Glasgow Blatchford score predicted intervention or death better than 4 other prediction models in upper GI bleeding

Commentary authors writing in ACP Journal Club said the findings add to a growing body of literature that supports the use of the Glasgow Blatchford score for selecting low-risk patients who can safely be discharged from the ED (with a minimal false-negative rate).


The Glasgow Blatchford score (GBS) performed best at predicting hospital-based interventions or death in a recent study of patients with upper BI bleeding. In 3,012 patients presenting with GI bleeding at six hospitals, the GBS most accurately predicted the composite outcome of transfusion, endoscopic treatment, interventional radiology, surgery, or 30-day mortality compared with the admission Rockall, AIMS65 (Albumin level <30 g/L, International normalized ratio >1.5, altered Mental status, Systolic blood pressure ≤90 mm Hg, and age >65 years), full Rockall, and PNED (progretto nazionale emorragia digestive). The authors found that a GBS ≤1 was the optimum threshold to predict survival without intervention, that is, to identify patients who could be directed to outpatient care.

The study was published online Jan. 4 by The BMJ and was summarized in the January ACP Gastroenterology Monthly. The following commentary by Nicholas Talley, MD, PhD, FRACP, and Michael Potter, MBBS (Hons), was published in the ACP Journal Club section of the April 18 Annals of Internal Medicine.

Scoring systems are increasingly useful in clinical practice as risk-stratification tools, and with smart phone technology they are readily available for use at the bedside. The large multicenter study by Stanley and colleagues supports use of the GBS, a simple scoring system that employs variables readily available in the emergency department, as the most useful of the many risk-stratification tools for patients presenting with upper GI bleeding. The findings add to a growing body of literature that supports the use of the GBS for selecting low-risk patients who can safely be discharged from the emergency department (with a minimal false-negative rate); until recently, early endoscopic triage has been the standard of care.

Previous studies addressing the accuracy of these scoring systems for predicting the need for hospital-based intervention have been limited by small sample size, retrospective or single-center design, failure to consider outcomes that would allow hospital discharge, and lack of comparison of scoring systems or head-to-head comparison of a smaller number of systems. The study by Stanley and colleagues is the first to use a large international cohort and prospectively and concurrently compare the use of 5 separate scoring systems to predict outcomes related to hospital admission.

A “low-risk” GBS has previously been difficult to define. If it is too low (GBS = 0), it identifies few patients and may not substantially influence admission rates. If too high (GBS cutpoint 2), it may misclassify an unacceptable number of patients who require therapy. In this study, a GBS cutpoint of ≤ 1 classified almost 1 in 5 patients presenting with upper GI bleeding as low-risk and, when applied, should be cost-saving.