Skip to main content

Table 3 Measures of discrimination, calibration and model fit for SHARF T0, Mehta and UK APACHE II

From: Case mix, outcome and activity for patients with severe acute kidney injury during the first 24 hours after admission to an adult, general critical care unit: application of predictive models from a secondary analysis of the ICNARC Case Mix Programme Database

  SHARF T0 SHARF II0 Mehta UK APACHE II
Discrimination (AUC [95% CI]) 0.632 (0.624–0.640) 0.668 (0.660–0.676) 0.693 (0.686–0.701) 0.74 (0.733–0.748)
Calibration (mortality ratio [95% CI]) 0.759 (0.747–0.771) 0.950 (0.936–0.965) 1.223 (1.205–1.242) 1.147 (1.129–1.165)
Hosmer-Lemeshow (C-statistic [P value]) 20,894 (< 0.001) 2,575 (< 0.001) 2,170 (< 0.001) 575 (< 0.001)
Cox's regression calibration:     
   Slope (95% CI) 0.247 (0.229–0.266) 0.415 (0.390–0.439) 0.444 (0.417–0.470) 0.753 (0.720–0.787)
   Intercept (95% CI) -0.195 (-0.243 to -0.153) 0.005 (-0.031 to -0.040) 0.314 (0.278–0.350) 0.306 (0.269–0.343)
χ2 (1) (P value)a 11,598 (< 0.001) 2,303 (< 0.001) 2,602 (< 0.001) 596 (< 0.001)
Fit (Shapiro's Q [R statistic]) 0.392 0.491 0.487 0.539
  1. aChi-squared statistic on one degree of freedom (and associated P value) from a likelihood ratio test of slope = 1, intercept = 0 in Cox's regression calibration model. APACHE, Acute Physiology and Chronic Health Evaluation; AUC, area under the receiver operating characteristic curve; CI, confidence interval; SHARF, Stuivenberg Hospital Acute Renal Failure.