University of Tartu research group publishes EST-Health-30 data resource profile in the International Journal of Epidemiology

Complete health records for more than half a million Estonians have been brought together in a single, standardised research dataset. Researchers from the University of Tartu Health Informatics Research Group and STACC have now published a Data Resource Profile of EST-Health-30 in the International Journal of Epidemiology, setting out how the dataset was built, what it contains, and how researchers can access and use it. It covers a random 30% sample of the population (509,856 individuals) from 2012 to 2024, with annual updates planned through 2026.

Routinely collected health data hold considerable research value, but they are rarely available in a linked, standardised and analysis-ready form. As a result, basic questions about how patients are actually diagnosed and treated often go unanswered. Earlier analyses of linked Estonian records have shown what becomes visible once systems are joined up: in heart failure, for example, many patients had received no diagnostic testing and only around half were on the recommended medication. These kinds of patterns remain invisible within any single database. EST-Health-30 is designed to make population-level insight of this kind systematic rather than exceptional.

Sulev Reisberg, first author of the article and Associate Professor of Health Informatics at the University of Tartu, commented: “Clinicians have long wanted this. A doctor knows how they treat their own patients, but not how those patients fare after the visit ends, nor how similar patients are treated elsewhere in the country. Comparison across Estonia has been almost impossible, and improving average care has been correspondingly hard. For the first time the complete patient picture is visible in one place, allowing unprecedented overviews of treatment patterns, costs and multimorbidity.”

The dataset links individual-level records across five nationwide databases: electronic health records, health insurance claims, prescription data, the cancer registry and the causes of death register. This yields rich, multimodal information — diagnoses, procedures, laboratory measurements, prescriptions, free-text clinical notes, healthcare utilisation and costs — across more than a decade and the full life course.

All sources are harmonised to the OMOP Common Data Model (version 5.4) using international vocabularies such as SNOMED CT, RxNorm and LOINC. This standardisation is what connects EST-Health-30 to the wider European research landscape. When data is following a shared model, EST-Health-30 can contribute to federated studies in which analyses run locally and only aggregated results are shared, with personal data never leaving the secure environment.

Privacy protection is built into the design. The dataset holds no names or personal identity codes, and records are pseudonymised so that individuals cannot be traced. All analysis takes place inside a secure environment at the University of Tartu that is isolated from the internet and meets EU data-protection and health-data standards; researchers never download the data, and only aggregated results that cannot identify anyone may be taken out. Quality checks found the data highly complete and consistent with 99.2% of applicable checks passed and its age and sex profile closely matches the Estonian population.

Access is deliberately rigorous. Before working with the data, researchers must sign a cooperation agreement with the Health Informatics Research Group and obtain national ethics approval and authorisation from the Ministry of Social Affairs.

Population-scale, standardised data of this kind underpin personalised-medicine research, allowing it to move from individual studies towards a more systematic evidence base. TeamPerMed is a co-funder of the resource through the Horizon Europe programme, and its development is a step towards making linked Estonian health data available for research at scale, in Estonia and across Europe.

The data resource profile is available in the International Journal of Epidemiology (doi.org/10.1093/ije/dyag113).