For our latest thought leadership webinar, we welcomed a panel of experts for a discussion on taking a data-driven approach to operational improvements, day-to-day management, and proactive care.
To bring a variety of expert insights and viewpoints to the discussion, we were joined by Stephen Slough, chief digital information officer at Dorset ICS; Paul Charnley, digital lead at Healthy Wirral Partners; and Esther Ocrah, director at GE HealthCare Command Centre Europe.
Stephen: I’m the CDIO for the NHS Dorset; I’m the accountable officer for digital technology and intelligence across the system, including the NHS organisation, the local authorities, and the voluntary sector. In terms of our model and the way we’re working, we’re looking at how we can get the best out of all those components for the population that we serve.
Paul: I work and live on the Wirral, I’m the digital lead coordinating between all the relevant organisations, trying to create a joined-up, place-level digital service for our population. Previously I was the chief digital information officer for Cheshire and Merseyside ICS. For a couple of days a week I’m contracted to work for NHS England, where I’m helping to steer the digital blueprinting programme. We are looking to see if we can gather data and evidence, so we can share that with other people in the hopes of crossing bridges faster and avoiding pitfalls.
Esther: I work for GE HealthCare Command Centre, and I’m the European lead for business development and delivery. At GE, we have a proprietary software that provides real-time information as well as predictive information for patient care coordination. That looks like a suite of applications that tell you what is happening now, what can you expect to happen tomorrow, and in the near future. In our team, we also have change management experts who are available to lead and support our customers with their process improvement, where required and the overall change management to ensure that the software is adopted and well-implemented into their everyday working routine.
Current use of data and programmes
Stephen: I started in the NHS about seven years ago – when I started here, they were using data quite well and there was lots of reporting happening, but we weren’t really getting the value out of it that we should be. It’s about asking the right questions. Where do we get the insights and intelligence out of the data to know if we are doing the right things? Are the services we are providing delivering the outcomes that we expect?
We started a project called the Intelligent Working Programme, which aimed to gather data from all the different partners and put it all into one place, turning the data into insights for the clinicians. It was clinically-led but implemented by the analysts that were on the ground here. We retrained and unskilled lots of people on how to do that, and that’s gone on from there. We use that for our population health management, our operational intelligence. It’s been a long journey, and it’s not over – you’ve just got to keep going. As technology improves we can do more and more with this stuff, draw more insights out, make it more intuitive.
Our ambition locally is to get more of the data about our population appropriately anonymised and made available to the public, so that when we’re talking about what we’re trying to achieve and how we’ll need them to work in partnership, they can see why. Collectively, we can try and move the needle on some of these indicators for our population to help them get more healthy life years.
Esther: I’ll use an example from the NHS. We’ve been working with Bradford Teaching Hospitals for a few years now. When we started, like many trusts they were having issues with congesting in ED, mismatch between capacity and demand, patients not being discharged on time and issues around standardising processes. As part of the programme, they envisioned a central function, a command centre; this was to be a department or a physical space with co-located teams who could make use of real time insights and analytics our technology provides, which would support them in making decisions.
For example, the patient placement team sits in the command centre. With our application they can see every single patient who is trying to access Bradford Teaching Hospital, be it from the emergency department, from a GP, or transferring over to another department. They have a holistic view of processes across the hospital, including status of discharge, so they are better placed to make decisions about patient placement. They’re efficient because they have that information at their fingertips, in real-time. If you have to question whether the information you’re seeing is a day old or even 30 minutes old, your decision-making may already be compromised.
In addition to the real-time application, we helped them develop a digital twin of their hospital. They used this to test out scenarios before implementing them, for example around bed mix and capacity planning. As winter approaches, they can ask the model questions like, how many beds they need to open, whether any need closing, which specialties need extra help.
We also do a lot around change management and the process improvement that goes along with that, but from a data perspective, that’s the angle we take.
