Decommissioning health care: identifying best practice through primary and secondary research a prospective mixed-methods study

Summary of the extant literature

At the time of commencement of the research, a number of structured evidence reviews had been completed or commissioned in which the authors identify practical constraints caused by poor indexing of decommissioning and its synonyms. 17 In order to build on work under way and to avoid these well-documented indexing problems, we opted to collate and synthesise learning from existing literature reviews.

Our approach to synthesising the existing literature therefore took the form of a ‘review of reviews’. The aim was to distil relevant messages from the evidence base on decommissioning and how this relates and applies to the specific context of the English NHS, as well as to identify knowledge gaps and unanswered questions to be pursued in subsequent phases of our research. To these ends we analysed existing reviews from the period 1990–2016 to address the following questions:

• How are terms such as ‘decommissioning’ and ‘disinvestment’ employed in the literature?

• What are the current and previous levels and types of health-care decommissioning as reported in previous studies?

• What are considered to be the main determinants of successful decommissioning programmes?

• What models and frameworks are available to guide decommissioning and how have these been evaluated?

• What are the remaining knowledge gaps in terms of evidence and practice?

Search terms were identified from a scoping search of the literature drawing on the expertise of the authors and project advisors (Table 1 and Figure 2). Searches were carried out at the beginning and end of the project to ensure that results were up to date at the time of writing and that early results were available to inform subsequent phases of research design and analysis. Existing reviews were identified in the first instance through searches of health databases. From the initial list of search terms, ‘termination’, ‘replacement’ and ‘closure’ were removed from the searches of MEDLINE as they generated unmanageable volumes of results. The term ‘health’ was added to searches of the non-health-specific database (Social Sciences Citation Index). Snowballing searches were also conducted, including citation analysis and bibliography scanning.

FIGURE 2.

Filtering flow chart.

Synthesis of the six included literature reviews took the form of structured data extraction and tabulation against the five research questions (see Appendix 1). As the included reviews contained studies with a range of methodologies, our approach to synthesis was primarily descriptive, involving presentation rather than translation of main findings. 19 Results are presented in narrative form in Chapter 4.

Included reviews were confined to health settings and this is reflected in our synthesis. Although the scope for inclusion was wide in other respects, included reviews were somewhat skewed towards a narrow range of decommissioning types and were drawn largely from cognate disciplinary fields. In recognition of this, additional literature items uncovered during the search and retrieval process were summarised and incorporated in an unsystematic fashion. These were selected for inclusion if they were judged by the research team to complement the included reviews, for example by focusing on a wider range of decommissioning types or on relatively neglected stages of the decommissioning process (e.g. implementation). However, this process was opportunistic and non-replicable and a more in-depth evidence review was beyond the scope of our research. We put forward recommendations for future evidence synthesis in the conclusions chapter of this report (see Chapter 8).

Mapping the decommissioning landscape

In order to develop our understanding of the current decommissioning ‘landscape’ we conducted a mapping exercise of the roles and remit of agencies that might be expected to play a part in decommissioning in the NHS. No further sampling logic was employed. A list of potential organisations was drawn up by the research team and the project advisory group. Telephone interviews and/or searches of websites and official documents were carried out by two of the authors (JH and IW) in relation to each organisation, with the aim of establishing:

• current roles and responsibilities with regard to decommissioning

• current and planned decommissioning-related activities

• perceptions regarding the challenges facing those leading decommissioning

• any available good-practice guidance and other resources.

One interview was conducted at short notice and therefore was not audio-recorded, with note-taking carried out instead by the interviewer.

A key intended outcome of the mapping exercise was to inform the design of subsequent work packages and to provide a context to results emanating from the survey and case studies. Table 2 presents a list of organisations included and the method of data collection employed in each.

Collecting qualitative decommissioning vignettes

To augment these assessments of the published research and the decommissioning landscape, we compiled a sample of retrospective accounts from individuals who had led decommissioning programmes within health and social care contexts in England. These narrative vignettes were collected through seven semistructured interviews carried out by two of the authors (JH and IW) with an opportunistic sample of local leaders of decommissioning processes identified through early mapping work and other suggestions of the advisory group. 20 The aims were to:

• explore the stages and activities involved in a sample of recent decommissioning journeys

• identify the actors and agencies involved in local processes of decommissioning to inform sampling for case studies conducted in work package 3

• gain additional insight into the challenges, contexts and determinants of contemporary decommissioning processes in the NHS and related service areas.

Interviews were semistructured so as to enable core issues to be covered while allowing new themes to emerge. Three interviews were not audio-recorded at the request of interviewees, although note-taking was carried out. Table 3 provides further details of data collection.

Data from the mapping interviews and from the narrative vignettes were gathered, stored and analysed in accordance with best practice. Coding was carried out by two members of the research team (IW and JH) and differences resolved in discussion with a third member (GR). The primary aim of the mapping exercise was to help understand the wider context within which decommissioning activities were enacted. The primary function of the vignettes was to sensitise the research team to themes and issues to be explored in subsequent work packages. Findings from these two activities are presented in short descriptive form in the following chapter of the report (see Chapter 4).

Delphi study of research, policy and practice opinion

To help fill some of the gaps in the published evidence as regards ‘what works’ in decommissioning, we carried out an international Delphi study of expert opinion. Delphi surveys build a consensus through iterative questionnaires sent to a panel of experts and are effective in establishing a consensus in complex topic areas. 21 Developed by the RAND Corporation in the 1950s, the Delphi method was used to forecast the emergence of new technologies22 and has subsequently been used to establish research priorities in health care,23–28 as well as to generate a consensus on policy issues. 29–31 Delphi studies on policy themes share key features of the original approach including multiple rounds in which data are analysed and fed back as the basis for subsequent rounds with at least one opportunity for participants to revisit and revise their judgements on the basis of wider group responses, and anonymity for the participants who never meet or interact directly.

The first round of the Delphi study usually involves participants suggesting factors or cues that form the basis of subsequent closed questions. Participants are then sent a second questionnaire asking for their individual views on the items that they and their co-participants suggested previously. Responses are collated and returned to the participants in summary form, indicating both the group judgement and the individual’s initial judgement. Participants are then given the opportunity to revise their responses in the light of group feedback. This process may be repeated a number of times before the judgements of the participants are aggregated. 27

In this study, individuals were purposively selected from three groups (total n = 30) with expertise on the topic of decommissioning. These were researchers, policy-makers and regulators, and commissioners and providers of health-care services (Table 4). In this context ‘experts’ were understood to be individuals with experience in decommissioning in one or more of these capacities.

Candidates for the ‘research’ group were identified through the previously described searches of the published literature. Candidates for the ‘policy’ and ‘practice’ groups were identified through desktop searches and nominations from an international advisory group for the research project. An initial list of approximately 100 target respondents from the UK, Europe, North America, Australia and New Zealand were invited via e-mail to participate (see Appendix 2). Of these, 30 agreed to participate. As Table 4 shows, participants were drawn from the UK, Australia, Canada and the Republic of Ireland, although the final sample contained an over-representation of UK respondents, especially in the ‘practice’ category.

Participants were requested to consider, define and rate criteria and factors that shape the process and outcomes of decommissioning programmes. They were invited to complete three online rounds (with 1 week for completion of each one) and to suggest examples of ‘best practice’ in decommissioning. Open comment responses were analysed by the research team using open coding and constant comparison. Similar codes were grouped to identify concepts emerging from the data. A consensus was statistically operationalised by measuring whether or not group ratings were strongly polarised (e.g. 50% of respondents strongly agreeing and 50% strongly disagreeing with any statement is a strongly polarised distribution).

In round 1, participants were asked for up to five nominations for each of the following:

• considerations that should inform decisions to implement decommissioning

• considerations that do inform decisions to implement decommissioning

• factors that positively shape the process of decommissioning

• factors that negatively shape the process of decommissioning

• factors that positively shape the outcome of decommissioning

• factors that negatively shape the outcome of decommissioning.

They were also invited to suggest best-practice recommendations for the implementation of decommissioning decisions. Open comment fields enabled respondents to explain and/or justify their suggestions, and to raise any other issues. The anonymised, cumulative responses were then fed back to the whole panel to inform the design of round 2.

For the second round, participants were requested to rank their level of agreement with statements derived from round 1, using a four-point Likert rating scale (strongly disagree, disagree, agree and strongly agree). They were also asked to rank relative importance of the factors proposed in round 1 in shaping the process and outcomes of decommissioning, again using a Likert rating scale (very low importance, little importance, high importance and very high importance). Open comment fields enabled respondents to explain responses. The research team calculated the level of consensus in relation to the total of 88 rating scale questions in round 2 using the following thresholds:32

• high = 70% of ratings in one category or 80% in two contiguous categories (e.g. agree and strongly agree)

• medium = 60% of ratings in one category or 70% in two contiguous categories

• low = 50% of ratings in one category or 60% in two contiguous categories

• none ≤ 60% in two contiguous categories.

Anonymised, aggregated responses relating to factors with no consensus or a low level of consensus were then fed back to the Delphi panel to inform the design of round 3.

For the third round, participants were invited to reflect and comment on the round 2 results for the 17 statements that achieved no consensus or a low level of consensus, and to rank their level of agreement with the 17 statements, having had an opportunity to review the results from the panel as a whole. The final outcomes from rounds 1 to 3 were fed back to all participants and an opportunity to make further open comments provided.

Case selection

We selected four case studies in which a planned and explicit approach to decommissioning had been adopted. These were at varying stages of progression (as mapped against the stages model) in order to enable us to follow decommissioning journeys from initiation and development to implementation and, where possible, direct and indirect outcomes.

