Skip to main content
. 2009 Aug 25;13(4):R137. doi: 10.1186/cc8016

Table 3.

Model fit – comparison of the SAPS II, APACHE II and ICNARC models

SAPS II model APACHE II model ICNARC model
Eligible admissions, n (%) 4,973 (64.6) 6,212 (80.7) 7,156 (93.1)
Observed deaths 3,030 3,579 4,237
Expected deaths 2,675.2 3,538.5 3,393.3
SMR (95% CI) 1.13 (1.11 to 1.16) 1.01 (0.99–1.03) 1.25 (1.22 to 1.27)
AUROC (95% CI) 0.74 (0.73 to 0.75) 0.74 (0.73 to 0.76) 0.79 (0.78 to 0.80)
Hosmer–Lemeshow Ca statistic
 χ2(10) 568.0 68.0 900.9
P value < 0.001 < 0.001 < 0.001
Cox's calibration regression
 Intercept (95% CI) 0.44 (0.38 to 0.51) 0.08 (0.02 to 0.14) 0.62 (0.56 to 0.67)
 Slope (95% CI) 0.59 (0.54 to 0.63) 0.80 (0.75 to 0.85) 0.77 (0.73 to 0.81)
 χ2(2) 469.4 54.3 708.95
P value < 0.001 < 0.001 < 0.001

The Hosmer-Lemeshow test divides the data into ten groups and compares the observed mortality in these groups to the predicted mortality given by the model. The C statistic is a chi-squared statistic for testing the hypothesis of perfect calibration. A significant value for the C statistic indicates that calibration is not perfect.

APACHE II: Acute Physiology and Chronic Health Evaluation II; AUROC: area under the receiver operating characteristic curve; CI: confidence interval; ICNARC: Intensive Care National Audit and Research Centre; SAPS II: Simplified Acute Physiology Score II; SMR: standardised mortality ratio. a