Paul: The Wirral began as an NHS vanguard in 2016, to start to pull together data from our solutions into an operational shared care record and also into a population health management solution. From this, we’ve been able to supply a fairly rich 360-degree view of our population, to allow us to analyse chronic disease, gaps in care, and the system-wide care for things like diabetes, COPD and asthma.
It’s challenging going from working as a place and developing things locally to working as an ICS. We need to do a lot of work on our infrastructure.
I’m very interested in Esther’s command centre concept – I think all ICSs are trying to create more coordination across the systems. We’re not there yet, but we have looked at Kent and Medway’s coordination centre and their blueprinting arrangements around that.
Whatever we do at operational level, real-time access to information on patient flow is a different challenge to the one that we are facing with population health management platforms. It’s taking slightly slower to get to that point where we can see where the ambulances should be going, for example. That kind of intelligence is going to be critical to balancing the system, not just at the acute emergency department end, but also in terms of coordinating the discharge and flow of patients.
Harnessing data to deal with immediate priorities and pressures
Stephen: The best example of this recently would probably be COVID. We had our local platform, we were getting it going quite well – lots of good data flows coming through and enabling decision-making. But I was having to find money from national sources every year, to pay for the team and the infrastructure. Then COVID hit, and very quickly ICSs needed to have an overview of what was going on operationally, updating as close to real-time as possible, from multiple different data sources. So the platform rapidly transformed and took us to a place where it was drawing the data to find out the latest situation – how many staff were off, where those staff were off, how many ventilators we had, whether we had enough PPE, the mortuary capacity, all of those things. And that’s on top of the test status, results, vaccine results and numbers.
When we reached the more proactive phase of managing the pandemic with vaccines, we created more pages that looked at who is eligible, where they live, the uptake, their demographics. We looked at some protected characteristics which allowed us to target specific communities within our population if the uptake was low, to identify why and try to dispel any myths that were appearing. That saw a massive take-off in the number of different analytic cases that we are asked to use and create, to support clinicians in what they’re doing. It was likely the best moment in a horrible situation, because people realised that what we’ve been saying about the capability of data.
We’re using data to identify rising risk and where we might see patients beginning to present because of characteristics that they share with other patients who have also gone on to present. We’re using it with proactive management – we’ve plugged in a number of digital tools around cardiology, asthma, COPD, so we can take readings from the devices in use. We can see if they’re being used at all, for example, and get in touch with people to make sure that they know how to use it if they aren’t. If they’re not getting the benefit, why not? Is there an element of education that’s needed to support them? Do they need a bit more care? We’re trying to drive more self-management through our public engagement groups too. We try to look at the right tools that will actually help people, but also we’re defining, explaining, providing in a way that makes them accessible. It’s no good having all of this technology if nobody can use it.
Esther: Situational awareness, often times all you need is visibility of what is happening. It can be as simple as number of available beds, that’s a good place to start. Which facility has an ICU bed open, who has an oncology bed open in the region? So many hospitals are crowded with patients when there may be some capacity in a hospital or care facility down the road. But currently, the only way you get that information is by ringing around, and by the time you get through to the right person, the information is out-of-date.
Then there’s visibility over capacity for services, looking at which hospitals are on divert for ambulances, for example, and visibility over equipment; who has ventilators?
I have a good example of where we’ve done this in the state of Oregon during the first weeks of COVID. We helped them develop a web-based application that allows them to track this information within every single hospital in the state. Currently we’re up to 62 hospitals, sharing non-protected patient information data around things like bed capacity, demand and equipment availability. So that was a very good example of how COVID really moved the discussion forward around using data.
Coming onto managing the inbound-outbound patient flow. Within our patient manager application, we have implemented a criteria based algorithm that identifies candidate patients for specific pathways and/or protocols. One example is the algorithm that identifies patients who are candidates for a discharge lounge. The algorithm scans patients in real-time to identify who is expected to go home today and if they’re clinically ready. So if you’ve got a good candidate, you can free up the bed, move them to a discharge lounge and have another patient come in to that bed. Another example of this in use could be on a particular ward; maybe you have five patients who are potential discharges for today, but they are waiting for physiotherapy. Having access to that information at your fingertips means that you can then work with the physio department and prioritise their workload. That can have a real impact on patient flow.
