Luke Edwards: Review into the operational productivity of mental health and community services

This content relates to the following topics:

Article information

  • Posted:Tuesday 31 October 2017

Speaking at The King's Fund's Community health services conference on 31 October 2017, Luke Edwards (Director of Sector Development, NHS Improvement) shares updates on the Carter review into community and mental health services.


Thank you very much. That was a fantastic introduction and has definitely set the context very clearly for what I am going to say today.

I have been doing this for about 12 months now, I have not got a background in health, I have got a background in policing offender management and possibly my most popular job was collecting student loans for 3 years so anything is a step up from there.  In terms of the work, when I started this role I had a very early conversation with stakeholders as you do, and one of the conversations I had was with The King’s Fund.  I ran the state of data and the understanding of community services across the sector so it’s a great privilege to be back today to be presenting on where we have got to and I would say this: what I am not going to present is a finished article, what I am going to present is our work so far and we are very interested in views both of providers and a broader group of stakeholders today. 

So in terms of what I am going to take you through. I am going to give you a bit of context around the Carter Review, the approach we take to delivering that.  I am going to share some early data that we gathered from the sector which has not been without its challenges both for us and providers and I know a few of you in the room and I am sure you would echo that and then illustrate highlights from the areas where we think we need to go next, and that is very much up for debate. I should also say it is has been fantastic working with Lord Carter over the last 12 months. I have been on the end of many of his reviews so it has been great to be in at the start of that process. 

So, the basic approach is very simple but before I go into it, it is worth putting community services in their context.  It is very difficult to really isolate how much we spend on delivering community physical health services within the NHS, partly because as many of you will know, a lot of it sits in block contracts which are delivered by mental health and community Trusts but we think there is about £5 billion worth of expenditure in the community sector and we think there is about a further £5 billion of expenditure in the private independent providers, just on community physical healthy.  If you take that and put that in any other bit of government, and I have worked in quite a lot of it, that is a huge number.  But actually, when you look at the focus we have in terms of the way the service is delivered I do not think they are commensurate with the level of spend and the profound importance the services that people deliver have an impact on people’s lives so I think it is a fundamental service and the first thing I am going to say is that I am going to echo the comments from colleagues.  This is an a really important area of public service and while there has been a lot of rhetoric around the form, I do not think anybody would disagree with the fact that we are not there yet.  So that’s just a little bit on the kind of context on why this is important.

Community Services clearly play a critical role in the wider infrastructure of the health system and again, I do not think we fully always recognise that role.

So what do we do, and how do we go about doing a Carter Review. Many of you will have read the review he published in 2016 in the acute and in some ways the approach is really pretty simple actually. What we first need to do is understand how organisations work. What configuration of services do mental health community Trusts deliver, how are they driven, how are they commissioned. Then we need to understand what good looks like in delivering those services, what is best in class, what are people proud of and we have had those conversations across the country and then we need to find and capture that information and benchmark that so we can understand how we take the work forward and that gives us insight into where the improvement opportunities are.  It is quite an intuitive approach and we have worked through it in different sections and at different paces and we do it by working with the sector.  I and my team do not have all the answers.  The answers are in this room and in providers and with commissioners.  So what we do is, we run a process of engagements, quite detailed engagement with 23 Trusts to generate that learning and understanding. We have been working on that process for about 9-10 months now, so we are at the point where we are starting to form some fairly, I think, clear views for those users subject to challenge around what we think the system looks like. 

So, when we then look and take a step back at our starting point, what did we find?  Well firstly, as colleagues in The King’s Fund said we have got a very complex system and we have highly fragmented provision. I mean this is really quite difficult to get your arms round. What community services are actually delivering and the value they add to the sector.  We know it’s there but it is pretty tricky to get hold of. We have very limited outcome data and relatively limited national standards and we have then possibly the biggest challenge when you are looking at it from a very efficiency perspective as there are no consistent service line definitions or specifications. Therefore, it is incredibly difficult to benchmark like with like and actually coming into the sector it is really genuinely quite difficult to understand what different organisations are delivering relative to each with each other within a population base.  So other than saying this is all a bit difficult, let’s not bother, what do we do.

