We were joined for a recent HTN Now panel discussion focusing on the move from reactive to proactive care by experts from across the sector, including Dan Bunstone, clinical director at Warrington ICB; Patrick Denston, PCN digital transformation and change manager at Frimley ICB; Pete Thomas, CCIO and executive director of digital development at Moorfields Eye Hospital; and Joseph Waller, director at Aire Logic.
The session looked at how NHS organisations are beginning to make the transition to proactive care, the kinds of data and digital tools required to make the change, and the impacts beginning to be seen on patient care, outcomes, and operational pressures. We also asked our panel to consider what the future model of proactive care might look like, what it should involve, and the steps that still need to be taken to get there.
“I’m a GP by trade,” shared Dan, “as well as clinical director for Warrington Innovation Network, and I chair NHS Confed’s Data and Digital Group. I’m super interested in proactive care, lifestyle, medicine, and digital, looking at how those things can interact together to make a real impact on patients and general practice.”
“I’m a founder and director at Aire Logic,” Joseph said, “and I’ve spent 20 plus years in healthcare technology. I’m just a humble techie, but have always worked in healthcare, particularly at the national level, but also quite extensively in trusts.”
Pete told us about his clinical position as a consultant paediatric ophthalmologist at Moorfields, and how he had also had “a very heavy technology and digital side” to his career as well, working for IBM in research and development, and building a research career specialising in extended reality and AI. “At Moorfields, I lead on things like our ophthalmic EPR, and I have national leadership roles around digital in eyecare,” he continued. “Recently I became founder and chief executive of a spinout involving Moorfields, UCL and Topcon Health, which looks to bring through products on the research we’ve done, including the work we’ve done with Google DeepMind.”
For our final introduction we came to Patrick, who shared details of his current role as PCN digital change manager. “I’ve got a keen interest in how we can harness the power of digital to improve population health, population behaviour and change management,” he said, “and I’ve spent the last few years working on how you can make most of your care scheduled and proactive rather than reactive in an acute setting.”
Current position and examples of projects in the proactive care space
Joseph talked about the importance of proactive care in helping the health system deal with mounting pressures relating to an ageing population, funding, and so on. “It feels like a lot of the things we’ve talked about as having potential over the last ten years are becoming possible,” he considered, “and I think that will make the next few years very interesting. I know when I worked on the Data Processing Service for quite a few years with NHSE, every year people would be asking when we would start using AI in that space, and now it’s finally feeling like that’s going to be possible at national scale.”
Dan gave an example of work with the ICB in Warrington which saw his PCN and surgery focusing on more proactively managing patients with known hypertension using an app. “It wasn’t about trying to find new people who we didn’t know about; it was around better managing and proactively treating those we did know about, and getting to really tight targets,” he explained. It involved reaching out to patients, looking at things like increasing compliance with current medication, lifestyle changes, and running an AF screening programme. “The app helped us find an additional 25 patients with AF, and we treated around 1,200 patients, reducing their risk of heart attack and stroke by 3 to 4 percent through reducing blood pressure by ten to 12 points,” he continued.
Patients loved it, according to Dan. “They were really impressed with the remote bit, and the fact we were flipping it so that rather than them having to call us and let us know something is wrong, we were calling them to have those discussions. That’s a really big success story, and we ran it with a completely remote team, so none of our clinical or admin team were based in Warrington at all.”
Frimley has a population health platform that takes in secondary care data, 111 data and primary care data, Patrick told us, allowing extraction at practice level to look at trends. “It’s currently going through a project looking for patients that might have CKD that’s never been coded, hypertension or diabetes,” he said, “and it’s looking at where they might have had an EGFR in secondary care and one in primary care, and the two have never been married up in actually having a CKD diagnosis.” It similarly looked at weight measurements that may have been taken in one setting but never followed up with a HbA1c, and hypertension where a similar lack of follow up is identified.
