Nigel Hughes: A Non-Identifiable Data Layer On Top of Clinical Systems That Retain Memory May Be The Future of European Health Data

Nigel Hughes has a thirty-six year career spanning the NHS in the UK (16 years), NGOs and patient organisations (10 years) and within the pharmaceutical industry (18 years). He has worked clinically in HIV and viral hepatitis, liver disease, and in sales & marketing, medical affairs, market access and health economics, R&D, precision medicine, advanced diagnostics, health IT and Real World Data/Real World Medicine. His experience covers clinical, education, as an advisor, consulting, communications and lobbying over the years. He is currently the Project Lead for the IMI2 European Health Data & Evidence Network (EHDEN), and was Platform Co-Lead for the IMI1 European Medical Information Framework (EMIF), as well as consulting on numerous projects and programmes in the domain of RWD/RWE. We talk today with Nigel Hughes about future of European health data:

Looking back, how would you describe the journey to the current state of European health data infrastructure? Where are we now?

In common with all regions and countries, our health data infrastructure is reflective of how we all organically developed systems over prior systems, apart from perhaps such countries as Estonia, who were afforded the opportunity to start afresh after the end of the Soviet Union.

No systems were developed de novo to meet current and future needs, but were adaptations of platforms and technologies to meet new needs, such as electronic health records (EHRs) being formerly administrative and billing systems. Furthermore, effectively every institution, district, country, and region have their own systems, platforms, and technologies, developed iteratively over time.

We of course also must contend with multiple human languages across Europe, but multiple machine languages, with a lack of incentives for many, including vendors, to support interoperability. Several projects and programmes, such as within the Innovative Medicines Initiative (IMI) have been addressing this, creating platforms that can facilitate interoperable data for research purposes. In particular, we have seen considerable enhancement of our understanding of our own biology, the massive increase of multimodal data of increasing complexity, so-called, ‘big data,’ and technologies like federated/machine learning, so-called ‘artificial intelligence (AI),’ which are avaricious in their need for data themselves.

So, there are technological challenges, but also political, as health is derogated within the EU Charter to Member States, and apparatus like the EU Commission can only operate cross-border or in the gaps as it were between the Member States. Movement to an European Health Union is perhaps a challenging one politically. More complicated, there is significant heterogeneity across Member States regarding healthcare provision, resources and digitalisation, impacting efforts to harmonise Europe-wide. This is a situation replicated in multiple countries and across regions.

In Europe, recently there has been more focus on research collaborations, as well as cross-border portability of summary data, and a common EHR exchange format, but the most significant move has been initiation of a legislative pathway for a European Health Data Space (EHDS) to support the use of clinical data (EHDS1) and research (EHDS2), linked to relevant legislation such as the General Data Protection Regulation (GDPR) and Data Governance Act (DGA).

Due to the pandemic, policymakers, healthcare providers, and public health are well aware of the data not being in the right place at the right time to answer the right questions, a situation that is the case for all diseases, not just COVID-19. This may well raise the impetus for an interoperable Europe, maybe even a Health Union.

What are the European opportunities or strong sides with regards to health data infrastructure and to health data, in general?

To date Europe has been an ocean of data, but a desert for analysis, hence we have an opportunity to rectify this via interoperability of data, standardisation of tools and methods for interoperability of analysis, but above all open science collaboration.

The European region has a population of ~640 million people, across EU Member States and associated countries, representing enormous diversity across peoples, health status and disease, care settings, and PEST considerations. This is one of the largest regional populations outside of e.g., India or China, and possibly with the most internal differences.

With the right technical framework, considerate of ELSI, and open science collaborative community across Europe, the region could be a hotbed of scientific enquiry on healthcare research down to a molecular level.

Which areas of health data collection, management, and use are likely to change most in the near future, in your view, and why?

Paradoxically, more complex data is perhaps being addressed more readily with reference to data collection and management than basic data contained in our medical records, but both are on a positive trajectory.

When discussing real world data, that is data derived from, e.g., our clinical care, this is not being collected for the primary reason for research, but in reference to the management of our care. Healthcare workers are not thinking about research use of personal health data when they are attempting to resuscitate a patient in an emergency room.

