases, using a receiver operating characteristic (ROC) analysis.

Materials and Methods

In this study, data from 58 Holstein parturient cows (16 primiparous and 42 multiparous) were collected from November 2019 to November 2021. The data were collected in an organic farm located in Pyeongchang, Gangwon-do. Every 2 weeks (from 4 weeks prepartum to 4 weeks postpartum) periparturient cows were physically examined and their body condition score was determined by two experienced veterinarians (7). Also, blood biochemical analysis was also performed. The average milk production recorded by the robotic system (DeLaval, Sweden) was 24.9 L. The robot milking machine (automatic milking system) was allowed milking from 1.5 to 3 times a day on average. The cows were fed as shown in Table 1. Although organic farms may have a narrow range of feed formulations and specifications, efforts have been made to feed total mixed ration (TMR) to meet the nutritional requirements of each period (pre- and postpartum) of cows according to NRC (2001).

Table 1 . Ingredient and nutrient composition of the prepartum and postpartum cow diet.

CompositionDry cow dietLactating cow diet
Ingredient (%)
Formulated feed mixture26.245.2
Silage-16.14
Alfalfa-10.90
Oat hay36.98.72
Meadow hay36.917.43
Kapok-1.61
Nutrient (%)
DM (%)88.078.61
CP (% of DM)10.4615.54
EE (% of DM)1.923.64
CF (% of DM)24.2920.08
Ash (% of DM)7.357.93
Ca (% of DM)0.520.92
P (% of DM)0.310.39
ADF (% of DM)31.2521.59
NDF (% of DM)48.5035.31
NFE (% of DM)44.5431.42
DM Intake (kg/day)12-1321.5


Samples of heparinized blood from the coccygeal vessel were sent to a laboratory. Plasma was separated and used for metabolic profile analyses, which included measuring the levels of total protein (TP), albumin (Alb), triglyceride (TG), T-chol, BHB, NEFA, glucose (Glu), total bilirubin (T-bil), direct bilirubin (D-bil), aspartate aminotransferase (AST), γ-glutamyl transferase (GGT), calcium (Ca), and inorganic phosphate (iP). An automatic analyzer BS-400 (Mindray, Shenzhen, China) was used. The data were divided into four time periods: 15-30 days before calving (BC-2), 1-14 days before calving (BC-1), 0-14 days after calving (AC+1), and 15-30 days after calving (AC+2).

Postpartum metabolic diseases were based on the occurrence from calving to 30 days in milk. Retained placenta was defined as the failure to excrete the placenta by 24 hours after calving (21,25,27). Subclinical hypocalcemia was characterized as the plasma concentration of total Ca < 2.0 mmol/L (29). Cows were considered to have subclinical ketosis when the BHB concentration was 1.2 to 2.9 mmol/L (22), and clinical ketosis if the concentration was ≥ 3.0 mmol/L (25). Displaced abomasum was diagnosed by a veterinarian, based on the auscultation of a characteristic tympanic resonance during percussion (21). Subsequently, the incidence of postpartum metabolic disease was calculated. Appropriate veterinary treatment was provided for each disease immediately after a definite diagnosis.

For statistical analysis, parturient cows were divided into two groups: cows with at least one disease and cows with no disease. Further statistical analysis was performed with 1) cows with subclinical disease vs. clinical disease vs. no disease 2) cows with one disease vs multiple diseases. Subclinical hypocalcemia and subclinical ketosis were included in the subclinical disease, and retained placenta, displaced abomasum, and clinical ketosis were included in the clinical disease.

