Transforming primary care with population health: moving to an outcomes-based model

This content relates to the following topics:

This free online event explored how patients at risk of certain conditions can be identified and supported to stay healthy, through the adoption of a population-level approach to health care.

Our speakers discussed how different areas are using population health management to enable more individualised care that improves outcomes for an increasingly diverse and ageing population.

They also talked about clinical commissioning groups that have tried this approach – their successes and the challenges of implementing new models of working, as well as how it can improve the delivery of care for those living with long-term conditions.

They discussed how they built relationships with stakeholders to address the pressures facing general practice, change clinician behaviour and enable service transformation.

Watch on demand


  • I: Interviewer (Becky Baird)
  • Paramjit Gill
  • Amy Rilance
  • Jonathan Behr

I:    A very warm welcome to today’s online event here at The King’s Fund in Central London, supported today by Vision. We’re going to be exploring population health today, looking at how people at risk of certain conditions can be supported to live healthier lives, and to be identified in the first place, using a population approach to health care. I’m joined today by an interesting panel, who I’ll introduce briefly. I’m Becky Baird, I’m a senior fellow in Health Policy, here at The King’s Fund, and I lead our work on general practice. I’m joined today by Professor Paramjit Gill, Amy Rilance and Dr Jonathan Bear. And I’m going to ask you to just introduce yourselves briefly, and say a little bit about what’s brought you here today. Paramjit, I’ll start with you.

PG:    Thank you. Hello everyone, my name is Paramjit Gill, I’m a GP in Coventry and a Professor of General Practice at Warwick Medical School. Now, what’s brought me here today is really, as we know, it’s the 70th anniversary of the NHS, and it’s a wonderful time to think about what that means for us. And in terms of population health, as we know that General Practice in primary care has been the core element of the NHS, and it remains so, but it needs to evolve, and we want to see where our population health can help us within General Practice to improve care for our patients.

I:    Thank you. Amy?

AR:    Hi. So, my name is Amy Rilance, and I head up Healthcare Engagement and Development for Diabetes UK. And I think what really interested me about this event is that we have amazing data on diabetes. The National Diabetes Audit is the largest clinical audit in the world, and it means that we really know what’s going on in diabetes healthcare. And the big thing we know about diabetes care is that there is huge variation between the best performing areas, and the areas that are really struggling to deliver good diabetes care. And I think today is a great opportunity to start to dig into why that happens, and what we can do about it.

I:    That’s great, thank you. And finally, Jonathan, if you’d introduce yourself?

JB:    Sure. I’m Dr Jonathan Bear, I’m a GP and Chief Medical Officer at Vision who provide an electronic records system to primary care in the UK. My involvement today really is borne out of the experiences I had as a newly qualified GP at the coalface, dealing with results for things like kidney disease coming in, and not really having the time to manage it. And so, getting involved in the clinical decision support element of the IT system, with Vision at that time. And that tool itself grew over the subsequent years, to become something that can be used as a population health tool. So, whilst you’re applying clinical decision support at the patient level, you can also replicate that at the practice level and indeed at the CCG, and indeed nationally if necessary, to try and standardise your approach to managing clinical conditions.

I:    Thank you. Welcome to you all, and welcome to you, our online audience. There are around … I think over 900 people signed up today, from a whole range of people working in the NHS, health and social care more widely, including GPs, academics, people from the charity sector, so welcome. I’m just going to say a few words about how it’s going to work today, we’ve got about an hour. I’ll start by asking a few questions of our panellists, just to set the context for our discussion. And then the remaining time, we’ll use to take questions from you, our online audience. You should be able to see on your screen, a box, which allows you to submit questions directly for us. They’ll be fed through to me on the screen here, so if I’m looking down, you’ll know why, that I’m looking at the questions and I’ll be able to feed those to our panellists. I think that’s everything for now. I think we’re ready to get started. So, Paramjit, I’ll come to you first, if I may? Can you tell me a little bit about what population health means to you, and how that approach has helped you in your work?

PG:    Sure. Well population health really means looking at health outcomes for a defined population. And that defined population may be your registered population at the general practice level, or it may be at a higher level that has been alluded to at the CCG level, or even at the national level. And what we’re looking at is looking … and Amy alluded to it here again, looking at variations in health outcomes, and what might be determining those variations, and what can we do to address them? Now an example would be, for example, we’ve had childhood immunisations, which we look after individuals, but really helps in the population within a health setting. And another example is looking at proactive case findings which has been … given it’s 70 years now, Dr Julian Tudor Hart in Glyncorrwg really highlighted the importance of looking after our individual patients, but also the practice and improving their cardiovascular health. I think those are the elements of population health, applied individually.

I:    And tell me a little bit more about the work that you’re doing at the moment?

PG:    Well one of the examples I can use is looking at frailty, which is an important area, that we’re all challenged with. Now within a cluster, which is a group of 50,000 to 70,000 population, group of ten or twelve practices, what we’re looking at is identifying patients with frailty, exactly using the data and our read codes, and then linking that data with the social care, the local authority data that they have, trying to find out if our frail patients are getting … what packages of care are they getting? If we assume that they’re getting a package of care, and we assume that somebody is doing whatever needs to be done for them, but we don’t understand if they live alone. We used to do home visits where we used to get that information but now we rely on data, and that data is held by somebody else. So, it’s bringing that together, and this particular student is collating all of that, and mapping it, and producing these heat maps, or hotspots, finding out where, if there’s a cluster of these frail, elderly patients, that we need to target with services that we can address these needs.

