Wanted: A platform for big data projects

The ability for researchers to share data from multiple sources is a basic requirement for the big data projects that are shaping the future of healthcare delivery across Europe. The timing of the emergence of OHDSI’s OMOP common data model could hardly be better.


The value of cross-institutional and healthcare system-wide big data research in translational and clinical medicine is clear. Finding a data model that meets scientists’ needs and fits standardised clinical record architectures, however, has been a major hindrance in the quest for a more streamlined approach in the ongoing quest to improve patient care. 

The obvious solution is an open data model that provides researchers from different institutions with communication and semantic standards, allowing them to share data that has previously been locked into their own databases.

Enter OMOP, a common data model developed by the Observational Health Data Sciences and Informatics (OHDSI, pronounced odyssey) consortium in the US, which is gaining traction in Europe thanks to its adoption by some large projects. 

The first annual European OHDSI symposium was held at the Erasmus University Medical Centre (MC) in Rotterdam in March, bringing together more than 200 scientists, data custodians, regulators and pharmaceutical companies to focus on the benefits of a standardised data model for the future of epidemiological research.

“In Europe we are currently in a perfect storm for the OMOP Common Data Model (CDM),” says OHDSI collaborator Peter Rijnbeek, from the Department of Medical Informatics at Erasmus MC. “There is large interest from many stakeholders to make this important transition to improve the interoperability of the data. Many data sources across Europe are in the process of mapping to the standard already, and many more will follow in the upcoming years.”

Rijnbeek describes the OMOP-CDM as a ‘federated’ approach: the code is brought to the data, which remains local and secure, and only the aggregated results are shared.

“To do this, you want to use the same code on all the data sources that participate in your study,” he says. “The problem is that in Europe we are using many different coding systems to store our data, and every data source has invented the ‘best’ data structure. We need to speak the same language and use the same tables to store our data, so we can perform much easier using common tools. It is the same as the adapter you use for your mobile phone when you travel across Europe. I need many to make my phone work everywhere – we do not want that for our research.”

Rijnbeek says the OMOP-CDM itself is just one piece of the puzzle that can enable transparent and fully reproducible research. Its deeper value will ultimately lie in the powerful tools developed by the OHDSI community to perform data characterisation, develop prediction models and perform risk estimation of side effects. These will enable evidence generation on an unprecedented scale, he suggests: “It is a big step forward if we can contribute to this at a European level.”

‘A network of collaborative researchers’

According to another OHDSI collaborator, Patrick Ryan, Senior Director of Epidemiology and Head of Epidemiology Analytics at Janssen Research and Development, this international community can have a major impact on the speed of research.
“One of the challenges every research group faces with observational data is going through data captured for clinical care purposes in order to draw insights about the general population,” he says.

“Data management is cumbersome, reiterative analysis by teams of people and it can take many months or even years to get from question to answer. OHDSI’s approach is not just about establishing a standard. It creates a network of collaborative researchers using common tools, harmonising the way they think about data, and answering questions in hours or days.”

Ryan says the data model can be the foundation for new opportunities to provide better care. When data is captured for a primary purpose, the value accrued to the patient is immediate. A common data model such as OMOP extends the range of that value to subsequent patients: the accumulation of patient experience has the capacity to create a new depth of knowledge across the research community.

“Most questions patients deserve answers to don’t currently have adequate responses,” he says. “This is a model where patients and clinicians can direct the research that needs to be done, and get answers en masse. It’s a paradigm shift from conventional research because it defies the status quo of territorial individuals.”

The will to start generating evidence at a faster pace is demonstrated by a number of large European projects that have selected OMOP as their data model, with governments increasingly keen to invest in an initiative that initially emerged from the academic research community and was developed in a public/private partnership in the US.

The European Medical Information Framework (EMIF), for example, mapped 10 European data sources, including the Health Informatics Network in the UK, to the OMOP-CDM. By the time it concluded in May, it had achieved significant technological advancements and tool development among 57 partners from 14 countries. These have enabled important new research, most notably in the fields of Alzheimer’s disease and metabolic complications of obesity.

Standardising patient records across Europe

In November, another five-year project, the European Health Data and Evidence Network (EHDEN) will begin with the mission to provide a new paradigm for the discovery and analysis of health data in Europe. It will build a large-scale, sustainable, federated network of data sources standardised to the OMOP-CDM. A large (€17m) harmonisation fund will be made available through open calls for data sources, to enable the translation of their data to the standard.

The goal, explains Peter Rijnbeek, is to standardise 100 million European patient records to the OMOP-CDM. A network of SMEs will be trained and certified across Europe to support the mapping process.

Rijnbeek says that while there are still optimisation steps to be made – for example, certain European drugs need to be added to the Standardised Vocabularies – the momentum to apply the OMOP-CDM to future research is strong.

“The OHDSI community is a highly motivated and very experienced group that is building fascinating tools for research to improve patient care,” he says. “I am convinced that because of projects like EHDEN, we will be able to make the European OHDSI community a strong and active group that will contribute to the global OHDSI goals.”  

Piers Ford

Related News

Hyland is addressing the clinical blind spot

Digitising patient information is a step in the right direction, but siloed repositories of digital information are not the desired outcome

Beyond EMRs: Health intelligence systems

EHR, patient care

‘We are not talking about replacing monolithic systems, but complementing them’