Data were expressed as mean ± standard deviation (SD). According to the normality test results, parametric data were analyzed with the Student’s t-test, and non-parametric data were analyzed with the Mann-Whitney U test to compare the significant differences in the parameters between diseased and non-diseased cows. To determine the cut-off values for predicting the occurrence of postpartum metabolic diseases, ROC analysis was performed on blood parameters from the two periods before calving. The point on the ROC curve with the highest combined sensitivity and specificity was regarded as the critical threshold (10). Interpretation of this cut-off point was based on the area under the curve (AUC) such that if AUC = 0.5 it was noninformative; if 0.5 < AUC ≤ 0.7, it was less accurate; if 0.7 < AUC ≤ 0.9, it was moderately accurate; if 0.9 < AUC < 1, it was highly accurate; and if AUC = 1, it was considered perfect (10). With all the parameters showing significant differences, the predicted probability model was obtained using logistic regression analysis. And, ROC curves were calculated to determine the sensitivity, specificity, and AUC of predictive models using logistic regression and presence or absence of postpartum disease. The BCS and blood parameters for subclinical vs. clinical vs. no disease cows were analyzed using one-way ANOVA followed by Tukey test at p < 0.05. All statistical analyses were performed using SPSS version 25.0 (IBM SPSS Inc., Chicago, USA).

Results

A total of 31 parturient cows among the 58 cows (53.4%) were diagnosed with at least one disease after calving. The incidence of subclinical, clinical and no disease was 26, 13, and 27 cases, respectively. Nineteen of the 31 cows with periparturient disease had one disease, and 12 of the 31 cows with periparturient disease had two or more diseases. The incidence of disease was 16 cows with subclinical ketosis (27.6%), 13 with subclinical hypocalcemia (22.4%), seven with retained placenta (12.1%), six with displaced abomasum (10.3%) and three with clinical ketosis (5.2%). There was no significant difference in incidence rate according to the parity through chi-test (x2 = 0.07 and p = 0.792): nine of the 16 primiparous cows (56.3%) and 22 of the 42 multiparous cows (52.4%) had postpartum disease.

At BC-2, there was no significant difference in blood biochemistry parameters between diseased and non-diseased cows except for BCS (p = 0.033) in Fig. 1. At BC-1, plasma concentrations of NEFA, T-bil, D-bil, and the NEFA to T-chol ratio were greater for cows in the periparturient disease group than those for cows in the non-periparturient disease group (p = 0.011, p = 0.017, p = 0.026, and p = 0.010, respectively). Interestingly, the aforementioned parameters of BC-1 were also significantly different between the groups at AC-1 and AC-2 (p < 0.01).

Figure 1. Comparison of body condition score (BCS), plasma total protein (TP), albumin (Alb), triglycerides (TG), total cholesterol (T-chol), β-hydroxybutyrate (BHB), non-esterified fatty acid (NEFA), NEFA to T-chol ratio, glucose (Glu), total bilirubin (T-bil), direct bilirubin (D-bil), aspartate aminotransferase (AST), γ-glutamyl transferase (GGT), calcium (Ca), and inorganic phosphate (iP) in cows with periparturient disease and non-periparturient disease at for four time periods. Data are expressed as means ± SD. Significant difference between differences between periparturient disease cows and non-periparturient disease cows is indicated by *p < 0.05 and **p < 0.01.

On the other hands, a significant difference between subclinical vs clinical vs non-disease with prepartum blood biochemical parameters were only in the case of NEFA, in which subclinical cases had significantly higher NEFA concentrations at BC-1 than non-disease cases (p = 0.038). Moreover, the levels of BCS and Ca of multiple disease cases was significantly lower than those of one disease cases at BC-1 (p = 0.019, p = 0.007, respectively).

There were significant group differences in the levels of BCS, NEFA, T-bil, D-bil, and NEFA to T-chol ratio before parturition. These differences were evaluated using ROC analysis to determine a cut-off value for association with the occurrence of periparturient diseases (Table 2). Each parameter included in the ROC curve analysis was less accurate, as AUC was less than 0.7.

Table 2 . Receiver operator characteristic (ROC) curve for the prepartum body condition score (BCS) at 15 to 30 days before calving (BC-2). The prepartum plasma non-esterified fatty acid (NEFA), total bilirubin (T-bil), and direct bilirubin (D-bil) concentrations and NEFA to T-chol ratio at 1 to 14 days before calving (BC-1) were measured as predictors of periparturient disease.