To identify potential case studies, we adopted a snowballing approach using existing contacts from the project principal investigator, advisory group, decommissioning narratives, survey responses and mapping exercise. Using these approaches, we developed a matrix of potential case study options structured around four sampling criteria (Table 5), intended to capture a diversity of decommissioning activities:

1. geography – including programmes implemented in both rural and urban settings in England

2. scale and complexity – including decommissioning programmes that vary from the relatively simple (e.g. implementation of National Institute for Health and Care Excellence (NICE) disinvestment guidance) to the highly complex (e.g. reorganisation of services across organisational and sector boundaries)

3. conflict – including programmes with a high degree of stakeholder buy-in and support and others for which there are currently (or it is anticipated that there will be) high levels of resistance and stakeholder challenge

4. programme instigation – including decommissioning programmes where national bodies play an important role and others that have been instigated and led entirely by local organisations such as CCGs.

Based on these provisional sampling criteria we selected the following case studies.

1. Reorganisation of specialist services for paediatric burn care in England.

   This case study was selected to fulfil the criteria of being a nationally led, complex reorganisation process involving planned service removal or reduction. The case was identified through the national mapping exercise, and access was negotiated between the project principal investigator and NHS England. In these conversations it was established that the burn care reorganisation would incorporate the decommissioning (by downgrading) of some services and that it would be closest to the ‘programme design’ phase at the commencement of fieldwork. As a national reorganisation it was anticipated that it would generate relatively high levels of conflict and that the specialised services sites would be in predominantly urban areas.

2. CCG-led decommissioning of an end-of-life (EOL) home support service.

   This case study was selected to fulfil the criteria of having relatively low complexity and low conflict (although we could not be sure of this at the outset) and of being in an urban area and closest to the implementation phase of the decommissioning process. The CCG was initially identified as a potential case study through networks of the research team and access was negotiated with the CCG programme manager for the decommissioning programme.

3. Decommissioning activity by an Area Prescribing Committee (APC).

   This case study was selected to fulfil the criteria of being rural, and having low complexity and relatively low conflict (although again this could not be fully established prior to commencing the case study). The intention was to observe implementation of NICE guidance by the committee and any additional decommissioning undertaken. The APC was identified following the mapping exercise through an e-mail invitation to a number of APC representatives in suitable areas, asking if they wished to showcase work in the area of decommissioning. Access was negotiated through the APC chair and research lead for the host CCG.

4. CCG-led review and planned reorganisation of local primary and acute care services.

   This case study was selected to fulfil the criteria of covering an area with rural parts, which had high levels of scale, complexity and likely conflict. This case study was closest to the decision-making phase of our stages model. Access was negotiated with the programme manager for the service transformation programme following an initial approach to the research team.

Data collection

For each case study, the primary unit of analysis was the decommissioning process itself and we compiled narrative accounts of the programme of work intended and/or under way. We employed non-participant observation techniques as used successfully by the applicants in previous research. 38 Detailed field notes were taken to record the processes through which decommissioning plans were identified and drawn up, and the role of decision-making tools and frameworks in this.

For each case study, semistructured interviews were conducted with a sample of those involved (see Table 6 and Appendix 4). In all case studies, the interview sample comprised individuals involved in the design and implementation of the decommissioning programme. For case studies in the early stages of progression, a second round of interviews was conducted approximately 12 months after the initial round. These were intended to update the research team on programme progress. For case studies in the implementation phase (e.g. case study 2) the primary focus of a first round of interviews was on design and enactment of the implementation plan with a follow-up round planned to explore outcomes.

Interviews focused on:

• the origins, aims and intended outcomes of the decommissioning programmes

• decision-making tools and other information used to inform decommissioning programmes

• the web of relationships between internal and external actors and influences in decommissioning design and implementation processes

• the role of key interest groups in decommissioning, including politicians, clinicians and the public

• outcomes, experiences and attitudes towards future decommissioning.

In summary, a total of 59 interviews were carried out in work package 3. This is somewhat less than the 90 interviews anticipated in our project plan. The main reasons for this discrepancy are as follows:

• We placed a greater emphasis than was initially intended on observation over interviews in the case studies, as these provided a rich source of data (Table 6).

• Data saturation was reached early in interviews for case study 3.

• We were unable to gain access to respondents for follow-up interviews in case study 2 as some individuals had moved on from their posts.

Data analysis

Informed by our theoretical framework, we inductively analysed interview data to explore participants’ perspectives and experiences. Analysis involved comparative case study methods and pattern matching. 36,37,39 In order to facilitate internal validity,40 all interviews were fully transcribed and we used qualitative coding software [NVivo version 11 (QSR International, Warrington, UK)] to support data storage and retrieval during the analysis phases. Two members of the research team (IW and JH) were involved in building coding frames for themes from qualitative data and compared independent coding of a subset of data to identify and address coding differences and ensure consistency. All identified themes were discussed at ‘analysis days’ attended by the core project team. External validity and transferability of analysis was addressed through detailed description and data triangulation between work packages. 40 Respondent validation was facilitated through sharing of draft case study reports with respondents (with four individuals taking up the offer to review and comment), and additional clinical expertise was drawn on in case study 1 to ensure the accurate presentation of clinical information.

In our approach to case study analysis we drew on our theoretical framework, including concepts derived from the ANT research tradition. ANT was developed by Bruno Latour, Michel Callon and John Law41 as part of the academic field of ‘Science and Technology Studies’ during the 1980s. Although it has ‘theory’ in its name, ANT is better understood as a range of methods for conducting research that aims to describe the connections that link together humans and non-humans (e.g. objects, technologies, policies and ideas). 42,43 In particular, ANT seeks to describe how these connections come to be formed, what holds them together and what they produce.

In the context of health-care service and delivery, ANT has typically been forwarded as a framework for investigating technology implementations in health-care settings (e.g. 44–46). Aside from technology implementation, ANT has also informed studies of implementing ‘Lean thinking’ in health-care organisations,47 exploring the effectiveness of quality-improvement interventions48 and the evolution of indoor smoke-free regulations (a policy innovation). 49

A key concept in ANT – particularly pertinent to the study of decommissioning processes – is what Latour called ‘the sociology of translation’. 50 By this, he meant that players interact to build heterogeneous networks of human and non-human actors, forming alliances and mobilising resources as they strive to convert an idea into reality. The central focus of ANT in the context of our study is, therefore, the process by which a decommissioning project is ‘brought into being’ through the process of translation and how it changes over time. The picture is always a dynamic one, as actors interrelate, define one another and realise their goals (or not) by mobilising intermediaries, such as technical artefacts, texts, human skills and other resources. 51,52 Translation is achieved by displacements that require discourse and the exercise of power, and it may or may not achieve the desired outcome. This continuous, organic realignment of people and non-human actors is what Latour50 calls the ‘chain of transformation’. What emerges may not be a unified, shared goal; indeed, the concept of the ‘network’ is that different actors often have conflicting goals, and outcomes are the result of struggles between different interest groups and the flow of power through the network. Callon51,52 summarised the process of translation as four ‘moments’ or phases:

1. problematisation – the definition of the nature of the problem in a specific situation by an actor (a group or an individual) and the consequential establishment of dependency

2. interessement – ‘locking’ other actors into the roles that were proposed for them in the actor’s programme for resolving that problem

3. enrolment – the definition and interrelation of the roles that were allocated to other actors in the previous step

4. mobilisation – ensuring that supposed spokespersons for relevant collective entities are properly representative of those entities.

Prior to beginning our study we felt that ANT and the concepts briefly outlined above would enable us to explore decommissioning processes in novel ways by helping to describe the multiple interacting actors and influences in our four case studies; to consider why such actors and processes appear to ‘behave’ differently in different settings or at different times; and to draw attention to the unintended consequences of decommissioning processes (as well as the anticipated ones). We also proposed that applying an ANT perspective would help us to explore several common themes that, from the literature, appear to impede successful outcomes of decommissioning processes, namely:

• one or more interest groups may feel threatened by substantive change

• external actors may feel insufficiently consulted and involved in decision-making processes

• stalemates between coalitions of interests

• the role of power and politics.

In drawing on ANT, we needed to ‘follow the actors’ in our case studies and to analyse how these actors themselves define what is going on. However, it should be noted that the empirical scope of the case studies was not as expansive as is usually found in the ANT tradition and precluded us from, for example, extending our analysis into all levels of health-care decision-making and activity. It would therefore be inappropriate to cast our case studies as full actor–network analyses or to make inferences about the full actor networks at play. Instead, we have drawn on Callon’s51 summary of the translation process to make sense of smaller-scale case studies that were primarily interview led and supported with observational and documentary analysis. We have recast the Callon framework in the following language to be more accessible to practitioner audiences. These form a structure for presenting and analysing findings:

• the role of evidence and other resources in identifying and framing a need for decommissioning

• alliance building (e.g. analysing context, attending to interests and power relations) as part of a decommissioning process

• social acceptance (analysing engagement strategies, attempts to gain wider acceptance) for the solution of decommissioning

• implementation and institutionalisation of a decommissioning decision.

It is important to note that several criticisms have been made of ANT that are relevant to our study of decommissioning processes. Building on Robert et al. ,53 these include the following:

• ANT fails to attend to the various ways in which macro-level structures (e.g. external regulatory bodies or the Department of Health in the context of our study) shape and modify the process of social interaction and practices; by ignoring such institutional sources of power, ANT is criticised for having little to say about the systematic exclusion that prevents some social groups from having a voice in, for example, decommissioning processes.

• How to delineate an actor network (with a view to studying it) presents a methodological problem. The network is open, and hence must be artificially defined by the researcher. There is an argument for not defining the actor networks in advance but rather seeing what emerges as key in any particular study.

• May54 suggests that ANT may also be limited in terms of accounting for everyday micro-level practice and assisting with practical problem-solving.

Focus group discussions

We carried out three deliberative focus groups, each with approximately 10 participants sampled with the intention of achieving a diversity of age, gender and ethnicity. 55 Focus groups involved combinations of citizens/members of the public, representatives of national citizen organisations (e.g. HealthWatch) and patient organisations (general and specific), community organisations and independent third-sector organisations. In the focus groups, some participants also self-identified as NHS patients/service users. The logic of this sampling was simply to recruit individuals more likely to be affected by (as opposed to being responsible for) decommissioning.

