Description

Hekmat et al reported the Cardiac Surgery Score (CASUS) for evaluating cardiac surgery patients in the intensive care unit (ICU) following surgery. Subsequently a logistic regression model was developed. The authors are from the University of Cologne, Friedrich-Schiller University in Jena, and the Catholic Clinic Koblenz-Montabaur in Germany


 

Patient selection: cardiac surgery patient in the ICU

 

Oriignal parameters of the CASUS::

(1) PaO2 to FIO2 ratio

(2) pressure-adjusted heart rate (PAR)

(3) serum creatinine in mg/dL

(4) serum total bilirubin in mg/dL

(5) lactic acid in mmol/L

(6) platelet count in 10^3 per microliter

(7) neurologic status

(8) intra-aortic balloon pump (IABP)

(9) ventricular assist device (VAD)

(10) renal replacement therapy (continuous venovenous hemofiltration, dialysis)

 

In addition there was addition of a categorical variable for postoperative day (new).

 

Of these, 3 have 0 beta-coefficients in the logistic model:

(1) PaO2 to FIO2 ratio

(2) pressure-adjusted heart rate

(3) plaletlet count

 

Of these, 3 multiplied a continuous variable versus the beta-coefficient:

(1) serum creatinine in mg/dL

(2) lactic acid in mmol/L

(3) serum bilirubin in mg/dL

 

points for serum creatinine =

= (serum creatinine) * 0.3

 

points for lactic acid =

= (lactic acid) * 0.2

 

points for serum bilirubin =

= (serum bilirubin) * 0.2

 

Of these, 4 were categorical variables:

(1) renal replacement therapy

(2) IABP

(3) VAD

(4) neurologic status

 

Categorical Parameters

Finding

Points

neurologic status

normal

0

 

confused in conversation

0.4

 

sedated

0.7

 

diffuse neuropathy

1.4

intra-aortic balloon pump

no

0

 

yes

0.6

ventricular assist device

no

0

 

yes

2.2

renal replacement therapy

no

0

 

yes

0.4

postoperative day

1 or 2

0

 

3

0.8

 

4

1.0

 

5

1.2

 

6

1.4

 

7

1.6

 

8

1.9

 

9

2.1

 

10 or 11

2.2

 

12

2.3

 

>= 13

2.5

 

X =

= SUM(points for all of the parameters) – 5.6

 

probability of ICU mortality =

= 1 / (1 + EXP((-1) * X))

 

Performance:

• Discrimiinatory power (area under ROC curve): 0.87 to 0.96

 


To read more or access our algorithms and calculators, please log in or register.