### Description

Schaefer et al developed models for predicting survival for a patient with acute renal failure in the intensive care unit (ICU). The authors are from the Free University of Berlin.

Models:

(1) at 0 hours (immediately before dialysis)

(2) at 24 hours after first dialysis

(3) at 48 hours after first dialysis

Parameters:

(1) mechanical ventilation before dialysis

(2) mechanical ventilation at 24 hours after first dialysis

(3) mechanical ventilation at 48 hours after first dialysis

(4) mean blood pressure before dialysis

(5) mean blood pressure at 24 hours after first dialysis

(6) mean blood pressure at 48 hours after first dialysis

(7) presence of liver cirrhosis

(8) blood glucose before dialysis

(9) blood glucose at 48 hours after first dialysis

(10) heart failure before dialysis

(11) heart failure at 48 hours after first dialysis

(12) septic shock before dialysis

(13) heart rate at 24 hours after first dialysis

(14) prothrombin activity (time) at 48 hours after first dialysis

Parameter

Finding

Points

mechanical ventilation

no

0

yes

1

mean blood pressure

>= 70 mm Hg

0

< 70 mm Hg

1

liver cirrhosis

absent

0

present

1

blood glucose

<= 10 mmol/L

0

> 10 mmol/L

1

heart failure

none, NYHA I or II

0

NYHA III or IV

1

septic shock

absent

0

present (hypotension and fever)

heart rate

<= 119 beats per minute

0

> 119 beats per minute

1

prothrombin activity (time)

>= 50%

0

< 50%

1

where:

• The patient with septic shock should have a low mean arterial pressure.

• Prothrombin is measured based on a plot of percent prothrombin vs prothrombin time. The less prothrombin activity the longer the prothrombin time.

• Mean blood pressure = ((systolic systemic arterial pressure) + (2 * (diastolic systemic arterial pressure))) / 3

X for predialysis score =

= (0.58711 * (points for mechanical ventilation)) + (0.51897 * (points for mean pressure)) + (0.48142 * (points for liver cirrhosis)) + (0.44313 * (points for blood glucose)) + (0.32542 * (points for heart failure)) - (0.34577 * (points for septic shock)) + 1.81243

where:

• The constant is given in Table 3 as 1.81243. In the implementation I will also include the calculation using (-1.81243) just in case there was a typo.

Y for 24 hours score =

= (0.5522 * (points for mechanical ventilation)) + (0.52621 * (points for mean pressure)) + (0.51229 * (points for heart rate)) - 1.78523

Z for 48 hour score =

= (4.45892 * (points for mean pressure)) + (2.25054 * (points for blood glucose)) + (2.56843 * (points for prothrombin time)) + (2.24830 * (points for heart failure)) + (2.22227 * (points for mechanical ventilation)) - 2.38171

where:

• The functional coefficients for the 48 hour score are 4-8 times greater than the previous scores. The presence of 3 or more factors would give a 100% mortality rate, which seems too high.

• There is no indication on Pubmed that corrections were made to the formulas.

mortality rate at a given time =

= 1 / (1 + EXP((-1) * (X or Y or Z)))

Performance:

• The discriminant functions were not considered accurate enough to make clinical decisions about starting or continuing dialysis.

• They still may be useful in gauging prognosis.