Abstract
Objective This retrospective study assessed the longitudinal prognostic accuracy of the Model for End-Stage Liver Disease (MELD), MELD-Na, MELD 3.0, and Child-Pugh scores in predicting mortality and rebleeding risk in patients with liver cirrhosis (LC) following successful endoscopic hemostasis for acute variceal hemorrhaging (AVH). Methods Time-dependent receiver operating characteristics and survival analyses were performed to predict mortality. The risk of rebleeding was analyzed using a competing risk model. Patients A total of 168 patients with LC who underwent successful endoscopic treatment, including ligation and sclerotherapy for AVH of the esophagus or stomach, were included. Results The MELD 3.0, which demonstrated the highest accuracy for predicting mortality, had an average area under the curve of 0.789, remaining above 0.8 for up to 18 months. The survival rate was significantly worse in the high-MELD group (3.0) than in the low-MELD group (3.0), with a hazard ratio (HR) of 3.23 and a shorter median survival time of 113 days versus 1,750 days. Advanced hepatocellular carcinoma and the need for red blood cell transfusion were also independent risk factors for mortality. A high MELD-Na level was the only factor associated with an increased risk of rebleeding (HR, 2.07), and the cumulative rebleeding rate was significantly higher in the high-MELD-Na group than in the low-MELD-Na group. Conclusion MELD 3.0 and MELD-Na serve as reliable non-invasive tools for predicting the long-term mortality and rebleeding risk after AVH. These scores can aid in clinical decision making, enabling the early identification of high-risk patients for targeted interventions and closer monitoring.