Virtual Festival of Evidence | Claire Cordeaux


Cross-boundary, cross-sector – using simulation to understand the impact of integration


Claire Cordeaux


Claire Cordeaux: I’m passionately committed to why people should be using simulation of modelling in healthcare to understand the impact of change, particularly in light of growing demand and probably less capacity of resource, and the need to make cuts.

Interviewer: In terms of evidence, you’ve used modelling to increase the early diagnosis of Hepatitis C. How did this come about?

Claire Cordeaux: One of the shocking things about Hepatitis C is that only about two percent of the population are diagnosed. Of course if you’re not diagnosed that means you go on to develop liver disease or cancer, and you probably die early. We set up a model to enable people to examine their current state – how many people they were currently diagnosing, against the number of people that there actually are in the area with Hepatitis C – to experiment with understanding the impact on their resources and their costs if more people are diagnosed and treated.

What we found was actually the more patients you diagnosed, the lower the costs. Although you had to pay for the patients to be treated for Hepatitis C, you were not paying for treatment for other conditions later on, and the overall benefit was a reduced cost overall. Also the more patients you diagnosed and treated, the cost per patient treated reduced.

We’ve run that model now in about 10 different clinical commissioning groups to enable people to understand how it might work in their system – including solutions like bringing the treatment closer to home, so people could be treated at a local hospital instead of having to travel 50 miles.

Interviewer: So from your perspective what can modelling and simulating change achieve that pilot projects can’t?

Claire Cordeaux: Well, it helps you work out the answers to those big questions in a virtual environment, so you’re not going to hurt any patients whilst you’re doing it – and that makes you a bit braver about the scenarios that you want to test. It also helps engage people in that decision making. The simulation model is very visual, which helps to engage stakeholders in the change that you’re envisaging, and it enables them to put in their own ideas to see what the impact might be. Also, doing a simulation usually helps you understand how the current system is working. That’s not always very clear to people who are working in the system when you just deal with one part of it. So that learning of itself can help people to develop improvements, which they might not otherwise have thought of.



Claire Cordeaux: Thanks Terry. So good morning everyone and I’ve started in the style that Terry’s explained by being late and if somebody could model traffic flow on the M25, it took me four hours what should have been a two hour journey. So I’m pleased to be here [0:08:00.0]. I’m delighted that there are people also here and I’m going to talk to you about cross boundary, cross sector work using simulation. I’m from SIMUL8 Corporation. We’re a software company. But my background’s NHS and before that social services. So I came into simulation not because I’m particularly a mathematician or in any way expert at modelling, but because I found that I couldn’t really do my job, which was strategic planning at the time, without having a model which showed me some of the answers.

Which helped me think through some of the problems that I was needing to solve. So it’s slightly odd to be sitting down, I usually wave my hands about. But perhaps you’ll bear with me. So what I want to start with is talking a bit about the policy agenda and then how simulation can help. I’ve split that into a few segments really. So starting with prevention, going onto segmenting populations, which for many people would mean long term conditions. But also single conditions, [0:09:00.0] which we’re going to talk about today. The whole of the emergency care flow, which we’ve already talked about.

Then the impact on the operational side of doing things differently. I’ve got an example around community services. Then we can have some discussion if there’s time. But there’ll be plenty of time across the five days. So you all know health policy extremely well. So I’m not going to dwell on this. But I just really wanted to set the scene around the importance of supporting people outside hospital and hence the importance as Terry said of getting this whole system model going. So that we really understand what the impact of that really is on the entire system. We know the rationale for it. We know that it’s supported, both in the UK and the US, with the ACO model.

It’s something that we need to do. We know that we’re actually doing it. When you look at the data you can see across [0:10:00.0] numbers of countries. Actually the overall trend is for admissions to be down and outpatients to be up. Obviously we’re looking at secondary care data here. So we know that our efforts whatever they’ve been and whatever the evidence has or hasn’t been actually that is the trend, which is a good thing. But we need to know why that’s working and how we can do more of it. Very often what I find people are concerned about is yes, we know there’s good evidence that this might be the best thing.

