Implementing a technical safety net: A scalable automated quality assurance tool for oncology patient navigation.

612 Background: Dana-Farber Cancer Institute (DFCI) implemented its Community-Focused Patient Navigation (CFPN) program in October 2021 with a manual quality assurance (QA) process. As part of the CFPN, Patient Navigators (PN) identify eligible patients in Epic, DFCI’s electronic medical record system, and document navigation activities in Quickbase (QB), a HIPAA-compliant application platform. Manual QA checks were performed to ensure all eligible patients were identified and documented accurately across both systems. As the CFPN expanded, the manual QA process grew more complex and resource intensive, increasing risk of human error and time demands. This underscored the need for an automated solution to improve QA efficiency. Methods: A business intelligence data visualization dashboard was developed to enhance QA by integrating Epic and QB data. Data housed within QB was ingested into the Enterprise Data Warehouse (EDW) using R scripts to make POST requests to the QB application programming interface, facilitating automated extraction and loading. Three static reference tables were created in EDW to streamline structured query language (SQL) updates and support scaling to additional treatment centers. Epic front-end fields were traced to corresponding EDW fields, and a single unified database query categorized patients by eligibility and identified in which systems (Epic or QB) their records existed. Three dashboard views were designed to visually highlight patient records requiring reconciliation, with an assigned staff member presenting the data to PNs. In treatment centers with CFPN active, Gastrointestinal, Breast, Thoracic, and Gynecologic oncology, the dashboard identified 124 missed patients out of 533 (23.26%) who were eligible from 9/25/2023 – 4/30/2025. Results: The dashboard centralized two data sources into one automated solution, ensuring all eligible patients were identified and documented accurately. Data within the dashboard automatically updates daily. It takes the user 5 – 40 minutes per disease center to identify any missed eligible patients, replacing the formerly multi-hour manual process. Moreover, the method for identifying eligible patients was created as a reusable data source and thus far has been used to build a second dashboard for a pre-post analysis of the CFPN. Conclusions: The dashboard functioned as a technical safety net for the CFPN, revealing and filling gaps in eligible patient identification. Next steps include exploring solutions for the current manual process of sending QA findings from the dashboard to PNs, using QA process insights to improve operational workflows, and refining the dashboard as needed. As the program expands, dashboard loading delays may become a limitation. This project was undertaken as a Quality Improvement Initiative at DFCI, and as such was not formally supervised by the Institutional Review Board per policy.