“It’s then going forward and getting those four-day, seven-day readings, and doing a full package of care to get patients on to lifestyle changes, like Dan mentioned, and not just coding them as having a long term condition,” Patrick explained. “It’s been very useful, and the idea is that we can upscale that to other conditions and case find from across the system, rather than just from primary care data.”
“What we’re doing at the moment that works really well is moving toward a digital or virtual hospital model,” Pete shared, “which is rather than forcing patients to come in and see us, trying to deliver that care to wherever the patient has been.” This model is supported by a video consultation service which allows patients to be put through directly to a Moorfields ophthalmologist for a consult via A&E without booking, he went on, “and we’re approaching about 100,000 patients there, 78 percent of whom never need to come in for a physical appointment, and 50 percent of whom we manage entirely remotely”.
Moorfields also has contracts with North Central London ICB and half of North East London to get involved early in an enhanced triage and an advice and guidance role, allowing them to manage 40 percent of patients having imaging of the back of their eye, without the need for a hospital visit. “To make that service really accessible, we’re allowing optometrists on the high street to get advice about patients early, allowing us to intervene much earlier,” he added.
Pete also shared details of the trust’s collaboration with Google DeepMind. “That has led to algorithms that are as good as one of our best ophthalmologists in diagnosing diseases at the back of the eye,” he told us, “and some of the work we’re doing now is converting that into products via our new spin-out, with the intention that those can run in optometry practice on the high street and ultimately pick up really early signs of disease.” Looking ahead, the team are beginning to publish research showing how signs of neurodegenerative diseases like Alzheimer’s and Parkinson’s can be picked up ” a decade before they become clinically detectable”, Pete said, “and that’s where the field is moving in general”.
Aire Logic has seen a lot of growth and investment in proactive care nationally, Joseph told us, “including in the Digital Prevention Services, which is where our vaccination programme sits, and in areas such as health checks”. Aire Logic places a large focus on innovation and not-for-profit projects, according to Joseph, “so we tend to fund projects which are good ideas for patients that wouldn’t get money from traditional routes”. One of those projects is the Lancaster Model, which assesses hundreds of thousands of pupils to predict difficulties with health or behaviour, he went on, “so that’s a really proactive area that is going to be impacted by AI as we start to use that across the data, as well”.
Agreeing with Dan and Pete’s emphasis on the remote aspect of proactive care and the importance of patients being empowered to take control of their own care, Joseph talked about another of Aire Logic’s projects, Lifelight, which allows patients to take vital signs from their phone using an app. “We’ve been looking at trying to extend that to allow blood pressure and heart rate to be taken through a webcam,” he noted.
Challenges around the move to proactive care
One of the challenges with providing this digital and/or remote model is the risk of excluding non-digital patients, Patrick considered. “It’s important to highlight, though, that getting 90 percent done using digital methods frees up a lot of time for the remaining 10 percent – it’s the challenge of allowing people to have time to resolve and embrace that ideal of change, and to think maybe that 10 percent will become eight or nine percent in the next year or two.”
Pete spoke from his experience with some of Moorfield’s projects. “Specifically for eyecare, the challenge is that a lot of the opportunity to be proactive and intervene early is based around high street optometry,” he said, “but that isn’t part of the NHS. Practices will have contracts with the NHS to provide sight tests and to refer, but they’re not part of the NHS in the sense that they have access to the Spine so they can look up or identify patients, or so we can see what appointments they have available.” This can create a silo, Pete explained, which in turn leads to challenges with linking all of that valuable data that would support early intervention together.
The second challenge Pete identified is operational and around the design of services, “trying to run digital services from organisations that were never designed to run digital services, which is a big cultural change”. Whilst he considers that Moorfields has been “quite successful at doing that, in some ways”, some of the fundamental building blocks such as EPRs aren’t always well suited to managing patients on remote pathways. “The ways we’re tackling it is that the digital hospital model is run slightly separately from our other services, with a big focus on customer relationship management with the optometrists using a platform to track engagement. We also look for educational opportunities, with the purpose being building relationships which will support us to break down data silos.”
