Health data - how to sell sharing
Time to tell a positive story about using our health records for research
Consider these two questions and imagine what your response might have been if you’d been asked them in the summer of 2020:
Would you feel happy about your NHS health data being sold to a company that would aim to make a profit from it?
Would you feel happy about your NHS data being used to help in the search for a safe Covid-19 vaccine?
Now, I’m pretty confident that most people would say no to the first question and yes to the second but essentially they are being asked the same thing with a different spin. After all, the successful search for new vaccines was undertaken by companies whose shareholders expect a healthy return on their investment, even if in AstraZeneca’s case profits were postponed for a while.
What I’m trying to highlight is the marketing challenge involved in convincing the public of the merits of sharing their health data for research. It’s a subject I addressed in a speech to a conference held this week by Health Data Research UK, the government funded national institute for health data science.
Talking to an audience absolutely immersed in this subject - doctors, statisticians, pharmaceutical researchers and policymakers - I had a severe case of impostor syndrome. But as someone with two long-term health conditions who is eager to share his data and encourage others to do so in the interests of research, I had a message that seemed to resonate - we have a brilliant community of health data scientists in the UK but it needs to get a lot better at talking to the public.
Looming over everything are two great health policy disasters, the care.data fiasco where an attempt to collect GP health records in one central database was abandoned in 2014, and last year’s attempt to do something similar, this time confusingly called GPDPR (General Practice Data for Planning and Research), not to be confused with the EU’s GDPR data protection law.
The first the public knew of GPDPR was last May when stories emerged of a row about it being foisted on GPs and their patients without proper consultation, and it was then kicked into the long grass where it remains. I showed the conference audience the headline of a Guardian story where the scheme was described as an “NHS data grab”, using the term coined by a privacy campaigner. The PR battle around the scheme was lost before it began, and millions of people opted out of having their records shared, probably not having heard about the benefits in terms of better treatment that access to their data could bring.
But amidst this record of failure, there have been successes of more modest data sharing schemes. In fact, the conference was held to highlight the achievements of nine Health Data Research Hubs, established as centres of excellence to show the way data could be shared effectively and transparently with academic and commercial researchers.
Each hub promised access to data from a particular area, from cancer to respiratory diseases to mental health. Each had involved patients on panels which reviewed the way they collected and used data. I was particularly interested in the Insight hub which focuses on eye health, getting much of its imaging and clinical data from University Hospitals Birmingham and from London’s Moorfields Hospital.
As a long-term customer of Moorfields, I’m very pleased to see my data employed in this way, but patients can opt out of having theirs used if they wish. Hospital data has long been available for research - it’s the pooling of GP records that has caused so much controversy.
But the other thing that I learned at the conference is that there is another big scheme that brings together GP data and has not attracted such division. It’s called the Clinical Practice Research Datalink and it collects data from, among other sources, a network of GPs across the UK. The key difference is that it’s voluntary - GPs choose to join it, whereas they were all going to be obliged to hand over their records to care.data or GPDPR. About a quarter of all GP practices have signed up to it.
All these health data schemes suffer, however, when public anxiety about how records will be used rises. Patient opt outs from CPRD, which had totalled around 1,000, soared to 75,000 at the height of the care.data brouhaha.
One speaker at this week’s conference had a clear message about the potential benefits of data sharing and the danger posed by mass opt-outs.
Dr Janet Valentine, who ran CPRD from 2015 until last autumn overseeing a period of major growth, now has the job of championing health data research at UK Research and Innovation. After a speech calling for lessons to be learned, both from the failure of care.data and the successes of CPRD and the data research hubs, she told me about the urgency of getting a message to the public: “We need to have a proper conversation about sharing data. People need to understand the benefits and values of doing it and what happens if there's opt out.”
She explained very clearly the following:
The benefits of getting it right - the regulator the MHRA ‘would not have been able to analyse the safety of the COVID vaccines without our data.”
What happened before we had this kind of data sharing - “the reason everyone was so slow to respond to thalidomide was that there was no centralised organisation that brought together all that data to say, oh, there might be something wrong with this drug.”
And the biases that were introduced when people opted out - “no one has explained to people the absolute importance of staying in the system, because if they want to know the drugs they're taking are safe, they need to be there. If they're from a minority group, they particularly need to be there because we need to understand their health.”
Dr Valentine says it is possible to change people’s minds, relating how she explained to a meeting that the fact that many of those opting out of data sharing were middle-aged women meant attempts to study any link between HRT and breast cancer could be hampered. “A woman jumped up and said, ‘Oh, I've opted out. I'm going to opt back in and tell all my friends to as well.’”
It strikes me that there are two main fears spurring people to opt out - concern about government surveillance and the idea that valuable and sensitive data is being handed to untrustworthy ‘Big Pharma’ and ‘Big Tech’ companies.
I had felt that the surveillance fears were overdone - it’s hard to see a scenario in which the police or civil servants would go through the laborious task of deanonymising your health data to find out about you. Then again, last week’s US Supreme Court ruling on Roe v Wade has highlighted a plausible malicious use of data by law enforcement - in America at least - to pinpoint women who had had or were seeking abortions.
Similarly, that’s served to heighten distrust of big tech companies which already know so much about us and have been known to spot a pregnancy even before a woman is aware she’s expecting. I happen to be delighted that Google’s DeepMind, in collaboration with Dr Pearse Keane of Moorfields Hospital, has developed an algorithm to spot eye diseases after being given access to millions of scans. But I can see why some might wonder whether their data was safe with a Californian tech giant.
But as for “Big Pharma”, I think that term is a lazy cliche. Sure, there are rogue companies in the pharmaceutical industry and there are drugs that don’t work or have severe side-effects. But if we block their access to our health records, we are not only hindering the search for new drugs, we are making it less likely that failings in existing products will be found quickly.
In this country we have one huge advantage when it comes to health data and that is scale. For all its faults, its poor record in adopting new technology, and the risk aversion which hinders collaboration between different parts of the service, the NHS has one of the biggest data pools of any of the world’s health systems, collected over many years.
Now all we have to do is get a simple message across to the public - the benefits of sharing that data outweigh the risks.