We’ve read the papers, we understand that this is the direction of travel. But actually what does it mean for us in our organisation? Are we going to lose money? Are we going to have enough resources? What’s the likely activity that we’re going to have? What’s it all about? Actually my experience is that it’s that which drives the decision making. Yes, you can have the evidence and you can have clinical evidence for whatever you want to do [0:11:00.0]. But actually before adoption any management team will need to make sure that the evidence is right for them in their situation and they’ll need to apply it locally.

That’s where tools like simulation and modelling, which help you get a handle on it at a local level are extremely helpful. Obviously integrated care is a key area at the moment, as part of that moving people out of the hospital. The house of care you’re probably all familiar with and how we support people rather than their individual conditions. The funding streams that are available for that. So it’s really with those things in mind that I wanted to move into this cross border, cross sector discussion. Because putting any of those things into practice has an impact on every organisation within the system. What we’re hoping is that’s obviously going to improve outcomes for patients [0:12:00.0].

So why would we use simulation for this? Well if it’s a service or a system redesign it’s very helpful to understand what that’s going to look like in practice and to understand the impact on flow, on cost, on capacity and resource. It’s particularly a useful technique when there’s no historic data. So the territory that we’re going into with whole system change is an area where there isn’t necessarily historic data. We haven’t been doing it before. We can look at trends, we can look at our own data. But there’s no particular blueprint and we know that there’ll be different impacts on organisations. Different people will feel where the costs are.

Different people will benefit from the revenues. Of course the patients is the key to this. So I just wanted to set that as a preamble to say why I think simulation is useful in this slightly nebulous world of trying to change things for the right reasons. But not really being quite sure how that’s all going to pan out [0:13:00.0]. Simulation can help you think that through and give you some useful results. So let’s look at the health and care system flow. This is the way that I understand it. I think you’ll probably see things that you’ll recognise here. But for me you start with the population. This is the patients who are likely to be sick and to need support and those who are well and might need prevention.

They’re supported within the community, they may go into hospital and they’ll then be discharged back into the community. So that’s the basic flow. But of course there are whole ranges of variables around that that are very useful to explore and that are causing particular problems. So for some people there might be a lack of access. For some people, like that long term conditions group and vulnerable groups there’s also particular services that need to be in place. Maybe the service isn’t 24 seven. What happens [0:14:00.0] when the doors are closed to you? What’s the difference between rural and urban populations?

How do different people access services and what happens when there’s not enough capacity? So those are just a few of the issues that I think we’re all grappling with. I think I’m going to use that as a concept as I’m talking through some of these models. So the first example I’m going to talk to you about is the prevention example, so where we’re tackling the population if you like at the front end. Now this is a case study of Hepatitis C screening. We developed this with one of our clients, who has actually used it in I think 10 CCGs now. So it’s a model that we can repeat. It takes about a day to make sure we’ve got it right for that particular area. But the key of this is Hepatitis C screening.

What we’re looking at is the impact of [0:15:00.0] future demand. In this particular case, this example, if a new Hepatitis C service is delivered locally rather than in a hospital 50 miles away. What we’re looking at is what’s the impact on the projected burden of disease? What’s the impact on projected treatment costs for the service? When we were starting this model I didn’t realise that actually only two percent of Hepatitis C is diagnosed. So that’s quite significant. Of course all those people who are not diagnosed are likely to go on to develop liver disease. If you were listening to the radio as I was this morning in my long journey on the M25, you will hear that they are talking about liver disease today and poor outcomes and how it’s increased hugely.

