LAUSR

Background and AimsCompared with ulcer bleeding (UB) in non-variceal upper gastrointestinal bleeding (NVUGIB), non-ulcer bleeding (NUB) is often considered to have a low risk of poor outcomes and is treated less intensively without any risk stratification. We conducted this study to assess the predictability of scoring systems for NUB and compare the outcomes of NUB and UB.MethodsA total of 1831 UGIB patients were registered in the database during the period from February 2011 to December 2013. Among them, 1424 patients with NVUGIB were divided into two groups: Group UB (1101 patients with peptic ulcer bleeding) and Group NUB (323 patients with non-peptic ulcer-related bleeding).ResultsThe most common cause of bleeding in Group NUB was Mallory–Weiss tears (51.1%), followed by Dieulafoy lesions (18.9%). A receiver operating characteristic (ROC) analysis revealed that the pre-Rockall score [area under the ROC (AUROC) = 0.798; 95% CI 0.707–0.890] and full Rockall score (AUROC = 0.794; 95% CI 0.693–0.895) were relatively good at predicting overall mortality in NUB. Glasgow–Blatchford score (AUROC = 0.783; 95% CI 0.730–0.836) was the most closely correlated with the need for clinical intervention in NUB. Those who had Glasgow–Blatchford score of 0 did not require any interventions, including blood transfusions. There were no statistical differences in overall mortality (p = 0.387), bleeding-related mortality (p = 0.447), or the incidence of re-bleeding (p = 0.117) between the two groups.ConclusionsScoring systems are useful to predict mortality and the need for clinical intervention in patients with NUB.