Potential participants for the focus groups were contacted through HealthWatch England, National Voices, the Department of Health Voluntary Sector Strategic Partner Programme, Carers UK, patient representative and advocacy organisations such as the National Association for Patient Participation and Shaping Our Lives, researchers with particular expertise and interest in patient and public experience and engagement and individuals involved in patient and public engagement identified in work package 3. No NHS organisations were contacted, in accordance with the terms of our research ethics approval. Participants were not required to have direct previous experience of decommissioning, although it transpired that a substantial proportion of those who consented to take part did have, having been directly affected by decommissioning.

The focus groups were intended to sensitise us to the issues and perspectives and so questions were open ended, encouraging wide-ranging discussion. At least two facilitators were involved in each of the focus groups (IW, SB and JH) and examples of decommissioning scenarios were used to prompt discussion (see Appendix 5). Each discussion lasted for approximately 1.5 hours and the final 30 minutes was taken up with the co-design of questions for a follow-up Delphi study of expert opinion, drawing specifically on the experiences and perspectives of citizens, patient/service users, carers, third-sector organisations and local community groups.

Delphi study of citizen, patient/service user representatives, carer and community groups, and third-sector organisations

Drawing on these insights we implemented a second three-round, online Delphi survey in order to compare views on drivers of decommissioning with those expressed by the first Delphi panel (work package 1) on drivers of decommissioning and to elucidate a consensus on best practice for the engagement of patients and the public in decommissioning processes, from the perspective of citizens, patients/service user representatives, carers and community groups and third-sector organisations. 28,30 The letter of introduction (see Appendix 6) incorporated the suggestions of participants in the focus groups about accessibility, as did the decision to replace the term ‘decommissioning’ with the phrase ‘move or take away services’.

The approach to analysis and consensus-building was as described in Chapter 3, Delphi study of research, policy and practice opinion. The final sample for the second Delphi study comprised third-sector organisations that provide support to, and advocacy/representation of, service users and carers (including, among others, Shaping our Lives, National Voices, Carers UK, National Development Team for Inclusion, the Mental Health Providers Forum, the Voluntary Organisations Disability Group, the National Care Forum and National Association for Voluntary and Community Action); other patient and public representative organisations (including HealthWatch England, National Association for Patient Participation, the Race Equality Foundation and patient expert groups); and selected academics specialising in public involvement and/or patient experience. Following advice from our advisory group, we restricted the scope to the English NHS rather than seeking respondents from other countries. More detail on respondents is provided in Table 7.

Of the 50 individuals and organisations invited, 26 agreed to participate. Female participants made up the majority of respondents (15, 60%). Eighteen described themselves as ‘white British’ or ‘white English’, with five describing themselves as ‘non-white’, ‘European’ or ‘other’. Ten respondents indicated that they had a disability and a further two provided no response to this question. Older age groups were over-represented within the sample, with two respondents aged between 25 and 34 years, four between 35 and 44 years, four between 45 and 54 years, seven between 54 and 65 years and seven > 65 years.

In round 1, participants were asked about their experiences of decommissioning, including good and bad aspects of these experiences. They were then asked ‘What if any do you think are good reasons to move or take away services?’ The same question was then asked in relation to ‘bad’ reasons and the reasons behind such a decision in reality. They were also asked how they would wish to be involved ‘if the NHS was thinking about moving or taking away services in your local area or nationally’. Open-text comment boxes were provided in relation to each of the questions. The anonymised, cumulative responses were then fed back to the whole panel to inform round 2.

In round 2, participants were asked to rank their level of agreement with statements derived from round 1, using the four-point Likert rating scale and consensus thresholds employed in Delphi study 1. The statements related to:

• good reasons to move or take away services

• bad reasons to remove or take away services

• the reasons why these decisions were made in practice.

Respondents were also asked to choose the three reasons that they considered were most justifiable/least justifiable/most influential in practice.

Two more questions asked respondents to review lists of statements deriving from their first-round comments about how they would like to be involved and which methods they thought most important to employ. They were then invited to select the three that they considered most important.

The anonymised, cumulative responses relating to those factors that achieved a low level of consensus or no consensus were fed back to the whole Delphi panel to inform round 3. In round 3, participants were asked to reflect and comment on the round 2 results for the 19 statements that achieved a low level of/no consensus and to rank their agreement with these statements, having had an opportunity to review the results from the panel as a whole. They were also given an opportunity to comment on the combined three highest and lowest ranking statements from each of the questions.

The final outcomes from rounds 1 to 3 were fed back to all participants and further open comments were invited.

How are terms such as ‘decommissioning’ and ‘disinvestment’ employed in the literature?

In their review, Niven et al. 60 document the terminology employed in the literature, identifying 43 separate terms currently in usage, which they go on to map on to a conceptual framework made up of de-adoption ‘phases’ (identification of low-value practices, facilitation of the de-adoption process, evaluation of de-adoption outcomes, and sustaining de-adoption). This mirrors to some extent the stages model that we outlined in Chapter 1, albeit without the ‘programme design’ stage, reflecting the authors’ focus on clinical rather than organisational interventions. ‘Disinvestment’ is reported to be the most prevalent term in the literature, although the authors advocate ‘de-adoption’ as more suited to a process rather than a decision point.

Four of five other reviews employ the more established term ‘disinvestment’,17,56,58,59 whereas Leggett et al. 57 employ the term ‘reassessment’ in keeping with their specific interest in this area. The earliest of the six reviews56 locates the term ‘disinvestment’ within the HTA movement and notes the tendency for prescriptions for practice following a HTA model. As noted, each of the reviews confines its scope to clinical/therapeutic interventions and, notwithstanding Niven et al. ’s60 preference for ‘de-adoption’, this area of decommissioning has become synonymous with the term ‘disinvestment’. In most definitions of the term, ‘disinvestment’ contains a normative component in its emphasis on improved health outcomes:

The complete or partial withdrawal of resources from health care services and technologies that are regarded as unsafe, ineffective or inefficient, with those resources shifted to health services and technologies with greater clinical- or cost-effectiveness.

Polisena et al. 58

Parkinson et al. 17 employ the term ‘disinvestment’ in a more expansive way to include restrictions on prescribing, price negotiations and the encouragement of generic prescribing, with ‘delisting’ used to refer to complete disinvestment. This very broad definition appears to signal a break with the narrower scope of previous usages of the term.

What are the current and previous levels and types of health-care decommissioning as reported in previous studies?

The scope of each of the six reviews is primarily confined to (1) activity that is commonly known as disinvestment (incorporating the principle of cost-effectiveness maximisation) and (2) health technologies (i.e. treatments and interventions). The reviews’ reporting of decommissioning activity rates is similarly circumscribed. Furthermore, the approach taken to reporting prevalence of decommissioning varies across the reviews. Leggett et al. 57 and Parkinson et al. 17 focus on formal programmes instigated at the national level. Polisena et al. 58 and Mayer and Nachtnebel59 identify and analyse case study decommissioning programmes (n = 14 and n = 7, respectively). Niven et al. 60 offer the most extensive synthesis of decommissioning (or disinvestment) levels, from candidate identification through to ‘sustaining de-adoption’. In summary, the following messages can be identified in relation to levels of decommissioning (with particular focus on the NHS in England):

• Some decommissioning has been instigated at national and local levels across Organisation for Economic Cooperation and Development (OECD) countries.

• Overall rates are reported as being somewhat low, although recent reviews appear to show an increase.

• At the national level, the focus has been on comparative effectiveness research and HTA, whereas for lower tiers of decision-making, economic frameworks such as programme budgeting and marginal analysis (PBMA) have been advocated.

• NICE is often the main focus of reported decommissioning activities in England.

• Safety (or harm reduction) is the most common driver of decommissioning programmes that are successfully implemented, with cost-effectiveness, for example, being far less likely to be cited as the primary driver.

• ‘Passive’ or ‘implicit’ forms of decommissioning are more common than ‘active’ or ‘explicit’ forms.

• Restriction (e.g. to patient subgroups) is more common than full withdrawal (or ‘delisting’).

• Considerable attention has been paid to the identification of decommissioning candidate interventions and the assembling of lists of those deemed suitable for either removal or reduction in provision.

What are considered to be the main determinants of successful decommissioning programmes?

The reviews report some common impediments (or ‘barriers’) to the implementation of planned decommissioning. These include a lack of resource for research into candidate technologies for replacement or removal. 56 Later reviews, for example by Mayer and Nachtnebel,59 note that attempts have since been made to address this evidence deficit.

The reviews also identify consistent and widespread ‘resistance to change’ from stakeholders, including patients and clinicians. 58 This can be because of ‘sunk costs’57 or ‘losses’ experienced by clinicians, patients and manufacturers of interventions considered for decommissioning. Parkinson et al. 17 elaborate on the logic of these impediments, referring to restrictions to ‘patient and prescriber choice’ and ‘perverse incentives’ created by payment regimes for clinicians. They observe that ‘there may be resistance to changing prescribing behaviours in the face of established clinical training and practice paradigms’. 17 The resulting lack of ‘will’ among political, clinical and administrative actors is the most consistently cited barrier. 56,59 Perhaps not surprisingly, early and effective stakeholder management is foregrounded in good-practice recommendations in the reviews.

What models and frameworks are available to guide decommissioning and how have these been evaluated?

Gallego et al. 56 report HTA and other methods for establishing lists of candidates for disinvestment as being one of the main sources of support for health systems and techniques, such as PBMA, as a means of putting these into practice. Beyond this, the authors identify a lack of ‘formal structures, processes or mechanisms’ to support practice. 56 Polisena et al. 58 also identify PBMA and HTA as two of the most prominent tools for supporting decommissioning, even though these are not deployed simultaneously in their case study examples. They also identify the application of the accountability for reasonableness (A4R) model – based on four process conditions for decision-making – in three case study disinvestment processes. Leggett et al. 57 focus on health technology reassessment (HTR) programmes and find one ‘model’ for reassessment.

