The overarching theme of the EHDEN virtual roadshow was the importance of strengthening Europe´s ability to generate, collect, and collate high-quality observational health data in the context of an increasing willingness on the part of regulators to factor in real-world evidence (RWE) into decision-making on the effectiveness of new biopharmaceutical products. As a cornerstone of its RWE gathering approach, EHDEN relies on the Observational Health Data Sciences and Informatics (OHDSI) network of over 300 institutional collaborators with access to data on nearly 600 million patients worldwide.
At the opening of the roadshow, OHDSI was briefly described by George Hripcsak, Chair of Columbia University´s Department of Biomedical Informatics, which hosts OHDSI´s global Coordinating Center. OHDSI´s contribution to the world of health research is its proprietary software-as-a-service Observational Medical Outcomes Partnership (OMOP) Common Data Model, which is transforming the way health data gets accessed and utilised with minimal disruption to data owners.
In a podcast leading up to the event, Matt Wiener, Director of Informatics and Predictive Science at Bristol-Myers Squibb, who is involved in one of the EHDEN work packages, walked listeners through what OMOP is: a data infrastructure that enables EHDEN´s “data translators” – a cohort of small and medium enterprises specialised in data analytics – to access and convert data from disparate observational databases to a standardised format and vocabulary. It lets them query the data and perform research of interest locally, so that administrative, EHR, or patient-reported data never leave the organisation where they are collected.
Kees van Bochove, founder of the open source software company The Hyve, which supports EHDEN in FAIR-ifying the data it works with, added colour to EHDEN’s description:
This capacity for crunching and analysing data locally is crucial for gaining stakeholders’ trust and buy-in, as EHDEN seeks to expand European healthcare organisations´ participation in the project. It currently has 65 participating data partners, 25 of which joined after submitting expressions of interest to a COVID-19-specific call EHDEN published last April (and were selected among a total of 75 applicants).
Against this background, Xavier Kurz, Head of Surveillance and Epidemiology Service at the European Medicines Agency (EMA), noted that EHDEN´s use of OMOP is important from a regulatory perspective of assessing the validity of data sources and the methodologies used to generate RWE in different circumstances and scenarios. He explained that while regulators are increasingly open to factoring in such evidence in addition to traditional RCTs, being able to evaluate the validity of the data source is a critical element of their work – a consideration EHDEN has embedded into its approval process for each new data partner, essentially guaranteeing the provenance of the data.
To illustrate the utility of the OMOP data model from a public health perspective, Dani Alhambra-Prieto, Professor of Pharmaco- and Device Epidemiology at the University of Oxford as well as EHDEN Research Coordinator and Lead for Work Package 1 on Evidence Workflow Development, said the routinely collected health data provided by EHDEN´s COVID-19 data partners has been used to track what happens to people who were infected or hospitalised with the virus. Based on that information, over time and with more data coming in, EHDEN aims to develop a prediction model that can anticipate what is likely to happen to patients with given clinical characteristics and social demographics, and eventually to perform causal inference on a population level.
“We´ve had a lot of bad COVID-related decision-making, both internationally and nationally, because we were not using observational evidence and everybody was waiting for some sort of massive RCT in relation to some of the decisions that had to be taken, and that at the policy level is not useful,” said Rafael Bengoa, Director of SI-Health, a health consultancy firm in Bilbao, and former Health Minister of Spain´s Basque Region. Providing a country-level perspective, he added that beyond the challenges of obtaining RWE at a European level, lack of interoperability is also a significant problem for Spain, whose national health system decision-making is devolved to the 17 autonomous regions. Therefore, EHDEN´s data collection approach with its built-in privacy protections could be useful for accessing real-world data without threatening regional and institutional data ownership, Bengoa suggested.
Indeed, the “train analysis,” privacy-by-design nature of EHDEN’s approach to collecting health data could be key to overcoming some of the institutional resistance to data sharing that researchers often report encountering. In the words of Hripcsak, “OHDSI is a voluntary network, so we ask people to participate and they participate if they so wish. That´s made them comfortable and that´s how we´ve been able to aggregate such a large dataset. The laws for data sharing between Taiwan and South Korea are very different and we can accommodate all the different cultures of privacy by enabling [organisations] to control their own data.”
This sentiment was shared by Prieto-Alhambra, who said that privacy-masking features such as those built into OMOP should help guide European health data policy going forward:
Anchoring the discussion around RWE curation, Raquel Yotti, Director of Spain´s Instituto de Salud Carlos III, reminded everyone that in the desire to advance on the data front, technology companies and decision-makers should not overlook the most precious asset of all: human resources. “It´s very important to strengthen scientific and technical capabilities within public health departments at the regional and central level, including at our institution, because we need talent and public health is an area that has not been considered enough in the last decade,” said Yotti.