Now a lot of that is due to alcohol. But some of it could be prevented if we were able to diagnose Hepatitis C earlier. So we used our tool scenario generator, which was created with the NHS Institute for this particular model [0:16:00.0]. Where we’re starting with an age banded population from a CCG. In fact this was a collective area of CCGs. We’re looking at the disease prevalence of Hepatitis C and we’re looking at the demand. In this particular case we’ve got an intervention which is for all of those who are not genotype 1. We’ve got the prevalence. So as the population ages, so you expect to see the increase or decrease in that disease with that age banding.

We have…there’s some good data from the RAND organisation about how Hepatitis C is likely to increase. We’re able to simulate that using that technique. But of course different populations are different. So if you haven’t got a large growth or even a small growth in the 25 to 34 age band, you’re likely to see that growth tail off because [0:17:00.0] that’s the age band who is most likely to have Hepatitis C. So that’s our simulated pathway. So we simulate our demand and then we move it into our simulated pathway. I’m not going to go through this in a huge amount of detail. But essentially we’ve divided people with Hepatitis C into different groups.

There are those groups who are…have a moderate Hepatitis C. There are those groups who already have cirrhosis and those who have liver disease. Then there’s the group that are undiagnosed and they are likely either to be any of those groups. They might go into [unintelligible 0:17:39.3] and more severe liver disease and then death. The idea is to model more people through the treatment pathway, so we can work out what the cost and the resource is involved in treating those people and what the likely impact is on the cost of not treating them. Because those people at the far end of [0:18:00.0] the model, on that right hand side here, are still going to cost money because they’re going to turn up in the health service.

So we’ve modelled that as a burden of disease and we’ve looked at the cost of that disease year on year using data from the RAND report. We’ve modelled that so that each year you have a probability of moving into the next disease state with an associated cost. Here’s the complicated treatment pathways. Again you won’t be able to see that. But it’s just to reassure you there’s a huge amount of detail that’s gone into this, including when you have your blood tests and when you’re seen by a nurse, how often that happens and what the likelihood of adverse events is. Also what your likelihood of cure is depending on the type of treatment that you’ve had.

We’ve applied costs to each part of the system. We’ve run a validation to make sure that the confirmed cases that we’re getting in our simulation are confirmed by local [0:19:00.0] data. So we’re sure that our model is representing broadly speaking what’s actually happening in the system. So then we looked at some of the results and we ran this over five years. What we assumed because this is what the local system was telling us was that if you have to travel 50 miles to have your appointment, probably about 50 percent of that group are unlikely to go to continue their treatment. That is a particular group that doesn’t necessarily complete treatment anyway.

So we wanted to improve that likelihood of compliance. What’s interesting is when you look at the different results. So we’ve got the current situation. We’ve got the future intention, which is to bring those cases locally. Then we’ve got some scenarios we want to run around let’s say we increase diagnosis by 10 percent, by 20 percent, by 30 percent. What does that mean? So obviously there’s a cost associated with that.

So can everyone see that better? Good, sorry I couldn’t see the screen. So the interesting thing about this is that the cost reduced with the [0:21:00.0] increased diagnosis. Although you’re obviously paying to treat people that you’ve diagnosed, you’re reducing your costs of them becoming sicker later on. The more people that you diagnose, the more those costs reduce. You can look at that by patient cured, the different scenarios and the number of patients cured, which is the graph on the left hand side. Then on the right hand side you can see the cost per patient cured. Actually the cost per patient cured is much higher in the current situation and it becomes much more economical the more people that you treat.

You can also look at those savings over time. Now we’ve used this model slightly differently in a number of different CCGs. But you get similar results as you’d expect. Obviously each system is a bit different. But what CCGs [0:22:00.0] are doing is they’re using this as the business case for change. They’re involving their clinicians in looking at the results, in validating the results, understanding what the impact is and in bringing in those decisions around screening for Hepatitis C. Not an easy thing to do. Where do you start? A lot of people start in the sexual health clinic, which involves us then looking at who is coming into the sexual health clinic.