In relation to implementation, Mayer and Nachtnebel59 compare top-down approaches based on mandated decisions and forced compliance, with bottom-up approaches where disinvestment is encouraged and facilitated and led by those working at the ‘coal-face’. They conclude that while top-down approaches can engender resistance, bottom-up approaches require voluntary engagement that may not be forthcoming in all cases. They therefore advocate a combination of both approaches.

As well as low-value lists, Niven et al. 60 identify the following ‘common mechanisms’ for supporting de-adoption processes:

Restructuring of funding associated with the given practice, changes to local and/or regional policies, and more consistent integration of health technology reassessment within existing health technology assessment programs.

Niven et al. 60

Parkinson et al. 17 identify a variety of means by which candidates for disinvestment might be identified, including assessment processes for new and existing drugs and consultation with stakeholders. In relation to implementation, they considered restricting the use of drugs and reducing the process to be more ‘acceptable politically’ than full withdrawal.

Mayer and Nachtnebel59 devote some attention to the need to sustain change and the need for financial/human resource capacity and organisational levers:

For government-initiated programs, tying a program to existing controlling tools (e.g., maintenance of a catalogue of benefits, conditional coverage, coverage under evidence development) and establishing new tools (e.g., coverage of new technologies provided only that ineffective technologies are removed concurrently) could facilitate the implementation of reassessment processes.

Niven et al. 60

Although none of the reviews set out to document existing guidance on decommissioning, we identified two such guides in the process of synthesis,16,61 although each of these reviews is somewhat narrow in focus. Overall, the evaluations presented in the reviews chart a realisation of the limits of tools such as low-value lists and HTA in addressing the full range of decommissioning stages. 60 To this we would add their limits as tools for the decommissioning of wider services and programmes. Of more obvious practical value is PBMA, which enables varied forms of investment areas to be compared from within fixed budgets. However, PBMA remains primarily an economic (normative) model and is not always aligned with the constraints of local decision-making contexts. 4,62,63

What are the remaining knowledge gaps in terms of evidence and practice?

Of the reviews that present recommendations for future research, Gallego et al. ,56 writing in 2010, note the prevalence of case studies in the grey literature and the absence of published research; Polisena et al. 58 focus on the need to support the process of identifying candidate technologies; Leggett et al. 57 note the ‘negligible’ focus on the monitoring and implementation of HTR and Mayer and Nachtnebel59 argue for further research on the impact of decommissioning programmes. Niven et al. 60 identify a lack of clarity over the threshold for agreeing to decommission:

There are multiple factors that likely determine when a practice should be de-adopted (e.g., nature of the intervention, lack of effectiveness or degree of harm, nature of the evidence) but the role of each factor and the interplay among them that ultimately determines when to de-adopt is not clear.

Niven et al. 60

They also argue the need for a ‘taxonomy of de-adoption terminology’ and to:

Summarise barriers and facilitators to de-adoption, and quantify the impact of past examples of de-adoption. We also need empirical research to examine optimal strategies for identifying candidate low-value practices, and to determine which de-adoption strategies are likely to have the most impact.

Niven et al. 60

All of the reviews contain examples of planned and successfully implemented decommissioning, although these tend to have been driven by safety concerns. The scope of the identified reviews means that it is difficult to draw conclusions about rates and outcomes of decommissioning programmes involving wider services and organisations. We would therefore add to the knowledge gaps identified in the reviews the need to take empirical research beyond discrete technologies (interventions) to wider management and service changes. There remains a need for process-based research into the organisational and political dimensions of decommissioning in general and specific investigation of programmes involving wider service withdrawal, closures or reorganisation. There is currently a lack of robust evidence and guidance to inform design and implementation of decommissioning programmes. Therefore, developing a better understanding of how decommissioning programmes unfold in the NHS and elsewhere is a crucial first step towards providing evidence-based guidance for the successful implementation of decommissioning in the NHS.

Summary of additional literature

These additional studies give some insight into the determinants of decommissioning in forms that go beyond those addressed in the six literature reviews. They suggest that such processes are influenced by organisational and institutional factors, and by staff and user attitudes, local politics and language and how this is deployed. This in turn suggests the need to develop explanations of the phenomena of decommissioning that draw not just on an evidence-based decision-making paradigm, but also take into consideration:

• how wider financial drivers and the availability of financial resources influence decommissioning programmes

• how organisational types and characteristics might influence levels of predisposition to decommissioning

• how staff attitudes might shape and be shaped by organisational factors

• how wider organisational networks and infrastructure can engender inertia among stakeholders

• how wider political factors and the absence of a shared language to discuss decommissioning might inhibit programmes at the local level

• how community groups might come together to oppose decommissioning plans.

Summary of findings from the mapping interviews

The mapping interviews show that at the time of research there was a range of bodies with responsibilities in relation to decommissioning. However, these national roles and responsibilities were neither joined up nor mutually informed, and not all respondents indicated a detailed awareness of the remit of the other bodies in relation to decommissioning. Many of the national bodies included within the mapping review, with the exception of NICE, were not providing active support for locally led decommissioning. Others had a responsibility to intervene, for example in a review role, as in the case of the Independent Reconfiguration Panel (IRP), National Clinical Advisory Team (NCAT) and Overview and Scrutiny Committees (OSCs), on a case-by-case basis. This national context was highly fluid during the lifetime of the project, with some of the bodies either merged or discontinued.

Round 1 results

Thirty respondents provided a total of 142 criteria that they considered should inform decisions to carry out decommissioning and 126 criteria that they considered did inform such decisions in practice. After grouping of similar suggestions by the research team (Box 8), 16 criteria were identified as informing decisions to carry out decommissioning, both as it should be done and as it is done in practice. An additional eight considerations were identified for the latter. Open comments in support of these suggestions included scepticism as regards the evidence base of decommissioning decisions and a belief that cost considerations are most influential:

[We] need to acknowledge that health services are delivered within a dynamic policy and political context that on occasions drive decommissioning contrary to evidence.

Australia, policy/practice

Whilst the ideal is decommissioning of services that don’t represent high quality care or value for money, the reality is that decommissioning decisions are generally made as a money-saving mechanism.

UK, practice

Cutting costs is a legitimate reason for change but it is often hidden from public view or presented within the case for change as a subsidiary factor.

UK, policy

Respondents also put forward 463 factors that shape the process or outcomes of decommissioning (either positively or negatively). These were grouped by the research team into the 30 factors, listed in Box 9. The sheer volume of identified factors meant that categorisation was challenging. In order to present findings back to respondents in an accessible format, we grouped the factors into three broad themes (below). However, we recognise that they are not entirely mutually exclusive:

1. change management and implementation strategy – this includes factors such as strategic planning, project management, communication and leadership

2. evidence and information – this includes formal evidence of effectiveness and cost-effectiveness of candidates for decommissioning as well as, for example, information from implementers elsewhere

3. relationships and political dimensions – this includes factors such as relationship with communities, national political influence and the media.

Participant comments included reflections on politics and evidence:

Decommissioning health care services is difficult unless you have clinicians onside. This also needs to be complemented by political support at the local and national level. Without these and a clear narrative of what these changes are and why these are necessary then decommissioning is difficult to bring about.

Australia, research

Lots of decommissioning I see seems to focus on a technical, rational decision to change services but doesn’t pay sufficient attention to the politics of such change (at its peril).

UK, policy

A very rigid ‘science of health care’ and therefore what is worth commissioning is causing the decommissioning of ‘non evidence-based’ services in a way that appears fair but is hugely damaging. The evidence-base used is flawed.

UK, practice

Finally, in round 1, participants generated a total of 125 ‘best-practice’ recommendations to guide effective decommissioning, which were grouped by the research team into 19 ‘best-practice’ recommendations (Box 10). Again, participant comments provided further explanation of these recommendations:

Once a decision has been taken it should be stuck to as long as the appropriate process has been followed. Many decommissioning projects fail to achieve what they set out to achieve because commissioners and organisations go back on their decisions. Among the main reasons why commissioners may change their minds are the influences of staff groups who may not have been fully engaged in the decision making process and politicians who may have been influenced by public opinion to oppose change.

UK, research

Decommissioning services is a taboo subject. In order to manage it effectively you need a good strong clinical leader so that the service is decommissioned from a clinical perspective on quality issues and ineffectual clinical outcomes.

UK, practice

Round 2 results

Of the 30 participants in round 1, 27 completed the round 2 questionnaire, which assessed the extent to which there was a consensus regarding which of the considerations identified in round 1 should inform decisions to carry out decommissioning, and which do inform decisions in practice. Appendix 7 presents these results. Respondents were also asked to rank the relative importance of the factors proposed in round 1 as shaping – either positively or negatively – the process or outcomes of decommissioning. The results are shown in Appendix 7. The politically contested nature of decommissioning and the importance of clinical buy-in featured in open-text comments:

Politicians may not instigate decommissioning decisions themselves but they can bully senior decision makers into changing their plans if they feel that they will adversely effect their election chances. Other decommissioning decisions are taken at short notice and they need to deliver savings quickly and effectively. If a decommissioning decision costs money in the short term or is hard to implement and will take too long then it won’t happen.

UK, research

Decommissioning projects fail when the clinical case is not clear or supported by clinicians and the public.

UK, policy

Calculation of the degree of consensus in responses to the 88 rating scale questions from round 2 found that 59 questions achieved a high degree of consensus among participants, 11 achieved a medium degree of consensus, 10 achieved a low degree of consensus and 8 achieved ‘no consensus’. The nine statements and three ‘best-practice recommendations’ with the highest degree of consensus are shown in Box 11.

Of the 18 questions with a relatively low level of (or no) consensus, 14 related to the factors that participants put forward as informing decisions to carry out decommissioning in practice. By contrast, the other five sections in round 2 generated much higher overall consensus levels. All of the questions in the section exploring the extent to which the nominated factors should inform decisions to carry out decommissioning achieved a medium or high degree of consensus.

