Predictors of endoscopic intervention in upper gastrointestinal bleeding patients hospitalized for another illness: a multi-center retrospective study

BACKGROUND: To characterize variables that may predict the need for endoscopic intervention in inpatients admitted for several causes who during the hospitalization developed acute non-variceal upper gastrointestinal bleeding (NVUGIB). METHODS: A retrospective analysis of inpatients who underwent upper gastro-intestinal endoscopy for acute NVUGIB while hospitalized for other causes from 1 January 2016 to 1 December 2017, was performed. In the primary outcome analysis, patients (N.=14) who underwent endoscopic intervention (group A) were compared to those (N.=87) who did not need for endoscopic intervention (group B). Secondary outcome analysis included patients who had significant endoscopic findings compared to those who did not have them. RESULTS: Multivariate regression analysis showed that in the primary outcome analysis, two parameters were significant: the number of packed red blood cells (PRBC) units transfused (odds ratio [OR]: 1.5, P=0.01) and Rockall Score (RS) (OR: 1.4, P=0.06) with receiver operator characteristic (ROC) curve of 0.7844. In the secondary outcome analysis, only the use of proton pump inhibitor drugs at admission was associated with protective effect for the development of significant endoscopic findings (odds ratio [OR]: 0.42, P=0.05) with ROC curve of 0.7342. CONCLUSIONS: In hospitalized patients, in case of de novo NVUGIB, the number of PRBC units transfused and RS are predictive of significant endoscopic findings.