ACCURACY OF GLASGOW-BLATCHFORD SCORE AND ROCK ALL IN PREDICTING CLINICAL OUTCOMES IN UPPER GASTROINTESTINAL BLEED

Abstract

Background: Upper gastrointestinal bleeding (UGIB) is a prevalent medical emergency associated with considerable mortality. The Glasgow Blatchford Score (GBS) and Rockall Score are recognised risk stratification instruments; nevertheless, their relative diagnostic accuracy for predicting the in-hospital mortality within specific populations is uncertain. This study sought to assess and evaluate the precision of the GBS and Rockall Score in forecasting in-hospital mortality in patients with UGIB. Methods: A cross-sectional validation study was conducted at the Department of Medicine, Fauji Foundation Hospital, Rawalpindi, from August 2024 to March 2025. A total of 441 consecutive adult patients with UGIB were enrolled. The GBS and Rockall scores were calculated at admission and after endoscopy, respectively. The primary outcome was in-hospital mortality. Diagnostic accuracy parameters (sensitivity, specificity, predictive values, likelihood ratios) were calculated against this clinical outcome which was in-hospital death. Stratification was done on the basis of age, gender, and BMI. Results: The mean age of the study population was 51.58±11.26 years, with a female predominance (61.5%). The actual mortality rate was 8.8%. The GBS demonstrated marginally higher sensitivity (71.8% vs. 64.1%) but similar specificity (51.2% vs. 51.0%) compared to the Rockall Score. Both scores had low positive predictive values (GBS: 12.5%, Rockall: 11.3%) but high negative predictive values (GBS: 94.9%, Rockall: 93.6%). Stratification revealed significant performance variations: scores were more accurate in males and patients ≤55 years but demonstrated catastrophically low specificity in patients of >55 years of age. Conclusion: The GBS and Rockall scores are effective instruments for excluding mortality in UGIB patients, as demonstrated by their elevated NPVs. Nonetheless, their limited overall accuracy and poor positive predictive values contribute to a considerable incidence of false positives, culminating in a substantial overestimation of risk, especially among the elderly. The GBS shown a little superiority over the Rockall Score. Clinicians ought to employ these ratings judiciously for risk classification, particularly in elderly patients, and should contemplate their integration with clinical judgement to prevent superfluous procedures