In descending order of importance, the following good-practice recommendations were identified as being the most important:

• identify and establish a strong leadership team

• engage and involve clinical leaders from an early stage

• establish a clear rationale and narrative for change

• ensure clear and thorough project planning and governance

• secure high-level political support (national and local) at an early stage

• base decisions on evidence of what works

• adopt a whole-systems perspective from the beginning.

Round 3 results

Of the 27 respondents in round 2, 25 participated in the third round of the survey, and Appendix 7 presents the full results. Because of the marked contrast between the judgement of what should happen and what happens in practice in relation to decommissioning decisions, we explored this theme further in the third and final round. Participants were invited to reconsider the 18 questions in round 2 that registered a low level of/no group consensus, 13 of which related to the ‘what happens in practice’ section. Participants also reflected on the possible reasons for the low consensus levels. For example, in relation to whether or not the costs of decommissioning do in practice inform decisions, participants reflected that:

I can remember 4 years of meetings with officers from a handful of other bodies all relating to a decommissioning which was mainly intended to secure a service improvement, rather than a cash saving. The time we spent on the exercise was never counted as a cost.

UK, practice

There is a major misconception of cost. The costs incurred are far more complex than I believe is considered.

UK, practice

I don’t really think the costs per se are relevant – it is more about skills, having the right framework in place, leadership and political will . . . the failure to see priority setting as a programme . . . which needs resources attached to it is one of the reasons we have poor priority setting . . . it is a lack of understanding and will to establish priority setting as a major programme needing a team and resources to support it.

UK, policy/practice

Results from the third round show a low level of/no consensus in relation to five of the original 88 statements generated in round 1:

• ‘The evidence base’ does inform decisions to carry out decommissioning.

• ‘Responding to changing demographics/population needs’ does inform decisions to carry out decommissioning.

• ‘Support from industry and other interest groups’ does inform decisions to carry out decommissioning.

• ‘Impact on workforce’ does inform decisions to carry out decommissioning.

• What is the relative importance of the ‘pace of change’ for shaping the extent to which decommissioning is implemented as planned?

The first of these generated the widest range of reactions to the round 2 rankings, and the role played by evidence was a source of notable differences of opinion:

Very disconcerting ‘disagree’ response rate. Evidence certainly ought to inform decisions and in my applied experience it is central. Please unpack and report the reasons for the ‘disagree’ responses.

Australia, research

I think we have all experienced decisions that were taken in the absence of a robust evidence base. I don’t think I have personally experienced decisions that were taken in flat contradiction of a robust evidence base. For that reason I choose ‘agree’ but recognise others may choose ‘disagree’.

UK, practice

I’m afraid I just don’t buy the standard textbook answer here – it feels more an issue of responding to national/local politics and a pragmatic sense of what we can get away with in practice.

UK, policy

In my experience, I still feel that a lot of decommissioning decisions are taken on ‘gut feeling’ rather than necessarily on the basis of hard evidence or facts.

UK, research

My experience is that a highly reductionist approach to evidence base and efficacy tends to negate what actually works/the effectiveness in real life . . . services that work on paper often don’t work in practice. The biggest block to truly evidence-based decommissioning is a lack of effectiveness data and a lack of review of the beliefs around efficacy.

UK, practice

Changes between rounds 2 and 3

Following round 3, 9 of the 13 statements moved from a low level of/no consensus to a medium or high degree of consensus with three statements added to the list of statements with a high level of consensus (Box 12).

After the third round there was a high level of consensus that ‘reputation of existing providers for shaping the extent to which decommissioning is implemented as planned’ had high or very high importance.

Overall results from the Delphi survey suggested areas of agreement and disagreement and a notable gap between what the panel believed should and does inform decommissioning. The panel agreed on six good-practice recommendations related to leadership: clinical engagement, creating a change narrative, project planning and governance, political support, evidence and whole-systems thinking.

Reorganisation of burn care services in England: a brief history

Burn severity is measured according to size, depth and location of the injury, and patient characteristics. Specialist paediatric burn care in England is low in volume and high in cost. Traditionally, burns services have developed incrementally across the country, and the national Burn Care Review of 200172 was the first time that a co-ordinated assessment was carried out of the effectiveness of these arrangements. This led to a series of national and regional initiatives intended, at least in part, to reconfigure services in line with changing patterns in demand (Box 13). A key outcome was the tripartite categorisation of burn services according to severity, with levels of care designated as ‘unit’, ‘facility’ or ‘centre’, with the most specialised (centre level) requiring multiprofessional specialist care support.

In 2003, the National Burn Care Group was set up with Department of Health funding and allocated responsibility for the assessment of services against the national burn care standards, and the designation of these according to the three orders of specialty. However, differential rates of progress led to the formation of four regional networks covering England and Wales (Northern Burn Care Network, Midlands Burn Care Network, South West UK Burn Care Network and London and the South East of England Burns Network) and charged with continuing the process of change under the oversight of a national network that is itself a subgroup of the predecessor to NHS England (the NHS Commissioning Board). In 2005 a Burn Injury Database was set up to enable monitoring of provider activity and outputs, and, in each of the regions, recommendations were subsequently made to reduce the number of centre-level providers. However, in two of these regions (the North West in 2007 and London/the South East in 2012) proposals were opposed by provider organisations threatened with decommissioning and no significant service changes resulted. In these regions, the landscape of paediatric burn care provision has therefore remained largely unchanged since 2001.

The Health and Social Care Act 20128 and subsequent establishment of NHS England saw responsibility for the commissioning of specialist burn services move from local and regional arrangements to a national commissioning contract, with clinical advice provided by the Burn Care CRG comprised of nurses, doctors, patient and lay representatives, NHS England and Public Health England, and intended to be geographically representative of the country. In 2012 the CRG led a review and revision of the Burn Care Standards, and confirmed the following minimum specifications for provider designation at (paediatric) centre level:

• on-site access to designated paediatric intensive care (PIC) and paediatric high-dependency care

• burn-competent paediatric anaesthetists

• sustainable 24/7 consultant burn surgeon rota

• 24/7 access to paediatric-related specialties (including paediatric surgery and paediatric renal medicine).

At the time of the research, assessing compliance against these specifications was the responsibility of NHS England Local Area Teams reporting to NHS England and the CRG. Regional operational delivery networks (ODNs), coterminous with the former regional networks, had responsibility for the implementation of national policies and plans. These arrangements afford two main mechanisms for delivering service changes, each of which involves the CRG. These are:

• identification of ‘derogation’, that is, a failure to comply with one or more of the minimum standards (in accordance with the legal origins of the term, derogation implies a temporary relaxing of the rules while arrangements are put in place to enable compliance)

• wider reorganisation initiatives intended to concentrate high-level paediatric burn care in fewer provider sites.

Both mechanisms had been activated at the time of the commencement of the case study. One centre – hospital A – had self-assessed as non-compliant with the requirement for a co-located PIC unit, and assessment of other centre-level providers was under way. Furthermore, the CRG was in the process of preparing an options statement for the reconfiguration of paediatric centre-level burn care in England. The target date for resolving the issue of hospital A’s derogation and gaining approval for implementation of the wider changes was 31 March 2015.

Case study activity

During the period of fieldwork the CRG met five times and observations were conducted by the research team in September 2014, September 2015 and April 2016. Key documents were retrieved and added to the analysis and a total of 17 telephone interviews were conducted (including four follow-up interviews). Access to interviewees was negotiated through the chair of the CRG and the NHS England programme lead. Consent was secured collectively from participants at the September 2014 CRG meeting and individually prior to each interview. The sample of interviewees included representatives from each of the four regions, eight CRG members and six non-members, including NHS England employees at national and local level, and clinical staff. Interviewees were identified on an iterative basis following initial advice from the CRG chair. All participants approached agreed to be interviewed in either first or second rounds. The chair was a regular point of contact for the case study and as such is not quoted verbatim in this report. Further details of data sources are provided in Appendix 8.

Defining the problem and mobilisation

Building alliances and networks

Overall, and notwithstanding these points of contestation, the case study was notable for the recognition among most interviewees of the need for some service change. However, there was significant resistance to the twin-track direction of travel and disputes over the evidence base were central to this. The espoused values of interviewees were apparently shared, but these were filtered through perceptions of unfairness with regard to implementation of the derogation process, and these perceptions appeared in turn to shape how resources such as evidence and theory were mobilised. Despite some agreement over the drivers for change, discussion of the specific decommissioning plans uncovered differences of emphasis, between, for example, a focus on safety (i.e. co-location with PIC), sustainability and organisational performance. There was also concern expressed at the apparent influence of institutional power and prestige in the derogation process. For example, representatives from hospital A felt that their status vis-à-vis more prestigious trusts was a factor in the decision-making, with one remarking: ‘they are all big institutions, teaching hospitals, strongly supported by their local health economy. We are a little pimple in [de-identified] aren’t we?’.

Acceptance of the problem and the solution

Institutionalisation and implementation

Reflections

The purpose of the case study was not to ascertain if or in what way the changes to paediatric burn care should proceed, and it is important to note limitations in our methodology. For example, we were unable to engage further with service user representatives on issues such as travel times and geography and our approach did not extend to modelling the effects of proposed changes in other parts of the system [e.g. ambulance services, accident and emergency (A&E), and so on] or assessing their resilience to major events. Instead, our focus was on the social processes involved and how plans were formulated and received. When considered over a longer time horizon, the network of actors advocating the reorganisation of burn care services had made progress in building a clinical case and mobilising such resources as were available. However, the period of case study fieldwork coincided with something of an impasse. One interviewee’s summation was that ‘there’s been review after review after review, and all of them come up with the same answer, but nobody can come up with a solution’. At the time of the case study, those advocating change had formulated a solution but appeared to lack access to the range of resources required to implement their plans. The work put in by a small but dedicated group towards profile and alliance-building, and mobilisation of evidence and other resources did not look as though it would henceforth be sufficient to overcome organisational and political counter-forces, and interviewees voiced fears that the network advocating change itself was at risk of atrophy in the light of slow progress. The political arena in which the CRG operated was felt to be either disengaged or non-committal in respect of its plans, reflecting a perception that the service itself was somewhat lacking in profile and voice. Ultimately, this apparent intransigence appeared to prevent the CRG from being able to overcome or circumvent the resistance of those in the immediate and extended network that opposed the changes. The problem of inertia was compounded by the absence of a mandated lead body with sufficient authority and resources to implement the changes. As a result, the CRG was hampered in its attempts to influence upwards through the confusing and ever-changing chains of NHS authority, and downwards through formal and informal implementation channels.

