2498 - Charting the Cosmos™: An Exploration of Radiation Oncology Encounters in the Epic™ Data Repository

Purpose/Objective(s): Cosmos™ aggregates de-identified, encounter-level data from millions of patients across diverse health-care systems. This dataset has not previously been examined in ambulatory radiation-oncology cohorts. Cosmos™ was queried to describe utilization patterns, payer mix, and sociodemographic characteristics of radiation oncology patients.

Materials/Methods: A retrospective descriptive query was run on 17 Feb 2026 for all radiation-oncology encounters recorded in Cosmos™ between 1 Jan 2023 and 27 Jan 2026. Data used in this study came from Epic Cosmos™, a dataset created in collaboration with a community of health systems using Epic™ representing more than 300 million patient records from over 1,903 hospitals and 42,000 clinics as of 2/21/26. Radiation oncology encounters were identified by department or provider specialty of “radiation therapy” or “radiation oncology”. Extracted variables included encounter type, primary billing diagnosis, payer type, age/sex/race, RUCA classification, and state. A separate longitudinal query captured annual encounter counts from 1 Jan 2010 to 31 Dec 2025.

Results: Annual radiation oncology encounters grew steadily from 58,817 in 2010 to 282 315 in 2025, with the steepest rise after 2018. Between 1 Jan 2023 and 27 Jan 2026 Cosmos™ reported 688,670 unique encounters. Outpatient /office visits comprised 288,712 (42%), telephone contacts 75,442 (11 %), and use of tele medicine was rare (1,560 < 1%). The three most common billed diagnoses were antineoplastic radiation therapy (Z51.0) – 102,949 (15%); malignant neoplasm of prostate (C61) – 54,680 (8%); and estrogen receptor positive status (Z17) – 29,255 (4%). Payer mix was dominated by commercial insurance (356,762, 52%), followed by Medicare (159,674, 23%), Medicaid (17,520, 3%), and VA (8,393, 1%). Age distribution was skewed older: 65–75 yr (200,266, 29%), 50–64 yr (187,640, 27%); pediatric encounters (<18 yr) accounted for 32,330 (˜5%). Females represented 52% of encounters. Racial composition was primarily White (497,877, 72 %); “Other” (102,391, 15%), Black (94,309, 14%), Asian (18,809, 3%), and Native/Alaskan (6,668, 1%). Geographically, 74% of encounters occurred in RUCA 1 (metro core). Small town (RUCA 7) and rural (RUCA 10) settings contributed 3% each. State level volume was highest in Ohio (138,711, 20,%), followed by North Carolina (65,802, 10%) and Florida (39,603, 6%).

Conclusion: Radiation oncology encounter volume accelerated after 2018, coinciding with broader health policy initiatives. The 2023-2026 snapshot shows a predominance of commercial payers, metropolitan patients, and minimal tele medicine use. Although the Cosmos™ dataset lacks procedural and outcome details, its breadth, timeliness, and de identified nature provide a robust baseline for potential future studies of benchmarking, predictive modeling, and health policy research in radiation oncology. Further work is needed to assess the representativeness of COSMOS to the US population.