A study from Reform finds a haphazard patchwork of data-sharing being led at a local level. The think tank’s director of research Eleonora Harwich tells PublicTechnology why a national strategy is needed

“It’s a polarised debate where, on one side, people might see it as privatisation through the back door and, on the other side, you have… incentives of private sector companies that might not always fully align with the NHS,” says Eleonora Harwich, director of research at think tank Reform.

The debate in question concerns NHS data and how – and by whom – it is used. 

And, for many people, particularly the general public, it is a conversation that is often informed solely by those instances that make headlines – invariably unwelcome ones. The most obvious example being that of the Royal Free NHS Foundation Trust’s work with Google-owned artificial intelligence firm DeepMind.

Although the partnership between the trust and the company led to the creation of an app to help diagnose acute kidney injuries, if people know anything about the engagement, it is probably that the Information Commissioner’s Office ruled last year that the sharing of 1.6 million patients’ personal data was in breach of data-protection laws.

A new Reform report, Making NHS data work for everyone, sheds light on what is currently a very uneven landscape of partnerships in which public bodies and commercial companies work together to realise benefit from healthcare data. These partnerships are typically led by individual trusts acting in isolation, and without any guidance or operating framework provided from above.

“It might be happening in some trusts where they might be lucky enough that they have a team that have the commercial savvy… [and] legal expertise,” Harwich says. “There might be some centres of excellence where, because they’re next to a very good university that has a unit specialised in spin-outs of academia, they might have access to [expertise in] how you go about apportioning intellectual property, and how you go about doing more complex commercial structures.”

She adds: “But that’s not the sort of skill that is equally distributed within the system, and I don’t think that there is an of overarching strategy looking at commercial models.”

“What’s mostly happening on the ground is that trusts are entering in very simple data-sharing agreements where there’s… no clarity as to what the value exchange is.”

The report provides an overview of the current environment and, over the course of 10 recommendations, sets out a blueprint for how Reform believes the NHS, the private sector, and citizens can all benefit from data innovation.

The first of these suggestions is that “the Office for National Statistics should provide a framework and accounting standards for measuring and reporting the value of knowledge assets such as healthcare data”.

Harwich says that ascertaining the value of NHS is data is “a very complicated question” that is yet to be properly addressed.

“There’s no single definition of value because some people might look at it with a narrow definition… of what is it worth in terms of money? But that’s not the primary value of NHS data – the primary value is really about direct patient care, that’s the reason why it’s collected,” she says. “That is its first purpose within the NHS – to make sure that you can provide the best care that you possibly can with the best type of information… there’s immense direct value in human terms.” 

Harwich adds: “But I think it would be naive to not look at the potential commercial value that this has. And understanding what that might be is difficult, because it depends on the type of product, and it depends on the quality of the data.”

Maintaining trust
The second of Reform’s recommendations is that the Department of Health and Social Care should work alongside NHS organisations and commercial entities to engage in “dialogue with the public… at a local level to discuss commercial models”.

As it stands, NHS bodies enjoy more public trust in their storage and use of data than organisations from any other sector, according to recent research from the ICO. 

But Harwich warns that “trust is something that can be easily eroded”. She adds that, if it wants to ensure it can make effective use of its data, the NHS needs to ensure it gets a large number of patients from a diverse range of backgrounds to buy in to the benefits of data-sharing. 

Endemic gaps in data – perhaps related to certain genetic diseases, or conditions that disproportionately affect one ethnic group or demographic – could reinforce inequalities.

Reform’s 10 recommendations

• The ONS should provide a framework and accounting standards for measuring and reporting the value of knowledge assets 
   
• DHSC should ensure that a dialogue with the public is set at a local level to discuss commercial models 
   
• The NHS should create a register of data sharing agreements between the NHS and commercial organisations
   
• DHSC should include in its formal review of commercial partnerships a macroeconomic study of the impact that different partnership models might have
   
• DHSC should include the results of this study in a clear national strategy which should seek to optimise the value of data held by NHS organisations when it is accessed for commercial purposes
   
• NHS England should create a Data Quality Service, with a tiered-fee system, to provide bespoke reports on data quality at the early stages of a partnership
   
• Procurement rules should include an agreement by digital providers that data generated within clinical applications should also be freely available for and interoperable with clinical information at the patient level
   
• Health Data Research should develop the appropriate data governance structures to ensure that its Digital Innovation Hubs are safeguarding patient data.
   
• NHS organisations should offer synthetic datasets to share with private sector organisations at the early stages of the innovation process
   
• DHSC create a new independent unit with legal and business experts to help NHS organisations negotiate fair and proportionate partnerships

“Biases within healthcare data or medical data… are not a new problem,” she says. “One of the appealing things of using tech in healthcare is the potential scale at which you can treat people and the potential speed at which you can do things – but that also means that you have the potential to replicate and propagate bias at a scale that was not possible before. So, that’s why it is extremely important that people trust that all of this is done for their interests and that it’s not only a question of blind trust – but a question of trustworthiness, of being able to verify that [a system is trustworthy].”

The model for patient engagement put forward by the Reform report is the “proportionate governance” model used in Scotland, via its Public Benefit and Privacy Panel for Health. The panel, which assesses requests from external parties to access NHS Scotland data, operates a two-tier model.

The panel’s first tier, which is there to examine the technical and security ramifications of any request, is made up of a trio of information-governance professionals. If these people cannot agree on an approval or rejection of a request, it is referred to a secondary panel.