Additionally, within any NHS hospital, we tend to start our day with a bed meeting, which often requires staff running around, capturing the information they need so that they can go and present that information at the meeting for decisions to be made. Bradford has advanced on that; they have an application that pulls all that real-time information. On every ward they can see the occupancy, available beds, whether they’re clean or dirty, how many discharges are planned for today, how many patients are being transferred in or out – they have all that information in a very simple overview. They have reduced the time that they spend on those meetings quite significantly through use of this application, from about 30 minutes to maybe 10 minutes. It’s easier to focus when you have visibility on the delays or pressure areas.
Paul: For us, in terms of immediate priorities, elective recovery is one of the things that’s very much at the front of our minds. We have a project using artificial intelligence to seek out patients who could be promoted up the waiting list or put back because of their conditions, so that we aren’t offering surgery to a patient who isn’t yet ready. We have a prehabilitation service for diabetic patients who are on the waiting list, for example, so having identified the elective list and the people who are diabetic, they can reach out to them and help get them to the managed state they need to be for surgery. That’s a great example of how we’ve been able to provide data flow into those services to enable them to reach out to people.
In terms of inbound and outbound, we also place focus on not getting patients into the services at all where possible. I think there’s a lot more we need to do around prevention, to allow people to stay healthy and out of the system altogether. One of the things we learnt relatively early on was the idea of gaps in area. If you take a diabetic person, for example, you generally need them to have a number of measures that will keep them well – say an eye test, or blood test. We’ve found that there is a direct correlation between the number of gaps in care for a person, and the number of times they will turn up in A&E. We don’t yet know which of those measures might be leading to gaps in care, that then leads to that person presenting at the emergency department. So we’re exploring the influence of these measures on our resources – maybe there are areas where we can provide a push that will close those gaps and ease pressure on A&E, and maybe there are some areas where we can save a little effort to be reallocated somewhere more useful.
Stephen: When we started our work in this area, we wanted it to become a system-wide solution from the outset, so it was very much designed in that way. We spent almost a year looking at how to architect that appropriately, working out how to get the data in – not just from NHS organisations, but from other system partners as well. Taking the appropriate amount of time at the beginning was a very important thing to do. It did mean that we weren’t necessary delivering with the pace that people wanted us to or expected us to, but we’ve recovered from that because it’s doing quite well. We’ve also spent time engaging with people, trying to illustrate the power of data to handle their expectations and to show them what data can do – how it might mean that their time would be better spent elsewhere, for example.
We’ve now got a predictive modelling solution, created by one of our data scientists, that allows us to model different opening times of community hospitals, different numbers of staffing and different staffing types. That allows us to work out where the most appropriate place for a patient is likely to be. Then you can look at the staff you require, the staffing mix, and so on. There are all sorts of new possibilities you could never even consider at the beginning which start to come out of this.
There can be a cultural challenge around the belief that digital people are trying to tell clinicians how to do things. To address that, we got chief clinical information officers involved. We also supported chief nursing information officers into primary care so that they could explain things to their colleagues and their peers in a technical but non-technical way. That aligned with the peer respect that colleagues have for each other. That has been successful, and it gets the focus right. Otherwise, there’s the risk of diving into what the technology can do, but it’s got to be there for a reason. It’s about practitioner needs; what answers are they looking for, which improvements to care do they want to see, how can we drive that of the data for them?
Esther: When you have different Hospital Information Systems involved, it becomes very difficult to bring all of the data together and truly integrate.
The second challenge is the focus and prioritisation. You need a problem-back approach, where you understand what you are trying to solve with the data, almost before you even start pulling data out. There’s so much information in the system with clinicians inputting throughout each day. You can pull that data out, but if you’re not clear on what you are trying to answer, it becomes very difficult. You end up with a system or solution that no-one uses because it’s unclear what we are using it for. We can build all the AI or machine learning that you want, but if there is no need, no clinical use, then it becomes useless.