I:    Thank you. Amy, I’ll come to you. Tell me a little bit about the work that Diabetes UK has been doing, and how a population health approach has helped you, and changed the way in which you’re providing your services?

AR:    Yes. So, I’m going to talk about a very specific example, where we looked at one of the really big challenges in diabetes care. So, in diabetes care in the UK at the moment we spend 10% of the NHS budget on diabetes, but actually eight billion of that is spent on treating complications. And those complications are preventable, those complications can be prevented by controlling three things, HbA1c blood pressure and cholesterol. But, when we look at the population data, and we look at people with diabetes, and we see that actually only about 40% of people with diabetes are within the safe parameters, that NICE sets of those things. And that prompts some really big questions for us as a charity trying to look at how do we improve care? And we really dug into understanding what was going on there. And we looked at things from a health care provider perspective, and we identified issues around primary care, around the level of clinical knowledge, around diabetes that not everyone in primary care has received the appropriate levels of training around diabetes. We also looked at issues around things like clinical inertia and whether that was meaning that people weren’t necessarily being proactively treated to reduce their cholesterol or reduce their blood pressure. But then we also flipped it and we looked at it from the perspective of people with diabetes, and we said, well what’s going on for people with diabetes? And the big thing that came across was around whether people were really genuinely engaged and understood why these three things mattered so much. And so, the approach we took was to work in a co production. So, we brought together people with diabetes, we brought together primary care professionals, psychologists, specialists, and we created what we call information prescriptions. And those are very short, very targeted pieces of information that are designed to really engage somebody with diabetes in their healthcare. But where we take a step back and we go back to population health is, how we’ve looked to deliver those. So, we’ve worked with Vision, we’ve worked with all the IT suppliers in primary care, to actually embed those into the system so that somebody walks into their GP practice, and they have diabetes, and they have high HbA1c. The GP, or practice nurse, gets a real time alert, alerting them to that fact, and if they click on it, it takes them straight to the information prescription, and it autocompletes that patient’s data for them so that it’s a genuinely personalised piece of information. And I think for me that’s the big thing, that we’ve got to use the tools of population health but actually each individual patient is an individual. And so we’ve got to find a way of translating the big data into something that’s real for the patient in front of you, and that’s how we’re working with the IT systems to create genuinely individual interventions at scale.

I:    Thank you Amy. Some really interesting examples there about how population health management can and is being used. So, please don’t wait until our speakers have all finished to submit your questions. Keep submitting them, sending them in to us, and we’ll be able to use those as we go to stimulate our discussion. Now I’m going to come to Jonathan.Tell me a bit about your work, and how you’ve been using a population health approach, particularly with the CCGs that you’ve worked with?

JB:    Sure. Really, it’s our responsibility to implement what people, such as Amy, and Professor Gill have done in terms of their research and trying to find the best evidence for what is a good intervention. Population health in general to us means that it really follows a biomedical model as in if you take, let’s say, 100 people, we’re all at risk of developing something, and those risks will vary within that population. The question is, how do you identify the higher risk group that you need to do something about? So, let’s say a condition X, and of those 100 patients there’s 30 who are at high risk, and it just so happens that there is a particular intervention that might actually prevent that condition from developing, and let’s say that by using population health techniques you can identify that say ten of those, address that risk factor and that they never go on and develop that condition. That’s a really good outcome from population health intervention. So, the remaining 20, depending on where they are in the country, the level of interaction with health services and so on, will have a different route through developing … first of all going through a diagnostic route, having the diagnosis made, and then follow up and management thereafter. So, it might be that ten of them just don’t get that kind of excellent service they need, for whatever reason, they’re not engaging, or there aren’t the services available, and so on. So, it could be that two, let’s say, happened to die without really getting the diagnosis, but there’s some sort of engagement. And then others, let’s say another two or three, start the diagnostic process, but don’t go through it in the best way that they could. There’s delays in making that diagnosis, and they start to suffer. So, then you suddenly get, you’re down to about three or four who could be managed well potentially, and others who may not necessarily die from this made-up condition, but go on to have comorbidity, and therefore aren’t really contributing back into society thereafter, and having a good quality of life themselves. And the other ten that I’ve been sort of alluding to, let’s say they went through perfect treatment and had everything at the right time, made a full recovery and it doesn’t really matter if they’ve got the condition, or if they’ve recovered from it. Now, every one of those is a touchpoint that population health can address. So, you can risk stratify, so the frailty that you talked about. You can identify your most frail patients and do something about them. What’s the reason that they’re frail? Is it because they’re on ten medications? Well do a medication review and stop half of them, and therefore they’re less likely to have side effects and less likely to fall over, fracture a hip and go into hospital. The diagnostic pathway itself, we did a study … well we did a project in Greenwich CCG to identify patients who really should have the diagnosis of some chronic conditions, including diabetes. So, all the data was there but the dots weren’t being joined up, and so they weren’t getting the right treatment at the right time, because of that delayed diagnosis. In other areas such as there’s a cluster in South Wales that we’re just about to, this week, get some data back on, they want to understand, what is the burden of a condition? Is it worth doing something about? And that particular condition is acute kidney injury. So, we get alerts in the patient records saying, ‘This patient is at risk of an acute kidney injury’ based on lab results that come through. They want to know, who are those patients? Are they all nursing home patients who actually we don’t want to go into hospital and therefore we don’t want to sort of create false positive results that are meaning that we’re overmanaging them. Or are they a different group? Are they diabetic patients who aren’t being managed well who are at high risk of developing true acute kidney injury? So, we’re going to look at the number of those alerts coming through to work out if there’s something we need to do for those patients. And just finally, an extension of that really, in Bury CCG over the past year and a half, we’ve been baselining treatment for acute kidney injury of patients who have had the diagnosis, they’re not at risk of it. The alerts have happened but they’ve actually had the diagnosis. And how are they being followed up thereafter? Are they having the right blood tests done at the right time? If they are, great, but what’s that subsequent result? Are they developing chronic kidney disease, are they being readmitted and so on? So, they’re baselining how 30 or so different practices are managing those patients, based on … well, no specific guidance at this point. We’re really baselining that and seeing where the gaps are in the follow up for those patients. So, it’s just a couple of examples there of again those, sort of, different touchpoints in a disease process that you can attack.