Usually projects expand to look at different sorts of questions. But I thought that was a useful example of how you might use simulation to look at the prevention end of the spectrum. Does that make sense to everybody? Of course what’s that doing is a number of organisations are working together in order to see what benefits those might bring. So if that’s the kind of intervention we might look at in terms of a whole system at the front end of the pathway. Another way and one that I think we’ll be talking about [0:23:00.0] a lot more when we talk about long term conditions is in segmenting populations.

Looking at how particular populations, particular disease groups, particular groups of disease groups move through a system and what interventions you might want to put in place. Then understand the impact of that change on your system for that group and how that applies across the piste. So I’m not going to talk too much about the long term conditions Year of Care model because I know we’ve got a particular session on long term conditions. I know that Whole System Partnership has also got some models as well. I’m really looking forward to that conversation because we approach things in slightly different ways.

But there’s a huge amount we don’t know. So I’m sure if we work together, we’ll help each other. But I just wanted to mention the Year of Care for long term conditions. This is a programme nationally, which you’re probably aware of. But I’ll just very briefly [0:24:00.0] mention it. It’s about looking at the impact of a capitated tariff, capitated fee for individuals who have multiple long term conditions. Understanding what that might look like, what behaviours that might drive if you were to have an annual capitated tariff for an individual with long term conditions. This is a project that has been going on for a couple of years.

First with the Department of Health, then with NHS England, now it sits with NHS IQ. They started off looking at re-stratification to identify those groups of patients with multiple long term conditions and look at those two questions there. What happens if they’re proactively managed or they’re unmanaged and what happens if we apply annual tariff? So it’s an interesting project because in this case we’ve used simulation not to get results particularly, although we have got results. But to help drive the thinking. Because [0:25:00.0] when we started this project the hypothesis was that you had patients who were at risk and you’d be able to pick that up from your re-stratification tool.

You clinically asses their need and that would correlate with their risk score. Then you’d look at the services they’d consumed during the year including the fact that they were likely to have an exacerbation which probably took them into hospital and there we are. We’ve got a pathway for each type of patient. It would all work fantastically. What we found in that first year is actually there is no real correlation between the risk score and the level of need. Which is interesting, but probably not surprising when you think that everyone is using slightly different risk stratification tools. Some of them are picking patients up actually after the exacerbation.

So there’s lots of different reasons why that might happen. Anyway, it didn’t work as a system. So we moved on to looking at the data. I know that Kent are coming some time during this week to talk about a lot of the data work that they’ve been doing, [0:26:00.0] which is really interesting. They’ve been part of this project. What they’ve been looking at is how, along with the other early implementer sites, is how people with long term conditions develop those conditions over time. So we know 30 percent of people over 75 years have multiple morbidity. But just look at how your numbers of long term conditions go up as you get older.

The research from the Scottish School of Primary Care pointed out very early on in this project that actually you’re much more likely to have more than one long term condition than you are to have only one. So why do we treat people in single stream pathways. So we were setting out to create this co-morbidity pathway if you like that would fit people at different stages of need. So this is some of the data from Kent, which very clearly outlines this [0:27:00.0]. We can look at the cost, total health and social care cost, which is strongly related to multi morbidity. I won’t go into this in too much detail because I think Kent might go into it a bit later.

So I don’t want to steal what they’re going to say. You can see where those costs tend to fall in terms of the services that people access. Then this very interesting crisis curve that occurs when you look at patients by their risk score and also by their multi morbidity over three years. Kent had three years’ worth of data for the same cohort of patients, which is fantastically rich to mine. What you see is this crisis curve in both of those areas. Very interesting to think about because…and it fits with that idea about exacerbation. So people have a long term condition and then they’re likely to have a crisis and the costs are likely to go up and then you [0:28:00.0] stabilise it.

I’m looking at Carol there because we’ve both also done work on end of life care and when you look at the trajectory for end of life care you’ve got cancer that goes along, slightly going down and then goes down at the end. You’ve got frailty that very gently goes down to death. Then when you look at long term conditions it does this. So it’s exactly the same thing. So really useful to know that when you’re trying to model people with long term conditions and how they’re likely to use services, because it’s going to be that variation. What Kent saw was that more community mental health and social care services are delivered to people following their crisis than before their crisis, which absolutely proves that case that people get picked up at that point and their treatments change.

