Analysis of the diagnostic utility of Glasgow-Blatcford Score, Rockall Score, AIMS65 Score, BUN-to-albumin ratio and BUN-to-creatinine ratio for predicting the need for transfusion in upper gastrointestinal bleeding

Abstract

Objective: This study aimed to compare the predictive accuracy of commonly used risk scores in predicting the need for transfusion among patients with upper gastrointestinal bleeding (UGIB). Material and Methods: This retrospective diagnostic utility study was conducted at a tertiary care academic hospital. The primary outcome was the diagnostic accuracy of the Glasgow-Blatchford Score (GBS), the Rockall Score, the AIMS65 Score, the BUN-to-albumin ratio (BAR), and the BUN-to-creatinine ratio (BCR) in the prediction of transfusion in UGIB. Results: Results from the study showed that 75% of the 104 patients included in the study received blood transfusions, with a median of 3 (IQR 2 - 4) units. Admission hemoglobin and hematocrit values were higher in the non-transfusion group (p<0.001 for both comparisons). The GBS (AUC 0.790 [95% CI 0.699 - 0.864]; p<0.001), AIMS65 (AUC 0.672 [95% CI 0.573 - 0.761]; p=0.001), and BAR (AUC 0.625 [95% CI 0.525 - 0.718, p=0.04) were found to be useful diagnostic indices in predicting transfusion administration, with ideal cut-offs of >0, >10.75, and >0, respectively. But Rockall and BCR were not found to be useful diagnostically. The study suggests that these indices can be used as decision tools for transfusion administration in patients with acute upper gastrointestinal bleeding. Conclusion: The GBS demonstrated superior accuracy compared to AIMS65 and BAR, while Rockall score and BCR were found to be ineffective. The GBS may therefore be useful to clinicians when assessing the potential need for blood transfusions in patients with UGIB.