Skip to main content
. 2021 Jun 3;16(1):84. doi: 10.1007/s11657-021-00957-y

Table 4.

Performance of the predictive model for predicting VFF and comparison with previous models

Cutoff Sensitivity Specificity AUC PPV NPV LR +  LR- Odd ratio
Model 1 (with BMD)
  HHL > 1.5 cm + BMD FN < -1.7 5 83% 52% 0.67 42% 88% 1.73 0.33 5.23
  Age > 65 years + HHL > 1.5 cm 7 79% 57% 0.68 43% 87% 1.82 0.38 4.84
  Age > 65 years + BMD FN < -1.7 6 51% 81% 0.66 52% 80% 2.65 0.61 4.36
  All parameters 9 43% 86% 0.65 56% 78% 3.09 0.66 4.67
  FRAX MOF with BMD 10 36% 85% 0.6 50% 76% 2.41 0.76 3.19
  FRAX HF with BMD 3 51% 73% 0.62 44% 79% 1.90 0.67 2.83
Model 2 (without BMD)
  HHL > 1.5 cm 3 93% 32% 0.63 36% 92% 1.38 0.2 6.68
  Age > 65 years 4 83% 52% 0.66 42% 88% 1.73 0.33 5.23
  All parameters 7 51% 81% 0.66 52% 80% 2.65 0.61 4.36
  FRAX MOF without BMD 10.00 12% 95% 0.53 48% 72% 2.23 0.93 2.4
  FRAX HF without BMD 3.00 30% 85% 0.57 44% 75% 1.94 0.83 2.33
  OSTA -1.00 76% 42% 0.59 35% 81% 1.29 0.58 2.21
-4.00 30% 82% 0.56 41% 74% 1.68 0.85 1.98
  KKOS -1.00 72% 46% 0.58 35% 80% 1.31 0.63 2.09
  ISCD 2019 model with BMD none 60% 68% 0.64 44% 81% 1.89 0.58 3.24

Cutoff value was calculated from sum of individual risk scores in Table 3 and according to original values reported for other models

AUC area under the curve, PPV positive predictive value, NPV negative predictive value, LR + positive likelihood ratio, LR- negative likelihood ratio