I would not imagine a large-scale change in this situation in the near future. Though there are improving systems and platforms, mass adoption remains a challenge internationally, and the majority of countries still have basic data capture capabilities. It will take not only financial resources, but also fundamental changes to a clinical workflow established for a century or more, as well as the political will of governments.

Future data governance models in healthcare and how they will be implemented infrastructure-wise, how would you describe them? What is lacking in the current ones?

This is a vexed area and is the key challenge vs. technical aspects. Unless we can resolve the central basis for secondary use of primary clinical and other data, it will remain so. Certainly, in Europe the focus has been on the role of consent, alongside for instance the stipulations and rights enshrined in the General Data Protection Regulation (GDPR).

Ironically, from a researcher’s perspective the identity of an individual is not necessary, we would need only be able to substantiate that they are an individual (e.g., for record-linkage). Meanwhile, this is precluded by the challenge of an ongoing debate on anonymous, anonymised, pseudonymous, etc.

In a model that is not focused on a need for consent (which is not scalable, and will introduce bias), but with optimal protections, such as privacy by design inherent in e.g., federated networks, we need to address a balance between benefits of data reuse against perceived risk, and actual harms.

A sizable proportion of data breaches are due to cybersecurity challenges to healthcare providers, though this does not negate the responsibility of researchers in ethical conduct and use of data, and ought to be a focus.

How do you see future infrastructure for an efficient health data exchange? What new possibilities will end users have?

Ultimately data exchange should support a quid pro quo for data reuse, with reference to patients being able to benefit from insights into e.g., their disease management and care, or treatment outcomes, while researchers should be able to derive new insights into that disease, or the R&D to treat and/or cure it. Meanwhile, clinicians should be able to benefit from portability of data in reference to decision-making.

Technologies exist already to support this, albeit built on top of archaic architectures. It would of course have been more helpful for this all to be designed from the outset, rather than sticking plaster initiatives to fix gaps in the infrastructure. For the most part the challenges with data exchange have been motivational and governance, and this is what needs to be tackled more vociferously if technical solutions are to be employed.

You are a scientific director; however, this question is more about policy shifts. What new policies towards health data, or changes in the current ones are desirable to address the challenges effectively?

A critical policy response is to rebalance the debate on societal benefits vs. perceived risk and/or actual harms. The pendulum has swung too far today in terms of risk-based debate, and with insufficient focus on the benefits we would see from large-scale data reuse.

It is a truism I believe that people are harmed, or indeed die every day, because we cannot derive optimal insights into how to prevent, treat or cure diseases. More than 6,000 rare diseases alone have no adequate response.

The concept of precision or personalised medicine remains science fiction for the masses that cannot be realised without the reuse of data. We suffer from our inability to have a meaningful societal debate about this in the open, and with the focus on risk, we have policy responses that err on draconian or punitive, or at least restrictive.

Ironically, legislative work around ‘AI’ is usually too lax considering the very real risks to individuals and society from this development path.

Your vision of future European health data infrastructure, please describe it. What are currently the main threats to this vision, and what are our strong sides in order to achieve it?

Very simply put, an egalitarian layer of non-identifiable data that is seen as a societal asset across the research domain, on top of clinical systems that retain memory, but are agnostic to location and time for clinical decision-making. Enshrined within federate data networks, and with equitable access by patients, care providers and researchers, aided by machine learning and transparent AI to interpret and ensure security.

Ultimately, Delphic Health Oracles that can interpret the vast troves of health data, supporting and augmenting human decision-making (but not replacing it) are achievable at a global, multiregional level. Critically this needs to be seen in the context of a learning health system(s), constantly learning, updating, and optimising decision-making for the best outcomes for patients, individually and at a population level.

There is though a lack of political vision to ensure this can be realised, as governments internationally have consistently failed in being able to manage a necessary paradigm shift from treatment to prevention, and emphasising cure, versus chronicity. A new public health agenda would need to address this also across the planet, and not just in individual care institutions, districts, countries, or regions.

We live in a global ‘hospital’ not just a village, but just as we see at a local hospital, the global one also does not ensure interoperability of communications and outcomes.