Risk Stratification With Glasgow-Blatchford Bleeding Score for Hospitalized Patients With Upper Gastrointestinal Bleeding Can Avoid the Needs for Urgent Endo-Therapy Nam Q. Nguyen*, Robert V. Bryant, Paul Kuo, Kate D. Williamson, Mark Schoeman, Richard H. Holloway Gastroenterology and Hepatology, Royal Adelaide Hospital, Adelaide, SA, Australia Glasgow-Blatchford bleeding score (GBS) effectively identifies patients presenting to hospital with upper gastrointestinal haemorrhage (UGIH) who can be managed safely as outpatients. Data regarding the utility of GBS in identifying the need for endoscopic therapy and related interventions in patients admitted to hospital for UGIH are limited. AIM: To evaluate and compare the performance of GBS, pre-endoscopy (Pre-ERS) and post-endoscopy Rockall score (Post-ERS) in predicting the need for endoscopic therapy and further interventions in patients admitted to hospital for UGIH. Methods: GBS and Rockall scores for consecutive patients with acute UGIH who were admitted to the Royal Adelaide Hospital for endoscopy over 18 months were prospectively examined. With receiveroperating characteristic (ROC) curves, the performance of these scores to predict the need for endoscopic and related interventions were compared. All patients received high dose acid suppression therapy. Results: 455/572 patients (295 M; 65.4 0.8yrs) who underwent endoscopy for UGIH, 188 (41.3%) required endoscopic therapy and 19 (4.1%) had surgery for haemostasis. Patients who required endoscopic or surgical interventions had higher GBS, Pre-ERS and PostERS (P 0.001). On ROC analyses, GBS and Post-ERS were superior to Pre-ERS in predicting the need for endo-therapy (AUC: 0.71 vs. 0.78 vs. 0.65, respectively) and the need for repeat endoscopy due to re-bleed or for further endo-therapy (AUC: 0.64 vs. 0.63 vs. 0.56). GBS was also superior to both Preand PostERS in predicting the need for transfusion (AUC: 0.83 vs. 0.72 vs. 0.70) and surgery (AUC: 0.750 vs. 0.67 vs. 0.54), with no patients with GBS 7 required surgery. None of the 32 patients with GBS 3 required endo-therapy, blood transfusion or surgery. Conclusions: GBS is superior to Rockall score in predicting the need for endotherapy, repeated endoscopy, transfusion and surgery in hospitalized patients with acute UGIH, and should be the preferred risk scoring system. A GBS 3 avoids the need for urgent endoscopy and such patients may be managed as an outpatient.