Lessons from modelling longer-term depression in Sheffield. A discussion paper.

ABOUT DEPRESSION Depression is a common mental health problem, which often has debilitating effects on the quality of life for the individual. It also has a very large impact on society, with yearly costs of about £9billion, of which just under half a billion pounds are spent by the NHS on the treatment of depression. Although depression has been historically viewed as an acute, one-off condition, for many individuals it is a chronic condition and they will experience multiple episodes of depression during their life. Because the treatment of depression can be very expensive and because many people experience depression for a long time, the effective organisation of health services is very important. Recent NICE guidance recommended the use of a stepped-care model (SCM), where individuals typically start by receiving the most cost-effective treatment, ‘stepping-up’ to more costly (and more effective) treatment if current treatment doesn’t work. However, the NICE guidance didn’t state how services for depression should be structured within an SCM, and how patient’s should interact with these services. The result of this is variation in the pathways-of-care experienced by patients with depression; with the implication that some pathways may be more cost-effective than others.

EXISTING WORK ON DEPRESSION Within South Yorkshire the Collaboration for Leadership in Applied Health Research and Care (CLARHC) is a programme of work aimed at translating innovative healthcare research into healthcare practice. The depression theme of CLARHC is running a project known as ‘Improving Quality and

Effectiveness of Services, Therapies and Self-management’ (IQuESTS). The IQuESTS project considered existing pathways of care for people with longer-term depression in Sheffield, and if these could be improved. Structured interviews with service users were conducted to gain an understanding of selfmanagement techniques and people’s experiences of NHS services for depression. Concurrently, a mathematical model was built to reflect existing NHS services of care in Sheffield for people with longer-term depression. This model, the ‘Sheffield Model’ was designed to be flexible – so that the effects of potential service re-designs could be explored – and to include outcomes that measure the overall cost-effectiveness of the entire pathways. These outcomes included the costs to NHS of treating depression, and patient experience in terms of health-related quality of life. Outcomes within the Sheffield Model are available for entire pathways of care, and for sections within the pathways (such as outcomes for the tertiary sector). The Sheffield Model was been developed through interviews with local service experts (managements, service users, clinicians and academics) as well as the relevant literature. These were used to first create a conceptual model. A conceptual model is a visualisation of current knowledge about the NHS services of care – it is meant to be transparent and easy to understand so that all service experts can contribute to its development even though they may come from very different backgrounds. The conceptual model is then used to help decide what should go into the mathematical model, and in how much detail. Inputs to the mathematical model came from Sheffield data wherever possible, if data were not available then the literature or expert opinion were used. The Sheffield Model was used to evaluate the following potential service re-configurations: 

Fast-track self-referral back to therapist,



Reducing drop-out rates from treatment,



Improving access to non-therapy services.

It was found that the cost of implementing any of these re-configurations would outweigh any future costs avoided (so each option would result in an increase in costs to the NHS), but they would all

result in improved patient outcomes (specifically, improved health-related quality of life). These three options are all currently being piloted in Sheffield.

MOTIVATION FOR THE CURRENT WORK Having a mathematical model of current NHS services of care for people with longer-term depression proved to be very useful for the IQuESTS project. Because there is variation in the implementation of SCMs across England, it was felt that it would be of use to other areas too. However, the current mathematical model is focused on services of care provided in Sheffield. Hence the aim of this piece of work was to explore the following:  How useful would the current Sheffield Model be for other places?  How easy would it be to adjust the Sheffield Model to accommodate the care pathways and data from other places? In addition, the Sheffield Model was built primarily from a service-provider point of view. Hence two further questions to be explored were:  How useful would the current Sheffield Model be for service commissioners?  How useful would the current Sheffield Model be for service users?

WORKSHOPS To date two workshops have been held. The first workshop included commissioners from Sheffield, Rotherham and Barnsley. The second workshop included commissioners and providers from Rotherham, Doncaster and South Humber (RDASH). Both workshops also involved core members from the IQuESTS project. Briefing materials were sent-out prior to each workshop along with a list of suggested questions for discussion (these are replicated in the appendices).

