S983 Development of an EPIC Smartphrase for Barrett’s Esophagus Screening: A Tool for Improving Early Detection

Introduction: Current Barrett’s esophagus (BE) screening strategies have not significantly reduced esophageal adenocarcinoma (EAC) rates, partly due to low screening rate—fewer than 40% of at-risk patients undergo screening endoscopy. Contributing factors include the absence of gastroesophageal reflux disease (GERD) symptoms in over half of BE patients, limited resources, and lack of recognition or referral from primary care. Offering esophagogastroduodenoscopy (EGD) for BE screening for patients at risk during their screening colonoscopy has been proposed to potentially improve detection rates. To support this, we developed and validated an electronic medical record (EMR)-based decision support tool that uses existing patient data to identify patients with BE risk factors undergoing screening colonoscopy. Methods: An EPIC SmartPhrase was developed to identify BE screening candidates based on risk factors including GERD, White race, male sex, age >50, smoking, obesity, and family history of BE/EAC. Patients with >3 risk factors or a family history of BE/EAC, excluding those with prior EGD, qualifies for screening. Consistent with American Gastroenterological Association Clinical Practice Update recommendations, GERD was not required among the 3 risk factors and family history alone prompted screening. To validate the SmartPhrase, we retrospectively reviewed patients aged 45-75 years who underwent screening colonoscopy (January 1, 2024-March 31, 2024) by 3 endoscopists, comparing SmartPhrase results to manual chart review. Results: Among 366 patients undergoing screening colonoscopy, the SmartPhrase identified 172 (47%) patients who qualified for BE screening. Manual chart review confirmed 147 (85.5%) as true candidates. Errors in 25 cases were due to prior EGDs at outside facility (48%) or missing records (52%). The SmartPhrase identified all true candidates with 100% sensitivity and 86.6% specificity when compared to manual chart review. Conclusion: A major barrier to BE screening is quickly and accurately identifying eligible patients. We developed an EMR SmartPhrase that is fast, highly sensitive and adaptable across systems. Compared to manual chart review, the SmartPhrase showed 85.5% accuracy, 100% sensitivity, 86.6% specificity, 85.5% positive predictive value and 100% negative predictive value. Errors (14.5%) were due to undetected outside EGDs or missing records, which could be addressed by confirming with patients. Ongoing research includes prospectively using the SmartPhrase to identify patients undergoing screening colonoscopy who are eligible for BE screening, enabling simultaneous BE screening via non-invasive screening or combined EGD during the same sedated procedure.