EMIF NEWSLETTER (MAY 2018)
Use the buttons below to jump to each article in this issue.
To Subscribe to the Newsletter, please complete the Enquiry Form.
Impact of the EMIF Project
Achievements, Lessons Learned, and Opportunities
The Final EMIF General Assembly Meeting
and Public Symposium
The Odyssey of EMIF
Bringing Light Where It is Needed
The space where we need to look for the answers to some of the most pressing questions in medicine today is in the alley, not under the lamppost. The European Medical Information Framework (EMIF) was formed to bring light where it is needed; we wanted to shed light on potential keys to liberating insights and evidence from real world data in Europe.
There is a transformation taking place, with healthcare becoming a truly data-driven sector, from clinical care through to R&D. Compared to many other industries in society, healthcare has been a latecomer to the digital realm. However, over the last five to six years since EMIF was first imagined, healthcare is truly moving into the digital realm, and our project has contributed to that change. In this newsletter—the last of this IMI project—we will be bringing together views from around the EMIF project on what we have achieved, what was a challenge, what lessons we learned, and where we could be going in the near future.
We Can Only Achieve Through Collaboration
EMIF has been a complex project with its three components: the socio-technical and governance framework (EMIF-PLAT), and the two disease projects, in Alzheimer’s disease (EMIF-AD) and in the metabolic consequences of obesity (EMIF-MET), both working on early biomarker development.
Collectively, we have brought together a remarkably diverse research and development effort within EMIF to address some of the key challenges for modern healthcare. Moreover, we have recognised that we can only achieve through collaboration, working together from when clinical data is generated to when it can potentially be analysed with the latest open-source tools. Our final meeting of the consortium and our public event spoke to the remarkable achievements of EMIF and all those who were involved in it, as well as the challenges and lessons learned. This is covered much further within the newsletter.
We have no doubt that working at scale—with increasingly large cohorts and in collaboration with data custodians internationally and eventually globally—is going to be the standard for real world observational health research. This will especially be the case when cohorts are linked to other complementary datasets, such as the molecular and imaging data used in parts of EMIF. We have learnt much about how to develop and nurture a federated data network, and about the value and indeed challenges of linking such observational data to basic science in the two disease areas. Importantly, we have been learning together. This has transformed not just our knowledge but led to productive and lasting scientific collaborations.
More on the results of EMIF is contained in this newsletter, but we also would want to take the opportunity to thank all those involved in and collaborating with EMIF. The project has set the stage for new projects that are impending, such as the IMI2 European Health Data and Evidence Network (EHDEN), which we hope will start later this year. Meanwhile, sustainable elements of EMIF will be accessible beyond the IMI phase for the research community. EMIF is already contributing to the infrastructure of other projects—providing the EMIF Catalogue and contributing to the harmonisation within Dementias Platform UK for example.
Setting New Precedents
EMIF has been an outstanding success, as evidenced by one of the highest citation impacts of any IMI project, and its scientific output in the three domains will continue to reverberate for many years. Insights generated into, for instance, the variable progression of Alzheimer’s disease, the liver-related consequences of obesity, and the utilisation of common data models in harmonising diverse European data, have set new precedents.
EMIF has done much to grow the community, platforms, infrastructures, and case for enabling the reuse of data in Europe. We look forward to this legacy being carried forward, not least through other IMI programmes and other initiatives as well as the EMIF continuation plans. All of you who have done so much to make EMIF the success that it is should be as proud as we are of the work we have done. We hope you enjoy reading some of the work of EMIF in this newsletter and many scientific papers to come, shining yet more light on where the keys to medical and societal challenges may be found.
Achievements, Lessons Learned, and Opportunities: Reflecting on the Past Five Years of EMIF
The European Medical Information Framework (EMIF) project has made major contributions towards promoting the reuse of healthcare data by developing tools for this purpose and by applying these tools in the fields of Alzheimer’s disease and the metabolic complications of obesity.
EMIF’s mission to improve access and re-use of healthcare data has been accomplished to a large extent, but sustainability will be crucial in order to leverage the full potential of EMIF.
EMIF has resulted in new collaborations that project leads hope will continue into the future.