FINDINGS FROM WORKSHOP 1 – THE COMMISSIONER VIEW

STRENGTHS OF THE EXISTING MODEL The work on conceptual modelling generated a lot of interest. The conceptual model of current services of care for longer-term depression was designed to facilitate discussions between commissioners, providers and clinicians. This was felt to be of particular importance since the NHS is undergoing (yet another) culture change. It was noted that the work on conceptual modelling was also very similar to work on process mapping that commissioners engage with. It was felt that results from the conceptual modelling would be useful in emphasising the following important points:  For many people depression is a chronic illness – there should be stronger ties with the Quality, Innovation, Productivity and Prevention (QIPP) long-term conditions work-stream.  There are a lot of people who are either never seen by NHS services, or who shimmer in and out of the system (due to drop-out).  The need to consider the entire pathway of care to integrate services and personalise them for the user. It was also noted that because it is possible to assess outcomes for specific parts of the pathway, The model could help with market testing and service specification, informing negotiations between commissioners and current providers. Because it provides a systematic approach to making decisions, the model could be used to help plan evidence-based performance metrics such as local CQUIN targets based on reducing DNA rates or improving referral rates into the system.

LIMITATIONS OF THE EXISTING MODEL The main limitation identified with the model was that it did not include mental health clusters, which are to be used within Payment by Results (PbR). Pathways for the mental health clusters are currently being implemented locally, whilst the model may help with this it is likely that model re-design would still be required. In addition, it was noted that the framework for mental health PbR still had many

unanswered questions (such as how IAPT services would be included). It was stressed that PbR was about delivering the right packages of care at the right time. Another important limitation to the model was that it did not model severity of depression. This is typically measured by PHQ-9 score. It was felt that including this could be very useful as there are likely to be regional variations in severity of depression – both for those presenting to NHS services and for those who fail to present. These variations are likely to affect both the design of services and their effectiveness. Including severity of depression would also make the model more realistic, as the decision to step-up is influenced by PHQ-9 score, and there is likely to be an association between depression severity and quality of life. A third identified limitation was that whilst the Sheffield Model was useful in drawing attention to the whole pathway of care for people with longer term depression, the model did not consider co-morbid disease which is a particular problem for this patient group. Co-morbid disease may mask the depression itself, or its treatment may take priority over that of depression. It was noted that any additional work would need to be responsive to the typically short time-scales involved in commissioning cycles. It was noted that the model could be further refined by modelling differences due to ethnicity and levels of deprivation.

FINDINGS FROM WORKSHOP 2 – USE OF THE SHEFFIELD MODEL IN RDASH GENERALISABILITY OF THE EXISTING MODEL Whilst there were similarities between the Sheffield Model, and pathways in RDASH, a number of differences were discussed:  To improve access to services, particularly for the elderly, the voluntary sector (such as Age UK) has ‘befriending’ services. Hence many people with depression may not enter primary care.

 Secondary and tertiary care are organised differently: there are no community mental health teams and some tertiary care is delivered locally whilst some is commissioned from Sheffield.  The main role of GP contact is to direct patients to the appropriate therapy (in the Sheffield Model it is to treat with antidepressants). It was felt that it should be possible to use the existing mathematical model (with its stepped care), but change the pathways within each step to reflect the RDASH experience. It was noted that the main requirement for this was the availability of data about effectiveness in RDASH to populate the model. If the model could be adapted then it would be possible to use it to help with the local implementation (creation) of care pathways for mental health clusters, in particular cluster 5, which relates to patients with non-psychotic disorders (very severe). Results from the mathematical model would inform a ‘Case for Change’, which would tie-in with QIPP initiatives: are the current resources in the right place? In addition, the following important points were noted:  It is important that any implementation involves GPs from the offset. This would make the resulting product more acceptable (instead of appearing to be an RDASH product forced onto the GPs). It would also provide an important perspective from the start of the project.  It is important to be aware of, and fit in-to, commissioning cycles. Results of the economic modelling would be required by October to inform the 2014/15 commissioning cycle.  Care should be taken with the data used in the mathematical model, as poor data can be cited as a reason for ignoring the results of the economic modelling.  Visual representation of the care pathways (as used in the mathematical model) would help to facilitate discussions between providers and commissioners.

