Centralized Digital Surveillance for Abdominal Aortic Aneurysm Detection, Longitudinal Tracking, and Management Within an Integrated Health System: Retrospective Cohort Study.

Authors

Abstract

Incidental detection of abdominal aortic aneurysms (AAAs) has increased with widespread cross-sectional imaging, while traditional surveillance remains fragmented and clinician-dependent. Real-world descriptions of centralized digital surveillance programs combining structured electronic health record (EHR) queries with natural language processing (NLP) of radiology reports remain limited. This study aimed to describe the design, implementation, and operational performance of the System to Track Abnormalities of Importance Reliably (STAIR), a centralized digital surveillance program combining structured EHR queries, an internally developed NLP model, clinician referrals, and automated lost-to-follow-up detection to identify, track, and manage patients with AAAs across a large integrated health system. This retrospective cohort study included all patients enrolled in the STAIR AAA surveillance program from December 2022 through December 2024. Patients were identified via 4 pathways: an internally developed NLP model applied to radiology reports, EHR problem-list queries, clinician referrals, and automated lost-to-follow-up queries. All cases underwent standardized centralized clinical review, and surveillance intervals were assigned using guideline-informed institutional protocols. Administrative status and ongoing surveillance were assessed through April 2026. Outcomes were descriptive. Cohort integrity was evaluated against an independent, automated, computed tomography (CT)-based measurement system among patients with CT imaging. A total of 8464 patients were enrolled (mean age 77.1, SD 8.8 y; male: n=6516, 77.0%; and White: n=83119, 8.2%). Identification was predominantly automated via problem-list queries (n=4994, 59.0%) and radiology NLP (n=2454, 29.0%), with clinician referral (n=592, 7.0%) and lost-to-follow-up queries (n=423, 5.0%). Following centralized review, most patients were assigned guideline-based duplex surveillance (biennial: 45.3%; 5-year: 9.5%); 20.6% were referred for vascular surgery evaluation, and another 20.6% had prior AAA repair at enrollment. Among all 8464 enrolled patients, 4188 (49.5%) remained under active surveillance as of April 2026, and 2709 (32.0%) had transitioned to management outside the health system. Duplex ultrasonography was the predominant surveillance imaging modality (6122/6951 patients, 88.1%); 351 elective AAA repairs were performed system-wide during the study period, including 260 in the enrolled cohort. In a large integrated health system, this centralized digital surveillance infrastructure was operationally feasible and supported large-scale, structured identification, guideline-based surveillance and management assignment, and documented administrative status through April 2026 for all 8464 enrolled patients, including those who could not be reached despite repeated outreach. These descriptive findings establish operational feasibility only and characterize a scalable, workflow-focused approach to population-level AAA surveillance and management that emphasizes structured clinical oversight rather than autonomous AI decision-making; clinical effectiveness is not demonstrated by this design.

Tags

Topics