RESUMO
BACKGROUND: Upper gastrointestinal (UGI) haemorrhage is a frequent cause of hospital admission. Scoring systems have been devised to identify those at risk of adverse outcomes. We evaluated the Glasgow Blatchford score's (GBS) ability to identify the need for clinical and endoscopic intervention in patients with UGI haemorrhage. METHODS: A retrospective observational study was performed in all patients who attended the A&E department with UGI haemorrhage during a 12-month period. Patients were separated into low and high risk categories. High risk encompassed patients who required blood transfusions, operative or endoscopic interventions, management on high dependency or intensive care units, and those who re-bled, represented with further bleeding, or who died. RESULTS: A total of 174 patients were seen with UGI bleeding. Eight of them self-discharged and were excluded. Of the remaining 166, 94 had a 'low risk' bleed, and 72 'high risk'. The GBS was significantly higher in the high risk (median = 10) than in the low risk group (median 1, p < 0.001). To assess the validity of the GBS at separating low and high risk groups, receiver-operator characteristic (ROC) curves were plotted. The GBS had an area under ROC curve of 0.96 (95% CI 0.95-1.00). When a cut-off value of > or = 3 was used, sensitivity and specificity of GBS for identifying high risk bleeds was 100% and 68%. Thus at a cut-off value of < or = 2 the GBS is useful for distinguishing those patients with a low risk UGI bleed. CONCLUSIONS: The GBS accurately identifies low risk patients who could be managed safely as outpatients.