Predicting Re-Bleeding after Peptic Ulcer Hemostasis

ABSTRACT

Objective: To identify independent risk factors for early (≤7-day) re-bleeding after peptic ulcer bleeding (PUB) and to compare the predictive performance of Forrest classification, Complete Rockall Score (CRS), and Glasgow-Blatchford Score (GBS).

Methods: We retrospectively analyzed adults with endoscopy-confirmed peptic ulcer bleeding from 2015–2020. Early re-bleeding was defined as ≤7 days after index hemostasis. We applied univariable and multivariable logistic regression and assessed discrimination with ROC curves (AUC).

Results: Independent risk factors for early re-bleeding included: heart rate (OR 1.054), hemoglobin (OR 1.878), erythrocyte distribution width (OR 1.171), degree of ulcer erosion (OR 1.191), and blood transfusion intervention (OR 12.296). Forrest showed the best discrimination (AUC 0.775; sensitivity 96.2%; specificity 58.8%), followed by GBS (AUC 0.670) and CRS (AUC 0.507)

Conclusions: Heart rate, hemoglobin, erythrocyte distribution width, ulcer erosion, and blood transfusion are significant risk factors for early re-bleeding in PUB. Forrest grading is the most effective predictor, while GBS can stratify risk and may benefit from modifications. CRS showed limited predictive utility.

Limitations: single-center, retrospective design; possible residual confounding; no external validation.

Clinical implications: Forrest can guide intensified monitoring/hemostasis; GBS supports pre-endoscopy triage; CRS adds limited value.

Keywords: Peptic ulcer bleeding; early re-bleeding; Forrest classification; Glasgow-Blatchford Score; Complete Rockall Score; risk prediction.