I:    That’s great, thank you very much everybody. I’m going to take some questions now from people in our online audience. First of all, from Emily Plain, from Barking and Dagenham, Havering and Redbridge CCGs. She asks the question that I think a lot of people would be interested in the answer which is, ‘How do you match up the data about frail elderly people from GP records to social care data and particularly, what information governance barriers were there that you had to overcome?’ And can you tell us a bit more about that? So, I think Paramjit I’ll come to you first, and then maybe Jonathan?

PG:    Well no, it’s a very important and pertinent question. It is a challenge with our new regulations come in. We actually are working with a cluster and working with the individual practices to get consent from them, and the local social authority. We’ve got consent and for our research purposes, we’re using what they call pseudonymised data. So, the practice needs that data for developing the services, so do the social services, they bring the data together in the NHS cluster, linking it up, and providing us with pseudonymised data, and we do not have the link to unlock that data. So, that’s how we’ve got around it. Obviously, given the current climate on the GDPR regulations coming along, the research councils are actually working with the information commissioner into looking at the obstacles and how these can be addressed as well.

I:    And Jonathan, did you have anything you wanted to –

JB:    Yes. Obviously, we have direct experience of our own tool for population health and we sidestep a lot of the issues, because we don’t actually move the data. So, if a practice has consented to their patient’s data being analysed, all that we take out of there for our population level dashboards are the counts of a particular parameter. So, alluding to my previous example, how many patients with an AKI warning, stage three alert, are there? That’s all we get is a number against that, there is no movement of clinical data. Patients can still be excluded from that process if the right read codes are in their system as well. I think what’s a really exciting development is the creation of the Lycra projects across England where you are getting data from the primary care. Vision have signed up to it in the London one recently. You’re getting the secondary care system signing up, they’re bringing in the social care and actually that data is being held centrally. And so, tools such as ours can be applied directly across that data in a way that that consent has already been given for that patient, by that patient, for that data to move, from any of those systems, into one place. And I think that’s a really interesting route that population health could go down.

I:    Thank you. So, I’m going to move on. We’ve heard some really interesting examples about diabetes and frailty. We’ve had a couple of questions in about whether you have a view about adopting this kind of approach to mental health and mental wellbeing, and what would good look like around a population approach to mental wellbeing? Amy, can I come to you?

AR:    I’d like to pick up on that one because actually Diabetes UK last year did our largest ever piece of engagement with people with diabetes and we spoke to 9,000 people with diabetes about what mattered to them. And actually, support around mental health was one of the biggest things that came out of that. And when we look at the data, around a third of people with diabetes will have a diagnosable mental health condition. And we feel therefore that it’s really important that we are proactively looking for that connection between people with diabetes and people who may need some support around their mental health. So, again, we have worked with the IT systems to look at those prompts, so that if you’re treating a patient with diabetes, you’re actually thinking about their mental health, that that’s actually the fact that they have diabetes is triggering a mental health alert. And we use the PHQ9 questionnaire to prompt the GP to have that conversation with a patient. And we have an information resource to support, really so that people know where to go with that. Because when we talk to clinicians about it, they said, ‘We often don’t ask those kind of questions because we’re afraid of opening a can of worms. We don’t have time in ten minutes, we’ve got to go through … if it’s eight care processes for diabetes we’ve got masses to get through, we don’t have time to deal with that.’ So, what we wanted to do was make it really quick and easy to have an effective and useful conversation that actually did signpost people on to some useful support.

I:    Thank you, that’s a really helpful example. Did either of our other panellists want to comment? Professor Gill?

PG:   Well just to come in a bit, like other long-term conditions, mental illness has other conditions attached with the issue of multimorbidity. So, what really goes down to the heart of it is good recording within general practice of all the conditions and using the appropriate codes so people can use that data to help the individual patient at the population level.

I:    Thank you. So, a question now which builds on that one and we’ve talked about the time available for general practice. Paula Bland from NHS Kernow has said, ‘The move from reactive to proactive care clearly offers better patient care. But can you offer some suggestions about how to engage already overworked general practice in this? And, is there any evidence around reduced workload for general practice, by taking a population health approach?’ So, I’m going to talk to my GP colleagues first. Actually, Jonathan, would you like to start?

