Skip to main content
. 2022 Jan 13;37(15):3877–3884. doi: 10.1007/s11606-021-07377-1

Table 2.

Performance of the SMR Models in Derivation and Validation Cohorts

Model and Cohort N Mortality, N (%) C stat H-L GOF chi-square Brier’s score
Acute Care SMR-30 derivation 769710 33180 (4.31) 0.871 187.7 0.035
Acute Care SMR-30 validation, full year 373791 16036 (4.29) 0.870 211.6 0.035
Acute Care SMR-30 validation, Q1 92338 4038 (4.37) 0.871 39.1 0.035
Acute Care SMR-30 validation, Q2 94640 4258 (4.50) 0.869 49.1 0.037
Acute Care SMR-30 validation, Q3 95269 4023 (4.22) 0.867 66.3 0.034
Acute Care SMR-30 validation, Q4 91544 3717 (4.06) 0.873 88.4 0.033
ICU SMR-30 derivation 151426 13635 (9.00) 0.871 155.3 0.063
ICU SMR-30 validation 72160 6638 (9.20) 0.864 110.6 0.066
ICU SMR-30 validation, ICU level 1/2 62806 5746 (9.15) 0.866 103.0 0.065
ICU SMR-30 validation, ICU level 3/4 9354 892 (9.54) 0.848 18.0 0.071
ICU SMR-30 validation, Q1 17981 1652 (9.19) 0.864 29.3 0.066
ICU SMR-30 validation, Q2 18704 1787 (9.55) 0.863 32.1 0.069
ICU SMR-30 validation, Q3 18276 1681 (9.20) 0.865 36.3 0.066
ICU SMR-30 validation, Q4 17199 1518 (8.83) 0.862 31.8 0.064
Acute Care SMR derivation 853194 15429 (1.81) 0.916 247.6 0.015
Acute Care SMR validation 413329 7173 (1.74) 0.914 218.6 0.015
Acute Care SMR validation, Q1 102049 1802 (1.77) 0.918 57.8 0.015
Acute Care SMR validation, Q2 104446 1960 (1.88) 0.911 35.4 0.016
Acute Care SMR validation, Q3 105513 1759 (1.67) 0.912 75.0 0.014
Acute Care SMR validation, Q4 101321 1652 (1.63) 0.914 77.6 0.014
ICU SMR derivation 152914 9641 (6.30) 0.895 229.7 0.046
ICU SMR validation 72752 4555 (6.26) 0.887 153.0 0.047
ICU SMR validation, ICU level 1/2 63873 4122 (6.45) 0.887 128.0 0.048
ICU SMR validation, ICU level 3/4 8879 433 (4.88) 0.887 41.7 0.039
ICU SMR validation, Q1 18125 1138 (6.28) 0.887 23.8 0.047
ICU SMR validation, Q2 18881 1259 (6.67) 0.886 49.4 0.050
ICU SMR validation, Q3 18451 1145 (6.21) 0.888 42.0 0.046
ICU SMR validation, Q4 17295 1013 (5.86) 0.887 61.3 0.045

ICU, intensive care unit; SMR, standardized mortality ratio; Q, quarter; C stats, C statistic; H-L GOF chi-square, Hosmer Lemeshow goodness-of-fit C statistic chi-square value where a lower value is better and the size of the value is linearly related to the size of the cohort; all HL GOF chi-square values were significantly different (p<0.05). Level 1/2, ICUs who can provide most subspecialty medical and surgical Care; level 3/4: ICUs in smaller hospitals which lack some or many medical and surgical subspecialty care