As an example of decommissioning, the reorganisation of paediatric burn services is perhaps unusual in not being driven to any great extent by the imperative to control spending. Indeed, the lack of a strong financial case may have led to the plan’s low profile in wider decision-making circles (although engagement with these bodies was beyond the scope of the research). In cases such as this, in which decommissioning is likely to require some prior investment, both in the change process and in the development of new service arrangements, the payoff is perhaps more difficult to promote.

The dual nature of the burn care plans, including system-wide transformation and redesignation of non-compliant services, may have increased disaffection as they appeared to rest on separate logics. The identification of specific units in derogation implied a case-by-case approach that seemed at odds with the parallel national plan and was interpreted as unfair by some. This is redolent of Charles Levine’s75 distinction between ‘organisational atrophy’ and ‘problem depletion’ as reasons for service ‘cutbacks’. The two following quotations exemplify the two approaches:

I’m happy for a visitor to come to me and say, ‘You don’t provide good enough care. Your outcomes don’t meet the standards of your neighbours’. Then that’s my real concern, because that is my primary focus. And if they can’t say that to me, but what they can say is, ‘Your wards are too far away from theatres’. That’s not good enough to me.

Interviewee 3

We’ve got a lot of really good burn units that have been burn units for as long as I’ve known them, that actually do not meet the criteria and actually if you were sitting down with a blank piece of paper you would never have put the burn unit there in the first place.

Interviewee 6

The aim of achieving a sustainable distribution of burn services is arguably in tension with the iterative designation of individual service providers. In the strategic review options document an explicit attempt had been made to reconcile the two approaches as sequential ‘phases’ of reorganisation. However, in the absence of an arbitrating body with authority and resources to ‘redraw’ the service map on strategic (rather than performance) grounds, progress on ‘phase 2’ had faltered.

Introduction

This case study report presents findings from a CCG-led decommissioning of an EOL home support service provided by an independent organisation in 2014. Decommissioning of this service took place over a period of 3 months, following a 6-month review by the CCG. This case study was selected to fulfil the criteria of being locally led and having relatively low complexity and low conflict. Drawing on Callon’s51 four processes as a framework, this report explains how the CCG was able to propose and subsequently manage the closing down of the service. Key to the success of implementing the decision to decommission was the management of the relationship with the provider. Wider lessons highlighted by this particular case study concern the availability of alternative patient pathways during service changes and the challenge of post-discharge patient monitoring.

Background: the end-of-life home support service

The EOL home support service was originally commissioned by the predecessor PCT in 2009 under preferred provider status for a 1-year pilot. The EOL home support service was a non-clinical service designed to provide support to patients, their families and carers at home during the last 12 months of life. The decision to commission the service by the PCT originally had three intended outcomes: to increase the percentage of patients dying in their preferred place of death, to improve the amount of support given to carers and to decrease the number of unnecessary admissions to hospital in the last year of life.

Under the EOL home support service, patients were provided with a named nurse co-ordinator, who provided regular contact and home visits as assessed as necessary by the co-ordinator, as well as telephone access to a ‘care bureau’ providing triaging and nursing advice and support, 24 hours a day, 365 days a year. The 12-month pilot was extended for 3 years in 2010 by the then PCT, and inherited by the newly formed CCG in 2013. The contract was extended for a further 6 months for existing patients in October 2013 to allow for a review of the service to be undertaken, followed by a further 3-month managed close-down process beginning in April 2014 (Table 19).

Case study activity

Our fieldwork took place between May and October 2014 in the final stages of implementation and review of the decommissioning process. Interviews with 13 individuals involved in leading and managing the decommissioning process were carried out from the CCG, CSU and provider organisations involved, and two meetings were observed between the CCG and provider. Some interviews occurred retrospectively. Further details of data sources are provided in Appendix 8. In order to protect the anonymity of individuals, identifiers are removed from verbatim quotations.

The following sections use Callon’s51 four processes to consider how the CCG was able to propose and then successfully implement decommissioning of the EOL home support service. Consistent with this approach, the activity of decommissioning in this case study has been divided into four overlapping processes: (1) defining the problem and mobilisation; (2) building alliances/networks; (3) gaining acceptance of (a) the problem and (b) the solution; and (4) implementation and institutionalisation. It is important to note that these processes are not necessarily linear or sequential; rather, they (re)occurred at various points throughout the decommissioning process.

Defining the problem and mobilisation

As explained in previous chapters, establishing current service arrangements as a ‘problem’ requiring a ‘solution’ represents the first stage in the process of institutionalising and implementing decommissioning. In this case study, this process began when a Local Commissioning Network requested that a review of the EOL home support service contract be undertaken by the CCG owing to concerns over value for money and clinical effectiveness of the service, with a potential £1.98M saving projected.

It transpired to be quite an expensive service in relation to what it provided . . . it was a service liked by patients, but not medically of paramount importance.

They were very, very keen to disinvest because they were needing to find huge savings. But they were quite right in what they’d identified as being a service which wasn’t providing value for money.

The fact that the contract was due to end in October 2013 after its 3-year extension was also a key factor for the CCG in mobilising the process.

In October 2013, the CCG extended the contract for the service (for existing patients only) for approximately 6 months (until 31 March 2014) so that a review of the service could be carried out, in order to decide its future. General practitioners (GPs) were notified of changes to the arrangements and no new referrals were made during the 6-month review period. A review team was established to oversee the process and included representatives from the finance, contracting, procurement, quality, patient engagement, communications and data intelligence teams within the CCG and CSU. A clinical lead was also appointed.

Building alliances and networks

Building alliances and managing relationships with key stakeholders was the second important process in ensuring the successful mobilisation and completion of decommissioning the EOL service in this case.

The review process and preparation of the review document helped to build a foundation for the successful completion of the decommissioning process by bringing together individuals internally to co-ordinate the process and building relationships with key decision-making bodies. During the review process the review team, driven by the CCG and led by the appointed clinical lead, presented review findings to the relevant commissioning board and approached Health Overview and Scrutiny officials early on to begin discussions about the proposed decommissioning of the service. This relationship-building was felt by interviewees to help facilitate the review process. In addition, two sets of relationships were reported retrospectively by the review team to have been important in managing and implementing the decommissioning process successfully: the relationships with the provider and with GP practices.

Acceptance of the problem and solution

As explained above, the final review document produced at the end of the review process was significant in defining and framing the issue as one of clinical effectiveness and value for money. After establishing this as the ‘problem’, three options for decommissioning the service were proposed in the review document, thus embedding decommissioning as the ‘solution’:

1. option 1 – do nothing (i.e. contract ends on 31 March 2014)

2. option 2 – allow the contract to continue for existing patients only until the wider EOL strategy had been developed

3. option 3 – a managed close-down process over a short period of time. Patients would be discharged to GPs for re-referral to alternative services as appropriate. The contract would be extended to allow that process to take place.

An analysis of pros and cons for each option was presented in the review document. The analysis suggested options 1 and 2 to be unfeasible in terms of patient risk and cost, and option 3 was recommended. This was presented by the CCG and debated with the commissioning board and Health Overview and Scrutiny officials. A consensus was reached with these governing bodies to extend the contract for 3 months at a cost of £370,000 to manage the close-down process and transition patients.

Interviewees reflected that involving and engaging key stakeholders and decision-makers early on was important to the consensus-building process, as well as having strong supporting evidence. This was felt to be particularly the case for building a consensus with GP practices. Here, the clinical lead took an active engagement role and no contestation was reportedly experienced, despite mixed views about the service expressed during the review process. Interviewees reflected:

Locally when we made the views early on it was done basically on the basis of the GPs first of all who said ‘yes we can save some money if we were not to continue to use this service, so let’s look at that’. So I think the GPs were inherently on-board to begin with which is better.

Keep GPs on-board – we’ll keep GPs involved in the decision making process so that they are happy with the decisions that they’ve got to make and be involved, and be happy with the position. They can’t have decisions made to them they need to be involved in it.

Institutionalisation and implementation

A project group, comprising two leads from the CCG, a senior project manager and the service manager from the provider, was set up to manage the closure and transfer of patients from the service. Discharging patients from the EOL service involved a series of negotiations between key stakeholders, which are described below.

Reflections

Overall, the management of the relationship between provider and commissioner was key to the success of the decommissioning process in this case study, and as an example of decommissioning this case is perhaps unusual in the preparedness of the provider to co-operate with the process. The process of preparing the review document was important in drawing together and maintaining the network of stakeholders involved and in informing decision-makers, and the strong financial case presented in the review document appears to have been instrumental to the decision to decommission. Meanwhile, the lack of strong patient and provider resistance and the lack of wider public or political voices and profile in this case meant that potential counter forces were not mobilised. The availability of financial resources to manage the 3-month close-down and transition of patients from the service also appears to have negated potential disaffection. This case illustrates the challenges of monitoring the impact of decommissioning on users of services and on alternative services in the health economy. Second-round interviews suggested that activity had moved on and capacity and resources were directed elsewhere in the commissioning chain; individual patient pathway monitoring happened elsewhere in the health system.

