Evaluating The Glasgow Blatchford Score for Upper Gastrointestinal Bleeding Risk Stratification in A Community Hospital: A Retrospective Study.

Abstract

Introduction: Upper gastrointestinal bleeding (UGIB) is the most common emergency in gastroenterology. The Glasgow Blatchford Score (GBS) is a validated tool used for risk stratification. The cutoff values for GBS to predict the need for clinical intervention, endoscopic treatment, and mortality, are not consistent. To determine the relationship between mean GBS score and the need for hemostatic intervention, and blood transfusion, and to evaluate quality of care and proper allocation of resources at our midwestern community hospital.

Methods: In this cross-sectional study, we retrospectively extracted records for patients ≥18 years who were admitted for UGIB and underwent esophagogastroduodenoscopy between July 2018 and July 2020. GBS was calculated for each observation. Multivariate analysis and a logistic regression model were performed to predict the GBS score, and the odds ratio, associated with the need for hemostatic intervention and blood transfusion while controlling for confounding factors.

Results: The GBS sample mean score was 11.17. Those who required hemostatic intervention and blood transfusion scored significantly higher GBS (13.18 versus 10.79) and (13.57 versus 9.21), respectively. A GBS of >10 was associated with higher odds at 21.84 (CI: 10.324,46.185, P<0.001) and 5.085 (CI: 1.864, 13.872, P=0.001) for receiving blood transfusion and hemostatic intervention, respectively. A cutoff of 10 was 22.41% sensitive and 95.41% specific for requiring hemostatic interventions and 66.67% sensitive and 89.91% specific for receiving blood transfusion.

Conclusion: There is a clinical role to using the GBS even at a score higher than 2 to further stratify the severity of UGIB and determine the need for intervention. The sensitivity of a score of 10 on the GBS in this dataset was low. A cutoff with higher sensitivity is needed to stratify a life-threatening condition such as UGIB.