JB:    I think the risk with population health tools that we employ is that you do over alert and you create sort of a tick box culture that drags patients in for no discernible benefit. I think QOF is probably an example to a certain extent of ticking things because you have to rather than why the patient is sat in front of you. So, I think our responsibility there is to really kind of make the practices want to call those patients in by turning that data that’s in the records into really meaningful information. So, I think risk stratification is a good one. If you’ve got frailty and you’re saying, ‘Well look, these ten patients are most likely going to go into hospital in the next six months but you could do this, rather than just kind of leave it hanging there, if you do this, this and this, that might reduce their risk’, then there’s a good reason to address that. So, another example would be, if I’m in the patient record and it’s saying, ‘At risk of …’, sorry I keep coming back to kidneys, but ‘At risk of AKI’, well okay. But actually, if it says, ‘Well the creatinine has jumped from 90 to 140’ and that’s a whatever percent rise and they’re on an ACE inhibitor, and they look like they’re unwell because of what you’ve just entered, then you might want to stop that ACE inhibitor whilst they’re unwell. And so, it’s just making use of the screen real estate, by giving them really key information, I think, to make it worthwhile. And if you encourage that, then I think the use of the tools will increase.

I:    Thank you.

PG:    I was just going to give you an example of digital technology being used within primary and general practice now. For example, the undiagnosed atrial fibrillation which we know there are a lot more people out there. Instead of just taking the pulse, we can use this technology, the live code for example, where within just taking the ECG recording, it takes a minute, and you’ve got the answer straightaway and it doesn’t mean the GP needs to do it, the practice nurse could do it, and if you train the healthcare assistant. So, that is really being proactive and distributing the workload as well.

I:    So, looking at the wider general practice workforce?

PG:    Exactly.

I:    That’s really helpful, thank you very much.

JB:    Can I just add to that?

I:    Of course.

JB:    So, I think involving the patient is also really important obviously but that forms a cornerstone of what you’ve been about. But, with the announcement of the NHS app, it’s all about booking appointments and things, well that’s great, well there’s ten other apps that do that already, but wouldn’t it be great if actually some of that information that I was talking about that was sitting in front of the GPs, can’t we get that in front of the patients, and involve them and get them to make the decisions to drive their own healthcare agenda, because surely that’s the biggest way of ensuring success here. We can be calling them and texting them and Skyping them, or whatever, but it has to come from the patients if they really want to improve, I think.

I:    Amy, I’ll come to you with a reflection on that, because I know that’s something … the co-production of care is something you’ve been working on a lot?

AR:    Absolutely, I couldn’t agree more, I think it’s fundamental. And the time thing actually is very related to that, because when we introduced information prescriptions, a lot of people worried about it was going to take more time and actually we went through a very thorough process of piloting and really understanding what was going on. And people said it doesn’t take more time, it enables us to do what we need to do anyway and we do it quicker, because we do it far more effectively. And I think when we use population health, when we use triggers well, they enable us to do what we should be doing better and faster. I think we’ve probably all had that experience of wasting half an appointment trying to find a piece of paper or trying to find resource to give somebody. Actually, if we can use these things more intelligently so that we are using the systems to do that searching for us, so we’re creating those alerts, so that our time with the patient is genuinely time with the patient, it’s genuinely about interacting with them. And I think one of the key bits of feedback that we have from patients, when they receive an information prescription is, this is the first time anyone has ever explained it to me. Now, when we speak to clinicians they say, ‘Well it’s not. I know that I’ve had that conversation lots of times.’ But the explanation that hits home is the explanation that matters. And in terms of time saving, why have that conversation over and over and over again ineffectively, when you can have it once, and it actually makes a difference?

I:    Thank you. I’m going to take a couple of, I think, linked questions, around older people, frailty and particularly loneliness, which has been brought up by Richard Foxwell, who’s the Chair of Wellmore, who was talking about how a population health approach can support clinicians and practitioners to address issues around loneliness, and also supporting … somebody else has raised a question about the supporting of people in care homes, particularly. So, it’s that link between social care and health that people are particularly interested in. Professor Gill?

PG:    Well it’s a very important area that, as I said, when I started out in general practice doing home visits, we used to know who our lonely patients were. And that was part of the reason for doing home visits, every three months or so, to take their blood pressure, but just to say, ‘Hello, how are you?’ But now, the home visits rates have gone down dramatically because of the other pressures. But, identifying them is the key element. And some of this data is, once people are admitted, then the data really is with the social care services. And although we may get a hospital discharge note, it doesn’t say, ‘This patient lives alone.’ You have to unpack that, you have to search for that. And sharing of the data is vital here, and then recording. I go back to this recording again, because that’s vital really. And where they are living, and what support there is, elements intervention uses such things as Age UK, such as social prescribing, or in Coventry they have used the Fire Service to go and help as well, you know? There are other services available, that you’ll have to go and engage with, and talk with, within your locality.

I:    Thank you. Amy, is this something you’ve come across at Diabetes UK?

