Health systems produce large volumes of information across electronic records, laboratories, imaging, devices and administrative platforms, making data standardisation central to interoperability and secondary data use. Yet adoption remains uneven. Political fragmentation, technical heterogeneity and the lack of shared implementation practices continue to slow progress, particularly around extract, transform and load processes. A systematic review of FHIR, OMOP-CDM and openEHR published in the Journal of Biomedical Informatics examined how these standards are being applied in real-world and real-like settings, where uptake is strongest and which barriers still limit wider use.
Where Adoption Is Taking Shape
The review screened records from five scientific databases and ultimately included 99 studies published between 2021 and 2024. Across those studies, OMOP-CDM appeared most often, featuring in 57% of the reviewed literature, followed by FHIR in 39% and openEHR in 8%. Most implementations were concentrated in research settings, which accounted for 87% of the included work. Data reuse was the most common purpose, representing 47% of applications, while clinical decision support accounted for 23%. In implementation terms, 42% of studies involved more than one centre, 32% were single-centre and 26% were national, indicating that adoption is no longer confined to small pilot environments alone.
Must Read: Interoperability Advances Cancer Research Collaboration
The distribution of use also pointed to clear patterns by domain. Clinical medicine emerged as the most prominent application area, while FHIR showed particular strength in public health and real-time exchange scenarios. OMOP-CDM was more strongly associated with large-scale and longitudinal research use. The review also identified five recurring use categories across the literature: data exchange, clinical decision support, vocabulary definition, data reuse and EHR design. Together, these findings suggest that standard adoption is becoming more operational and more specialised at the same time, with each framework tending to be selected for a different part of the data lifecycle rather than as a universal solution.
Strengths and Limits of the Main Standards
FHIR’s position rests on its modular, resource-oriented design and its ability to support flexible system development, real-time data exchange and incremental implementation. It has been used across clinical care, public health and research, and the review linked it to applications involving patient-reported outcomes, genomic data and broader interoperability frameworks. Its adaptability also appears useful in lower-resource settings and in environments shaped by strong regulatory drivers. At the same time, that flexibility creates tension. FHIR implementations often depend on extensions and local customisation, which can undermine consistency between deployments. The review also noted fragmentation across resources, limited native support for complex multi-criteria searches and the need for preprocessing modules or advanced server capabilities when query demands rise. Under the 80/20 principle built into the standard, some clinical granularity may remain outside the core model, increasing the risk of data loss.
OMOP-CDM was strongly associated with secondary use, clinical decision support and complex longitudinal analysis. Its relational structure and analytical ecosystem make it attractive for multi-centre research, predictive models and machine learning workflows. The review described it as particularly well suited to large datasets and observational health research. That strength, however, comes with a heavy implementation burden. Comparative mapping work identified the ETL phase as the most resource-intensive part of OMOP adoption, especially where institutions must reconcile different coding systems, terminologies and encounter structures. The review also noted that OMOP-CDM can lack some detailed clinical elements, can be difficult to use for administrative or institutional data and is limited in real-time ingestion and patient re-identification. Analytical tools within its ecosystem may also require substantial computational resources, which can create barriers for organisations with more limited infrastructure.
Why Fragmentation Still Persists
One of the clearest findings was not simply that the standards differ, but that implementation practices remain inconsistent. Only 27% of the included studies reported coverage of the standard, meaning the extent of mapping from source data to the target model often remained unclear. The review also highlighted gaps in ETL reporting and extension practices, limiting transparency around how standards are operationalised in practice. This weakens the ability of health systems to compare implementations, reproduce successful approaches or judge whether a standard has been used comprehensively or only partially.
The review did not frame FHIR, OMOP-CDM and openEHR as competing options. Instead, it positioned them as complementary components of a modern interoperability architecture: openEHR for data persistence, FHIR for transfer between organisations and OMOP-CDM for analytics and insight generation. Examples in the reviewed literature showed practical combinations of these models, including openEHR-to-OMOP mappings and ETL pipelines that harmonise FHIR-based hospital data with OMOP-CDM for participation in research networks. openEHR remained the least represented standard in the review, and no limitations were reported in the studies that used it, but its uptake was still limited. That pattern suggests that maturity is advancing unevenly, with the broadest momentum around FHIR and OMOP-CDM and a more restricted evidence base for openEHR.
The review showed that health data standard adoption is moving forward, but not yet in a coherent or fully scalable way. Real-world use is increasingly visible, particularly in research, public health and clinical decision support, yet adoption remains fragmented by domain, technical model and implementation practice. FHIR stands out for exchange, OMOP-CDM for analytics, and openEHR for persistence, but none resolves the full interoperability problem alone. Incomplete reporting on coverage, persistent ETL burdens and continued reliance on local customisation remain major obstacles. The overall direction is therefore not towards a single dominant standard, but towards hybrid ecosystems supported by clearer implementation methods and more consistent operational practice.
Source: Journal of Biomedical Informatics
Image Credit: iStock
References:
Marfoglia A, Arcobelli VA, Moscato S et al. (2026) Challenges of health data standard adoption and usage: a systematic review. Journal of Biomedical Informatics: In Press.
