E0295 - Glasgow-Blatchford Score vs Glasgow-Blatchford Score Plus Additional Variables to Predict High-Risk Peptic Ulcers in Acute Nonvariceal Upper Gastrointestinal Bleed in a United Arab Emirates Tertiary Center

Introduction: Acute nonvariceal upper gastrointestinal bleeding (NVUGIB) is a medical emergency and one of the most common reasons for emergency GI care. The Glasgow-Blatchford score (GBS) has been developed for risk stratification. Our group has previously reported that GBS has very limited accuracy in Middle East North Africa (MENA) region. Our aim was to evaluate whether additional clinical variables improved the accuracy of GBS to predict actual high-risk stigmata or active bleeding during emergency endoscopy and to validate this in a large volume tertiary care setting in the MENA region.

Methods: Data were collected retrospectively for a cohort of patients over the age of 16 years who attended the emergency department or were inpatients and underwent esophagogastroduodenoscopy (EGD) for acute NVUGIB from January 2020 through September 2021 at a tertiary hospital in the United Arab Emirates. Our predictor variables included GBS and other potential risk factors as listed in Table 1. GBS included patients' hemoglobin by gender, blood urea, heart rate, and systolic blood pressure from the time of admission as well as presentation with syncope or melena, and evidence of hepatic disease or cardiac failure. After assessing ability of the GBS to predict Forrest I- II classification by estimating the area under the receiver operating characteristic curve (AUROCC), we examined if other patient demographics and clinical characteristics might improve upon the ability of GBS to detect Forrest I-II classification by fitting several logistic regression models and calculating the AUROCC.

Results: Among the 153 patients in our cohort, the median age was 58 years (IQR XMR 11 (IQR 7 to 13). In our cohort, 28 (18.3%) patients had Forrest I-II classification identified during endoscopy. The AUROCC for the ability of GBS to detect Forrest I-II classification was 0.49 (95% CI 0.37 to 0.62). AUROCC for the ability of each patient characteristic to detect Forrest I-II classification is shown in Table 1. Models that contained age, diastolic blood pressure and nationality/immigration status in addition to GBS provided the highest overall AUROC with confidence intervals significantly higher than GBS alone.

Discussion: Our study did not find evidence that GBS by itself was a useful tool at identifying patients with Forrest I-II classification during endoscopy. Our data suggested younger age, nationality / immigration status along with lower DBP may be better predictors of Forrest I-II classification when considered with GBS.