The Area Prescribing Committee

The focus of the APC’s work was on the implementation of coverage and medicines management guidance from national bodies such as NICE, as well as the development of care pathways and the governance of local prescribing behaviour and responsibilities across primary and secondary care sectors. To this end, the APC was engaged in the production of policies and guidance and oversight of a joint formulary for the area. The formulary itself was formally described as ‘output of processes to support the managed introduction, utilisation or withdrawal of healthcare treatments’ (see the APC website),76 and the APC was accountable to the governing body of the CCG. The formulary application form employed by the APC included an option for removal, and the committee operated a traffic light system that placed restrictions on usage (e.g. confinement to secondary care initiation). At the time of writing, the APC did not employ a formal multicriteria decision tool in their decision-making.

In order to encourage joint ownership from the main provider, the APC had a rotating chair between heads of pharmacy at the CCG and acute trust. In relation to decommissioning, the APC operated as both a vehicle for implementation of NICE disinvestment guidance and a source of further local decisions on replacement and withdrawal:

The Area Prescribing Committee is our principal sort of governing meeting which therefore makes the investment/disinvestment decisions in terms of medicines management across the patch and it’s very much instrumental in terms of setting up and devising our sort of medicines/drug formulary that we have in operation within our health economy.

Interviewee 8

The work of the APC was supported by three members of the CCG pharmacy team who attended meetings and were responsible for drafting policies, supporting governance of medicines management and transference of prescribing responsibilities, working on care pathways and formulary development.

Case study activity

The APC was recruited to the study following conversations with the clinical lead for medicines management and the head of medicines management for the CCG (and cochair of the APC). The APC convened every 2 months and, during the fieldwork period, three such meetings were attended and observed by the research team, along with two meetings of the acute trust ‘medicines optimisation committee’ (MOC). Relevant documents were retrieved and added to the analysis and a total of 10 telephone interviews were conducted, including one follow-up interview with the current APC chair. Consent was secured collectively from participants at the November 2014 meeting of the APC and individually prior to each interview. The sample of interviewees included clinical and non-clinical representatives, members of each of the participating organisations and patient representatives on the APC. All participants approached agreed to be interviewed. During the fieldwork period no instance of implementation of NICE decommissioning guidance was observed and no locally identified programmes were instigated. As a result, the data presented here are drawn from general observations and comments in regard to the committee’s decommissioning work. Further details of data sources are provided in Appendix 8.

Defining the problem and mobilisation

Building alliances and networks

An important strand of the committee’s work in the time since its inception was securing buy-in from the range of prescribing organisations and individuals in the area. In general this was considered an area of relative strength, with high levels of ‘trust’ cited and increasing levels of adherence to formulary guidance. The medicines team published both APC decisions and the formulary so that these were accessible, and engaged in a range of activities designed to encourage compliance with these. However, interviewees regularly referred to some lack of adherence among both primary and secondary care prescribers, and also identified gaming strategies sometimes employed in order to gain access to non-approved treatments. In general, this was considered to be of greater immediate concern than relatively low rates of explicit and planned decommissioning:

I’m not really seeing huge amounts of disinvestment as such. Clearly, the emphasis for my commissioning colleagues is about getting the pathway streamlined for the activity out of secondary care where it doesn’t need to be. That seems to be the sort of tactic to date so I’ve not really seen any complete disinvestment.

Interviewee 8

There hasn’t been a particularly strong focus on disinvestment as yet. Although there are workstreams in process and de-prescribing, they’re some of the things that have come much later down the line. With trying to get the formulary in place and staffing in place to manage it, and the engagement of the secondary care colleagues as well.

Interviewee 2

Gaining acceptance of the problem and the solution

As noted, the APC had been unsuccessful in previous attempts to implement decommissioning, including as a result of media interventions. This had induced some risk aversion and reluctance to take a lead. Interviewees referred to ‘protests’ creating ‘disincentives for others to do the same thing’:

We had a big problem with trying to change the commissioning of [musculoskeletal services], which caused a lot of public outcry locally. Although there was nothing for them to be worried about – we were trying to improve the service, not actually take it away. But the hospital trust ran a very good media campaign to try and stop that. And it was thought at that time that approaching something like coeliac disease and decommissioning was not a . . . it was too sensitive to do that at that particular time.

Interviewee 9

The media are always going to give the sad story about, you know, poor Flopsy who can’t get this drug. And they’re always going to focus on the individual case, often with inaccurate clinical data and what have you. So you’re always going to have that problem and the only defence against that is really authoritative well-based guidance.

Interviewee 6

Institutionalisation and implementation

A key mechanism for decommissioning was the formulary. The process of establishing it had provided an opportunity to place limits on prescribing and to both support this with evidence, and persuade clinical stakeholders of the importance of this. Not surprisingly, however, it was rare for decommissioning options to be put forward by prescribers. A more common approach was the ranking of treatments:

So there’s always going to be that small group that fall into that problematic basket of ‘OK we take this away, what do we then do with them?’ So more often than not, rather than going for a drug that is a formulary choice whether it be secondary or primary care prescribed to then making it non-formulary, it would more often than not be recoded or knocked down the preference list so there was still the option to prescribe for that small group of patients.

Interviewee 2

Adherence to the formulary was not mandatory at the time of writing, although this was under review by the APC. Interviewees identified a tension between allowing for reasonable exceptionality while standardising care where this was warranted. They were also conscious of the relative lack of contractual levers to enforce GP adherence.

A small number of respondents questioned the approach of focusing on single medicines:

I’ve been talking lots of in terms of disinvestment as in probably stop using one drug or change drug A for drug B. For me, there is a whole bigger agenda that we sometimes forget about with disinvestment and it’s a more holistic approach to the patient and it’s the whole thing of are we just using too much of everything in the first place?

Interviewee 7

Implementation of decommissioning, and of APC guidance more generally, was seen as dependent again on the co-operation of patients and their clinicians. Transfer of patients was invariably a gradual process at best and rarely imposed as a blanket restriction:

When we’ve tried to change people as prescriptions come up, some of them are very reluctant to actually change it, and some do return back. And they’re willing to take on an element of safety risk because the drug is actually effective for them.

Interviewee 9

It doesn’t often happen that you will actually go out and actually switch everybody over to a new drug. I think it has happened once or twice but most times it’s when you start new drugs, or if the patient comes in for another reason, you might open them up to that suggestion.

Interviewee 5

As well as foregrounding patient preferences, a small number of interviewees cited institutional inhibitors to implementation:

They have very different procurement mechanisms in the secondary care environment in that they have things they’d procure centrally via the central hub which they have no control over. They also have more local bulk purchase contracts, so the more that they use the greater the discount. But unfortunately those savings aren’t then passed onto primary care. So that then poses a problem if they don’t want to adopt a specific drug because it’s not what they’re getting their bulk discount on.

Interviewee 2

Reflections

Overall, the formulary appeared to be the main mechanism for enacting replacement or removal within the APC. In particular, during the initial setting up of the shared formulary, the team had worked to reduce the number of similar agents in use for the same condition. Implementation of such changes was typically tackled through a gradual process of incremental patient initiation and some historical prescribing was permitted. Although the APC was proactively engaged in managing medicine and prescribing across the patch, planned decommissioning had not emerged as a mechanism in the committee’s work, and this can perhaps be understood as a response to the competing priorities they faced and the unique challenges that decommissioning poses. This was encapsulated in the following quotation:

Occasionally there is a fairly clear cut disinvestment decision where it might be as a result of a NICE recommendation or appraisal or it might be as a result of a change in the market availability where it becomes fairly clear cut that one sort of medicine should not be used in favour of a newer or a different medicine that would give the same or better benefits for the patients. Those aren’t all that common because most often a formal disinvestment decision doesn’t need to be made, it’s a matter of custom and practice and it almost happens by default. But occasionally it might be necessary to put the added weight of a formal policy decision behind something.

Interviewee 10

Previous experiences of decommissioning had reinforced the reticence of the APC, and some pessimism was expressed over the likelihood of future success. Although decommissioning, especially when informed by safety concerns, was seen as legitimate, the consequences of taking a more proactive approach were considered to be prohibitive, especially in light of the other challenges faced by the APC. Where decommissioning had been implemented, this often coincided with NICE guidance recommending replacement of a treatment, albeit this scenario did not arise during the case study. Interviewees felt that leadership or greater support would be required from bodies such as NICE if decommissioning was to be tackled more substantially.

Introduction

This case study report presents findings from a CCG-led review and planned reorganisation of local primary and acute care services, involving decommissioning of a local hospital and centralisation of acute services at alternative sites. This case study was selected to fulfil criteria of covering both an urban and a rural area and being high in levels of scale, complexity and probable conflict. The review and strategic planning for the reorganisation took place over a period of 18 months between 2014 and 2016. Drawing on the four processes described by Callon,51 this chapter explains how the CCG was able to propose and plan the reorganisation of local services, and why the implementation and mobilisation of plans subsequently stalled. Key issues highlighted by this particular case study are relationships between providers, system complexity, and the scale and scope of the proposed reorganisation.

Background: the proposed reorganisation of local primary and acute care

The geographical area overseen by the CCG in this case study was made up of four localities, with a population of approximately 592,600. In 2012, the CCG employed mainly new staff, as the majority of staff from the predecessor PCT went to CCGs in other parts of the local area. A strategic review to inform a reorganisation of local primary and acute care was proposed by the newly established CCG in summer 2014, following the publication of its own clinical strategy in 2013 and driven significantly by long-standing financial and estates difficulties in the local hospital trust.

The initial plans in summer 2014 were for a 12-week review/consultation period employing an external consultant company to draft plans for a reorganisation of health services locally (to include social care), and a programme manager, who was seconded from another part of the NHS, to co-ordinate and deliver the review. However, a longer consultation and review period (18 months) was eventually decided on, with one external consultancy providing expert advice to the process, and another employed to deliver the communications and engagement work for the review. This change was due in part to concerns around the feasibility of a 12-week period, and also to the legacy of several previous controversial attempts at review and reorganisation of acute services in the area, including the loss of A&E services from one locality. The programme management team was also extended to include two supporting officers to the programme manager in January 2015 (Table 20).