Firstly, as I said we focused on engagement.  We got out there, we talked to people, we wanted to understand what they felt good really genuinely was and we focused on primary data capture. Inevitably, given the state of data within the sector we have been able to establish at this point relatively limited opportunities for improvement where we can evidence that through clear benchmarks. That is not to say they are not there, but the data that is available in the system makes it really, really quite challenging to absolutely pinpoint where a provider can potentially drive forward improvement on a technical efficiency basis.  I should pause as well just to say our main focus of this work is technical efficiency. It is about the detail and it is about how you translate more of your inputs into additional outputs. It should be done so in the context of outcomes and those are critically important but it is that relationship between inputs and outputs and how we drive that forward is really the focus of our work. 

So, what have we found?  Well, the first thing to say is what data did we look for? I thought this was fairly naively coming into the sector a relatively straightforward request.  We asked people what services did they deliver, how many FT did they have and what their key outputs were. After about doing 4 months of analysis on that data return we got differential returns from providers. Some people were declaring 100 service lines for let’s say a £200 million organisation. Other providers gave us about 1000 service lines. So that helped us really understand the level of consistency or the lack of consistency in understanding the service lines across the sector but it did give us a good sense of the variation in delivery models at provider level and it really brought home actually the extent which mental health Trusts play a critical role in the provision of these services.  We then took the care of the patient day methodology and had a look at inpatients to see whether it would apply and I will talk about bit more about that later and then we started the productivity of community services.  This was probably the most challenging area and the area where we have got furthest to go. Then because we are an efficiency programme at its heart, we did look at the enablers, we looked at the state, we looked at procurement corporate services.

I think where we are in summary is we see lots of variation across the system.  That’s not a surprise.  Any system will have large amounts of variation.  What I think, and we have not yet fully proved this is there are large amounts of significant unwarranted variation.  Warranted variation is good. We support that, it helps innovation and drives improvement and drives patient outcomes. Unwarranted variation is really where we need to focus to drive that out with support to make those savings and actually deliver better outcomes.

So, hopefully you can see this, fairly big screens.  So what we did it is, we did an initial data collection on care hours per patient day and broadly this looks at the number of hours we are putting in on our wards relative to the number of patients we have.  What we found both in adult mental health and in particular community, was there was a large variation in the levels of staffing on ostensibly similar types of wards. There was a variation both in terms of the registered staff and HCA staff within those wards. On community was particularly interesting because we found there was no standard nomenclature, there was no standard definitions of what a ward did and what it was trying to achieve, which meant actually comparing like for like was even more challenging.  What this does clearly demonstrate is there is a large variation in the amount of hours and resources we are putting.  I have not actually got the picture up on the slides today.  The other big finding from that, and the one we are taking action on immediately because we do think there are opportunities for relatively quick efficiency to be made, is around rostering.  So when each ward of the 550 that we put together, we asked providers to give us a manual data collection of the number of staff they actually put into that ward against their planned establishment and we got them to do that every single day.  We also asked people to give us the flow through the wards and the sense of the acuity on the mental health side.  Interestingly, that data was not robust enough for us to use but it is something we would like to come back. What we found when we looked at that is there was significant use of bank and agency and depending on the type of ward, often small isolated wards were running well under their core establishment, often wards were running over and you saw quite heavy reliance of bank and agency, particularly sometimes on the weekend.  We also deliberately chose to run it across a half-term and we saw a big marked difference between the week pre-half-term and the week during half-term. What that indicated we felt is your cost per care hour across the sector was relatively high given the amount of hours you were putting into that ward so by using bank and agency, which is more expensive cost per hour, the overall cost of running that system was too high.

We then put in place a rostering collaborative with the cohort, those people who had inpatient wards and we have been working with them to improve their rostering. We started to have some really positive goals, lots of people were not locking down their rosters at 6 weeks.  Lots of people, the classic was local rules so there were so many local rules that the auto-roster system would fail to operate, thereby requiring ward managers to spend hours re-planning their rosters on a very local basis.  We think, and we are getting some fairly positive results from working with the collaborative that by paying attention to rostering both on inpatient and actually in the community as well, we can strip out some of those costs associated with high levels of bank and agency, simply through better planning.  So that’s the first bit we are looking at in terms of care as per patient day. 

