Scoring and Classification Systems for Upper Gastrointestinal Endoscopy (UGIE)
Primary Classification Systems
The Forrest classification is the primary endoscopic grading system used to stratify bleeding risk in peptic ulcers visualized during upper GI endoscopy, directly guiding decisions about endoscopic hemostasis. 1
Forrest Classification Categories
The Forrest classification stratifies ulcers based on endoscopic appearance and correlates directly with rebleeding risk 1, 2:
High-Risk Stigmata (Require Endoscopic Therapy):
- Forrest Ia (Active spurting bleeding): 55% rebleeding risk without treatment 1
- Forrest Ib (Active oozing bleeding): 55% rebleeding risk without treatment 1
- Forrest IIa (Non-bleeding visible vessel): 43% rebleeding risk without treatment 1
- Forrest IIb (Adherent clot): 22% rebleeding risk without treatment 1
Low-Risk Stigmata (No Endoscopic Therapy Needed):
- Forrest IIc (Flat pigmented spot): 10% rebleeding risk 1
- Forrest III (Clean ulcer base): <5% rebleeding risk 1
Clinical Application of Forrest Classification
Endoscopic hemostasis is strongly recommended for Forrest Ia, Ib, and IIa lesions because these high-risk stigmata carry substantial risk of persistent bleeding or rebleeding 1, 2. For Forrest IIb lesions (adherent clot), consider endoscopic clot removal; if underlying Ia, Ib, or IIa stigmata are revealed, perform hemostatic therapy 1, 2.
Patients with Forrest IIc or III lesions do not require endoscopic hemostasis and may be safely discharged on standard oral PPI therapy in selected clinical settings 1, 2. However, a critical caveat: patients with Forrest IIc lesions at second-look endoscopy who have Rockall scores ≥6 face an 18.6% rebleeding risk during days 4-14, compared to only 2.9% in those with Rockall scores <6 3. This highlights that endoscopic appearance alone is insufficient—clinical context matters.
Prognostic Scoring Systems
Glasgow-Blatchford Score (GBS) - Preferred for Risk Stratification
The Glasgow-Blatchford Score is the recommended prognostic tool for pre-endoscopy risk stratification, with superior sensitivity (99%) for identifying patients requiring intervention. 4, 2
The GBS uses only pre-endoscopic clinical and laboratory data 1, 4:
- Hemoglobin level
- Blood urea level
- Pulse rate
- Systolic blood pressure
- Presence of syncope or melena
- Evidence of hepatic disease or cardiac failure
A GBS score of 0-1 identifies very low-risk patients who do not require early endoscopy or hospital admission (sensitivity 98.6% for detecting high-risk patients) 4, 2. These patients can be safely discharged with outpatient follow-up, though they should be counseled about recurrent bleeding risk and maintain contact with the hospital 2.
Rockall Score - Better for Mortality Prediction
The Rockall score exists in pre-endoscopic and complete (post-endoscopic) forms 1. The complete Rockall score incorporates five variables 1:
Clinical Variables:
- Age (<60 = 0 points; 60-79 = 1 point; ≥80 = 2 points)
- Shock status (no shock = 0; tachycardia = 1; hypotension = 2)
- Comorbidity (none = 0; cardiac failure/IHD = 2; renal/liver failure/malignancy = 3)
Endoscopic Variables:
- Diagnosis (Mallory-Weiss/no lesion = 0; other diagnoses = 1; GI malignancy = 2)
- Stigmata of recent hemorrhage (none/dark spot = 0; blood/clot/visible vessel = 2)
A total Rockall score <3 indicates excellent prognosis, while scores >8 indicate high mortality risk 1. The Rockall score has better discriminative ability for predicting mortality than rebleeding, with sensitivity of 93-96% but may misclassify 4-7% of high-risk patients as low risk 4.
AIMS65 Score - Not Recommended for Low-Risk Identification
The AIMS65 score is not recommended for identifying low-risk patients suitable for early discharge because it has lower sensitivity (78-82%) for identifying high-risk patients and is designed primarily to predict mortality rather than safe discharge 4.
Integration of Classification Systems in Clinical Practice
Use the Glasgow-Blatchford Score first for pre-endoscopy triage decisions (hospitalization vs. discharge), then apply the Forrest classification during endoscopy to guide hemostatic therapy, and finally use the complete Rockall score to refine mortality risk assessment 1, 4.
The evidence consistently shows that combining clinical scoring systems with endoscopic classification provides superior risk stratification compared to either approach alone 1. For example, even a low-risk Forrest IIc lesion warrants closer monitoring if the patient has a Rockall score ≥6 3.
Common Pitfalls to Avoid
Do not rely solely on pre-endoscopic Rockall scores for discharge decisions—they lack validation for this purpose and the modified Rockall score (without endoscopic findings) has not been established as reliable 1. Do not use AIMS65 to identify low-risk patients for early discharge—its lower sensitivity makes it unsuitable for this purpose 4.
Inter-rater agreement for Forrest classification is generally good (0.639-0.859) but varies with endoscopist experience 5. The distinction between Forrest IIa (non-bleeding visible vessel) and IIb (adherent clot) can be challenging and significantly impacts management decisions, so when uncertain, err toward more aggressive classification and treatment 5.