HUGHESEFPIA Coordinator of EMIF-PLAT,
Janssen Pharma R&D
SMITHProfessor, University of Gothenburg,
Public lead of the Metabolic topic
STREFFERVP and Head of Early Clinical Neurology, UCB, EFPIA lead of the AD topic
JELLE VISSERAssociate Professor, Maastricht University,
VU University Medical Center, Public lead of the AD topic
WATERWORTHDirector of Genetics, Cardiovascular,
Metabolic and Dermatology, GSK, EFPIA lead of the Metabolic topic
Achievements Made Possible by EMIF
Nigel Hughes, the European Federation of Pharmaceutical Industries and Associations (EFPIA) Coordinator of EMIF-Platform, says that EMIF created “a community of disparate and diverse stakeholders with common goals around real-world, observational research.” These stakeholders then fostered the scientific and technical achievements of the project. One of the major achievements of the project is a common information framework, the EMIF Catalogue, to link up and facilitate access to diverse medical and research data sources. Over the past five years, EMIF has created a working platform and mapped multiple databases to the OMOP common data model.
Diagram representing the process of data standardization using the OMOP Common Data Model.
According to Johannes Streffer, the EFPIA lead of EMIF-AD, EMIF fosters and strengthens relationships between and among partners and data providers, which connects research centers more effectively. EMIF-AD has also developed tools such as the EMIF-AD Catalogue (with the cohort and patient selection tools), tranSMART, and Switchbox. Pieter Jelle Visser, public lead of EMIF-AD, notes that these tools are only possible through EMIF because of the combination of different types of expertise. These tools and pooled datasets have enabled EMIF-AD to perform large meta-analyses for a fraction of the cost and in a fraction of time that it would have taken otherwise.
The EMIF-AD Catalogue.
The EFPIA lead of EMIF-Metabolic, Dawn Waterworth, says that “without EMIF, we would not have been able to access electronic medical record (EMR) databases in Europe to run parallel analyses in four databases. We also were able to bring together a multi-disciplinary team that had the appropriate skills and knowledge to take advantage of this data, which would have been very difficult otherwise.” According to Ulf Smith, public lead of EMIF-Metabolic, having scientists with complementary expertise work with a carefully phenotyped cohort has been crucial in validating the importance of novel biomarkers.
EMIF has achieved its mission of developing common technical and governance solutions and facilitating access and use of healthcare data. “EMIF is helping to liberate insights within Alzheimer’s disease, metabolic diseases and other co-morbidities, from within the project through a series of use cases, and externally via the Catalogue,” says Nigel. The EMIF-AD Catalogue has over 400 registered users, clearly indicating that it meets a strong need.
There remain challenges, especially around governance and provenance of local data and external engagement with stakeholders. All of the topic leads agree that sustainability is essential to continuing EMIF’s mission and to leverage the full potential of EMIF.
The topic leads agree that the time spent building communities within the three areas of EMIF (Platform, AD, and Metabolic) was very valuable. In EMIF-AD, the topic leads are happy with how the teams achieved deliverables clearly and transparently, created cohorts and infrastructures, and re-used existing data. The EMIF-Metabolic topic leads say that their projects have led to some great collaborations, and they are pleased with the access to complementary expertise and large cohorts with disease outcome data.
For future similar projects, the topic leads have some aspirations. One of the keys to building a federated data network for real world, observational research in Europe is gaining an understanding of data sources’ local governance and provenance requirements. Future projects should also focus more on stakeholders, communicating the opportunity and possible rewards to partner organizations and data providers. This would allow better utilization and access of both data and infrastructure. For a long-term project like EMIF, topic leads would advise EFPIA partners to hire people specifically to work on the project, to make sure that they have time available for it.
All the topic leads agree that the EMIF project resulted in new high-level collaborations that they hope will continue beyond the five-year span of the project. In EMIF-Platform, there was collaboration with the OHDSI community around the OMOP common data model. Within EMIF-AD, there was linkage between, for example, European Prevention of Alzheimer’s Dementia Consortium (IMI EPAD), Dementias Platform UK (DPUK), et al, as well as the wider Alzheimer’s research community. In EMIF-Metabolic, there were collaborations with a number of EFPIA and academic participants that will continue into the future.
Johannes says, “Collaborations within existing partnerships have been strengthened and will be strong drivers of further research initiatives. New partnerships have been built and will hopefully further develop.” Nigel suggests that such developments are already taking place: “I believe that EMIF cemented relationships between academia, industry/EFPIA partners, and data sources in a pre-competitive state. I would suggest that this paved the way for the forthcoming European Health Data and Evidence Network (EHDEN) IMI2 project.” It will be exciting to see the legacy of EMIF in action in the form of projects such as EHDEN.