The next phase we have gone out to all community providers and all mental health providers to ask for all their date for all wards. We are currently working our way through 1500 wards and we will have similar snapshots, not from a manual collection but from an automated collection showing people’s actual rostering pattern. That is not something we will do routinely and that is not something we will do all the time but we are using it to generate an evidence base to really understand the capacity in the system.  One of the starkest things as I mentioned earlier was when you have an isolated ward, be it mental health or be it community physical health, the costs of running that ward are extremely high and it is very difficult to staff.  That is not to say we should immediately close or isolate wards but that is to say that we should have a look at the structure of the inpatient news across the system with a view to both delivering efficiency and the right patient outcomes. We have also found one ward that has fundamentally broken care hours per patient day.  I always ask this question and I never get an answer, so I am not going to bother this time. The reason why it broke it was because they had no patients, but we had an inpatient unit without any patients at all but appeared to be staffed.  So there are fundamental issues around the configuration of the sector.  It was on a small island to be fair, so there is some legitimacy to that but it does underline the point that we need to think about how we use our capacity in the system. 

Probably the most difficult bit that we have been working on the for the last 9 months is to get a sense of what actually community productivity looks like. There is obviously a great amount of information from NHS benchmarking and we have been working closely with benchmarking to understand and get access to that data.  What we wanted to do was look at a much more granular level than NHS benchmarking allows. So, we cohorts I think broadly willing acceptance we pulled all the date from the core IT system to that community productivity. That gave us something like 37 million lines of data to work through and then we had to map ostensibly similar types of services together to understand actually what is happening in those services.  It was really interesting, but we do not show it on this particular chart was that we actually got data back from around 18/19 Trusts. We could only process data for around 14 so the data quality, even in the core systems simply is not there at the moment really to support a proper view of productivity.

So, with all the caveats, what does that say?  So community nursing we had a look at, you see a variation from around 10/11ish contacts per day at their height down to about, well it says 3 at the bottom.  I think it is fair to say we discussed it with that Trust.  What was clear is they only managed to load a quarter of their data on so probably the better benchmark is let’s 9-6 contacts per day. What we also found is the face to face contact duration varied quite a lot.  We kept the provider on who had no time because that provider chose to switch off actually its time recording system.  Then you get patient time per clinical work day.  Again, you see that variation between 56 to around probably reasonably 39%.  So we are working through a validation exercise but I think that tells me a couple of different things.  That level of variation is too broad to be reasonable. The variation in contact time and contact duration is probably partly explained by the way people are recording it.  But actually, we need to get a much tighter understanding of what people are delivering when they are in somebody’s home and the patient time per clinical work day I think is really, really interesting because where we see services which are mobile enabled where people have ideally the ability to make notes when they are in their day when they don’t come into base in the morning, they start at the patient’s home you see a bigger increase in that patient time per clinical work day.  So where you see what I would describe as a much more modern operating model as opposed to a traditional small team model, you get more value out of that system and when we compare it to patient outcomes, particularly patient experience we do not see noticeable differences so we think there is scope to drive this.  I suppose my question to the sector is as we work through it will be, not only can we shut that benchmarking converge around the average and push people upwards, but we need to change fundamentally the benchmark and how do we think about actually making it easier for people to do their jobs, reducing the administrative burden going forward.

So that is kind of where we have got to on community nursing.  We have similar types of graphs around MSK and we have the capability to produce them across a range of community services.  What this of course does not tell you is the outcomes that are delivered. That is the next phase of the work but at the moment, we are, as I said, focusing on the nuts and bolts and really getting underneath as if you are reducing your amount of administrative time you are planning your routes effectively, you maximise your client care and at the moment we do not think that has a significant enough focus within the system.  I have put it on there because people might be interested.  We did a similar thing in adult mental health teams and we have got similar for older adults and CAMHS, we see a very similar story in terms of variation although the benchmark is perhaps understandably a little bit low.

In terms of the wider enablers, we had a look at the estate status from the gloriously named Eric data return, I cannot remember what that particularly stands for, but it is the return that all provider Trusts have met around their estate.  I think what we found was a huge level of variation.  At the moment, I am not able to say what is warranted variation and what is unwarranted because people’s bed base is quite different between different organisations. So if you are a Trust with no inpatient beds I would expect you to be right down the bottom. If you have a significant inpatient, I would expect you to be much closer to the top.  What we do see however, is some Trusts have been successful in consolidation their estate quite significantly as part of a wider transformation programme.  Others continue to have a very large number of sites which we think are probably unproductive. Corporate services across the sector, most Trusts are relatively small and what we see is a general relationship between size and cost.  So, the smaller you are, relatively speaking, the higher costs you have.  There are choices people can make about whether they make or buy their transactional services and we think there is scope to drive more savings and corporate services through collaboration and basically becoming a bigger entity.  Producing a payslip is producing a payslip.  It is not right that some people cost £5.66 to produce that payslip and some people spend £2 producing that payslip. That is money I think we need to drive out the system to ensure we put more into patient care. We see a similar story in procurement.  So the example we have is some people are spending £22 on a bandage, other people are spending about £4 on exactly the same bandage so we think there is stuff to go out which is cash releasing either in the short term or the medium terms which can be recycled into provider care. 

