P2452 - A Novel Model for Predicting the Need of Endoscopic Intervention in Patients With Upper Gastrointestinal Bleeding

Brian Lee, DO¹, Gian Galura, MD², Jeff Taclob, MD³, Swati Mahapatra, DO², Akanksha Togra, MD², Shivangini Duggal, MD⁴, Vishwajeet Singh, PhD², Jose W. Sotelo, MD², Monica Botros, MD², Alejandro Robles, MD³, Alok Dwivedi, PhD⁵, Marc Zuckerman, MD², Sherif E.. Elhanafi, MD^2
1Paul L. Foster School of Medicine, Texas Tech University Health Sciences Center, San Antonio, TX; ²Texas Tech University Health Sciences Center, El Paso, TX; ³Paul L. Foster School of Medicine, Texas Tech University Health Sciences Center, El Paso, TX; ⁴TTUHSC, El Paso, TX; ⁵Texas Tech University Health Sciences Center, El Paso, TX

Introduction: Endoscopy is a cornerstone of management of upper gastrointestinal bleeding (UGIB) that may constitute endoscopic interventions for hemostasis. Several scores have been developed to predict clinically relevant outcomes such as mortality, need for medical intervention, and length of hospital stay. However, these scores were not powered to predict the need for endoscopic intervention. The objective of this study was to develop a new model that predicts the need for endoscopic intervention. Additionally, we aimed to assess the performance of other known prognostication scores in predicting the need for endoscopic intervention in patients with UGIB.

Methods: A retrospective cohort study was performed on adults who presented to a tertiary care center with UGIB from July 1, 2017, to July 31, 2022. Demographics, clinical characteristics, endoscopic interventions and outcomes were reviewed. The model was developed through logistic regression. Variables were identified using a backward stepwise selection procedure at removal probability of 0.2. The performance of the model and the other UGIB scores to predict the need for endoscopic interventions were assessed using the area under the receiver operator curve (AUC-ROC).

Results: A total of 1029 patients presented with UGIB, of which 708 (68.8%) underwent endoscopy and were included in the study. Using logistic regression to develop a predictive model, six variables were identified as independent risk factors of endoscopic intervention. These included age (OR: 1.01; 95% CI: 1.00, 1.03), hematemesis (OR: 1.95; 95% CI: 1.29, 2.95), presenting heart rate (OR: 1.01; 95% CI: 1.00, 1.02), anticoagulation/antiplatelet use (OR: 0.49; 95% CI: 0.24, 0.99), hemoglobin level at presentation (OR: 0.92; 95% CI: 0.86, 0.98), and cirrhosis (OR: 3.77; 95% CI: 2.50, 5.69). The AUC of the new model including these variables was 0.740 (95% CI: 0.693, 0.788), while the AUC of other predictive scores were such as GBS (Glasgow-Blatchford Score) (0.635), MAP (0.625), ABC (0.611), AIM65 (0.571) and pRS (Pre-Endoscopy Rockall Score) (0.517) (Fig 1).

Discussion: In this retrospective cohort study, we developed a novel model for predicting endoscopic interventions with superior performance compared with other available bleeding scores. The new model might be an effective tool to help triage and manage patients presenting with UGIB. Prospective studies are needed to further validate the utility of the new model.



Disclosures:


Brian Lee, DO¹, Gian Galura, MD², Jeff Taclob, MD³, Swati Mahapatra, DO², Akanksha Togra, MD², Shivangini Duggal, MD⁴, Vishwajeet Singh, PhD², Jose W. Sotelo, MD², Monica Botros, MD², Alejandro Robles, MD³, Alok Dwivedi, PhD⁵, Marc Zuckerman, MD², Sherif E.. Elhanafi, MD². P2452 - A Novel Model for Predicting the Need of Endoscopic Intervention in Patients With Upper Gastrointestinal Bleeding, ACG 2024 Annual Scientific Meeting Abstracts. Philadelphia, PA: American College of Gastroenterology.