Table 3 The performs of prediction accuracy of Pre-model and Post-model.

From: Perioperative NT pro BNP can predict severe postoperative complications in elderly patients undergoing noncardiac surgery

Characteristics

Pre-model

Post-model

Area under the curve

0.899 (95% CI 0.845–0.940)

0.956 (95% CI 0.902–0.985)

Sensitivity

0.87

1.00

Specificity

0.92

0.81

Youden index

0.785

0.814

Cut off value

348.3 (316.5–446.5)

339.5 (95% CI 234.1–429.0)

Positive predictive value

0.33

0.57

Negative predictive value

0.98

0.99

Total accuracy

0.93

0.97

Hosmer–Lemeshow test

 χ2

8.321

1.298

 DF

8

8

 P value

0.403

0.996

  1. χ2, Chi-square; Pre-model, The model based on preoperative NT-Pro BNP; Post-model, The model based on postoperative NT-Pro BNP; CI, Confidence interval; Hosmer–Lemeshow test: it was used to evaluate the calibration degree of the model, when P > 0.05 means calibration degree of the model is good.
Back to article page