The utility of scoring systems in critically ill cirrhotic patients admitted to a general intensive care unit

Purpose: This study aimed to establish which prognostic scoring tool provides the greatest discriminative ability when assessing critically ill cirrhotic patients in a general intensive care unit (ICU) setting. Methods: This was a 12-month, single-centered prospective cohort study performed in a general, nontransplant ICU. Forty clinical and demographic variables were collected on admission to calculate 8 prospective scoring tools. Patients were followed up to obtain ICU and inhospital mortality. Receiver operating characteristic curve analysis was used to determine the discriminative ability of the scores. Univariate and multivariate analyses were used to identify any independent predictors of mortality in these patients. The incorporation of any significant variables into the scoring tools was assessed. Results: Fifty-nine cirrhotic patients were admitted over the study period, with an ICU mortality of 31%. All scores other than the renal-specific Acute Kidney Injury Network score had similar discriminative abilities, producing area under the curves of between 0.70 and 0.76. None reached the clinically applicable level of 0.8. The Sequential Organ Failure Assessment score was the best performing score. Lactate and ascites were individual predictors of ICU mortality with statistically significant odds ratios of 1.69 and 5.91, respectively. When lactate was incorporated into the Child-Pugh score, its prognostic accuracy increased to a clinically applicable level (area under the curve, 0.86). Conclusions: This investigation suggests that established prognostic scoring systems should be used with caution when applied to the general, nontransplant ICU as compared to specialist centers. Our data suggest that serum arterial lactate may improve the prognostic ability of these scores. (C) 2014 Elsevier Inc. All rights reserved.