PeriodParameterCut-off valueSenSpeStandard errorAUC+LR95% CI
BC-2BCS≥ 3.4480.648.10.0740.643
p = 0.061
1.560.498-0.789
BC-1NEFA (mmol/L)≥ 0.3945.285.20.0680.697
p = 0.010
3.050.563-0.831
BC-1T-bil (μmol/L)≥ 15061.374.10.0720.683
p = 0.017
2.360.542-0.823
BC-1D-bil (μmol/L)≥ 85.564.570.40.0720.670
p = 0.026
2.180.529-0.812
BC-1NEFA to T-chol ratio≥ 0.01258.177.80.0690.697
p = 0.010
2.610.563-0.831

Sen, sensitivity; Spe, specificity; AUC, area under the ROC curve; +LR, positive likelihood ratio; CI, confidence interval..



With BCS at BC-2 and plasma T-bil, D-bil concentrations, and NEFA to T-chol ratio at BC-1 were calculated for the predictive models using logistic regression and the presence or absence of postpartum diseases (Fig. 2). The sensitivity, specificity, and AUC were determined as 71.0%, 77.8%, and 0.769, respectively (Table 3).

Table 3 . Receiver operator characteristic (ROC) curve for to the predicted probability model using logistic regression analysis, that was measured as a predictor of periparturient disease.

Predicted modelCut-off valueSenSpeStandard errorAUC+LR95% CI
p = 11+ez≥ 0.52371.077.80.0630.769
p = 0.000
3.190.646-0.893

Sen, sensitivity; Spe, specificity; AUC, area under the ROC curve; +LR, positive likelihood ratio; CI, confidence interval..

z, –12.874 + 3.376x1 – 0.013x2 + 0.021x3 + 103.893x4..

x1, Body condition score (BCS) 15 to 30 days before calving (BC-2)..

x2, Total bilirubin (T-bil) 1 to 14 days before calving (BC-1)..

x3, Direct bilirubin (D-bil) BC-1..

x4, Non-esterified fatty acid (NEFA): total cholesterol (T-chol) ratio BC-1..



Figure 2. Comparison of diseased and non-diseased cows with regard to the predicted probability model using logistic regression analysis. Data are presented as mean ± SD and minimum and maximum values to non-periparturient disease and periparturient disease groups.

Discussion

In this study, 53.4% of cows experienced one or more diseases after calving. This is consistent with several studies showing that more than 50% of cows experience one or more periparturient diseases in conventional farms (21,32,33). The incidence of ketosis (27.6% for subclinical ketosis and 5.2% for clinical ketosis) and displaced abomasum (10.3%) in this study were higher than in those from two conventional farm studies in Korea (15,16). However, the incidence of placental retention (12.1%) was lower than that reported in the aforementioned two studies (16.7% and 30.4%) (15,16). In view of these results, by increasing the feeding amount from 1 week before parturition on this organic farm, retained placenta that checked within 24 h after calving may be less. However, as it progresses to the NEB state, it may be judged that the incidence of ketosis and displaced abomasum is high due to the limitation of organic supplementary feed to supplement the insufficient energy for peak of milk yield in the early lactation period.

Prepartum blood data between cows with one disease and multiple diseases were significantly different for Ca in this study. Periparturient diseases are not single entities but are interconnected (33). Indeed, cows with hypocalcemia are more likely to develop displaced abomasum, ketosis, and retained placenta (5,8,9). Eleven of 12 cows with multiple diseases showed subclinical hypocalcemia in the present study. Subclinical hypocalcemia may result in decreased intestinal motility due to decreased muscle contractility, which may lead to decreased appetite. In other words, it increases the simultaneous risk of abomasal displacement, ketosis and even hepatic lipidosis. Thus, subclinical hypocalcemia is a significant health risk for dairy cows and should be taken preventive measures to reduce potential economic losses to farms.