That’s just a bit of a snapshot of the data and I am conscious of my time and Anna will probably start looking at me in a second on time so I will whip through some of the next steps. 

One of the things we have done is try to locate ourselves within the wider NHS. I think we play a key role in supporting the broader vision because if we can improve operation efficiency of the community sector and actually demonstrate its value, including in developing a clear view of what is working and what is not working, we can start having a proper evidence based data led conversation about the value of community and we think that plays very heavily into the 5 years for review, local STPs and of course financial improvement I am from an HSI so therefore I think I am contractually obliged to mention that at least once.

In terms of building approach, we want to drive out CHPPD, we think we have the evidence base and we are looking to role that out across the sector.

On community productivity, I think this is a long game.  I think we need to focus on definitions and we need to get some consensus around the definitions, the data standards we need to apply and then the position of the sector to respond to those at the moment.  If came out with a very clear data request I do not think the sector could consistently have the ability, technical capability to respond to those data request and we need to work over that over a few years and we are also looking at the areas I mentioned earlier.

Medicines and pharmacy, I think is a key one and we are looking at work to see both how do we optimise the clinical model and actually how do we drive some value in the procurement chain. 

So how are we going to get this out to the provider sector?  We are working on a model hospital, it is there for acutes and we are building the acute. We have now made it capable of, so previously when I started it was not accessible with community mental health Trust, it now is accessible and over a period of time we are looking to extend relevant compartment as they apply to the community and mental health. We have just loaded the corporate services data on, we will load the next round of the Eric database and we will build out.  We also need to do a lot of work to expand the clinical service times to cover the community based clinical service lines. That will be done as and when we start generating that reliable date but it is work in progress and it will done intuitively.

Outcomes and pathways: we are working with the getting it right first-time programme and we are interested in exploring how the work they are doing in predominantly the acute sector needs to start thinking about community based care.  That will be a challenge because of the data, and I think it is the right challenge to present back to them and we are also starting to set some work up around mental health.  We have got a specific programme of work around wound care. I think I have found about 10 different numbers around how much money we spend on wound care across the system.  One of the things we are very clear on and we asked a lot of providers about it, was their grip on that particular pathway.  I think it is fair to say was varied.  The amount of people that routinely collect the numbers, type, the heal rates and had very clear clinical pathways when we talked to our Trust was very limited and we are fairly sure there is quite a lot of scope to do work across that pathway to drive out money and deliver better outcomes.  The way we are thinking about it is, we can do some pre-work as a precursor to a full getting it right first-time exercise. 

The final bit I wanted to touch on before opening out to questions was around enabling productivity.  We have, in all of our conversations, inevitably had lots of conversations about commissioners rather than with commissioners, though we have started to talk to commissioners, obviously, there is a huge range of commissioning specifications and KPIs.  A particular example I always use is there was one KPI for one and half FTE.  That particular organisation we are talking about had 9000 staff, so you can probably do the maths yourself about how many KPIs that organisation was having to respond to. That costs money and is almost impossible to work through.  So, we are talking to NHS England and how we get a smarter, simpler commissioning model and get more standardisation because without that standardisation it is going to be hard to drive efficiency through the sector in my view.  We need to get smarter on demand in capacity modelling.  There are some Trusts out there with some brilliant demand in capacity models.  Other Trusts do not have that capability and we think that is fundamental in terms of running an effective efficient business and then finally, the focus on information technology and digitally enabled. I do not come from the NHS but I was surprised when I joined the NHS around the state of digital roll out, particularly within the community sector.  Some Trusts have got it, they are delivering it, even those Trusts without mobile and digital capabilities say they have got further to go to derive out the benefit, yet I see other services that are still operating on paper, the patient experience is incredible poor and the productivity I think is very poor and it is not as uncommon as I was anticipating.  That I think is probably the key to unlocking operational efficiency within the system overall.

All are showing conclusions before I open it up.  We are part way through it, getting the data has been a real challenge but we feel like we are making some progress towards that. I think there is a broad support from providers to the work we are doing though inevitably some challenges and I am sure some of those providers we are working with in the room may want to make a comment on that and we think we are on the right track but this is nowhere near finished so I really welcome views from you guys about the direction of travel and where you see the key challenges