Objectives The admission of patients with liver cirrhosis to the intensive care unit (ICU) due to infections is a frequent occurrence, often leading to complications such as hepatic encephalopathy, renal failure and circulatory collapse, significantly elevating mortality risks. Accurate and timely diagnosis and intervention are critical for improving therapeutic outcomes. In this context, medical scoring systems in ICUs are essential for precise diagnosis, severity assessment and appropriate therapeutic strategies. There are no specific models for the prediction of mortality in ICU patients with liver cirrhosis-associated infections. This study aims to develop an improved prognostic scoring system for predicting in-hospital mortality among liver cirrhosis patients with infections in the ICU. This scoring system is designed to enhance the predictive accuracy of in-hospital mortality complementing existing sepsis and liver-specific prognostic models.

Methods A retrospective analysis was conducted in 620 patients with liver cirrhosis, treated for infections in the ICU of a German university hospital during 2017–19. Advanced statistical techniques were employed to develop and validate the LIVERAID (LIVER And Infectious Diseases)-ICU score, a novel scoring system specifically tailored for liver cirrhosis patients in the ICU with infections. The development of the multivariable logistic regression model involved selecting variables with the highest prognostic efficacy, and its predictive performance was assessed using calibration plots and the concordance statistic (c-index) to evaluate both calibration and discrimination.

Results The LIVERAID-ICU score integrates Child-Pugh class, serum urea levels and respiratory metrics. It is designed for bedside calculation using basic clinical and laboratory data, with no need for additional tools. In the validation cohort, the LIVERAID-ICU score exhibited enhanced sensitivity and specificity (AUC=0.83) in forecasting in-hospital mortality of patients with liver cirrhosis-associated infections when compared with established scores like Sequential Organ Failure Assessment (SOFA) (p=0.045), Model for End-Stage Liver Disease (MELD) (p=0.097), Child (p<0.001) and CLIF consortium acute-on-chronic liver failure (CLIF-C ACLF) (p<0.001).

Conclusion The newly developed LIVERAID-ICU score represents a robust, streamlined and easy tool for predicting in-hospital mortality in liver cirrhosis patients with infections, surpassing the predictive capabilities of established liver or sepsis scores like SOFA, MELD, Child and CLIF-C ACLF. The reliance of the LIVERAID-ICU score on fundamental clinical and laboratory data facilitates its global application in ICUs, enabling immediate application at the bedside for patients with liver cirrhosis during episodes of suspected or confirmed infections.