KEY POINTS
- This multidisciplinary Special Report compared five healthcare data standards—OMOP, CDISC SDTM, FHIR, NAACCR, and the Operational Ontology for Oncology (O3)—alongside the OBO Foundry ontology framework. The investigation used a 90-item benchmark derived from radiation oncology operations and clinical trial reporting, including the phase III NRG-RTOG 1308 photon-versus-proton lung cancer trial as a representative use case.
- O3 provided the broadest coverage of radiotherapy-specific information, representing 46/62 elements (74%) without qualification, compared with 15/62 (24%) for OMOP, 12/62 (19%) for NAACCR, 11/62 (18%) for FHIR, and none directly for OBO Foundry. However, this partly reflects the underlying purpose of each system: O3 was designed specifically for detailed oncology treatment representation, whereas the other standards prioritize different reporting or interoperability tasks.
- The relative ranking reversed for general clinical information. Among 28 non-RT-specific elements, OMOP represented 79%, FHIR 71%, NAACCR 46%, O3 25%, and OBO Foundry 4%. This supports the authors' argument that a single comprehensive standard is not necessarily desirable: specialized RT data models offer granularity, while established clinical standards provide broader healthcare interoperability.
- Rather than copying every radiation treatment data element into every system, the authors proposed a three-tier propagation framework. Twenty course-summary elements should be shared across standards, 26 planning and adaptive treatment elements should be adopted selectively according to use case, and 16 voxel-level or machine-level artifacts should remain in their authoritative systems and be linked using DICOM unique identifiers rather than duplicated.
- The framework addresses major differences in how standards represent treatment courses, prescription and fractionation, anatomical targets, imaging, adaptive interventions, delivery details, toxicity, and trial-specific endpoints. The authors recommend using O3 as a radiation oncology semantic reference, with mappings to FHIR for exchange, OMOP for observational analysis, and CDISC for clinical trial submissions, while preserving each system's existing functions.
- Four principal implementation barriers were identified: structural incompatibility, missing or inconsistent terminology, excessive adoption burden, and insufficient validation tools. The proposed solution assigns each gap to a responsible standards organization and an existing terminology or governance process rather than creating another competing universal model. Reliable extraction from treatment planning systems, oncology information systems, and electronic health records remains an essential prerequisite.
- The comparison is a structured framework and coverage analysis, not a demonstration that interoperable data transfer already works across hospitals or vendors. The benchmark is oriented toward radiation oncology trial and operational requirements, and several authors actively participate in the standards being evaluated, introducing potential perspective bias. No prospective implementation study measured reductions in manual data collection, error rates, cost, or improved clinical outcomes.
CLINICAL TAKEAWAY
Radiation oncology data harmonization may be more achievable through a shared core of treatment-summary elements and selective mappings between existing standards than through development of one all-encompassing data model. This framework provides practical direction for multicenter registries, automated trial reporting, and AI-ready datasets, but its real-world interoperability and resource requirements still require implementation testing.
SOURCE
International Journal of Radiation Oncology, Biology, Physics