PIRO-CIC model can predict mortality and futility of care in critically ill cirrhosis patients in the intensive care unit

Background Dynamic assessment of critically ill patients with cirrhosis (CICs) is required for accurate prognostication. Objective Development of a dynamic model for prediction of mortality and decision on futility of care in CICs. Design and setting In a prospective cohort study, we developed the PIRO-CIC model (predisposition, injury, response, organ failure for critically ill cirrhotics)] in a derivative cohort (n = 360) and validated it (n = 240) for patients admitted to the Liver ICU. Patients Decompensated cirrhosis admitted to ICU. The model was developed using Cox-regression analysis, and futility was performed by decision-curve analysis. Results CICs aged 48 +/- 11.5 years, 87% males, majority being alcoholics, were enrolled, of which 73.5% were alive at one month. Factors significant for P component were INR [hazard ratio 1.12, 95% confidence interval 1.07-1.18] and CystatinC [2.25, 1.70-2.97]; for I component were sepsis [4.69, 1.90-11.57], arterial lactate[1.40, 1.02-1.93] and alcohol as etiology [2.78, 1.85-4.18]; for R component-systemic inflammatory response syndrome [1.97, 1.14-3.42] and urine neutrophil-gelatinase-associated lipocalin [ HR 2.37, 1.59-3.53]; for O component-low PaO2/FiO2 ratio and need of mechanical ventilation [7.41, 4.63-11.86]. The PIRO-CIC model predicted one-month mortality with a C-index of 0.83 in the derivation and 0.80 in the validation cohorts. It predicted futility of care better than other prognostic scores. The immediate risk of mortality increased by 39% with each unit increase in PIRO-CIC score. Limitations Not applicable for acute-on-chronic liver failure and patients requiring emergency liver transplant. Conclusions Assessment and stratification of CICs with the dynamic PIRO-CIC model could determine one-month mortality and futility in the first week. Targeted and aggressive management of coagulation, kidneys, sepsis, and severe systemic inflammation may improve outcomes of CICs.