One study found that a lower BCS in primiparous cows during peripartum influences the occurrence of postpartum health problems (34). However, other studies have shown that dry cows with a better body condition had a greater depression of DMI and a deeper NEB than cows with a lower body condition in the peripartum period (31). According to the results of this study, cows with at least one disease had a higher BCS at BC-2, and when cows with one disease and multiple diseases were compared, cows with multiple diseases had significantly lower BCS in BC-1. Through these studies, prepartum BCS screening itself is vital in predicting the disease, but the degree of prepartum decrease in BCS also seems valuable.

Several studies have indicated that monitoring prepartum NEFA concentration is crucial for predicting postpartum diseases such as displaced abomasum (21,25), retained placenta (27), and ketosis (25). In this study, there was a significant difference between no disease and subclinical disease cows in the case NEFA in BC-1. Elevated NEFA concentrations convert to TG, which must bind to very low density lipoprotein (VLDL) to be transported to the liver (26,28). Low cholesterol levels limit VLDL production and TG accumulates in the liver, increasing the potential for fat infiltration (13,18). Hence, a decrease in plasma T-chol concentration may indicate a decreased DMI and lead to fatty liver due to a decrease in TG transport. Some investigators have suggested the need to assess the NEFA to T-chol ratio for the association between NEFA and T-chol (14,17). The NEFA to T-chol ratio of postpartum diseased cows was significantly higher than that of non-postpartum diseased cows at BC-1 in the present study (p < 0.05).

A previous study reported that the accumulation of fat and changes in cytoplasmic organelles in the liver of fasted cows was accompanied by impaired hepatocyte function, as indicated by the elevated serum T-bil concentration and AST activity (28). Another study also showed that in cows with hyperbilirubinemia, the frequently described signs were anorexia and rumen stasis (23). Hyperbilirubinemia has also been observed in cattle with fatty liver associated with ketosis (26). Bilirubin concentrations were significantly higher in cattle with hepatic lipidosis than in those with non-fatty liver (24). It is believed that increase in T-bil and D-bil is related to a decrease in the hepatic absorption of bilirubin due to competition with other anions such as free fatty acids and an impaired hepatocyte function (23). Our findings regarding the association of high NEFA and T-bil concentrations with the postpartum metabolic disease before calving are consistent with those of previous studies (26,28). Starting with the increase in NEFA, which reflects the status of NEB, which showed a significant difference between subclinical disease and the absence of disease, changes in various blood levels can be viewed as related rather than independently increasing or decreasing levels.

To predict the risk of developing postpartum disease, the cutoff values of prepartum BCS, NEFA, NEFA to T-chol ratio, T-bil and D-bil of dairy cows were determined using ROC curve analysis. However, there no parameter demonstrated an AUC greater than 0.7. ROC analysis based on logistic regression with all parameters with significant differences revealed a higher AUC (0.769), sensitivity (71.0%), and specificity (77.8%). One study confirmed that the results of an ROC analysis with a logistic regression model that combined milk yield and composition for the presence or absence of ketosis had a higher AUC than that of one composition (19). In another study, a high level of accuracy was achieved using a predictive model that combined multiple variables to predict calving time (6). It is suggested that better results that disease can be prevented can be obtained by analyzing each parameter and developing a predictive model by multiplying the logistic regression coefficients according to the negative and positive correlations of the parameters for the presence or absence of postpartum metabolic disease, which is a better direction than predicting postpartum disease through each variable.

In conclusion, this study investigated the incidence rate of postpartum metabolic diseases in dairy cows in an organic farm. We confirmed the presence of significant differences between the diseased and non-diseased group in prepartum BCS, NEFA, T-bil, D-bil, and the NEFA to T-chol ratio. ROC analysis using these parameters could be used to predict disease occurrence with moderate accuracy. However, this study was conducted on only one organic farm, and since the number of individuals used in the study is small, additional research is needed to increase the number of individuals and farms to create an index that can be applied to various farms.

Acknowledgements

This study was partially supported by the Research Institute for Veterinary Science, Seoul National University.

Conflicts of Interest

The authors have no conflicting interests.

Fig 1.

