Short-term prognosis in critically ill patients with liver cirrhosis: an evaluation of a new scoring system

Objective The mortality of patients with liver cirrhosis admitted to an int ensive care unit (ICU) has been found to be high. This study was performed to assess the physiological and laboratory parameters which are able to ide ntify on ICU admission the cirrhotic patients who are most likely to die. Design Prospective clinical trial. Methods Two groups of patients were analysed. Group A consisted of 196 cons ecutive cirrhotic patients admitted to our medical ICU for various reasons. For the detection of independent outcome predictors, we used a multiple lo gistic regression model. Based on these variables, the 'intensive care cirr hosis outcome (ICCO) score' was calculated. The ability to discriminate bet ween survivors and non-survivors was determined by receiver operating chara cteristic curves, and the area under the curve was calculated. Group B cons isted of 70 consecutive cirrhotic patients for prospective validation of th e ICCO score. Results Applying multiple logistic regression analysis, bilirubin, choleste rol, creatinine clearance and lactate were found to be independently associ ated with the hospital mortality. The ICCO score was 0.3707 + (0.0773 x bil irubin (mg/dl)) - (0.00849 x cholesterol (mg/dl))-(0.0155 x creatinine clea rance (ml/min)) + (0.1351 x lactate (mmol/l)), giving an area under a recei ver operating characteristic curve of 0.9. Increasing score values were ass ociated with an increase in mortality. All patients with an ICCO score > +2 .6 died. Conclusions Application of the ICCO score is rapid and available at the pat ient's bedside, and its application is simple and reproducible. In cirrhoti c patients, the ICCO score has a high ability to discriminate between survi vors and non-survivors. The ICCO score may facilitate estimation on ICU adm ission of the prognosis of critically ill cirrhotic patients. (C) 2000 Lipp incott Williams & Wilkins.