Introduction: The present study intends to validate the Rockall Score in patients with upper gastrointestinal bleeding (UGB) in our current medical setting and to find the value that best discriminates between patients with high or low risk of mortality, rebleeding and the need of more than two units of packed red blood cells (PRBC).
Materials and methods: A descriptive prospective study was made from patients who arrived to Cayetano Heredia Hospital's emergency department between February 2007 and January 2008 due to UGB symptoms (hematemesis, coffe ground remit melena or hematoquezia). The Rockall score was used to determine severity of UGB and to stratify patients with higher risk of mortality or rebleeding. All patients were interviewed and any additional information was gathered from medical history records and emergency and hospitalization endoscopic procedure reports. During the study all patients were evaluated for rebleeding, the number of units of PRBCs needed and mortality rate.
Results: 163 patients were included in our study, 107 (65.64%) were male and 56 (34.36%) female, 8 were excluded due to lack of an endoscopic procedure. The remaining 155 patients were studied to evaluate the discriminative ability of the scoring system, and to determine which value best distinguishes high and low severity patients using Receiver Operating Characteristic curve (ROC) and calculated area under the curve. The data analysis showed patients with a Rockall Score e 5 had an increased mortality rate compared to lower score with an area under the curve of 0.807, meaning an accurate relationship between mortality and a score e 5. For rebleeding and the need of two packed red blood cells for transfusion, the area under the curve were 0.65 and 0.64 respectively showing a low predictive value.
Conclusion: The Rockall scoring system is useful to identify patients with high mortality risk, but not to predict rebleeding or the need for blood transfusion in our hospital.