Figure 1.Comparison of body condition score (BCS), plasma total protein (TP), albumin (Alb), triglycerides (TG), total cholesterol (T-chol), β-hydroxybutyrate (BHB), non-esterified fatty acid (NEFA), NEFA to T-chol ratio, glucose (Glu), total bilirubin (T-bil), direct bilirubin (D-bil), aspartate aminotransferase (AST), γ-glutamyl transferase (GGT), calcium (Ca), and inorganic phosphate (iP) in cows with periparturient disease and non-periparturient disease at for four time periods. Data are expressed as means ± SD. Significant difference between differences between periparturient disease cows and non-periparturient disease cows is indicated by *p < 0.05 and **p < 0.01.
Journal of Veterinary Clinics 2022; 39: 199-206https://doi.org/10.17555/jvc.2022.39.5.199

Fig 2.

Figure 2.Comparison of diseased and non-diseased cows with regard to the predicted probability model using logistic regression analysis. Data are presented as mean ± SD and minimum and maximum values to non-periparturient disease and periparturient disease groups.
Journal of Veterinary Clinics 2022; 39: 199-206https://doi.org/10.17555/jvc.2022.39.5.199

Table 1 Ingredient and nutrient composition of the prepartum and postpartum cow diet

CompositionDry cow dietLactating cow diet
Ingredient (%)
Formulated feed mixture26.245.2
Silage-16.14
Alfalfa-10.90
Oat hay36.98.72
Meadow hay36.917.43
Kapok-1.61
Nutrient (%)
DM (%)88.078.61
CP (% of DM)10.4615.54
EE (% of DM)1.923.64
CF (% of DM)24.2920.08
Ash (% of DM)7.357.93
Ca (% of DM)0.520.92
P (% of DM)0.310.39
ADF (% of DM)31.2521.59
NDF (% of DM)48.5035.31
NFE (% of DM)44.5431.42
DM Intake (kg/day)12-1321.5

Table 2 Receiver operator characteristic (ROC) curve for the prepartum body condition score (BCS) at 15 to 30 days before calving (BC-2). The prepartum plasma non-esterified fatty acid (NEFA), total bilirubin (T-bil), and direct bilirubin (D-bil) concentrations and NEFA to T-chol ratio at 1 to 14 days before calving (BC-1) were measured as predictors of periparturient disease

PeriodParameterCut-off valueSenSpeStandard errorAUC+LR95% CI
BC-2BCS≥ 3.4480.648.10.0740.643
p = 0.061
1.560.498-0.789
BC-1NEFA (mmol/L)≥ 0.3945.285.20.0680.697
p = 0.010
3.050.563-0.831
BC-1T-bil (μmol/L)≥ 15061.374.10.0720.683
p = 0.017
2.360.542-0.823
BC-1D-bil (μmol/L)≥ 85.564.570.40.0720.670
p = 0.026
2.180.529-0.812
BC-1NEFA to T-chol ratio≥ 0.01258.177.80.0690.697
p = 0.010
2.610.563-0.831

Sen, sensitivity; Spe, specificity; AUC, area under the ROC curve; +LR, positive likelihood ratio; CI, confidence interval.


Table 3 Receiver operator characteristic (ROC) curve for to the predicted probability model using logistic regression analysis, that was measured as a predictor of periparturient disease

Predicted modelCut-off valueSenSpeStandard errorAUC+LR95% CI
p = 11+ez≥ 0.52371.077.80.0630.769
p = 0.000
3.190.646-0.893

Sen, sensitivity; Spe, specificity; AUC, area under the ROC curve; +LR, positive likelihood ratio; CI, confidence interval.

z, –12.874 + 3.376x1 – 0.013x2 + 0.021x3 + 103.893x4.

x1, Body condition score (BCS) 15 to 30 days before calving (BC-2).

x2, Total bilirubin (T-bil) 1 to 14 days before calving (BC-1).

x3, Direct bilirubin (D-bil) BC-1.

x4, Non-esterified fatty acid (NEFA): total cholesterol (T-chol) ratio BC-1.


