 ### Description

Van Bommel et al evaluated patients with acute renal failure admitted to the surgical intensive care unit (ICU). They used the ratios of 2 APACHE II scores to predict the risk of mortality in these patients. The authors are from University Hospital Dijkzigt and Erasmus University in The Netherlands.

Patient selection:

(1) admitted to the surgical ICU

(2) has a maximum APACHE II score during the first 24 hours that exceeds a certain threshold. A low APACHE II score on admission and a minor deterioration could be associated with a high ratio and so give a false positive mortality prediction. Based on the data shown I would say this should be set at >= 12.

(3) requires dialysis

(4) dialysis is not started within the first 24 hours of admission

Parameters:

(1) maximum APACHE II score during the first 24 hours of admission to the ICU

(2) maximum APACHE II score on the day that dialysis was started

ratio of APACHE II scores =

= (APACHE II on day of dialysis) / (APACHE II during first 24 hours)

Interpretation:

• The higher the ratio the greater the risk of mortality.

• The probability of mortality based on the ratio is shown in Figure 3 on page 739.

Model 1: Line approximation in Minitab

 Ratio Percent with Mortality < 0.9 < 2.5% 0.9 to 1.05 (2420 * ((ratio)^2)) - (4371 * (ratio)) + 1976 1.05 to 1.2 (-1914 * ((ratio)^2)) + (4581 * (ratio)) - 2643 > 1.2 > 98%

Model 2: Estimate of logistic regression equation (since a single variable involved)

X =

= (25.42 * (ratio)) - 26.52

probability of mortality =

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