WIREDGOV

Keynote by NHS Digital's CEO Sarah Wilkinson (delivered for Public Policy Projects on Saturday 10 October).

There are two significant areas in which we have seen much faster progress during this pandemic: the build of digital systems for health and care and the use of data within the health and care system.  

I want to share a few very short thoughts on the former and a few slightly longer thoughts on the latter.  

What is undeniable is that, across the NHS, digital products have been built and deployed at record pace this year.  

Within NHS Digital we have built a huge number of new services during this pandemic and modified and scaled many others.  

Some of our services became much more critical very quickly and usage escalated in a fairly radical way. The 111 Online system, for example, experienced a peak utilisation in April that was 95 times the highest ever peak utilisation. This was fantastic news in that we were able to take the load of the 111 phone lines in a way that we had always aspired to do but always struggled to achieve, but it was quite a shock technically to experience that level of demand escalate over a few days.  

We also:

• built, from scratch, a system to identify vulnerable patients in need of shielding 
• we made significant modifications to the electronic prescription systems 
• we have built the digital infrastructure for the new ‘Pillar 2’ testing facilities 
• we built new systems for predicting demand for ICU and ventilators 

• we built a huge number of new dashboards 
• we built new trusted research environments 
• we deployed Teams to 1.3m people 
• we deployed virtual smartcards 
• we upgraded network infrastructure in hundreds of places  

What made this work much easier, of course, was the urgency and focus associated with the situation which resulted in 3 significant shifts …  

1. Much clearer articulation of demand (none of what I call ‘vague visions’ which have been characteristic of some historical programs in this area)
2. Fewer parties involved 
3. Very different risk appetite: the focus was on delivering capability, even if slightly imperfect, with a stronger sense that bumps on the road would be dealt with in a pragmatic and collegiate way

These factors, taken together, have had a dramatic enabling impact.  

But I think what’s even more interesting, and has had an even greater impact during this pandemic, is the much greater and faster sharing of data within the health and care system.  

For example, there is a central NHS system called the Summary Care Record which is a centrally-accessible national record of the most critical aspects of each patient’s record including, for example, allergies and medications. As a result of temporary changes in government policy to respond to the pandemic, we were able to add Covid-related information into the summary care record system, as well as various other key health information that had only previously been shared into that central record when we had explicit patient consent, and we were able to make the system available to key players who were managing the response, in particular 111 call centres, ambulances and pharmacies.  

Another very different example: We were also able to provide data quickly into key medical research programs. For example, NHS Digital provided the data that underpinned the RECOVERY Trial, led by Oxford University, which has made a number of critical findings, in particular the finding that dexamethasone can cut the risk of death by a third for patients on ventilators and by a fifth for those on oxygen.  

Of course, beyond celebrating the impact of this data sharing the question that arises is … given the enormous benefits we have seen from data sharing, why was it so hard to share this data before Covid?  

Essentially there were two key factors making it harder:  

• The first is that there is significant anxiety about the extent to which the public trusts government and the NHS to use their health data.  
• The second is that the law which governs its use today is very complex, and in some areas, needs to be more fit for purpose to reflect how data is used in 2020 and beyond. 

Taking those items in turn:  

Public trust is obviously absolutely critical and we can’t deny the fact that there have been a number of historical initiatives (especially the care.data program in 2014) which have had a very detrimental impact on trust.  

Of course, in the middle of a global pandemic, the public are much more accepting of the fact that government must make judgements about such things without consultation and the law makes specific provision for data sharing in such circumstances, which we were able to lean on.  

Such usage must still be appropriate, proportionate and secure, of course, and within NHS Digital we worked intensely hard to ensure that was the case even when we were under great time pressure. We have also been very diligent about being completely transparent about usage.  

Pre-pandemic, I think we were struggling to find an effective way to consult with the public on health data usage, though various new approaches such as citizen juries were making good progress. It’s critical that as we move forward with these new modes of data sharing and use, which will now be in place for some time given the extended duration of the pandemic, we find effective ways to engage the public, listen to their feedback and ensure they continue to be supportive.  

And we need new legislation. There are too many aspects of the current legal framework around health data sharing that are extremely labyrinthine, very imprecise or require complex judgements.  

There are key matters that are just not clearly addressed or on which there are very different views and interpretations: for example, there are different views as to whether personal demographic data held within the system is confidential. What we know is that patients’ views on this, and clinician’s views, and government’s views, often don’t align and the law is interpreted differently by different communities.    

There are also some concepts and states of data that are poorly defined: For example, the law talks about anonymous and pseudonymised data as though those are precise privacy states, but the reality is that de-personalised health records exist on a privacy spectrum: they become increasingly re-identifiable depending on how many attributes are disclosed and what other data they can be linked with.  

And added to the challenges of interpreting the legal text is the sheer complexity of the health system, with its multitude of legal entities who are each in law individually responsible for their own data which, of course leads to a lot of overhead in interpreting who can share what data safely with whom, even within the system.  

So the two key messages I really want to convey are that:  

• We have a real opportunity now to have a dialogue with the public about how health data can be put to much greater use, safely and appropriately, to their individual and collective benefit, given they (and we) have now directly experienced the benefits and risks of these new modes of use.  
• And we need to ensure that as the pandemic resolves, and the more streamlined legislation currently available to us can no longer be leveraged, we have new, clearer legislation in place, which still respects patients’ rights and choices.