Study population (n = 1943) | External validation (n = 14,234) | |||
---|---|---|---|---|
ESC model with eGFRMDRD4* | ESC model with eGFRMDRD4* | |||
No | Yes | No | Yes | |
Models (OR, 95%CI) | ||||
ESC model | 2.2 (1.6–2.7) | 1.8 (1.4–2.4) | 1.9 (1.7–2.1) | 1.9 (1.7–2.2) |
ESC model with eGFRMDRD4* | ‒ | 2.3 (1.5–3.4) | ‒ | 2.1 (1.7–2.6) |
Overall model fit | ||||
Bayes information criteria | 812.3 | 804.4 | 3458.3 | 3386.5 |
Akaike information criteria | 801.2 | 787.6 | 3443.2 | 3363.2 |
Nagelkerke’s R2 | 1.6% | 2.3% | 0.6% | 1.2% |
Discrimination | ||||
Harrell’s c index | 0.631 | 0.676# | 0.617 | 0.667# |
Calibration | ||||
Adjusted χ Hosmer–Lemeshow goodness of fit across deciles of risk | 1.15 | 1.21 | 18.0 | 18.2 |
P Hosmer–Lemeshow | 0.009 | 0.06 | < 0.001 | 0.12 |
Risk reclassification between ESC model and ESC model with eGFR* | ||||
IDI | 1.1%(95% CI 0.5–1.7; p < 0.001) | 0.5% (95% CI 0.3–0.6; p < 0.01) | ||
Continuous NRI | 46.9%(95% CI 27.6–66.2; p < 0.001) | 45.2% (95% CI 35.1–55.3; p < 0.001) | ||
User category NRI | 19.5% (95% CI 6.3–32.6; p = 0.004) | – | ||
% of 30-day mortality correctly reclassified | 18% | – |