Table 3.

Performance indicators for prediction models of mortality after hip fracture surgery

	Discrimination	Calibration	Clinical usefulness
	AUC	H–L	PPV (%)	NPV (%)
30-day mortality
NHFS	0.80 (0.72–0.87)	p = 0.164	14.9	100.0
Holt et al.	0.86 (0.78–0.94)	p = 0.293	17.2	99.1
HEMA	0.74 (0.63–0.85)	p = 0.025	17.9	96.8
1-year mortality
NHFS	0.76 (0.70–0.82)	p = 0.322	40.5	95.0
Holt et al.	0.79 (0.73–0.84)	p = 0.709	40.9	95.4
HEMA	0.73 (0.66–0.80)	p = 0.331	48.7	89.1
8-year mortality
NHFS	0.90 (0.86–0.93)	p = 0.063	98.6	72.4
Holt et al.	0.89 (0.85–0.93)	p = 0.413	94.5	74.7
HEMA	0.79 (0.75–0.84)	p < 0.001	97.4	40.5

Values in parentheses are 95% confidence intervals. AUC, area under the curve. H–L, Hosmer–Lemeshow goodness-of-fit test. Bold typeface indicates statistically significant results (p < 0.05). PPV, positive predictive value is the probability of death for a patient in the high-risk group. NPV, negative predictive value is the probability of survival for a patient in the low-risk group. NHFS, Nottingham Hip Fracture Score. HEMA, Hip fracture Estimator of Mortality Amsterdam