Data is the "new oil.” Huge amounts of value, societal and financial, can be derived from information about people. Advanced data processing could create a much-improved health system—allowing for quicker services, better diagnoses and the discovery of new drugs. Last month, the NHS announced plans to transform how it extracts a particularly valuable oil: demographic, health and social data on up to 55m people, recorded from the first day an individual interacts with the NHS (usually on the day they are born). A predictable backlash has forced a two-month postponement.
The NHS is right to delay its plans. It should now revise its approach by putting in place better safeguards to ensure people’s data is processed securely and transparently, with benefits shared widely and fairly.
The proposed new system will extract data from GP practices in England and share them with NHS agencies and third parties, such as research institutions and businesses. The scale is huge. It will pool detailed health and social data on 55m people into a new centralised database, which would be made accessible in various forms to a broad tent of organisations for vaguely defined “research and planning” purposes.
Initial plans only gave people until 23rd June to go through a rather cumbersome, and little advertised, opt-out process if they wished to prevent their health records being extracted.
NHS Digital already extracts some data from GP practices. The new service will improve the quantity, quality and speed of this data flow. It has been developed under the direction of Matt Hancock, who is seeking to put onto a permanent footing some of the useful data-sharing measures enacted during the pandemic.
The potential benefits of data sharing are indeed enormous. For example, relaxed rules during the pandemic helped researchers at the University of Oxford discover that the drug dexamethasone improves the chances of survival of patients hospitalised with Covid-19. This discovery is estimated to have saved one million lives worldwide. In the future, better use of patient data will transform healthcare. It will reduce errors, such as wrong prescriptions and misdiagnoses, both of which are still too common. And it will deepen our knowledge of diseases, not least the still poorly understood long Covid.
However, this promise can only be realised if the public trusts the data-sharing process. To achieve this, NHS Digital needs to improve in three areas: privacy, purpose and public value.
A much stronger emphasis on privacy is paramount. Under the current plan, data is “de-identified” before it is sent to third parties, meaning individuals’ names and other key markers are removed. But without water-tight safeguards, large firms with broad data capabilities could easily re-identify who people are. More must be done to guarantee privacy. Instead of data being exported straight to third parties, it should first be handled and processed in a trusted research environment governed by rules developed through regular public consultation.
The question of who gets access is bound up with purpose, what they are using that data for, and why, and here there is an absence of clarity and transparency. Existing proposals are not clear about precisely who will get access to the new GP database in the future, on what grounds and for what purposes. People may feel comfortable with academic researchers using their data to discover new drugs, but less comfortable with commercial researchers assisting NHS service planning while also making a profit and developing market advantages. For example, could multinational companies, operating in advertising and insurance markets, use NHS data to develop algorithms that predict future health conditions for certain social and demographic groups?
At the moment, the public cannot even see when and where algorithms are informing NHS decision-making. The NHS has been clear that it employs tech firms for a wide range of its operations, including giving them “unprecedented access” to patient data during the pandemic. But there is no straightforward way for the public to inquire how their data is being used by these companies, what inferences are being made, or even for experts to audit the algorithms. Without better oversight and regulation, people may not trust NHS Digital with their data. But with greater transparency, researchers and businesses could innovate safely and consensually to improve our health.
Finally, NHS Digital needs to ensure that the public have a stake in the financial gains derived from their data. Big tech companies are the most valuable in the world because they capitalised on the benefits of big data. NHS databases contain immense commercial value. That is why several large corporations have offered their services to the NHS for free. But there is no agreement on how the royalties developed from public data processing will be shared with the public, for example through windfall taxes or forms of collective data ownership.
The concern and confusion ignited by NHS Digital’s data extraction plans is a sign of inadequate transparency and public participation in the process. A useful comparison is care.data, a 2013 GP data-sharing initiative that was expected to transform healthcare for the better, but ultimately was scrapped because of inadequate public engagement in its development.
The NHS has not demonstrated it has learnt lessons from care.data. Leaflets and posters placed around GP practices, at a time when many are not attending appointments in person, do not amount to appropriate public engagement for the new proposals. The opacity and uncertainty around third party NHS data deals does not engender trustworthiness.
History risks repeating itself, unless NHS Digital can address concerns around privacy, purpose and public value. That is a tall order to complete in just two months.
The NHS is at a fork in the road. Hidden in its data are enormous benefits for patients and society. But the proposed data-sharing plans have been rushed without important democratic checks and balances. If data is the new oil, we should better regulate its use—before it’s too late.