Kent have got some evidence that shows that if you have an integrated care plan, the pattern of services change. So there is an impact [0:29:00.0]. So we use that and the data provided by the other early implementer sites. There were eight to create a Year of Care long term conditions model. So you can see there is this state chart going on where people can move from state to state as they gain additional long term conditions. People can use this model [unintelligible 0:29:22.5]. It’s nationally available. You can download it and play with it. The idea is that you put in your population, you look at the proportion of people that fall into this category and then you look at the services that they’re likely to access during one year.

Then you look at the frequency of services that they’re going to access. So every patient in each state has got a different likelihood of accessing all those different services. Then there are costs associated with those services. Underneath, for those of you who would like to know, there’s [0:30:00.0] a discrete event simulation model, which is modelling all of that and just pushing the results out to the front end. The idea is that it’s used by people who have probably never used simulation before and are just looking at the input and then the results. The results you can find are looking at numbers of patients in each state by year.

What the average cost per patient is. Then you can experiment with setting your annual capitated tariff. If all my people with six long term conditions or more, I’m going to set that at 15,000 that’s what I think. Let’s just run that through. Let’s see what that really means when I put in all those individuals, how they use services, all the variation, am I going to be about right? Then again you can also see the cost of each area of service by each organisation or service type. So you can see, you can play about with that, test your different scenarios [0:31:00.0], what you might want to do with increasing community services for example.

Understand where the financial impact might lie. You can compare the tariffs and you can see the average cost per patient and compare your results. So the scenarios that people can run at the moment in the model are, are your patients known to your integrated care team or not? So you can then look at the different in results that produces. You can test against the proposed tariff. You can change the variation in cost for different services. You can decrease transitions through state. So really one of the things that you want to do is to stop people moving up into those multiple long term conditions. You want to be able to stabilise them so they don’t transition into the next stage where they’re going to be heavier users of services.

So how is this being used at the moment? It’s just been released to the world and people are starting [0:32:00.0] to use it. The early implementer sites have tested it. One of the things that was very…they’re using it in slightly different ways. They’re defining their groups in different ways and testing improvement scenarios. But one of the sites I was talking to recently is using this to negotiate costs between their healthcare providers. Also to the extent that they’re prepared to think about paying a provider a cost for a patient that they might not see in order to give some incentives to the system to look after those patients in the way that the commissioner has decided. It feels like groundbreaking stuff actually.

If you look at the clinical research around long term conditions, so much of it is single long term conditions, not necessarily this particular group in this particular way. So we’re really trying to use all of the evidence we’ve got, the experience of the early implementer site [0:33:00.0]. Our understanding about how transitions occur between diseases to test out what the impacts might be. Allow people to understand what the early implementers have been understanding over the last two years and to apply that thinking themselves. Okay, does that make sense to everyone? You’re very welcome to have a play with this on the NHS IQ website.

So that’s a multi morbidity look at the world if you like and how people use services across organisations. Here’s a specific look at mental health. Now this model is only just starting to be used. I was just going to give it a bit of a run through, but I’m aware of time and I’ve got some of the results on the screen to show you. So I might just wait and show you a later one if we’ve got a bit more time. But essentially this is [0:34:00.0] looking at people with schizophrenia and understanding how they use services, depending on their diagnosis. Understanding what would happen if they used more community services what the impact would be on acute services.

In this model you’ve got almost every service that people might get, including what happens if they go out of area. So the logic in this model is that there’s a disease model behind it and it’s then been validated with activity data. What happens is if somebody needs a bed and can’t get access to a bed they go out of area and there’s an additional cost applied. So that’s the reality. One of the things people want to do is to bring people back in area because they’re not necessarily using their inpatient bed so much. Because they’ve got better supported services in the community [0:35:00.0]. So reducing pressure, this is the purpose of it.

