The ICNARC model is predictive of hospital mortality in critically ill patients supported by acute dialysis

To compare prediction power between ICNARC model and RIFLE classification in postoperative patients receiving acute dialysis.

Between January 2002 and December 2008, 529 patients received acute dialysis during their ICU stay were enrolled. Patients' demographic, clinical and laboratory variables were analyzed as predictors of mortality. The RIFLE logistic regression and the ICNARC model on ICU admission were evaluated to predict the patient's hospital mortality.

Hospital mortality for the study group was 29.3%. Between two score systems, the ICNARC model showed better mortality prediction in this patient group by using the area under the receiver operating characteristic curve (ICNARC 0.836, RIFLE 0.702, p < 0.05). Multiple logistic regression analysis indicated that age, surgery category, metastatic carcinoma, ventilator use, and previous history of hypertension were also affecting factors for hospital mortality.

The RIFLE classification and the ICNARC model were both correlated with mortality in critically ill patient with acute dialysis. However, the ICNARC model was a better mortality predictor compared to the RIFLE classification.