EU Bids to Overcome Data Fragmentation in Pursuit of Real-World Evidence

Aug 01, 2017
Volume 26, Issue 8

To say “healthcare data are very fragmented” is an understatement, and a very obvious one at that. But that is the observation that the European Union’s Innovative Medicines Initiative (IMI) has chosen to introduce its latest bid to overcome the problem. On July 19, the EU launched a call for proposals, backed with hard cash, for developing a European Health Data Network (EHDN).

Creating this new network is closely linked to the growing need to get a better European grip on real-world evidence. The project is intended to provide a harmonized model that can tackle the current deficiencies that IMI identifies: data even within one healthcare center are typically spread across different repositories; across entities, different standards are used to code diagnosis, lab results, drugs, or procedures; and in most healthcare systems, a majority of the core clinical data is buried in unstructured text notes, making data analysis even more challenging. The aim is also to address the diversity in governance processes, with a framework that can be used to accommodate cross-center studies.

IMI knows this isn’t cheap. It notes that the costs of data harmonization can vary greatly between different data sources. While the harmonization of existing, highly structured, and integrated research databases may be relatively cheap, harmonizing unstructured or semi-structured data is a resource-intensive effort, with the cost to perform such a conversion estimated at between $30,000 and $100,000 per data source.

But this is more than just an exercise in the evident need to tidy up the data picture in Europe. It is directly connected to a much bigger dynamic: the push toward outcomes-focused healthcare services in Europe. Everyone is in favor of more outcomes-based healthcare, in principle—but there are many variations on the theme of how that should be achieved, and the EHDN project reflects some of those.

“The European market for health outcomes research is limited to commercial providers and a limited number of academic health science centers with funds available to develop secondary use platforms for research,” says IMI. “This both biases the research that can be undertaken, as only data collected by these providers can be used, and in some cases, creates a monopolistic environment that prevents health outcomes research from gaining more traction. It would likely be true to say that not one data source provides the whole truth in the real world, and as such, collaboration is critical to supporting quality evidence.”

So once the project has developed a standard methodology, it will then work with data providers and data users to stimulate new services that can improve the use of health outcomes-focused models in healthcare delivery and regulatory and health technology assessment decision-making. “Exploiting the opportunities offered by the wealth of emerging data from many evolving data sources via the generation of methodologies with real-world data will inform European decision-making in healthcare and policy debates,” says the invitation to participate. Ultimately, it is intended to help in defining “new business and health funding models (including incentive models) that will allow for healthcare systems transformation.”

The growing use of multi-center observational studies, with their increasing complexity, requires organization and a broader Europe-wide strategy, says IMI. “It is recognized that reuse and analysis of healthcare data holds the key to the transition” to outcomes-driven, sustainable healthcare systems. So the EHDN initiative seeks to address this critical challenge by converting a large number of relevant datasets across Europe to a common format and standard.

But EHDN is not going to be a big new central system. To respond to the sensitivities about movement of data, it will take a federated network approach. There is no intention of creating a centralized repository of patient-level data. The data will remain local, on the premises of the data owner/custodian, and under their clear control and governance. EHDN envisions each data owner implementing a harmonized, standardized version of their data set, so that research and reuse of data can be executed more efficiently. In essence, the “analysis is brought to the data” and only aggregated results are returned, so no patient data leaves the premises.

With deliberate pragmatism, EHDN will focus on facilitating three domains. The first is research—shaping and leading a community of interested data sources and data scientists and engaging with the global community. Topics may include discovery, pharmacovigilance, monitoring of effectiveness and safety, outcomes research, identification of variability in care delivery, disease background related information, or epidemiology of disease.

A second area of focus will be health services efficiency—in essence, how best to deliver real-world data that is relevant to evaluating real-world outcomes for therapeutic interventions. Activities could cover outcomes-based contracting, optimizing patient pathways, or quality improvement of health services.

A third domain will be individual patient care, and how the application of the federated data network can support patient level decision-making in clinical care. Aspects to cover include an interoperable data standard to stimulate a market in digital health solutions, expert systems, predictive algorithms, and integration with mobile health. It is hoped that the adoption of common enabling technology will stimulate a new generation of digital providers to develop and deliver services in data transformation, data semantics, and analytical capabilities, and a certification process for providers is already planned.

Central to the rationale for the EDHN initiative is that personalized medicine's more focused treatment options, driven by rapid technological and biological innovation, increase the difficulty of demonstrating risk/benefit in the real world. But it is also intended to help clinicians who currently have to make treatment choices based on short-term, surrogate, and often not-comparable data, and payers with finite budgets making reimbursement decisions on life-prolonging options on the basis of limited data. "Collaboration among healthcare systems and relevant stakeholders is necessary to capture and aggregate data, analyze it, and extract relevant insights," says IMI.

EHDN will apply two standards, the Observational Medical Outcomes Partnership Common Data Model (OMOP CDM) and the International Consortium for Health Outcomes Measurement (ICHOM) standards. The OMOP CDM is the result of a public-private collaboration, currently under the umbrella of the international Observational Health Data Sciences and Informatics project, whose Common Data Model specifies how to encode and store clinical data at a fine-grained level, ensuring that the same query can be applied consistently to databases around the world. The ICHOM standards identify specific outcomes metrics for a number of diseases. In cases where data elements are lacking (such as patient-reported outcomes), novel approaches can be developed to capture data.

As might be expected for any European project, the issues of data privacy are also taken into account. "Besides standardization and technical aspects, there is also a paramount need for further shaping a trusted environment for data sharing in Europe. To move the data sharing agenda forward, creating benefits for all stakeholders in the eco-system, several non-technical dimensions are of critical importance. These are, for example, legislative aspects, data security and privacy, or data quality improvement," says IMI politely.

EHDN isn't expected to be an easy exercise. Or inexpensive—the outline budget is around $30 million. Or rapid—the expected timeline is five years. The destination may be clear enough, but the journey will have to take IMI and the winning consortium through some hazardous territory. Not least, as the invitation notes almost casually, because it will be necessary, among many other things, to "seek input and definition from regulatory and HTA agencies as to what constitutes valid ‘real-world evidence.’" Which is, very much, where we came in.


Peter O'Donnell is a freelance journalist who specializes in European health affairs and is based in Brussels, Belgium

lorem ipsum