Particularly looking at the impact of additional referrals to early intervention services, rehabilitation and looking at the costs and resource utilisation. So these are the results that you get. So there’s a number of different scenarios that people are running here. On the first column there you’ve got the baseline. We’ve just compared that with increased referrals to the community. So we can compare what happens with the different sorts of cost and see how that happens over the period of time. You can also run what happens if you decrease the discharge to…if you decrease the length of stay and discharge people more quickly.

A range of other scenarios that people can test. Again this is a model that’s been created. It’s been created with Janssen actually, a pharmaceutical company who is [0:36:00.0] interested in working in partnership. They’re not using it to look at any particular drug. They’re interest is engaging with the NHS to understand the key issues in the health service when people are trying to plan for schizophrenia. This is being rolled out and used in a number of different trusts now. Their feedback from their partners so far is encouraging they say. Their trusts are interested. They want to put their own data in the model.

The important thing is that it’s also generating that engagement and discussion and enabling people to understand how the system works. When we first developed the simulation with people and tested it out with them there were lots of areas that people were a little bit hazy about what actually happened and how many patients and why and what were the referral routes. What did that depend on [0:37:00.0]? This has helped them understand what’s happening so that they can test their service redesign. Of course it enables them to put a business case for change without actually having to run an initial pilot or jeopardising the patient in that pilot.

So the feedback is this is an efficient way of looking at the numbers. You could see that it could generate cost savings. But without this it was hard for people to understand how the system worked and how it would change. I don’t know if other people have done mental health modelling. There’s probably quite a lot of experience in the room. So we’ve looked at now the prevention end. We’ve looked at the whole pathway right across four particular patients. What I’m going to look at now is what I’ve called a whole system flow. But it’s really looking at that part of discharge from hospital into the community [0:38:00.0].

This is a case study that comes from North Staffordshire. Martin Ware was involved in developing this simulation, which was also sponsored by NHS Improving Quality. What North Staffordshire wanted to do is look at how they might increase out of hospital services and what the impact would be on cost and capacity. North Staffordshire is quite an interesting place in that there’s a lot of community hospitals. I think in this country, it may be true elsewhere, it’s quite random the number of community hospitals which you might find. It very much depends on the history. If you have a community hospital you tend to use it, if you don’t, you probably don’t.

We’ve all seen the press if you try and close one. So you try and make the best of what you’ve got. But in North Staffordshire they have a lot of community hospitals and what they wanted to understand was how is their unscheduled care flow looking like? What’s the impact in five years’ time [0:39:00.0], taking into account population change. What’s the impact if we increase referral to domiciliary care direct from the hospital? So if instead of using the community hospital for some patients they go straight to the care of social care and have support at home. So this is a scenario generator model. That’s really just to show you how very complicated it is. I’m not expecting you to look at the entire flow.

But what we’ve got is people coming in at the front end who have urgent care needs, being picked up by all of the service pre hospital, then going to hospital. A number of different types of patient needing different types of bed for different periods of time. The important bit for us is that we’re looking at the discharge from hospital, which happens here. Then they’re discharged into social care. Again we ran the baseline model and looked at what the impact of [0:40:00.0] that flow was and validated it with the local systems. So again we were getting about the right numbers for the simulation when you compare it with NHS data.

Made sure through a number of workshops that people were happy with that and they believed the model. Then some of the things which I just wanted to mention was the costs in the model and also the length of stay assumption. So what happens is that patients go out of hospital into a community bed. They would tend to stay there for about 21 days before being discharged. Part of the rationale for that was if you stay in hospital and have to wait for a social services assessment you could be staying in hospital additional time. Actually if you get people out into a community hospital you free up the hospital bed, which is a good thing and you do the assessment while people are in the community hospital [0:41:00.0].