OTHER POINTS DISCUSSED

 Rotherham and Doncaster are one of the leading centres for the adoption and dissemination of cognitive behavioural analysis system of psychotherapy (CBASP) in the United Kingdom. This is not currently included in the model.  The importance of PHQ-9 score (which is not in the model) in directing patient’s pathways was again stressed. It was also noted that outcomes (such as drop out or response to therapy) are likely to vary with PHQ-9 score. However, it was noted that this measure has a number of limitations:  It can be ‘gamed’ – i.e. people responding to get high scores to receive benefits.  It can be overly sensitive - e.g. people with economic worries may be classified as depressed.  It doesn’t relate very well to health-related quality of life.

LESSONS LEARNT The objectives of this piece of work were three-fold: 1. Find out how useful the Sheffield Model is to areas outside of Sheffield, 2. See if the focus on service-providers is useful to service-commissioners, 3. Decide if the Sheffield Model may be of use to service users.

Work on the last objective is on-going, and so not reported here. To find-out the usefulness of the Sheffield Model for other areas a workshop was held with commissioners and providers from RDASH. Results were promising and suggested that whilst the specific implementation and organisation of care pathways differs between Sheffield and RDASH, the commonality of having a stepped-care model (resulting from the NICE guidance) makes it likely that the Sheffield Model may be adapted to be of use to other areas in a relatively straight-forwards manner. The mathematical model is flexible, and because it explicitly models each NHS service of care delivered during a patient’s pathway it is easy to identify where changes need to be made.

A limitation that was encountered during the development of the mathematical model was the lack of data for certain parts of the pathway – in particular no local data were available (at the time) for secondary and tertiary care. Similarly, the availability of local data is one of the main difficulties that may limit the usefulness of the Sheffield model when adapted to other areas. Current work is underway exploring how best to identify and acquire the data needed for a local adaption. Another potential difficulty in adapting the Sheffield model to other areas is the need for local service-expert engagement. Pathways of care for people with longer-term depression involve multidisciplinary teams including service providers, commissioners, and general practitioners. It is important that representatives from each of these groups, along with service users themselves, are included at as early a stage as possible. Finally, it has been noted that within the NHS, and especially commissioning in the NHS, timescales are generally short. It is necessary that any local adaptions fit-in with these timescales.

A separate workshop was held to explore the usefulness of the Sheffield Model to commissioners. The conceptual modelling work was identified as being very useful and interesting. Conceptual modelling is not routinely carried out alongside mathematical modelling, especially when looking at the cost-effectiveness of care pathways. However, some pioneering work is being done in Sheffield on creating a systematic framework for conceptual modelling, and so it is recommended that this be done more frequently. With regards to the actual mathematical model, it was recognised that the operational issues that it addresses were different to those faced by commissioners. In particular, mental health clusters (and more generally PbR) and depression severity were key omissions. In contrast to a local adaption of the existing model, adapting the model to incorporate mental health clusters and depression severity would represent a much larger task. This suggests that it may be preferable to have two separate mathematical models. One model would – like the existing model – focus on care pathways from a

provider perspective, whilst the other model would focus on cluster pathways from a commissioner perspective.

Appendices.

Agenda; Workshop 1

SESSION 1: 09:00 – 10:30 Introduction to the workshop: aims and objectives. Overview of IQuESTs and Work Package One. Introduction to the Sheffield service model. Short case-study: evaluating a service-change. Initial discussions: uses for the Sheffield service model.

COMFORT BREAK: 10:30 – 10:45

SESSION 2: 10:45 – 12:00 Feedback on the initial discussions. In-depth discussions: uses for the Sheffield service model. Feedback on the in-depth discussions. Details of follow-up work.

Agenda; Workshop 2

Agenda item

Time

Introductions (who we are and what we do).

9:30-9:40

The IQuESTS project – work to date.

9:40-10:00

Group discussions. For the existing model:

10:00-10:45

 What are its strengths and limitations?  What would be useful to disseminate? To who?  What are the differences between care pathways in Sheffield and Rotherham?  What are the possible uses for this model in Rotherham? Extending IQuESTS – work to date.

10:45-10:55

Group discussions. For any future work:

10:55-11:25

 How best to incorporate co-morbid physical disease?  Would this approach be useful for other mental health conditions?  What else would you like to see included? Summary, next steps and finish.

11:25-11:30

Briefing materials circulated.

