Compared with post-endoscopy RS and GBS, pre-endoscopy RS was less effective in predicting the need for endoscopic therapy (area under the curve, 0.76 and 0.76 vs. 0.66, respectively). Compared with the full RS and GBS, pre-endoscopy RS was less effective in predicting rebleeding (AUC, 0.64 and 0.71 vs. 0.57, respectively). Compared with both pre-endoscopy RS and post-endoscopy RS, the GBS more accurately predicted the need for both blood transfusion (AUC, 0.70 and 0.68 vs. 0.81, respectively) and surgery (AUC, 0.64 and 0.51 vs. 0.71, respectively).

COMMENT

The authors conclude that the Glasgow-Blatchford score is as accurate as the full Rockall score in predicting rebleeding and the need for endoscopic therapy and superior to the RS in predicting the need for transfusions or surgery in patients hospitalized for upper gastrointestinal bleeding. The results of this study may be affected by the inclusion of patients with variceal bleeding, the unblinded decisions of physicians on the need for endoscopy and transfusions, and the absence of posthospitalization follow-up. Also, the authors used a higher GBS cutoff value for interventions than that used in previous studies (3 vs. 1). Nonetheless, these results support those of multiple studies in which the efficacy of both the GBS and RS in risk stratification of patients with upper gastrointestinal bleeding, as well as that of a simplified system called AIMS65, have been demonstrated (NEJM JW Gastroenterol Feb 22 2013 and NEJM JW Gastroenterol Dec 9 2011). Overall, I believe that the importance of using any risk stratification tool in this setting is still of primary concern compared with the question of which scoring system to use.