Bridging Europe: The First Annual OHDSI Symposium
Over 200 participants attended the first annual European Observational Health Data Sciences and Informatics (OHDSI) symposium, hosted in Rotterdam, The Netherlands, by the EU OHDSI chapter.
During the first day of the symposium, presenters spoke about the adoption of the Observational Medical Outcomes Partnership (OMOP) common data model (CDM), the need for an EU OHDSI initiative, the importance of OHDSI tools, and more.
During the second day of the symposium, two workshops gave participants hands-on experience with the OMOP CDM and OHDSI tools.
PRAET,PhDLife Science consultant at CMAST bvba, subcontractor at Janssen Pharmaceutica
RIJNBEEK, PhDAssistant Professor at Erasmus University Medical Center
Importance of the OMOP CDM
George Hripcsak (Vivian Beaumont Allen Professor and Chair, Biomedical Informatics, Columbia University Medical Center; Director, Medical Information Services, NewYork-Presbyterian Hospital) was the first speaker of the day. He presented about the worldwide journey of OHDSI and the adoption of the Observational Medical Outcomes Partnership (OMOP) common data model (CDM). Currently, over 400 million patients and 1.26 billion patient records have been mapped to the OMOP CDM. Among its users are large institutes like the Food and Drug Administration (FDA), the Electronic Medical Records and Genomics Network (eMERGE), the National Institutes of Health (NIH), and the Innovative Medicines Initiative (IMI).
Peter Rijnbeek (Assistant Professor, Department of Medical Informatics, Erasmus MC Rotterdam) elaborated on the need for an EU OHDSI initiative. There are many different database structures and coding systems in Europe, and healthcare systems differ considerably across EU countries. Harmonization to the Standardized Vocabulary of the OMOP CDM is critical to the generation of real world evidence in a fast and reproducible way. The current “one study – one script” approach is not scalable and not transparent; moreover, it is expensive, slow, and prohibitive to non-expert use. The EU OHDSI initiative will stimulate the adoption of the OMOP CDM in Europe, providing a platform for European collaboration to further improve the CDM and vocabularies for the European setting.
Patrick Ryan (Sr. Director and Head, Epidemiology Analytics, Janssen R&D; Adjunct Assistant Professor of Biomedical Informatics, Columbia University) gave the last talk of the first session.
The last talk of the first session was given by Patrick Ryan (Sr. Director and Head, Epidemiology Analytics, Janssen R&D; Adjunct Assistant Professor of Biomedical Informatics, Columbia University). “We want our science to be reproducible by others, which is even harder than being able to repeat your own results,” said Patrick. If we want our science to become reproducible, we need to ensure that every step along the research process is identical. This is exactly why the OMOP CDM and OHDSI tools are so important. They are the product of a community working together to define community standards and tools, share evidence, and define best practices.
Adoption of the OMOP CDM
After the coffee break, different cases of the adoption of the OMOP CDM were shown. Nigel Hughes (Scientific director, JCI Patient Data for Research, Janssen, Belgium) presented the upcoming IMI European Health Data and Evidence Network (EHDEN) project, which has set the goal of mapping 100 million health records across the EU to the OMOP CDM in the next five years. While this goal is ambitious, Nigel Hughes expressed confidence that they could reach it. EHDEN is not a blank canvas, but builds on what has already been developed in other projects (like, for example, EMIF).
Martin Sedlmayr (Professor, Institute for Medical Informatics and Biometry, TU Dresden, Germany) presented about the Medical Informatics for Research and Care in University Medicine (MIRACUM) consortium.
Martin Sedlmayr (Professor, Institute for Medical Informatics and Biometry, TU Dresden, Germany) introduced the Medical Informatics for Research And Care in University Medicine (MIRACUM) consortium. Recently launched and supported by the German government, the MIRACUM consortium currently represents a quarter of all German university hospitals. It aims to stimulate the reuse of data by adopting the OMOP CDM and the OHDSI tools.
