Abstract A089: Implementation of an early detection program for pancreatic cancer using electronic health record-based risk prediction models

Abstract

Early detection remains a major challenge in pancreatic cancer (PC). We evaluated the feasibility of implementing an electronic health record (EHR)-based screening program that uses a validated risk prediction algorithm to identify patients at elevated risk for PC and offer targeted cross-sectional imaging. From December 2025 through May 2026, a previously validated EHR-based prediction model was applied every two weeks across two medical centers. Patients with a predicted PC risk ≥1% based on age, one-year changes in weight and alanine aminotransferase (ALT), current hemoglobin A1c level, and abdominal pain were contacted through the patient portal and invited to undergo a one-time pancreas-protocol computed tomography (CT) examination. Feasibility outcomes included patient interest, CT completion, and reasons for nonparticipation. Among 1,146 eligible patients contacted, 201 (17.5%) expressed interest in screening, 17 (1.5%) declined participation, and the remaining patients did not respond to outreach. Of the 201 interested patients, 105 (52.2%) completed CT screening, corresponding to 9.2% of all eligible patients contacted. Among CT completers, median (IQR) age was 72.3 (67.1–77.7) years and 53 (50.4%) were female. Among interested patients who did not complete screening, six ultimately declined because of CT-related costs. In addition, primary care physicians declined outreach for 16 patients, most commonly because of concerns regarding patient anxiety. This pilot implementation study demonstrates the feasibility of integrating an EHR-based risk prediction model into routine clinical practice to identify and engage patients at elevated risk for pancreatic cancer. Approximately one in six eligible patients expressed interest in screening, and over half of interested patients completed CT imaging. Key barriers included imaging costs and concerns regarding patient anxiety. Future efforts will focus on scalable outreach strategies, including automated text messaging and reminder systems, to improve participation and support broader implementation of risk-based pancreatic cancer screening programs. Tiffany Q. Luong, Jessica E. Moore, Qiaoling Chen, Susie Flores, Wansu Chen, Bechien U. Wu. Implementation of an early detection program for pancreatic cancer using electronic health record-based risk prediction models [abstract]. In: Proceedings of the AACR Conference on Pancreatic Cancer: New Frontiers in Biology and Therapeutic Development; 2026 Sep 25-28; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(18_Suppl_2):Abstract nr A089.