This second tier brings together a much broader range of people, including NHS leaders, as well as social care workers, government statisticians and, crucially, four representatives of the general public. 

This model could work well “at a regional level” across England, Harwich believes.

The third recommendation made by the think tank is that NHS England and NHS Digital “should create a register of data-sharing agreements between the NHS and commercial organisations”. This should include details on what data is being shared, and the type of partnership model adopted.

Partnership models
Reform has identified nine core models of partnering, the first of which is a “consortium agreement” in which an NHS organisation and a commercial enterprise or academic institution embark on a joint venture.

An “equity share” model would see the NHS take a minority stake in a private sector partner, while a “grant-funding model” would see public sector money invested in external research and development work.

The “golden share” model is a joint venture in which the public-sector entity holds a guaranteed equity share – or majority voting rights. “Intellectual property sharing” would see the NHS retain ownership of its data, with private sector firms required to pay a licensing fee to use it.

In an “open-access” partnership, both parties would agree to make the results of research openly available, while a “profit- or revenue-sharing” model would see the NHS receive ongoing annuity payments for any money made from products and services developed via the use of its data.

A “royalty payment” engagement would see the two parties agree on “a fixed percentage of profits from a specific asset or a fixed price per unit that is sold”.

The final model, “local asset-backed vehicles”, is predicated on a private-sector partner providing resources of expertise and time equal to the value of the asset being provided by the NHS. 

The fourth of Reform’s recommendations is that the DHSC – alongside the Crown Commercial Service, Office for Life Sciences, and HM Treasury – should conduct a review of existing commercial partnership models that includes “a macroeconomic study of the impact that different partnership models might have”. This is necessary to “avoid reinforcing a postcode lottery or other negative externalities such as exclusive data access”, the report says.

The results of this study should be included in a national strategy “to optimise the value of data held by NHS organisations when it is accessed for commercial purposes”, says the report’s fifth recommendation. 

For trusts who are approached by potential commercial partners, there is “no guidance” from the top as to potential commercial models and their relative merits, according to Harwich.

“What’s mostly happening on the ground is trusts are entering in very simple data-sharing agreements where there’s… no clarity as to what the value exchange is [between the two parties],” she says.

Typology of potential commercial models

• Consortium agreements
• Equity shares
• Grant funding models
• Golden share
• Intellectual property sharing
• Open access
• Revenue- or profit-sharing
• Royalty payment
• Local asset-backed vehicles

Reform recommends that NHS England “should create a ‘Data Quality Service’, with a tiered-fee system dependent on factors such as company size and global profits, to provide bespoke reports on data quality at the early stages of a partnership”. 

The think tank also advocates that government procurement rules should be amended to ensure that private-sector firms that create clinical applications make the data generated via their products freely available, and interoperable with other systems and patient records.

The report’s eighth recommendation is that Health Data Research UK should work with the NHS and the National Data Guardian to ensure that its Digital Innovation Hub programme has “the appropriate data governance structures”.

The penultimate suggestion put forward by Reform is that the NHS should release synthetic data sets – which are cloned replicas of real data sets, but with personally identifiable information stripped out to protect anonymity – to private sector partners “at the early stages of the innovation process”. 

Harwich says that using synthetic data would expedite the process of ascertaining the potential for innovation. As it stands, commercial organisations often end up dedicating “70% of the time of a project” to just cleaning the data to a state where it is fit for use, before any real examination of its potential value can begin.

The Reform research director stresses that, even for use of synthetic data, governance and approval measures would still be enforced before access was granted to a commercial partner.

“But at least this would allow them to play around with it a bit and to test the product… before it is validated on actual patient data,” she says. “You can have a real discussion about ‘this is where we see the potential, and here is how much market value this could generate’. So then both parties understand what they’re talking about, and you can have a bit more information as to what might happen in the future and have a little bit more of an idea of how much would the NHS data be worth as a contribution to that innovation.”

The final recommendation of the report is that the department of health and social care “should invest in creating a new independent unit with legal and business experts to help NHS organisations negotiate fair and proportionate partnerships”. 

Such an entity – which would likely be comprised largely of lawyers – should operate as an independent body with a similar standing to a regulator, and charged with examining the legal detail of contracts and ensuring equity for both parties.

“The primary value of data is really about direct patient care… but I think it would be naive to not look at the potential commercial value that this has.”

“You want to make sure that it is as impartial as possible and not taking either the side of the NHS, or the side of the private sector, and that it tries to get a fair deal for both organisations. There would obviously be a framework of principles, [to ensure] that you would never interact with the NHS if you can’t prove that your innovation or product actually leads to patient benefits… but so long as that is tangible, then you can have discussions around the type of commercial models that you might want to have.”

She adds: “Once you prove that it’s for patient benefit, you can be a bit more impartial – and should be, because the private sector is still investing lots of money and taking on a huge amount of risk to develop these things.”

Many of the themes touched on in the Reform report also featured in the Life Sciences Sector Deal 2 published by the government last week. As the topic of NHS data innovation comes in sharper focus over the coming months, Harwich says that the think tank hopes that its work can help “a frame for this conversation, or at least a stepping stone for different bits and elements can be further explored by different policymakers”.

Such a polarised environment could, perhaps, do with a few more stepping stones to help bring the two sides together.

Click here to read the full Reform report ‘Making NHS data work for everyone’