Hepatol Int. 2022 Oct 13. doi: 10.1007/s12072-022-10426-4. Online ahead of print.
ABSTRACT
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.
PMID:36227516 | DOI:10.1007/s12072-022-10426-4
