FROM EHR TO EVIDENCE: A FEDERATED OMOP-BASED ONCOLOGY DATA NETWORK DELIVERING SCALABLE REAL-WORLD EVIDENCE ACROSS EUROPE
Author(s)
Lauren J. Revie1, Beatriz Rubio Huete, BSc2, Dionisio Luis, BSc3, Jessyca Gil, BSc2, Irina Yakovleva, BSc2, Giovanni Tonon, PhD4, Xose Fernandez, PhD5.
1Senior Data Scientist, IQVIA, London, United Kingdom, 2IQVIA, Barcelona, Spain, 3IQVIA, Lisbon, Portugal, 4DIGICORE, Brussels, Belgium, 5IQVIA, London, United Kingdom.
1Senior Data Scientist, IQVIA, London, United Kingdom, 2IQVIA, Barcelona, Spain, 3IQVIA, Lisbon, Portugal, 4DIGICORE, Brussels, Belgium, 5IQVIA, London, United Kingdom.
OBJECTIVES: Access to high-quality oncology real-world data (RWD) is constrained by fragmented electronic health records (EHRs), inconsistent data structures, and governance requirements. The Digital Oncology Network in Europe (DigiONE) aimed to operationalise a standardised, analysis-ready oncology data network by transforming hospital EHR data into a common data model and enabling privacy-preserving, multi-centre analytics at scale.
METHODS: DigiONE was developed through European Commission co-funded programme to harmonise oncology EHR data across 18 hospitals. Sites implemented ETL pipelines guided by the Minimal Essential Description of Cancer (MEDOC) to standardise oncology variables. Natural language processing and patient-finder tools were used to structure key concepts from unstructured EHR data and standardized in a central research platform before being mapped to OMOP. A federated analytics architecture using open-source tooling enabled centrally developed analytical packages to be distributed and executed locally against OMOP databases. Results were returned as aggregated summaries, enabling cross-site pooling without transferring patient-level data. Version-controlled study packages ensured reproducible execution and methodological alignment across sites. A proof-of-concept retrospective pan-cancer analysis was conducted across 11 sites for diagnoses between 2019 and 2024.
RESULTS: Harmonised OMOP datasets were generated across all sites, enabling federated analyses of 140,000 patients spanning 13 tumour types. Despite heterogeneity in EHR systems and digital maturity, the ETL approach and MEDOC framework produced comparable, analysis-ready cohorts. The federated framework supported consistent derivation of key outcomes, including time from diagnosis to treatment and survival at 12, 24, and 36 months, without centralising patient-level data. Standardised study packages ensured reproducible analyses and consistent aggregation. Outputs were stratified by tumour type and geography, demonstrating scalable, multi-centre RWE generation.
CONCLUSIONS: DigiONE shows that OMOP-based harmonisation combined with reproducible federated analytics enables scalable, governance-compliant oncology RWE generation. This approach supports efficient multi-centre studies while maintaining local data control, providing a reusable infrastructure for pharmaceutical and healthcare stakeholders.
METHODS: DigiONE was developed through European Commission co-funded programme to harmonise oncology EHR data across 18 hospitals. Sites implemented ETL pipelines guided by the Minimal Essential Description of Cancer (MEDOC) to standardise oncology variables. Natural language processing and patient-finder tools were used to structure key concepts from unstructured EHR data and standardized in a central research platform before being mapped to OMOP. A federated analytics architecture using open-source tooling enabled centrally developed analytical packages to be distributed and executed locally against OMOP databases. Results were returned as aggregated summaries, enabling cross-site pooling without transferring patient-level data. Version-controlled study packages ensured reproducible execution and methodological alignment across sites. A proof-of-concept retrospective pan-cancer analysis was conducted across 11 sites for diagnoses between 2019 and 2024.
RESULTS: Harmonised OMOP datasets were generated across all sites, enabling federated analyses of 140,000 patients spanning 13 tumour types. Despite heterogeneity in EHR systems and digital maturity, the ETL approach and MEDOC framework produced comparable, analysis-ready cohorts. The federated framework supported consistent derivation of key outcomes, including time from diagnosis to treatment and survival at 12, 24, and 36 months, without centralising patient-level data. Standardised study packages ensured reproducible analyses and consistent aggregation. Outputs were stratified by tumour type and geography, demonstrating scalable, multi-centre RWE generation.
CONCLUSIONS: DigiONE shows that OMOP-based harmonisation combined with reproducible federated analytics enables scalable, governance-compliant oncology RWE generation. This approach supports efficient multi-centre studies while maintaining local data control, providing a reusable infrastructure for pharmaceutical and healthcare stakeholders.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
RWD20
Topic
Patient-Centered Research, Real World Data & Information Systems, Study Approaches
Topic Subcategory
Data Protection, Integrity, & Quality Assurance, Distributed Data & Research Networks
Disease
Oncology