Protecting people while using their health data for research, a framework for understanding relative responsibilities and roles

Introduction & why this paper was written

A publication in Frontiers in Big Data was posted on 16 September, 2022: “A concentric circles view of health data relations facilitates understanding of sociotechnical challenges for learning health systems and the role of federated data networks”. The authors set out a model of data systems that they call “Concentric Circles View” (CCV) of data relationships that is meant to help researchers and others associated with using health data to understand how to balance protection of patients and the value in using their health data for research.

The aim was to enable a consistent understanding of the fit between the local relationships within which real world data (RWD) are produced and the extended systems that enable their use. This can help understand and tackle challenges associated with the use of real world data in the health setting by improving understanding of not only how but why federated data networks (FDNs) may be well placed to meet the evolving needs of health research.

This article is meant to offer non-experts an overview on some of the historical issues facing health data-sharing and provide a short and easy-to-understand summary of the full paper.


Every day across Europe people are not receiving optimal healthcare because data that can be used to generate evidence is not being used due to the social and technical challenges of finding it, accessing it, and then analysing it. One difficulty that prevents this data from being shared for research is that it exists in multiple languages, systems and structures with challenging policy restrictions and technology considerations. But there is a solution to harmonise all this data: the OMOP Common Data Model. When data sources have their data mapped to this standard, it opens the door to shared research on an unprecedented scale and speed. What used to take years and months, can now be done in weeks and even days.

This large-scale use of RWD is central to hopes for learning health systems. Learning health systems are systems in which knowledge generation processes are embedded in daily practice to improve healthcare. However, while the OMOP Common Data model is helping to tackle many of the technical issues, such as health data security and data format incompatibility, there are also social, ethical or legal issues, including how to ensure adequate informed consent for data use, how to maintain public trust in data systems and how to meet legal and regulatory requirements such as GDPR in Europe.

Using the CCV to conceptualise data relations


The CCV approach aims to show that the contexts in which data are collected and used, and the relationship between these contexts and the person, can be characterised by their social, ethical, legal, and technical qualities and provides a model for thinking about these together. It suggests that how we think about the challenges of data use flows from the relationships associated with data, and that these relationships can be understood as a series of concentric circles. Within each circle, these relationships are relatively stable, but they change as one moves through the circles. For example, relationships closer to the data subject may tend to prioritise security over sharing, be based primarily on duties of confidentiality and relations of trust and involve localised systems for sharing health information. These duties, relations, and systems and the access controls associated with them change as one moves outwards into other circles. This change is both quantitative, in terms of intensity or scope, and qualitative, in the nature of the data and controls, and differences in duty and trust.

Overall, as one moves outwards through the model and becomes further “removed” from the data subject, data become less detailed and less easily identifiable. The model helps to show, however, how and why making data less detailed requires work. It also shows the work associated with moving “inwards” from an outer circle to a position closer to the data subject. This work to match the conditions of an inner circle is both social and technical and may involve building systems for privacy protection associated with more identifiable data, obtaining direct consent from the data subject and establishing closer or even personal relationships of trust. The paper illustrates these challenges through examples of trust, informed consent, and the nature and role of public and patient involvement.

In summary, an awareness of the qualities of a data context, the authors suggest, can help to better understand the work involved in moving or sharing data, and capture the value of frameworks that maintain the sociotechnical integrity of these contexts, while allowing these data to be accessed to achieve the maximal clinical and societal value – in other words, by translating into improved treatment and outcomes for patients.

The CCV and the promise of federated data networks

The use of the CCV model helps us to understand the possibilities associated with federated data networks (FDNs) in terms of their potential to enable data use without violating local norms, values, and governance arrangements, and without requiring undue work that changes the position of use within the CCV and in relation to the data subject. A federated data network is a network model in which a number of separate data sources share data via a central management system that follows consistent rules and regulations. It ensures anonymous patient data stays at the source, rather than being stored in a central data repository. By design, FDNs are well-positioned to meet the challenges of enabling access to differently-located data, diverse data sets for research and healthcare system improvement – datasets that exist, in the CCV model, in different data contexts. As a result, FDNs are well placed to meet challenges related to both the social and technical qualities of a data context.

In summary, the management of health data relationships has evolved significantly over the last decades. There are, however, still technical and social issues that pose challenges. Raising awareness among people, regulatory authorities, researchers and data owners about these issues and challenges is needed to help move data-sharing forward. To this end, the CCV model combined with the federated data network model provide a clearer understanding of strengths and weaknesses, and in so doing, has the potential to foster more trust and support of data-sharing at an unprecedented scale.

If you would like to read the paper in its entirety, it’s available here in Frontiers in Big Data.

The following short videos for non-experts are also available:


You can also register at the EHDEN Academy for access to the above videos or a range of others that have been designed for anyone working in the domain of real world data and real world evidence.