Pete’s team has spent time looking at what a good product would look like for running these services, he continued, “and we’re using a platform called CrossCover for our digital hospital, which is based on the idea of clinical pathways, with branching decision trees and decision support in there”. Ultimately, it’s about a mixture of finding the right tools and the right ways of working within an organisation to support this approach, he told us, “as well as building relationships with the people you need to so you can intervene early”.
From a technical perspective, Joseph started out by reiterating Patrick’s point about digital technology making time for those who aren’t familiar with digital. “The other thing we’ve noticed is that in some areas technology serves underprivileged areas better than you would think – the data from the NHS App shows that it’s been really successful in some of those areas, and that people might be happier to engage with technology than with a clinician in some cases.”
A product created by Aire Innovate, Aire Logic’s sister company, has been designed with data sharing and patient-facing services in mind, Joseph noted. “It’s called Aire Suite and it uses a concept of building blocks when creating an EPR, so those can be reused. In a way, as a clinician is building their patient record system, they’re accidentally building the system for patients too, because those building blocks are reused in the patient-facing service – they’re building up a library of building blocks which can gather data about the patient record, that can then be easily reused for unanticipated purposes or other settings.”
The focus for his team has been on finding ways to create systems that are better at sharing data, without having a single purpose in mind, according to Joseph. “These days, if you use an EPR, it’s in the cloud and it’s only the way it’s designed that stops it being instantly usable by the patient. If you build all of that in from the start, then it’s a natural side effect of building the clinician system that you’ve got an online and accessible system out of the box.”
Dan talked about how many of the challenges are deep rooted and longstanding. “We’re beyond the point of trying to work out whether proactive care is a good return on investment – you get between 10 and 20 times your return on investment, which is super important for a health system that is deeply entrenched in debt.” The challenge comes when trying to prove that return in-year, he continued, “and using the example of hypertension, the payout is probably many years from now – one of the ways we got around that was running two parallel projects, one looking at COPD and reducing admissions to free up money for the system. Let’s say you save £100,000 this year, and then ring fence that to proactively invest into next year for your hypertension work.”
Long-term commitment to do something not as a one-year pilot, but as a project that can run beyond that, is another facet of this challenge, Dan evaluated. “Then the reality is that as you blend from traditional primary care, treating disease and illness, into public health, proactive care inhabits a void where nobody really is, and it can feel like it’s everybody else’s responsibility. The pushback is those saying it’s a public health responsibility, which maybe it is, but we walk in a continuum where I will treat disease, public health will do a whole host of things around exercise and lifestyle, but in that middle bit we both work together, and that synergy could be fabulous, but we haven’t achieved it yet.”
Scaling proactive care interventions
Patrick shared with us some of his experience with scaling in the proactive care space, noting his “fortunate position” to put things in place at a practice or PCN level, or to suggest it at an ICB level. “Frimley is incredibly supportive, with a connected care team that are using the Microsoft Power apps to pull that data into one place and have population health platforms that are very advanced,” he explained. “From a micro level within the surgery, you could ask what the point would be in reducing ED admissions, because what they’re worried about are the patients who are calling up or coming in.”
What happens, though, is that in making that proactive switch, reactive care becomes “very predictable”, Patrick continued. “Over the last two years, we can see we had a variation previously of 180 percent in on-the-day demand, and now we have a variation of 10 percent. If you can predict your acute needs, you can then organise your capacity and capabilities to match that, which frees up more time, so it ultimately becomes almost self-sustaining.” Another effect of this is a happier workforce and patient population, he considered, “as you have GPs that have the time to actually see the patients, and who can finish at five and go home to their family, along with patients who are seen four or five times over the years for things like bloods and education, who feel cared for and looked after”. Seeing patients more regularly means advice on lifestyle changes has a bigger impact as a result of more regular check ins.
