Background/Aims: Acute-on-chronic liver failure (ACLF) is associated with high short-term mortality, and early prediction is critical to reduce the deaths of ACLF patients. To date, however, the prognostic accuracy of current models for ACLF is unsatisfactory, particularly, in patients with hepatitis B virus (HBV) infection. This study aims to develop novel prognostic models based on the dynamic changes in variables to predict the short-term mortality of HBV-associated ACLF (HBV-ACLF). Methods: A retrospective cohort study was conducted, with the population comprised in whom ACLF was confirmed.319 patients were enrolled and their clinical data were collected on Days 1 and 7 following hospital admission. Univariate and multivariate analyses were performed to identify risk factors for 28 and 90-day mortality. The dynamic alterations in the risk factors were further analyzed, and Days 1 and 7 prognostic models were constructed. Receiver operating characteristic (ROC) analysis were used to identify and compared the predictors of prognosis among our model. Results: Univariate and multivariate analyses revealed significant risk factors at Days 1 and 7, which when combined with the clinically important parameters, were used to establish the Days 1 and 7 prognostic models. For 28-day mortality, the predictive accuracy of the Day 1 prognostic model was significantly higher than that of the albumin-bilirubin (ALBI) model. For 90-day mortality, the predictive accuracy of the Days 1 and 7 prognostic models was significantly higher than that of the Model of End-Stage Liver Disease (MELD), MELD-sodium (MELD-Na), and ALBI prognostic models. Conclusions: The prognostic models established in this study were superior to the existing prognostic scoring systems to accurately predict short-term mortality, and therefore, could be potential novel prognostic tools for HBV-ACLF.
Elsevier, Heliyon, Volume 10, 15 April 2024
