Abstract
(1) Background: Acute variceal bleeding (AVB) represents an important cause of upper gastrointestinal bleeding (UGIB). Several prognostic scores may be useful for assessing mortality and rebleeding risk, with the Glasgow-Blatchford score (GBS) and Rockall score being the most commonly used for non-variceal bleeding. Scores assessing liver failure (MELD and Child) do not reflect bleeding severity. The neutrophil-to-lymphocyte ratio (NLR) increases in UGIB and can predict survival and rebleeding. (2) Methods: We analyzed the predictive role of NLR, GBS, Rockall, AIMS65, Child, and MELD for mortality (48 h, 5-day, in-hospital, and 6-week) and rebleeding in AVB patients admitted to our hospital from 2017 to 2021. ROC analysis was performed, and a multivariate analysis with logistic regression was used to construct a simplified model. (3) Results: A total of 415 patients were admitted. NLR exhibited fair accuracy for 48-h mortality (AUC 0.718, 95% CI 0.597-0.839, p < 0.0001), with limited predictive value for medium-term mortality. The NLR accuracy was better than that of the GBS and Rockall score, similar to that of the AIMS65 and Child scores, but inferior to that of MELD. The value for all scores in predicting rebleeding was poor, with the highest AUC for the NLR. (4) Conclusions: The NLR exhibited reasonable accuracy in predicting short-term mortality in AVB. Our model (including NLR, age, creatinine, bilirubin, albumin, INR, platelet count, HCC, and etiology) demonstrated 80.72% accuracy in predicting 6-week mortality.