Ben Hughes (Senior Vice President Real-World & Analytics Solutions, IQVIA) discussed why IQVIA, formerly Quintiles IMS, is continuously investing in the OMOP CDM. IQVIA helps many large-scale institutions (such as FDA and the European Medicines Agency) to answer both simple and complex questions about drug safety, adverse-event reporting, and more. Following a trial where 12 databases were harmonized to the OMOP CDM, IQVIA is convinced of the OMOP’s huge potential to help with better research, but IQVIA also remains realistic about the associated costs and time requirements.
Rae Woong Park (Professor, Ajou University School of Medicine, South Korea) presented the journey of how South Korea successfully adopted the OMOP CDM. There are currently many participating hospitals in South Korea with a total of 55 million patients. Despite the efforts of Professor Park, who had been promoting the Korean OHDSI initiative, it took more than 18 months before the first hospitals expressed interest in joining the Korean OHDSI initiative. This change came about in May 2016, due to the Korean National Health Insurance system which started using the OMOP CDM. From this story, we can learn that we need to get EU payers and regulators on board in supporting the EU OHDSI initiative if we want to be successful.
Daniel Prieto (Associate Professor and NIHR Clinician Scientist, Oxford University, UK) introduced the work of his group at Oxford, showing the audience how epidemiological work using patient records can help to study the risks and benefits of medical devices.
Erica Voss (Associate Director, Epidemiology Analytics, Janssen R&D, US) gave an overview of the type of research done at Janssen epidemiology analytics. She explained how real-world data can be used to conduct clinical trial feasibility studies, where you design your trial and then use real world data to validate if your trial is feasible or needs tweaking. She mentioned that “adoption of the OMOP CDM has required a substantial investment of time and money, but it has a very strong return. The more you invest, the more you get out of it.”
Sir Simon Lovestone (Professor of Translational Neuroscience, Oxford University, UK) spoke about the UK Clinical Records Interactive Search (UK-CRIS) system, which holds data from over 2.5 million mental health patients. Most of the data from mental health patients is written down in long text, which poses a tremendous challenge when harmonizing data. Currently, UK-CRIS is implementing a consent for re-contact model where 74% of patients agreed to be contacted again, which will tremendously speed up clinical trials.
OMOP CDM Tools in Action
After lunch, there was time for several demonstrations of tools related to the OMOP CDM and 40 research posters. The demonstrations included a patient-level prediction package in R, the ARACHNE research network platform, new ATLAS features, the AEGIS tool, and the EMIF Catalogue. The latter was presented by José Luís Oliveira (Associate Professor, University of Aveiro; Vice Director of Institute of Electronics and Informatics Engineering of Aveiro) who showed the audience the complete EMIF Catalogue pipeline from discovery, to assessment, all the way to the re-use of health data.
Prof. Dr. José Luís Oliveira gave a real-time demonstrations of the EMIF Catalogue, an online platform which integrates and enables the discovery and reuse of heterogeneous biomedical databases.
During a panel discussion, Jim Slattery (Senior Statistician, European Medicines Agency) mentioned that regulators use observational data more and more to guide decision making. Christian Reich (Vice President of Real World insights, IQVIA) stated that “the lack of reproducible research on which to base decisions is a problem no company, institution, or government can solve on their own, no matter how rich they are.” When asked how the OHDSI community can make sure it is compliant with the new EU General Data Protection Regulation (GDPR), Michel van Speybroeck (Director Data Sciences, Janssen, Belgium) replied that the exact interpretation of the GDPR is debatable, and that these new rules and regulations might be interpreted differently at different levels. However, as the OHDSI approach is to use a federated data network (versus centralized), it will be much easier to be compliant with the new GDPR.
Peter Rijnbeek (Assistant Professor, Department of Medical Informatics, Erasmus MC Rotterdam) closed the first day of the symposium with a hilarious show about the closures of the past OHDSI symposia in the US, including a singing section greatly appreciated by participants.
Hands-on Experience with OHDSI Tools
The second day of the conference hosted two workshops, allowing participants to get some hands-on experience with the OMOP CDM and OHDSI tools. By the end of the day, participants had a better understanding of the details of the OMOP CDM structure and the Standardized Vocabularies used to harmonize data. The tool ecosystem of the OHDSI community was also shown, with a strong focus on the ATLAS tool, as this allows browsing vocabularies, defining cohorts, estimating risk effects, making patient-level predictions, and more.
For more information about the OMOP CDM and the OHDSI community, and to stay up to date about the next edition of this symposium, please visit OHDSI Europe.