Utilisation of the Rockall Score in the Development of a Clinical Pathway for the Management of Non-Variceal Upper Gastrointestinal Bleeding

Abstract

Background: The Rockall scoring system (based upon co-morbidities, haemodynamic stability and endoscopic appearances) has been validated to predict outcome from non-variceal upper GI bleeding. We believed that the post-endoscopy Rockall score could be used to determine a length of stay recommended by a clinical pathway for non-variceal upper GI bleeding. Methods: Patients presenting with a upper GI bleeding were admitted to high dependency (HDU) or the gastroenterology ward depending upon their pre-endoscopy Rockall score. After endoscopy they were assigned a post-endoscopy Rockall score. Patients with a post-endoscopy Rockall score of greater than 7 (or who had required endoscopic therapy) were managed initially in HDU, with an expectation of discharge to the ward in 36 hours, those less than 7 were admitted to the ward with an aim for discharge within 48hrs. Endoscopic high risk lesions were treated with adrenalin injection +/- argon plasma coagulation followed by PPI infusion. Non-parametric, two-tailed Spearman coefficient calculation was used to determine the correlation between Rockall score and length of stay (total and HDU) and transfusion rates. Results: 165 patients (mean age 65.1, r:15-94 years) were entered over 10 months. Total length of stay, HDU stay and transfusion rates correlated with post-Rockall score (r=0.313, p<0.01:r=0.186, p<0.02:r=0.308, p<0.01). Patients with a post endoscopy Rockall score between 1-2 had a mean total length of stay of 2.8 days (and 0.92 units transfused), compared with 4.2 days (1.8 units) and 4.3 days (2.8 units) respectively for those with scores between 3-4 and greater than 5. Significant rebleeds (defined by a 20g/L fall in Hb) occured in 14 of all patients (8.4%), with 10 of these in the 45 (22%) patients with pepic ulcer disease. Three of the bleeding peptic ulcers (6%) proceeded to surgery, with one death (2.2%) due to an acute MI. There were 3 other deaths (2 due to presumed acute myocardial infarction (AMI) prior to endoscopyand 1 from sepsis) leading to an overall mortality of 2.4%. Discussion:We have demonstrated that the post endoscopy Rockall score predicts total and HDU legth of stay, as well as transfusion requirements.This enables us to predict expected length of stay in the patients clinical pathway immediately following their endoscopy for non-variceal upper GI bleeding. This will enable the easier detection of variance from the clinical pathway with respect to length of stay.