. 2020 May 7;67(8):981–991. doi: 10.1007/s12630-020-01686-5

Table 3.

Predicted risk and model performance statistics for external validation of the entire CCIS cohort and for ICU subtype groups

	n	Observed ICU mortality	Predicted risk of death in ICU before intercept recalibration	Predicted risk of death in ICU after intercept recalibration	Calibration in the large before intercept recalibration (95% CI)	Calibration in the large after intercept recalibration (95% CI)	Calibration slope (βrisk)	Discrimination (C-statistic) (95% CI)
All Level 3 units	121,201	13,594	11.2	14.4 (14.7)	8.7 [3.9–18.7]	11.2 (12.5)	6.3 [2.8–14.0]	− 0.343 (− 0.362 to − 0.324)	0.001 (− 0.018 to 0.021)	1.019 (1.001 to 1.037)	0.807 (0.804 to 0.811)
Teaching hospitals – medical/surgical units	28,894	4711	16.3	15.5 (15.2)	10.4 [4.2–20.8]	16.3 (15.7)	10.8 [4.6–22.2]	0.070 (0.036 to 0.105)	− 0.000 (− 0.035 to 0.034)	0.909 (0.878 to 0.941)	0.781 (0.774 to 0.788)
Community hospitals – medical/surgical units with high ventilator patient day rate*	33,653	5398	16.0	16.4 (17.1)	10.0 [3.7–22.3]	16.1 (16.9)	9.6 [3.6–21.6]	− 0.037 (− 0.069 to − 0.004)	− 0.006 (− 0.038 to 0.027)	0.982 (0.953 to 1.011)	0.816 (0.810 to 0.822)
Community hospitals – medical/surgical units with low ventilator patient day rate*	23,597	2100	8.9	11.6 (12.7)	7.5 [3.5–16.3]	9.1 (10.7)	5.5 [2.5–12.5]	− 0.350 (− 0.399 to − 0.302)	− 0.032 (− 0.080 to 0.016)	1.048 (1.003 to 1.093)	0.810 (0.801 to 0.819)
Cardiac /cardiovascular units	18,929	608	3.2	14.6 (12.4)	11.4 [7.3–16.3]	3.0 (3.8)	1.9 [1.3–3.0]	− 1.799 (− 1.883 to − 1.716)	0.064 (− 0.019 to 0.147)	1.062 (0.980 to 1.144)	0.768 (0.747 to 0.789)
Coronary care units	15,159	746	4.9	12.4 (12.8)	7.7 [3.5–17.1]	5.0 (6.6)	2.6 [1.2–6.3]	− 1.148 (− 1.225 to − 1.070)	− 0.011 (− 0.088 to 0.067)	1.232 (1.156 to 1.307)	0.850 (0.836 to 0.863)

Observed ICU mortality

Predicted risk of death in ICU before intercept recalibration

Predicted risk of death in ICU after intercept recalibration

Calibration in the large before intercept recalibration (95% CI)

Calibration in the large after intercept recalibration (95% CI)

Calibration slope (βrisk)

Discrimination (C-statistic)
(95% CI)

(n)

Mean (SD)

Median
[IQR]

Mean (SD)

Median
[IQR]

All Level 3 units

121,201

13,594

11.2

14.4

(14.7)

8.7

[3.9–18.7]

11.2

(12.5)

6.3

[2.8–14.0]

− 0.343

(− 0.362 to − 0.324)

0.001

(− 0.018 to 0.021)

1.019

(1.001 to 1.037)

0.807

(0.804 to 0.811)

Teaching hospitals – medical/surgical units

28,894

4711

16.3

15.5

(15.2)

10.4

[4.2–20.8]

16.3

(15.7)

10.8

[4.6–22.2]

0.070

(0.036 to 0.105)

− 0.000

(− 0.035 to 0.034)

0.909

(0.878 to 0.941)

0.781

(0.774 to 0.788)

Community hospitals – medical/surgical units with high ventilator patient day rate*

33,653

5398

16.0

16.4

(17.1)

10.0

[3.7–22.3]

16.1

(16.9)

9.6

[3.6–21.6]

− 0.037

(− 0.069 to − 0.004)

− 0.006

(− 0.038 to 0.027)

0.982

(0.953 to 1.011)

0.816

(0.810 to 0.822)

Community hospitals – medical/surgical units with low ventilator patient day rate*

23,597

2100

8.9

11.6

(12.7)

7.5

[3.5–16.3]

9.1

(10.7)

5.5

[2.5–12.5]

− 0.350

(− 0.399 to − 0.302)

− 0.032

(− 0.080 to 0.016)

1.048

(1.003 to 1.093)

0.810

(0.801 to 0.819)

Cardiac /cardiovascular units

18,929

608

3.2

14.6

(12.4)

11.4

[7.3–16.3]

3.0

(3.8)

1.9

[1.3–3.0]

− 1.799

(− 1.883 to − 1.716)

0.064

(− 0.019 to 0.147)

1.062

(0.980 to 1.144)

0.768

(0.747 to 0.789)

Coronary care units

15,159

746

4.9

12.4

(12.8)

7.7

[3.5–17.1]

5.0

(6.6)

2.6

[1.2–6.3]

− 1.148

(− 1.225 to − 1.070)

− 0.011

(− 0.088 to 0.067)

1.232

(1.156 to 1.307)

0.850

(0.836 to 0.863)

*Ventilator patient day rate = (ventilator days / patient care days) *100; based on fiscal year 2016–2017; mean rate = 43.61%; β = standardized regression (beta) coefficient; CCIS = Critical Care Information System; CI = confidence interval; ICU = intensive care unit; IQR = interquartile range; SD = standard deviation