AR:    Yes. Funnily enough, in my previous role, I did a lot of work around social prescribing in Bradford district so it’s something that’s very close to my heart. And I think we shouldn’t underestimate the social support systems that are out there. And I think also there are different ways that people can access those and that sometimes that is about face to face going to things but actually, some of the online resources, and online forums and things, can be incredibly powerful in terms of supporting people to feel connected and to feel supported by people who understand what they’re going through. I also … picking up on care homes and kind of going back I suppose to the theme around using data well, we were very aware of some serious concerns being raised around insulin delegation in care homes and a lot of anxiety around who’s allowed to give insulin, and how does that all work? And we did a piece of work where we brought together experts from across the country to look at the issues around this. And one of the key things they said was, to start with your population data, that sometimes we have areas where people are thinking that they have X number of patients who you have to send a district nurse out to do insulin, all these kind of things. And actually, if you’re looking at frailty, if you’re looking at whether somebody really should be on insulin at this point in their lives, that can dramatically change the kind of service that you need to be offering, and so, using those population data tools, before you go in there and start drastically redesigning your services. And we have a guide to insulin delegation on our website, if that’s of interest to anyone.

I:    Thank you. So, I’m going to take a couple of questions that take us slightly wider than the GP surgery and the patient interaction. A couple of people have talked about … so, I have one here from Louise Bevan from Ageing Well at Public Health England, who talks about the role of population health in primary prevention. And then another question around the wider kind of implementation of population health to inform public health agenda around, for example, better environments, and how we might use that data to inform local authority decision making. Jonathan, can you look outside the GP surgery and tell us a little bit about how population health management might help in these areas?

JB:    Well in terms of primary prevention, we are already actively involved in that, it is within the sphere of primary care and partly because it’s the only data we’ve got access to, in that sense. So, the health checks programme, for example, is all about preventing diabetes ostensibly, and identifying hypertension and so on. And so, yes, we’re involved in an active programme with our practices to enable that. However, again, at the moment, from my own experience I’d say we’re limited to that at this stage but with the advent of these Lycras, for example, I do think that does really give an opportunity to enable the sort of cross data approach to this that’s needed.

I:    Thank you. Anybody else? Dr Gill?

PG:    Yes. In terms of childhood obesity, for example, if we think about the number of food outlets which are near schools, now this is where the local authority and the health services need to work together, and the education, everyone coming together, including the parents as well, and children, coming together to think about do we need all those outlets, just outside the schools, you know? And what sort of outlets, food outlets do we need? So, that’s working, that’s a longer term. Obviously, it will involve not only the local government, also national government here, but that is the wider perspective. we are really to tackle the major problem.

I:    Yes. Amy?

AR:    And I think there are twelve and a half million people at high risk of Type 2 diabetes in the UK, so risk is something that’s very pertinent to us and I think what’s key is that we don’t stop at telling people their risk, but actually we tell people what next, what do you do about that? And it’s very exciting that we have the National Diabetes Prevention Programme, which is working with hundreds of thousands of people now, to try and prevent Type 2 diabetes. But I think when we’re looking at population health, we need to always be asking the kind of so what questions, that we can’t just stop at the data, we need to think about what does that mean to people? And I was working when heath checks first started and I had a lot of conversations with GPs, who felt very frustrated that all they were doing was telling someone their risk, and there wasn’t the so what. And I think we have moved on that in the intervening years, but I think that’s still an area where there could be more work around how do you go on to … well, what does that mean for an individual?

I:    Thank you. So, I’m going to move on to take a question a little bit around outcome from Sally Fowler-Davis from Sheffield CCG. Can you talk a little bit about if you think population health requires us to more carefully choose the outcomes available? We might, for example, in a frailty study she says think about nutrition, mobility and activation outcomes, and use that as a proxy measure. Have you done work on what might be appropriate outcomes to measure the impact of population health approaches, in any of your work? Jonathan, I’ll come to you first?

JB:    Yes, we have. Again, it’s following those touchpoints I alluded to at the beginning, and those touchpoints are really the outcome that you then measure. So, for example, in the Greenwich project, the challenge was to increase the prevalence, bizarrely, of the chronic conditions that we were searching for and helping the GPs to go down the diagnostic pathway appropriately, make sure the right tests are done at the right time to reach those diagnoses. So, the outcome there was quite an obvious one which was, have we increased the prevalence of diabetes, hypertension, stroke, atrial fibrillation, and so on, in the year of the study, beyond that expected of a natural incidence of those conditions, simply because there’s this whole pool of patients that weren’t diagnosed? So, the outcome really depends on what it is you’re trying to improve. So, the outcome for the AKI one is, what is the burden of the disease already established? And therefore, that will help them, in the cluster, to work out whether they need to provide a service, to address those patients, or whether actually it’s not particularly needed. So, it really does depend on the question you’re asking, will define the outcome. And at the moment, as Professor Gill has alluded to, we are very dependent on codes, certainly in primary care, read codes, and soon to be snomed codes. So, if things aren’t coded properly, patients get missed and we’ll perhaps talk about it later, but I think there’s an opportunity to address the way that population health is done through the current systems.

I:    Professor Gill?

PG:    Again, it’s coming down to what you said, it’s what are the outcomes that you within either the practice, or the cluster, for your particular population group you’re focusing on? So, if it is loneliness, the outcome may be, well are they less lonely now, six months, or a year? But you’ve got to agree on that outcome and you’ve got to have the appropriate read code, or whatever code you want to do. So, it’s standardisation, so you’re all working from the same hymn sheet.