IQuESTS Work Package 1: Modelling of care pathways

Frequently asked questions The Improving Quality and Effectiveness of Services, Therapies and Selfmanagement (IQuESTS) project is the joint effort of health researchers,

WHAT IS IQUESTS

service users, care providers and NHS funders. We are looking to improve the quality of care for people with longer-term depression in South Yorkshire by translating innovative research and knowledge into action. Work Package 1 (WP1) is a mathematical model of the whole NHS service of

WHAT IS WORK

care for people with longer-term depression. The model can be used to test

PACKAGE 1?

out changes anywhere in the system, and see the overall impact that these may have on both the NHS and on people with longer-term depression WP1 is based on the care pathways experienced by people with longer-term

WHAT DOES WP1 MODEL?

depression in Sheffield. This takes the form of a ‘stepped care model’, with people starting in primary care, and stepping-up to secondary care followed by tertiary care, if necessary. We used data from the literature for the epidemiology of longer-term depression. Where possible, data on the effectiveness of existing services

WHAT DATA ARE USED?

came from the services themselves. However, these data are incomplete, so the literature and expert opinions were also used including that from services users and those who experience depression.

Estimates of costs and usage generated by the system (or alterations to it),

WHAT ARE THE OUTPUTS?

broken down by provider (for example; GP, CMHT). Patient outcomes are also modelled, such as the average number and length of depressive episodes, time to relapse and quality of life measures. As with any model, the main limitation is the quality of data and evidence

WHAT ARE THE LIMITATIONS?

available. For example, the data covering secondary and tertiary care are very limited, and differences due to ethnicity or deprivation are not modelled. WP1 also currently only models the Sheffield service of care. The model may be used to estimate the likely costs and impact of any

HOW CAN I USE THE INFORMATION?

proposed changes to existing care pathways. By modelling the whole pathway it is possible to see how changes targeted at one level (such as primary care) will affect other parts of the system. It has the ability to weigh up potential options for change and their impacts. Visit us at: http://clahrc-sy.nihr.ac.uk/theme-iquests-introduction.html or contact one of the theme leads for more information on a particular topic:

HOW CAN I FIND OUT MORE?

IQuESTs: [email protected] and [email protected] WP1 modelling: [email protected] WP1 implementation: [email protected]

Improving NHS services of care for people with longer-term depression. The Improving Quality and Effectiveness of Services, Therapies and Selfmanagement (IQuESTS) project is the joint effort of health researchers,

WHO ARE WE?

service users, care providers and NHS funders. We are looking to improve the quality of care for people with longer-term depression in South Yorkshire by translating innovative research and knowledge into action. We have developed a mathematical model of the current NHS services of care for people in Sheffield with longer-term depression. Using this model we can experiment with changes to the care pathway, assessing their impact

WHAT CAN

on outcomes such as total costs to the service, average number and length

WE OFFER?

of depressive episodes per person, rates of presentation to and drop-out from the service, and the effectiveness of current services. This means we can ‘test out’ ideas about changes in the care pathway, and their potential impacts before putting changes in place in the real world. A ‘stepped care model’ is used; people start in primary care (GPs and IAPT), stepping-up to secondary (CMHT) and tertiary (SPS) care if required.

WORK TO DATE.

To date we’ve modelled the current service of care, and 3 potential changes: 1. Self-referral back to therapist after discharge. 2. Better management and prevention of drop-out. 3. Widening access to non-therapy services.

We are eager to hear your views. In particular, we would like to know:

HOW CAN



What information would you like to receive from the model?



What potential changes would you like to see modelled before putting

YOU HELP?

them in place in the real world? 

How can we make the model and its output easier to understand?



What improvements could we make to the model?

We will be holding a half-day workshop at the University of Sheffield on the 3rd of July. Further details about the workshop will be sent out closer to the event.

THE

During this workshop we would like to hear your views and questions about

WORKSHOP

the Sheffield service model, and the information that you would like to know about it. In particular we would like to know what service changes you would like to see modelled and how these might help with planning future care pathway plans, local CQINS targets etc. Along with this overview you should have also received a ‘frequently asked questions’ document and a brief report summarising our work to date.

FIND OUT MORE…

If you have any questions please email: [email protected] More information is also available at:

http://clahrc-sy.nihr.ac.uk/theme-iquests-introduction.html