“As Dan was saying, we do have to talk about finances, but what we find now is that we have a much more manageable and flexible workforce that can change and do other bits, so we try and get our long term conditions managed as quickly as we can within the year, get 90 percent of our QOF work done by September,” Patrick shared. “Then we have three months where we can upscale, work on other proactive care models, or even start working on next year’s QOF.” With things like COPD, managing the condition in the summer months when it’s quieter and when patients are less likely to have an exacerbation, means patients are less likely to need that in the winter, he went on, and ensuring measurements of weight, blood pressure and other information ready for clinicians when they walk into an appointment “makes their time more efficient, as well”, he added.
“Traditionally, eye referrals tend to start from the high street and go through the GP, which means losing all information about where a patient is coming from, as well as the ability to communicate directly with the referrer,” Pete reflected. “As we move toward direct referrals, like covering around 2.5 million patients in North London where referrals flow directly through us, we open the opportunity to have those conversations right at the start.” This is helping to prevent a situation where 45 percent of patients would come for an appointment and be discharged because they didn’t really need to be there, he added. “We’re also starting to do deprivation mapping across our referral flow, understanding the spread of services and getting that data to tell us more about the current status of primary care for eyes.”
Interoperability, integration, and more
Joseph talked about Aire Logic’s work on EPRs and building EPRs for trusts, as well as the challenges that being tied into a contract with a specific EPR vendor can pose for organisations looking to introduce new technologies or solutions. “There are pros and cons to both approaches, but obviously if you’ve bought a big EPR, you are somewhat tied to that vendor – there’s theoretically nothing stopping you going elsewhere, but the organisation has spent so much money that they need to see that return on investment.” At the same time, it’s important to look further afield at some of the great things being done by smaller startups, he considered, “and I think we need to be careful not to lose that potential”.
Trusts choosing to go with hybrid models and smaller vendors might have more freedom to integrate AI and other technologies, Joseph continued. “I think it’s worth noting that a lot of the AI we’re working with has solved discrete problems, but those small startups aren’t necessarily in a place where they can fit them into a pathway – we’re looking at how we can make sure that’s holistically integrated into a patient pathway.”
A lot of integration and interoperability takes place “in name only”, observed Pete. “Vendors will show you all of these interfaces they have, and all their capabilities to interoperate, but actually translating that into a real integration is sometimes impossible,” he said. Even where data has been successfully integrated, there remain questions about whether a workflow exists within an EPR to allow it to be surfaced in a way that it can be useful. Real world interventions “end up being very time intensive and costly”, he went on, “and I think there’s a real need for people who are developing clinical AI to actually ask themselves how the decision they’re supporting fits into the entire workflow, because if you’ve got a workflow that requires ten decisions to be made and you’ve automated one of them, you really haven’t achieved anything, you’ve just moved the bottleneck one bit down the production line”.
“Something we see a lot as part of our integration work is that we’re often asked for support after products have been purchased, and there tends to be a habit of looking at the requirements and seeing HL7, and thinking that’s enough, without realising that it doesn’t necessarily mean something will be compatible,” Joseph shared. “We try and emphasise to our customers the importance of not just thinking about functionality from a clinical perspective, but to look at whether something will be good at sharing data.” This is something Aire Logic considers actively when building its own product systems, he told us, which are designed to inherently share data and have open APIs. “I’d encourage any organisation to think hard before signing a contract, to make sure vendors definitely share the data and it’s not just lip service.”
Dan agreed with Joseph’s points around the challenge with being tied-in with a vendor, saying: “We’re 20 to 30 years behind retail with things like order tracking, but they’ve had the imperative to do that, we haven’t. There’s a lot we can learn – yes, we’ve got the GDPR and the DPIAs and all the things that complicate that, but we can’t keep using that as a blocker, and we have to start working a lot more like retail to open those doors.”
The role of AI and emerging technologies
“Like Dan and Pete have said, it’s about using the information we already have, integrating it, and not constantly having to find workarounds to extract it or use it,” Patrick agreed. “The platform we have at Frimley is a treasure trove, and I remember being overwhelmed when I first got access to it – it felt like there were so many things we needed to do, and there was this responsibility that once you’ve seen it, you can’t unsee it. You know there might be 1,000 undiagnosed hypertensives, but even if you’re not the person that’s ordered or gathered that data, it’s where you find that responsibility and where you work it in.”