I:    Thank you. So, I think we’ll pick up on that point now around implementation. I have a number of questions about how you implement this kind of approach. Starting with, if you’re going to do this in primary care, there’s a question from Chris Maguire, ‘Who would be responsible for doing the kind of data analysis that you’ve talked about? And would it be a central pool, or would it be somebody in each practice, or how do you envisage that practices can best use their data, and have the skills to analyse that data?’

JB:    So, we’ve found that the smallest units of measure really would be a federation, so a group of practices, called clusters, in Scotland and Wales, and Northern Ireland. So, they want to work together, they have regular cluster or federation meetings. Say ‘Right, we’ve got perhaps a certain amount of budget to address something that locally is a problem.’ And so, they would then commission somebody like Vision to access the data, provide them with the dashboards, they have access to find the specifications of what it is they want measured and we simply enable that. So, it’s very much driven from that level, and by the cluster or the federation. We’ve had other examples where it has been at the CCG level where we’re then suddenly looking at 30 to 50 practices, and pulling all that data centrally, and then feeding that back to the practices in order to enable a change based on the information that’s coming back.

I:    Thank you. And Professor Gill, what has been your experience?

PG:    Well partly it’s again what we call the GP Alliance, or federation level, that seems to be it. But the actual lead, the actual person doing the data analysis, the core bit may be in the CCG, but the actual person leading on it, could be one of the GPs who are actually quite experienced, and some of them are very savvy at IT systems. Or even practice managers, which we’ve got within one of the clusters is brilliant he is.

I:    Thank you. And tell me a bit more about the issues around coding? That seems to have come up a couple of times from your conversation? I’m interested to know. And it feeds into a question here from Steve Marriott who says, ‘What are the barriers about implementing population health, and how can they be overcome?’

JB:    Yes, coding definitely is an issue. So, in that case finding project in Greenwich, for example, one of the obvious quick wins was to find patients who had actually been diagnosed, probably correctly, with … or definitely correctly, but not with a code that was actually being counted. So, QOF, for its group of diabetes read codes, doesn’t have every single code in there, that has anything to do with diabetes, bizarrely, it’s a very defined list. And so there could be outlying patients who aren’t on that and therefore aren’t getting recalled in the right way. So, the first step was to do sort of a data tidy up, and look for the likely other codes, things like ‘Seen in diabetic clinic’ isn’t … it’s because they’re diabetic, but they haven’t got the right code in their system to get the right recalls. So, quite a scary example of a patient who … it was a colleague of mine actually who a friend had a diagnosis of Leukaemia. But the wrong code had gone in, it was the morphological sort of lab type code, not the clinical diagnosis code. So, they went on the cancer register and weren’t getting recalls. So, coding is really important. And the decision support tools that form part of the population health management, in any system really, part of what they should be doing is constantly looking for those, that sort of data quality issue and flagging it when appropriate, to make sure that patients are captured, and therefore treated properly.

I:    Thank you. Professor Gill?

PG:    Yes. I agree about the coding. But, let’s not forget the quality of coding has vastly increased over the past few years, when we think about ten/fifteen years ago. Given whatever we think about the QOF, the QOF has actually improved the standardisation. And that’s why our general practice primary care data sets are so highly valued for research and planning as well. So, we do want to improve, because like you say, there are thousands of codes, and which one to use? And that’s where it comes to training your local practices and emphasising that everyone is involved. The receptionists, the manager, the practice nurse, the GPs, and our future GPs as well, the trainees, we’ve got to get them using the right and appropriate codes.

I:    Thank you. I’m going to come back to some more issues around barriers. But Amy, we’ve had quite a lot of questions coming in around information prescription and I think it has generated some interest. So, can you share a little bit more detail about information prescription, and how they’re generated?

AR:    Yes. So, each individual prescription is based on a specific topic. And we started with the triple targets of diabetes, so HbA1c blood pressure and cholesterol. But, almost immediately, once we had these in the system, people were saying to us, ‘I love these, can I have one on X, can I have one on Y?’ And we’ve been quite tentative, because what we didn’t want to do is just flood GP practices with a load of new resources that they just couldn’t cope with. But we’ve really looked at what are the areas of greatest needs, what are the areas of greatest risk, where are the things that cause the biggest confusion? And, to give a very specific example, this week we’re launching a new resource around contraception and pregnancy. Now, if you look at women with diabetes who become pregnant, only 8% of them meet the NICE criteria for a safe pregnancy and that places them, and their unborn child, at huge risk. In almost half of all women with diabetes when their babies are born, there are complications. So, it’s a relatively small population, but a very high-risk population. And when we looked at it, we found that the specialist services that are there to support those women, they don’t see them until they’re pregnant. And actually, the things that these women need to do to be safe happen before they’re pregnant. It’s about getting on a high dose folic acid, which you need a prescription for, before you ever fall pregnant. It’s about coming off drugs that might be harmful to you, when you’re pregnant, and it’s about getting your HbA1c as low as you can safely get it. And all of those conversations have to happen with a woman long before she ever considers falling pregnant. And so, by creating some information that was connecting pregnancy and contraception, we felt that we could get that to women at the point where they really needed it which was when they were thinking about how not to get pregnant. And so, we’ve launched that in the system so that it now triggers, for any woman of child bearing age, who has diabetes, so that when they’re going in to see their practice nurse, and they’re going to have the conversation of getting the pill, or whatever it might be, that practice nurse can see this little alert that talks about contraception and diabetes. They can open that, and it means that that information is being conveyed to the woman, when they really need it, not when they’re ten weeks pregnant and having a scary conversation with somebody. So, it’s about using that to really make sure that we’re understanding the problem, and we’re designing solutions that can address that.