References

  1. Asl AN, Nazifi S, Ghasrodashti AR, Olyaee A. Prevalence of subclinical ketosis in dairy cattle in the Southwestern Iran and detection of cutoff point for NEFA and glucose concentrations for diagnosis of subclinical ketosis. Prev Vet Med 2011; 100: 38-43.
    Pubmed CrossRef
  2. Bell AW. Regulation of organic nutrient metabolism during transition from late pregnancy to early lactation. J Anim Sci 1995; 73: 2804-2819.
    Pubmed CrossRef
  3. Caixeta LS, Omontese BO. Monitoring and improving the metabolic health of dairy cows during the transition period. Animals (Basel) 2021; 11: 352.
    Pubmed KoreaMed CrossRef
  4. Chamberlin WG, Middleton JR, Spain JN, Johnson GC, Ellersieck MR, Pithua P. Subclinical hypocalcemia, plasma biochemical parameters, lipid metabolism, postpartum disease, and fertility in postparturient dairy cows. J Dairy Sci 2013; 96: 7001-7013.
    Pubmed CrossRef
  5. Curtis CR, Erb HN, Sniffen CJ, Smith RD, Powers PA, Smith MC, et al. Association of parturient hypocalcemia with eight periparturient disorders in Holstein cows. J Am Vet Med Assoc 1983; 183: 559-561.
  6. Fadul M, Bogdahn C, Alsaaod M, Hüsler J, Starke A, Steiner A, et al. Prediction of calving time in dairy cattle. Anim Reprod Sci 2017; 187: 37-46.
    Pubmed CrossRef
  7. Ferguson JD, Galligan DT, Thomsen N. Principal descriptors of body condition score in Holstein cows. J Dairy Sci 1994; 77: 2695-2703.
    CrossRef
  8. Goff JP. The monitoring, prevention, and treatment of milk fever and subclinical hypocalcemia in dairy cows. Vet J 2008; 176: 50-57.
    Pubmed CrossRef
  9. Goff JP, Horst RL. Physiological changes at parturition and their relationship to metabolic disorders. J Dairy Sci 1997; 80: 1260-1268.
    CrossRef
  10. Greiner M, Pfeiffer D, Smith RD. Principles and practical application of the receiver-operating characteristic analysis for diagnostic tests. Prev Vet Med 2000; 45: 23-41.
    CrossRef
  11. Grummer RR. Etiology of lipid-related metabolic disorders in periparturient dairy cows. J Dairy Sci 1993; 76: 3882-3896.
    CrossRef
  12. Grummer RR. Impact of changes in organic nutrient metabolism on feeding the transition dairy cow. J Anim Sci 1995; 73: 2820-2833.
    Pubmed CrossRef
  13. Herdt TH. Ruminant adaptation to negative energy balance. Influences on the etiology of ketosis and fatty liver. Vet Clin North Am Food Anim Pract 2000; 16: 215-230, v.
    CrossRef
  14. Holtenius P, Hjort M. Studies on the pathogenesis of fatty liver in cows. Bov pract 1990; 25: 91-94.
  15. Jeong JK, Choi IS, Kang HG, Jung YH, Hur TY, Kim IH. Postpartum reproductive tract recovery and prevalence of health problems in dairy cows. J Vet Clin 2015; 32: 168-173.
    CrossRef
  16. Jeong JK, Hur TY, Jung YH, Kang HG, Kim IH. Identification of predictive biomarkers of peri- and postpartum disorders in dairy cows. Korean J Vet Res 2019; 59: 1-8.
    CrossRef
  17. Kaneene JB, Miller R, Herdt TH, Gardiner JC. The association of serum nonesterified fatty acids and cholesterol, management and feeding practices with peripartum disease in dairy cows. Prev Vet Med 1997; 31: 59-72.
    CrossRef
  18. Katoh N. Relevance of apolipoproteins in the development of fatty liver and fatty liver-related peripartum diseases in dairy cows. J Vet Med Sci 2002; 64: 293-307.