The key has been focusing on one small bit at a time, Patrick went on, “not trying to tackle the whole of population health in one go”. There is a big fear, however, once you do get large amounts of data, about who is responsible for it, he shared, “and it’s having that maturity and looking beyond that, seeing having the data and understanding it as a positive thing, and looking at what we can do to fix things or provide a long term benefit”. It might be that you find an extra 1,000 diabetics, he considered, “but does that then move the bottleneck and create a shortage of diabetic preventative drugs? You can’t fix things too quickly, you’ve just got to keep going on that journey and evolving as you go along.”
Describing the current situation in primary care as a “wild west” of new and exciting things when it comes to technology and available solutions, Patrick also moved on to talk about the importance of working with suppliers and letting them know what’s needed. “If you have those sensible conversations, often they’re grateful and will make the changes you suggest,” he said. “I’ve spoken to a few suppliers about AI scribes, and about how GPs don’t want to spend their day copying and pasting, that needs to seamlessly go across, so we need those partnerships with scribes, communication software, your telephony and so on.” There’s potential for AI to be used in surfacing historical data to look at trends in population health, he shared, “and from that we can make much larger models, predicting problems that could occur in 10, 25 or even 100 years”.
The future of proactive care
Contemplating the future of proactive care, Pete highlighted the importance of getting the foundations right, ensuring things like integrations are in place, and exploring the potential of partnerships. “My colleagues on the research side spent a lot of time building a research environment called INSIGHT, which contains about 35 million ophthalmic images with lots of associated metadata,” he shared, “which is incredible to train, test and validate algorithms. So part of it is having the capability to develop those, but you also end up with a massive implementation gap, publishing boatloads of papers with realistically no chance of ever getting through to clinical impact.”
Identifying things at an early stage and bringing them into an environment where they can be developed as a medical product is integral, Pete went on, “and I think that partnerships are very much the solution”. For Cascader, the spinout Moorfields, UCL and Topcon Health have been working on, this approach has been key, he added. “Moorfields have the data and the clinical expertise, UCL has the AI expertise, and Topcon Health give us access to imaging devices and cloud based imaging platforms. It’s about forming the right partnerships, and about getting that data across – I think digital scribes are where we’ll see it first.”
The NHS is good at doing things on a small scale, considered Dan, but isn’t as “brave” when it comes to going beyond that. The ideal proactive care system would involve creating a “virtuous cycle”, he explained, “so you better manage people with hypertension, prevent some heart attacks and strokes, save £150,000, then reinvest it to do this on a larger scale”. It would then be possible to move on to diabetes and other conditions, Dan explained, “until eventually you get to the point where the amount of money being spent on proactive care becomes the norm – it’s not proactive care any more, it’s just care”.
Education and government policy need to be part of that change, Patrick said, “and your transport policy, your education policy, have to have a health element in them, to be working together, and to get that education in place around how your data is being used”. Passing ownership onto patients and populations, and empowering them over their own health, giving them the tools and understanding they need to track it such as smartwatches and apps, is essential, he continued. “Even looking to the Lidl app and people’s shopping habits, how many coffees they are buying in Starbucks – if we can start getting all the digital platforms to work together we can really start helping people change their entire lifestyle for health.”
Whilst colleagues in the health sector have been looking to the financial and retail sectors for years, Joseph said, “we’re all aware of the barriers, and we do a lot of work on information governance, DPIAs and things like that, as we figured out we couldn’t deliver the whole thing unless we delivered the whole thing”. That meant the commercial, legal and even clinical safety side became very important, he noted, “because you can only go so far with just the technology. I think if you give patients a greater sense of control, they see benefits and that can encourage adoption.”
We’d like to thank our panellists for taking the time to share their insight with us on this topic.
