Results and lessons learned from a federated OMOP study

The Nordic countries offer significant potential for clinical research. They maintain extensive, high-quality health data registers covering entire populations over long periods, supported by national personal identifiers that enable linkage across different registers, and a culture that trusts public institutions and values science and research.

However, for certain studies, the population of an individual country may be too small. Ideally, researchers would combine data from all Nordic countries into a single dataset.

This is easier said than done. Health datasets are sensitive and do not like to travel – especially across national borders.

The OMOP common data model (CDM), federated analysis, and anonymisation provide a practical solution to this challenge. In a federated analysis, data remains in its original location. Instead of transferring datasets, researchers send an analysis script to data holders, who execute it locally and return only anonymised, aggregated results.

But does this approach work in practice? To answer that question, the VALO – Value from Nordic health data project conducted a pilot study.

The pilot study compared the quality of care for patients with metastatic non-small cell lung cancer (NSCLC) in the Nordic countries.  A Nordic consortium was established to facilitate collaboration and implement a multinational federated analysis using the OMOP common data model across three university hospitals. Standardised analysis scripts were executed locally at each site, and aggregated results were compiled centrally.

The consortium was led by IQVIA and included data partners from Helsinki University Hospital, Oslo University Hospital, Rigshospitalet and observers from Karolinska University Hospital, Karolinska Universitet and Landspitali.

Principal Investigator Åslaug Helland from Oslo University Hospital stated: “We are very pleased with the opportunities demonstrated by the VALO pilot study. It has shown that real-world hospital data can be reliably used for both research and care quality assurance. At the same time, the project has highlighted areas where we need to strengthen data collection and storage practices.”

In essence, this was a study about a study – an assessment of the feasibility of federated analysis. The outcome confirmed that federated analysis works and can be applied in the Nordic context.

However, the process revealed several limitations. The actual study population was smaller than anticipated. Data access timelines were long, and there is no clearly specified process for data access in university hospitals. Coding practices vary, and some data elements relevant to research are not captured at all. These factors constrain the practical application of the federated model.

The VALO project continues its work strengthening a unified Nordic health region. In its second phase, another feasibility study is underway. Health datasets still do not cross borders; instead, they are imported into secure processing environments – either a national one or one operated by the data holder. Researchers receive remote access to review source data, understand its granularity and structure, and use that insight to refine analysis scripts.

Further insights and future recommendations from the first pilot study are described in a separate lessons learned report. The report identifies gaps in current data registries and processes and outlines concrete actions to create an environment that better supports medical research and enables OMOP based federated analysis.

Please take a look at the results of the first study and follow the VALO project website for more information and opportunities to contribute to the project.