Risk Stratification for Upper Gastrointestinal Bleeding
Use the Glasgow Blatchford Score (GBS) as your primary risk stratification tool for patients presenting with acute upper GI bleeding, and discharge patients with a score ≤1 without hospitalization or urgent endoscopy. 1, 2
Why Glasgow Blatchford Score is the Gold Standard
The GBS has superior sensitivity (99%) for identifying high-risk patients, meaning it misclassifies ≤1% of high-risk patients as low risk 1. This exceptional performance makes it the safest tool for discharge decisions. The score incorporates:
- Blood urea nitrogen level
- Hemoglobin concentration
- Systolic blood pressure
- Heart rate
- Presence of melena
- Presence of syncope
- Evidence of hepatic disease
- Evidence of cardiac failure 3
A GBS of 0-1 identifies patients at very low risk who can be safely discharged without hospitalization or inpatient endoscopy, with outpatient endoscopy arranged as needed 1, 2. Studies demonstrate that extending the threshold to GBS ≤2 could safely double the number of eligible patients for early discharge (from 15.2% to 32.5%) 4.
The Rockall Score: Post-Endoscopy Risk Assessment
After endoscopy, use the complete Rockall Score for comprehensive risk stratification and mortality prediction. 1 The Rockall Score requires endoscopic findings and includes:
- Age
- Shock parameters (heart rate >100 bpm and systolic BP <100 mmHg)
- Comorbidities (renal failure, liver failure, ischemic heart disease, heart failure, disseminated malignancy)
- Endoscopic diagnosis
- Stigmata of recent hemorrhage 5, 6
A complete Rockall score <3 indicates excellent prognosis with very low rebleeding and mortality risk 1. The Rockall Score has good sensitivity (93-96%) but may misclassify 4-7% of high-risk patients as low risk, making it less suitable than GBS for initial discharge decisions 1. However, it has better discriminative ability for predicting mortality than rebleeding 1.
Do NOT Use AIMS65 for Discharge Decisions
The AIMS65 score should NOT be used for discharge decisions because it was designed to identify high-risk patients for death rather than low-risk patients for safe discharge 1. Even at low cutoff values, AIMS65 misclassifies approximately 20% of high-risk patients as low risk, with a sensitivity of only 78-82% 1. While one study suggested AIMS65 was superior to GBS for predicting inpatient mortality 7, international consensus groups prioritize sensitivity for identifying patients who can be safely discharged—where GBS excels 1.
Clinical Implementation Algorithm
Step 1: Calculate GBS at Presentation
- GBS ≤1: Patient can be safely discharged with outpatient endoscopy follow-up 1, 2
- GBS >1: Admit for inpatient management and endoscopy within 24 hours 6, 2
Step 2: Post-Endoscopy Risk Stratification
After endoscopy, calculate the complete Rockall Score:
- **Rockall <3**: Low risk—consider early discharge after 24 hours if hemodynamically stable, hemoglobin >100 g/L, and low-risk endoscopic findings (clean-based ulcer or flat pigmented spot) 1, 6
- Rockall ≥3: Higher risk—continue inpatient monitoring for at least 72 hours after endoscopic therapy 6
Step 3: Contextual Factors for Discharge
Even with low scores, consider these factors before discharge:
- Access to hospital or ambulance services
- Access to out-of-hours endoscopy
- Patient preferences
- Adequate social support and accessibility to hospital 1
Common Pitfalls to Avoid
Do not use clinical judgment alone without a standardized scoring system, as this cannot be standardized and may lead to inconsistent risk assessment 1. The GBS provides objective, reproducible risk stratification that outperforms subjective clinical assessment.
Do not confuse pre-endoscopy and post-endoscopy risk scores. The GBS is calculated before endoscopy using only clinical and laboratory parameters, while the complete Rockall Score requires endoscopic findings 5, 1. Using the pre-endoscopy Rockall Score (without endoscopic findings) is inferior to GBS for identifying low-risk patients 1, 4.
Implementation of GBS-based early discharge reduces hospital admissions and healthcare costs without increasing adverse outcomes 1. One study demonstrated that patients with GBS=0 had zero incidence of the composite endpoint (blood transfusion, endoscopic therapy, interventional radiology, surgery, or 30-day mortality) 4.