I:    Thank you. So, it’s about using the big data to inform an individual approach?

AR:    Yes, absolutely.

I:    Great. I’m just ploughing through my questions here. We’re getting so many coming in, it’s quite hard to manage these here. So, I’m really interested, there’s quite a lot coming in around the role of engaging wider groups of professionals, and indeed the third sector. A comment here that it’s quite a technical, or clinical approach, to doing care. How can you stimulate that conversation with the wider group of health and care professionals around what they might do with this kind of data? Jonathan, have you been thinking about that at Vision?

JB:    We’ve worked with charities who don’t necessarily have the terminology expertise of the record system to be able to implement something that they feel would make a difference. And I think we’re keen to increase that. So, if somebody has a good idea, but they don’t know how to get started on it, I think there are people who can sort that element of it for them, if there’s a particular outcome they know they want measured, but don’t know how to go about implementing it. I think that’s eminently possible.

I:    Thank you. And Amy, in your experience?

AR:    I think our approach was always to look at this whole system and look at if from the perspective of people with diabetes, to look at if from the perspective of the primary care clinicians, but also to bring in the specialists. We were very keen to involve psychologists, behavioural health experts, and really kind of understand what the benefits were to everyone within that system. And I think when we approached this in terms of everybody winning, we found that people were very receptive and very open, and actually it wasn’t difficult to work with the likes of Vision, because people could see the clear benefits, and that has been really brilliant for us, you know? We know that over half of UK GP’s surgeries are using these, so that’s a real high impact. And obviously we want that number to grow, but it is getting things out there at a scale that traditionally I think the third sector has thought it’s very, very hard to work with the NHS, the NHS is thousands of small businesses, and small units, and you have to win each one over. Actually, by taking this whole system approach, we’ve been able to spread much further, and much faster than we ever could on our own.

I:    And Professor Gill, I’m going to ask you about your experiences with working with that wider team. I’ve had a question here around occupational therapy for example, and the role of the wider primary care team in supporting general practice to manage patients. What success have you had in engaging that team in population health management approaches?

PG:    Well it depends on the particular group that we’re dealing with. With the frailty, because there’s a need for it, and there’s an agreement from the managers, because I think that’s where the talking, and the conversations have to start as well. You do need a champion within the organisations to get them on board. Once you have that, then it’s much easier. But our occupational therapy colleagues are very sparse in number as well, and they’re overwhelmed, so we do have to be mindful of that, when we’re asking for their help and support, that it’s going to lead to something sustainable. It’s not just a one off, sort of a six month, or a year, then oh, actually another funding stream is going to come on board. So, we have to look at the long term.

I:    Thank you. And there’s a question here from Gill Currie, who’s a Dietetic Services Manager talking about the fact that if other professionals are involved, they’re often using multiple information systems. This might get into too technical a question, but Jonathan, tell me a little bit about how you overcome those linkages between different types of information and information systems?

JB:    Yes, that’s an age-old problem ever since IT came into the NHS, from what I can work out. But, the three or four major IT suppliers in the UK have their own ways of getting at the data, and that complicates things potentially if an area has a mix of those systems. Now, in our particular tool we’ve overcome that, so it doesn’t matter which underlying system is being used. Everything I’ve spoken about can be applied to that cluster, that CCG, it’s system agnostic in that sense. And so, anybody can, as long as they’ve got an interface into that data, so the Dietetic Service might operate out of one of those practices, but actually want to see the overall picture for the CCG. Yes, they’ll have that data and increasingly there are solutions that allow the real time clinical data, the patient record itself, to be viewed irrespective of the underlying clinical system and I think there are really changes afoot that will sort this out. And I keep coming back to the Lycra project, and I think it really is exciting. The data is being sorted, it doesn’t matter where it’s coming from, and as long as the right data sharing agreements are in place, the dietetics could plug into that using a system with an interface that suits the dietetics, not that it suits a GP to kind of render these interfaces entirely to suit the user of that data.

I:    So, just give me a brief overview about the Lycra project? We’ve had a couple of questions asking you just to say a little bit about exactly what it is, for those who aren’t aware?

JB:    Yes. I think it has been kept quite quiet, I don’t know why. They’re not confidential discussions we’ve been having, in the sense that I believe there are going to be ten of them across England, and it’s all about getting a data pipeline from the various sources of these data. I think up until now, there’s been efforts to force suppliers such as Vision, and the others, to create a common, what we call … this is getting technical here, but an API common route into the data. But I think that’s changing, that actually the data is going to sit as part of these Lycras and that this data well, as it were, is then going to be plugged into by whoever wants to provide the right kind of interface, as I was saying. So, we’re obviously in the primary care systems business, so we can provide an interface to a GP who could consume that data from across London. And I think that’s the route I understand it’s going down. I believe it’s at an early stage, at this point, but there’s real funding behind it, it’s not an aspirational thing.