    Pubmed CrossRef
  19. Kayano M, Kataoka T. Screening for ketosis using multiple logistic regression based on milk yield and composition. J Vet Med Sci 2015; 77: 1473-1478.
    Pubmed KoreaMed CrossRef
  20. LeBlanc S. Monitoring metabolic health of dairy cattle in the transition period. J Reprod Dev 2010; 56 Suppl: S29-S35.
    Pubmed CrossRef
  21. LeBlanc SJ, Leslie KE, Duffield TF. Metabolic predictors of displaced abomasum in dairy cattle. J Dairy Sci 2005; 88: 159-170.
    CrossRef
  22. McArt JAA, Nydam DV, Oetzel GR. Epidemiology of subclinical ketosis in early lactation dairy cattle. J Dairy Sci 2012; 95: 5056-5066.
    Pubmed CrossRef
  23. McSherry BJ, Lumsden JH, Valli VE, Baird JD. Hyperbilirubinemia in sick cattle. Can J Comp Med 1984; 48: 237-240.
  24. Mostafavi M, Seifi HA, Mohri M, Jamshidi A. Optimal thresholds of metabolic indicators of hepatic lipidosis in dairy cows. Revue Med Vet 2013; 164: 564-571.
  25. Ospina PA, Nydam DV, Stokol T, Overton TR. Evaluation of nonesterified fatty acids and beta-hydroxybutyrate in transition dairy cattle in the northeastern United States: critical thresholds for prediction of clinical diseases. J Dairy Sci 2010; 93: 546-554.
    Pubmed CrossRef
  26. Puppel K, Kuczyńska B. Metabolic profiles of cow’s blood; a review. J Sci Food Agric 2016; 96: 4321-4328.
    Pubmed CrossRef
  27. Quiroz-Rocha GF, LeBlanc S, Duffield T, Wood D, Leslie KE, Jacobs RM. Evaluation of prepartum serum cholesterol and fatty acids concentrations as predictors of postpartum retention of the placenta in dairy cows. J Am Vet Med Assoc 2009; 234: 790-793.
    Pubmed CrossRef
  28. Reid IM, Harrison RD, Collins RA. Fasting and refeeding in the lactating dairy cow. 2. The recovery of liver cell structure and function following a six-day fast. J Comp Pathol 1977; 87: 253-265.
    CrossRef
  29. Reinhardt TA, Lippolis JD, McCluskey BJ, Goff JP, Horst RL. Prevalence of subclinical hypocalcemia in dairy herds. Vet J 2011; 188: 122-124.
    Pubmed CrossRef
  30. Ro Y, Choi W, Kim H, Kim D. Prepartal decrease in plasma total cholesterol concentration in dairy cows developed subclinical ketosis. J Vet Clin 2017; 34: 222-224.
    CrossRef
  31. Rukkwamsuk T, Kruip TA, Wensing T. Relationship between overfeeding and overconditioning in the dry period and the problems of high producing dairy cows during the postparturient period. Vet Q 1999; 21: 71-77.
    Pubmed CrossRef
  32. Sepúlveda-Varas P, Weary DM, Noro M, von Keyserlingk MA. Transition diseases in grazing dairy cows are related to serum cholesterol and other analytes. PLoS One 2015; 10: e0122317.
    Pubmed KoreaMed CrossRef
  33. Van Saun RJ. Metabolic profiles for evaluation of the transition period. Am Assoc Bov Pract Conf Proc 2006; 39: 130-138.
  34. Wang Y, Huo P, Sun Y, Zhang Y. Effects of body condition score changes during peripartum on the postpartum health and production performance of primiparous dairy cows. Animals (Basel) 2019; 9: 1159.
    Pubmed KoreaMed CrossRef

Vol.39 No.5 2022-10-31

qrcode
qrcode
The Korean Society of Veterinary Clinics

pISSN 1598-298X
eISSN 2384-0749

Stats or Metrics

Share this article on :

  • line