Skip to main content

Table 3 Model area under the curve (AUC) compared to the area under a random and non-informative receiver operating characteristic (ROC) curve (AUC = 0.5)

From: Evaluation of the performance of register data as indicators for dairy herds with high lameness prevalence

Indicator

Predefined cut-offs

Optimized cut-offs

Se (%) [95% CI]

Sp (%) [95% CI]

DPRa

PPV [95% CI]

NPV [95% CI]

Se (%) [95% CI]

Sp (%) [95% CI]

DPR

PPV [95% CI]

NPV [95% CI]

AUC(P)

Cow mortalityb

40 [12;74]

67 [47;83]

0.07

29 [8;58]

77 [56;91]

100 [69;100]

67 [47;83]

0.67

39 [20;59]

100 [77;100]

0.76 (0.001)

BMSCCc

40 [12;74]

77 [58;90]

0.13

36 [11;69]

79 [60;92]

70 [35;93]

67 [47;83]

0.47

41 [18;67]

87 [66;97]

0.73 (0.01)

Lean cowsd

40 [12;74]

83 [65;94]

0.23

44 [14;79]

81 [63;93]

90 [56;100]

20 [8;39]

0.10

27 [13;46]

86 [42;100]

0.53 (0.82)

SD age 1st calvinge

40 [12;74]

60 [41;77]

0

25 [7;52]

75 [53;90]

80 [44;98]

23 [10;42]

0.03

26 [12;45]

78 [40;97]

0.50 (0.95)

  1. Sensitivity (Se), specificity (Sp) and predictive values (positive predictive values = PPV, negative predictive values = NPV) with 95% confidence intervals for indicators assessed at optimised and predefined cut-offs for the identification of high lameness prevalences (≥ 16%)
  2. aDPR = Differential positive rates = Se + Sp –1 as a measure of test Se and Sp strength in combination
  3. bAnnual mean cow mortality rate
  4. cBulk milk somatic cell count × 1000 cells/mL
  5. dProportion of lean cows at slaughter out of total number slaughtered per herd (%)
  6. eStandard deviation (SD)