The third challenge, I would say, is around capacity and workforce. When using data you want to make sure that you are working with the end users to ensure that they have an influence and are a part of your design process. We find that this really helps in the change management further down the line, because it’s no small feat keeping all the key stakeholders engaged from the get-go. Making even the smallest design decisions really helps them with their adoption in the end.
The last challenge that I would flag are the admin and data sharing challenges, especially within the UK. If you work in a GP practice, or in an acute, you’re part of that ICS network but we can’t really share patient information in this way, not until all the paperwork is in place. It’s a smaller challenge in comparison to others but it takes a significant amount of time. With that in mind, it’s important to ensure that you allow enough time in your planning phase to look at the admin requirements around data.
Paul: There’s something to be said for challenges around definitions. Even defining what a bed is and when it’s open and when it’s not – we still don’t seem to have quite cracked that. It’s about understanding the available resources.
Then there’s the challenge of getting enough data science and analytics capability to exploit what we’ve got. People are getting it now, but they want it in bucketfuls, and we just can’t get it out fast enough. The solution to that is probably a long-term workforce management development programme. In the Wirral, we’ve done a lot of work with Liverpool University in this area – we have honorary contracts for a number of data scientists from there.
Another problem lies in the financial incentives to make the changes that we know should be made in things like prevention, and the sustainability around that. If a GP or hospital isn’t going to be paid for something that they know they should do, they have to think twice about whether they should be doing it, even if it’s obvious to everybody that they need to. We’ve used some of our digital funding to incentivise in situations like this – for example, we’re running a project locally to find the means by which we can get blood pressures from house-bound hypertensives. We’ve incentivised GPs to try and create a digital channel to that house-bound individual, and if they can’t, to try and find another way of getting the blood pressure from them. We built that into our digital project, but when it comes to sustaining that kind of change, we need to think as a system. That’s something that ICSs will need to do, to think about moving resources from one part of the system to another.
Finally, a big challenge is general capacity for everyone – clinicians especially – to be involved in these projects. Again, you can provide some funding for some resources, but it does mean that people have to find the time to do it.
The road ahead: what does good look like in using data?
Stephen: I think what good looks like is a continuation of the trajectory we’re on. More and more detailed analytics, more use of data science.
Looking to the future, I need to increase the capacity of the team. In our five-year forward plan, we talk about our ambition to become a data-led system. To date, we’ve harmonised a number of our platforms or GP practices using the same platform, so we’re taking some massive steps forward. But I need more people at the sharp end. I need people who can take all of data and apply the skills they bring, turning it into insights and the usable visualisations for clinicians to use in action. I’d like us to be able to put some of that into the public space over the next 12 months, so that the public can really see what the NHS and the local authorities are doing for them.
I also want to get more of the new technologies in there – how can we make more use of machine learning, how can we use AI to help look for insights that we haven’t even considered?
There’s so much we could do, but I need some more people. If we can do that, then we can do all sorts of good things in the future.
Esther: I think good looks like more accessible data – more easily-read data at your fingertips to support decision-making. How can we give information to the team within the emergency department, within the care homes, or general practice, so that they can make operational decisions to help their day-to-say, rather than having to click into three different screens to pull reports? The only way to achieve an integrated care system is through digital connectivity, through data.
Paul: I’m lucky enough to have been able to start to look at the digital maturity assessment reporting that is based on what the NHS regards as what good looks like. Of the seven dimensions, the two that we are marking ourselves down for in my area are empowering citizens and population health. Moving forwards, we are looking to develop in both of these areas. Data is very heavy in the population side of things, and we need to join up data as well outside of our own domains – looking at education, income, pollution, and lots of other things that have a big bearing on people’s health. That’s where I’d like us to look next.
Also, it’s worth saying; we need to keep the focus on the patient as the customer of our digital services, and make sure that we are focusing on their needs rather than organisational needs.
Many thanks to our panellists for joining us.