I:    Thank you. So, Professor Gill, we have a question here around risk stratification, and more broadly around implementation. Andrew Griffiths asks about … says that it has been promoted for a number of years, risk stratification, what do you think is preventing its implementation scale and pace, to use a phrase, and what do you think might be done to support GPs particularly, in the implementation of these kinds of models?

PG:    Well it’s again a very pertinent question, because in the QOF there’s a risk stratification for identifying people with atrial fibrillation, then doing something about them, if they’re at high risk, and get them anticoagulated. Why is that inertia? Now that’s the real question that we haven’t really got an answer to at the moment. Because incentives have been applied through the QOF, to get people involved and also through the local schemes, and even through … you mentioned the Greenwich scenario there as well. But it’s still not being applied by all the practices, for all their eligible patients, there’s something else that needs to be done.

I:    So, yes, I was going to come to Amy, because you’ve had some experience with your information prescriptions, which has been a successful implementation. Tell us about how you did that, that might give us some insight?

AR:    Actually, I was thinking more broadly actually around risk and how you incentivise people to do that. And I think there has got to be the sense that it’s useful and I think that at the moment, a lot of people are feeling so overwhelmed. I spent some time last week with the Diabetes UK clinical champions, who are really focusing on improving diabetes care. And I mentioned right at the beginning the variation in care and we know that there’s a 71% gap between the best performing and worst performing CCGs. And we know that actually a lot of that is about systems and processes. And I think that in primary care, there is so much to do, there is so much to get through, that a lot of this, why things are and aren’t done, isn’t necessarily about conscious choices, it is about systems and processes and as an organisation we do some really interesting work with areas where we try to help them to improve on those. So, we do what we call best practice modelling where we put a bunch of practice nurses from different practices in a room, we show them their data and we say, ‘This is interesting, let’s look at the variation going on here. They’re doing really well on that one. What is it that you do? Why is it that you get 100% of your patients remembering their urine sample? What are you doing differently?’ And I think those kind of approaches, that aren’t about berating people, that aren’t about saying how do we put in more financial incentives, or whatever it might be, but that actually understand that this is about fundamental system change, which is about people, and it’s about actually getting the people working within those systems to care and have a stake in changing them. And what are we doing at scale, to be able to make those changes, and support those CCG areas that are only hitting 18-20% of their patients up to the very least the kind of medium point.

I:    Thank you. And Jonathan?

JB:    I’ll just come back to a point I made earlier that I think involving the patient is really key. So, at the moment we’re basically providing them with their data and if we can turn that into meaningful patient-level information that says something like, ‘It would be a good idea if you went and had these blood tests with your GP’ and let them drive the agenda. And actually, develop some really good gatekeeping rules around those so that only the right patients are seeing that information, we’re not worrying the worried well, as it were, and that it’s the patients at a high risk of getting that information through. I think it’s disingenuous to say that elderly people don’t use technology these days, a lot of elderly people do. And so, trying to get their involvement at the centre of all this I think would make the right patients be seen, at the right time, rather than a sort of random mix that we kind of see in general practice at the moment.

I:    Thank you. I think finishing on that view about co-production seems a good place. We’re coming to the end of our session now, so I’m just going to turn to each of our panellists starting with you Professor Gill, if I may, about your hopes for the future of this tool. What do you see as the next step for population health management, and where it might go?

PG:    I think we’re on the road there with our digital revolution, or evolution that’s going on and also involving our patients which has been highlighted here today, plus the wider sectors such as the charities, Diabetes UK, and the other charities. And we have to use what’s available within our localities to help us to manage our patients’ health.

I:    Thank you. And Amy?

AR:    So, I think population health is a really exciting thing in that it enables us to focus on where we need to focus. But I think if we’re going to make that real for patients, we have to find ways to move from the population level to the individual and actually focus on the patient in front of you, and if used well population health makes that easy.

I:    Thank you very much. And Jonathan, to finish with you?

JB:    From a technological perspective, I think … and we’ve talked about read codes and snomed codes, I think the majority of the golden data is in the narrative of a clinical record. And at Vision we’re making huge steps towards tapping into that narrative, using the modern sort of buzzword techniques like machine learning, to actually get that data out, and do something with that, not just the data that has been sort of consciously recorded by clinicians.

I:    Thank you very much. I’m sadly coming to the end of our time. Thank you all so much for your contributions. There have been so many questions, I haven’t been able to ask them all, but we will share those with our panellists here, so that they can see the kind of things that you’re interested in, that you were asking about. If you’ve got colleagues you think might be interested in our discussion today, please do ask them to register, and they can watch on demand, at a later date. And please do join me in thanking our panel Dr Paramjit Gill, Amy from Diabetes UK, and Dr Jonathan Bear. And hopefully you will join us again in the future at one of our online events at The Kings Fund. So, thank you very much and goodbye.

Event partner

This event is being run in partnership with Vision.

If you're interested in partnering with us on an online event please email Chris Lazenby or call him on 020 7307 2490.




Amy Rylance

Head of Healthcare Engagement and Development, Diabetes UK


Paramjit Gill

GP and Professor of General Practice, Warwick Medical School


Jonathan Behr

Doctor and Chief Medical Officer, Vision


Beccy Baird

Senior Fellow, Policy, The King's Fund