Although actually the cost of the community hospital bed is quite expensive. 21 days at that probably is paying slightly more than you need to. Also there’s the argument that if the patient’s staying in a community hospital for 21 days, maybe they’re getting institutionalised. Maybe the rehabilitation isn’t happening quite as you’d wish. So here are our results looking at population increase. So we could see the demand going up and what that impact is likely to be on the emergency department and admissions over nine years. We then moved on to looking at this domiciliary care scenario. So the idea would be that you’d take 30 percent of those community hospital referrals.

For that we said let’s say those patients would probably spend two more days in hospital because they’d have to have that assessment [0:42:00.0]. Then we had where the referrals were going. So 10 percent to complex, 38 percent to maintenance and 51 percent to reablement. We knew what the sorts of packages that people tended to have were. We had all the assumptions from social services about what the discharges looked like, how long they take, how many weeks they last for et cetera. So what the results were was that as a result of doing this you would make a 2.6 million saving overall to the system, which is fantastic.

But four million additional cost to social care. So obviously social services wanting to have a bit of conversation. But also what happened was as a result of that additional length of stay they’ll be a 1.3 million cost to the hospital when you looked at it in cost terms. Your maximum bed occupancy would go up by 10, with [0:43:00.0] an additional one percent utilisation. The community health service would save, or the commissioner actually would save 7.6 million as a result. Their utilisation of beds would reduce by 25 percent. With their maximum bed occupancy less 90, going down by 90, which effectively would have been a whole community hospital.

So some very interesting useful findings to start that conversation. As you know with the Better Care Fund these conversations are going on all over the country where people are trying to think about where the funding needs to go, how to flex it, what the likely impact is. So when I last left them they were off to have that conversation about how they were going to use the fund to look at that expenditure. But they could see from this the impact on each of the organisations, so the hospital [0:44:00.0], the community trust, the commissioner and social services from running that simulation. So we’ve looked at the whole system, we’ve looked at prevention, we’ve looked at managing patients, which we know have very specific needs and probably a high call on the health service.

So we’ve prevented disease, we’ve managed those conditions effectively in the community, we’ve reduced hospital admissions. But one of my queries always is well what happens to community services? Because a lot of the conversation around these system reform changes are about well that’s great, we’ve reduced admissions, fantastic. We’ve prevented [unintelligible 0:44:48.2], that’s fantastic. But in fact for every admission that’s reduced, community services are picking up those patients or those patients are not being picked up. But theoretically community services are [0:45:00.0] going to be picking up those patients. So the final simulation I wanted us to have a look at is.

Okay, so this last example is looking at the operational impact on community services. Because for every high level strategic whole system change, each organisation is going to have [0:46:00.0] to think about how to manage their day to day operations in a different way in order to cope. So the final case study I wanted to look at was the impact on a community team. We’ve had a number of people working with us on this. It was initially sponsored by Tissue Therapies, who are a medical device company, looking at wound care. One of their concerns is I can make the case for my lovely wound care intervention.

At a high level the health economics look really good. But actually who are the people who mostly deal with wound care, well they’re the community nurses. If I can’t convince them that this is going to be a really great thing and really help their workload, then probably even though the commissioner’s agreed it there’ll be resistance from people who are having to try and implement it. So if we could have a model that helped community services understand what the impact would be, that probably would be [0:47:00.0] helpful. It would be a helpful model anyway. But at the same time we were lucky to work with a Kent team who were able to tell us how the community team actually worked, also with the Isle of Wight.

I know there’s somebody from the Isle of Wight coming later on. So I’m not sure if there’re here now. But to look at how they would deploy this in their community team. So the questions were what’s the impact of an improvement intervention on the workload. As an example what’s the advantage of faster healing wounds on the workload. Would nurses have more time to care? Would they have more time to see other patients? Also the aim was to engage with the community team to understand actually how it worked on a daily basis. So that as you made these changes and activity came out of hospital or more referrals.

Actually how were people coping? How are they coping [0:48:00.0] at the moment? It’s not something that’s terribly well understood I don’t think. Partly because the way that we tend to report this is much of community service is covered by a block contract. So actually the activity and how it’s managed is a little bit hidden. So the way that this works is we assume that there’s an ageing population. But you can change that, which is probably going to mean that there are increasing referrals. So you start the model with a case load and then you have referrals coming in each week. You can sub divide your population into your wound care only patients, your multi morbidity and those that aren’t wound care.

So basically you’re trying to sub divide your patients by the groups where you’re likely to want to have a service improvement. We’ve started this with wound care, but actually we think it will apply to any type of patient in a particular condition area where you might want to employ an improvement [0:49:00.0]. So the patient comes into the team. They have a clinical assessment and then there’s a daily allocation to staff where we’re matching the patient need to competencies of the staff. We’re looking at multiple visits. So what’s the frequency of visits, how long do they last, how many people are not in when you go to see them.

What’s the travel time? How long does that patient tend to stay with that team and where do they get discharged to? So that’s the logic going into the model. We’ve just been testing out some graphics. I don’t think we’ll probably end up with these ones looking at the [unintelligible 0:49:43.4] one. But the scenarios that we’re wanting to look at are around adjusting the length of stay and looking at some of the detail of how that works. So each community team could put in their own data and get the results out [0:50:00.0]. So as you localise it to test your improvements, you can change your referrals. You can change your patient type and priority.

So the rule is if a patient has priority, they will always be seen first. Then one of the really interesting results we’re starting to get out of this is actually how many patients do you not get to see in that day, because they’re lower down on the list. You can change those visit times and frequency. You can include the travel times and you can test the impact of improvements. You can see there’s a number of dialogues where people…there’s a default in there. So you can run it to play with it to test the concept. But also you can enter your own data as well. The results that we’re starting to get out now are looking at what happens to resources by staff type as a result of running different scenarios [0:51:00.0], looking at their utilisation.

Also looking at the patient results and how long it is until they’re discharged and how that changes with the different scenarios that you’re running. So I wanted just to outline that one as an example of an operational model which helps you understand. Whilst you’ve [unintelligible 0:51:29.4] cross border, cross sector, at a high level all the executive teams are brought in. It all looks good. Actually at grassroots level it’s the people operating the system who are going to have to manage this and probably the people at the higher level don’t necessarily see how…what their daily life is like. Certainly the people we’ve been talking to are saying, yes it is.

The case mix is changing. We’re getting…actually some of our patients are just as acute as the ones in hospital. So now we’re going to have to step up [0:52:00.0] the frequency, the length of time that we spend with each patient. We get delayed. We’re working overtime. We’re not seeing patients. There’s a point, we’re sort of managing it now. But we’d prefer the quality of care that we’re able to deliver to be better. This will enable us to test out what our best scenario would be in terms of staffing and to discuss that as a business case. It’ll also enable us to test out what some of the strategy implications might be for our team.

So we can really evidence what resources we need to run this effectively. So I haven’t run a live simulation. But I’ve got them all here. I’ll do that. But I realised I was late and we need to have coffee and keep on time. So I’m very happy to run through any of those simulations with you [0:53:00.0] during the week, if you’re around. But just some final thoughts then is that one of the things that we don’t necessarily always talk about when we’re talking about simulation is how it can support understanding of a system. Quite often people think well until we’ve got the final model absolutely perfect we can’t use it to do anything with.

Actually I think a lot of the power of the simulation is in thinking through the problem, understanding what the key issues are. Testing them out in workshops with your stakeholders, getting them engaged with that. Changing your simulation, as they say, no it doesn’t work like that. It happens in a different way. Making sure that everybody from the senior decision makers to the people who are actually doing the job really understand how the system’s working. In that way, you actually drive improvement as you do your simulation [0:54:00.0]. As well as creating that final business case with all of the evidence you need to support the dialogue, to make the decisions. But I think essentially what’s so important about it is helping to define as well as.