Case study activity

Fieldwork took place between September 2014 and March 2016 during the review of local services. Interviews with 13 individuals involved in leading and managing the reorganisation process were carried out from the CCG, local acute and other provider organisations involved, the county council, and local HealthWatch. A second round of interviews was carried out with six of the individuals towards the final stages of the fieldwork. A total of 11 key meetings and public engagement events were observed during the review period. Further details of data sources are provided in Appendix 8. In order to protect the anonymity of individuals, identifiers are removed from verbatim quotations.

The following sections draw on Callon’s51 four processes to explain how the CCG proposed and planned the reorganisation of local health services, involving the proposed decommissioning of a local hospital. Consistent with this approach, the process of reorganisation in this case study has been separated into four overlapping processes: defining the problem and mobilisation; building alliances/networks; gaining acceptance of (1) the problem and (2) the solution; and implementation and institutionalisation. It is important to restate that these processes are not necessarily linear or sequential, but that they (re)occurred at various points throughout the review period to significantly shape the journey.

Defining the problem and mobilisation

Building alliances and networks

Gaining acceptance of the problem and the solution

Implementation and institutionalisation

Reflections

As an example of reorganisation, this case study was ambitious in its scale and scope, involving a range of providers and stakeholders in the strategic review and reorganisation process. Recognition of the severity of the financial and estates situation initially appears to have underpinned commitment to the strategic review process among providers and other stakeholders. However, at the time of fieldwork completion, input levels were becoming more difficult to sustain as the review progressed past original time scales and friction increased as providers ‘jostled for position’ in the new system.

System complexity appears meanwhile to have been the most significant barrier to gaining agreement for proposed solutions and moving towards implementation. This case study has revealed a number of tension points and variables that must be reconciled in any reorganisation process, including travelling distances for services users and equity of access to care; provider and commissioner boundaries of provision; care pathways and clinical design of the system; and system stability through the transition. Affordability was meanwhile an underlying driver throughout the review and strategic planning and is likely to shape the process and outcome of any attempted reorganisation.

An intensive engagement process was carried out by the review team in anticipation of strong public and political opposition, and an external consultant was recruited to assist with the development and communication of key messages and to add capacity to the engagement process. Unexpectedly, however, patient and public controversy over proposals was not experienced to levels anticipated by the review team. Observed events and interviews with patient representatives suggested moderate support for the review process in order to improve the quality of services, especially hospital provision, and to help integrate care. However, the timing of fieldwork means that limited data collection took place as more concrete proposals for future models of care were being shared and it was not possible to garner public opinion. Efforts to secure local political support were meanwhile initiated at the beginning of the process and a pause in the review was scheduled to coincide with the elections in May 2015 to avoid undue political attention or ‘hijacking’.

Overall, learning from this case study suggests that even planned and well-resourced change programmes may be subject to delay and lengthy time scales and that reconciling provider, commissioner and patient perspectives is likely to prove challenging. As an example of reorganisation, this case study has most relevance and learning for STPs, suggesting that pre-identified implementation mechanisms and honesty about the financial drivers are important prerequisites of change.

Participant views and experiences of decommissioning

Some key themes of the focus group discussions included mistrust of decision-makers, the importance of patient and public involvement in decommissioning processes, experiences of loss and uncertainty associated with decommissioning and views on the impacts and outcomes of decommissioning. These are discussed in turn below. Where excerpts are included verbatim from focus group discussions, ‘F’ denotes a female participant, ‘M’ denotes a male participant and ‘I’ denotes the interviewer (or facilitator).

Designing the Delphi survey

The final part of the focus group discussions focused on the design of the subsequent Delphi survey. This involved crafting questions and sense-checking language with participants, identifying potential participants and recruitment strategies and generating suggestions for the dissemination of findings among participant groups.

Round 1 results: the views and experiences of participants in relation to health and social care decommissioning

In round 1, respondents were asked to briefly describe their experiences of services being ‘moved or taken away’. Five respondents indicated having no direct experience. The remaining 18 referred to a range of decommissioning types (Box 14).

Some participants indicated having been affected by, or involved with, these decommissioning projects – for example petitioning decision-makers or submitting views in the consultation process. Some indicated that they had not sought involvement, whereas others expressed frustration and/or resentment at the limited opportunities to do so:

No account was taken of views of those affected by closure. Insufficient information given to those concerned. Decision taken to close in spite of overwhelming views to the contrary from patients and staff. Not enough time or skill to explain the alternative and reassure.

Member of the public/carer/service user/patient

Denied services. Went private but not everyone has the option to do that.

Patient/service user/member of the public

Exhausting and hard work trying to find the truth.

Carer

Each of these experiences and perspectives is likely to have informed the responses that participants provided to the subsequent questions. In total, the 26 individual respondents put forward 42 criteria that they believed were good reasons to carry out decommissioning, 33 criteria that they believed were bad reasons to carry out decommissioning and 44 criteria that they believed inform such decisions in practice. Unlike in the previous Delphi study (work package 1), the research team did not combine these lists as we wished to ensure that respondents were able to recognise the language employed in rounds 2 and 3 as deriving from their round 1 responses. However, as Box 15 shows, within each question we were able to combine and aggregate similar suggestions to arrive at lists of 14, 6 and 14 criteria. Two of the statements relating to ‘good’ reasons reflect the view of some participants that there are no such reasons (in italics in Box 16).

In addition, in round 1 individual responses were combined into a total of nine recommendations relating to how participants would like to be involved in decisions to move or take away services (Box 15).

Finally, individual responses were combined into a total of seven methods for involving people in decisions to move or take away services (Box 17).

Round 2 results: developing a consensus and comparing with the views of policy-makers, practitioners, health-care leaders and researchers

Of the 26 respondents in round 1, 23 participated in the round 2 survey that assessed the extent to which there was a consensus among the respondents regarding drivers of decommissioning decisions. Appendix 9 shows levels of consensus in relation to good reasons, bad reasons and reasons for moving or taking away services in practice. In each case, respondents were asked to identify the three drivers that they considered to be (1) most justifiable (2) least justifiable and (3) most influential in practice, in order of preference. Results are shown in Table 21.

Open-text comments provide some elaboration on these preferences. The majority of comments focused on the reasons for moving or replacing services in practice and the dominance in this of political and financial factors. Other comments emphasised the cost-effectiveness of patient empowerment, and some expressed cynicism about the motives of decision-makers and clinicians:

Before taking away a service it’s important there is a session with the people who regularly use the service to have input. As this could effect there health and well-being! Then how much would it cost you, when other services have to be put in place? If a good service was done then maybe you wouldn’t need to keep going back. That saves money. Or making people be in more control with the service they use is more cost-effective.

Patient

Care needs to be patient led and not organisation led.

Member of the public

Political and resource consideration outweigh patient need in practice.

Patient

Do not think that evidence or patient/carer/community opinion is taken into account as fully as needs to be. More commissioner led and finance driven.

Carer

One comment ran counter to the others with regard to the role played by politics in decommissioning:

Political pressures are usually a reason not to move or take away services (i.e. local opposition to reconfiguration, etc.).

Member of the public

An analysis of the degree of consensus in the panel’s responses to the 34 rating scale questions included in round 2 found that eight questions achieved a high degree of consensus among participants, seven achieved a medium degree of consensus, eight achieved a low degree of consensus and 11 achieved no consensus. The statements with a high degree of consensus among respondents are shown in Box 18.

Overall, compared with the Delphi study in work package 1, there was less of a consensus expressed over the factors put forward in each category for round 2. No consensus or a low degree of consensus was found in 8 out of 14 factors that participants put forward as good reasons to move or take away services; 3 out of the 6 factors put forward as bad considerations; and 8 out of 11 of the factors put forward as influential in practice.

Statements derived from round 1 relating to approach/methods of involvement showed a preference for in-depth and substantive involvement, for example through co-production, patient panels and focus group discussions (Table 22).

In open-text comments, participants confirmed the general lack of agreement with the statement ‘Involving the public and organisations such as charities can lead to bias and therefore should not be done’ (e.g. ‘I agree with all the above statements except for . . .’, ‘All statements other than . . . make sense’). Other comments elaborated on respondent preferences in relation to the other statements:

The method is less important that the commitment to genuine engagement, and building trust. All the above methods are fine, but there are countless examples of where they have been done well, and where they have been done badly. So I’m reluctant to rank ‘top three′ methods, because this suggests that particular methods are better than others and this isn’t the case.

Researcher working in the field of user and public involvement

Where a decision will affect someone’s health, there should be a patient who has recently received treatment present in regards to involvement.

Patient

One respondent expressed views not fully captured in the statements submitted and categorised during round 1:

These aren’t ways of involving patients in these decisions, these are ways of engaging patients with decisions that have already been made. What about patients actually having a seat at the table, being members of the bodies that actually make the decisions on an equal basis with commissioners and clinicians?

Patient/representing a community/third-sector organisation

Round 3 results

In the final round we asked participants to reconsider the 19 questions in round 2 that registered little or no group consensus. All of the 23 respondents in round 2 participated in the round 3 survey. Appendix 9 presents the full results.

Comments on statements with a low level of or no consensus from round 2

Fourteen of the 23 respondents left additional comments in relation to each of the factors brought forward from round 2. Across the three areas of questioning related to drivers of decommissioning some common themes were identified from these comments.

Analysis of overall responses on good and bad reasons for decommissioning and reasons in practice

When fed back to participants, the results of the ‘top three’ ranking questions in relation to reasons for moving or taking away services were generally confirmed in open-text comments (e.g. ‘No surprises there’, ‘makes sense to me’, ‘some I agree with, others not’, ‘would fit with my assessment’). Overall, the Delphi study found two factors that might constitute good reasons for moving or replacing services. These were ‘if other similar services are available’ and ‘if there are changes in patient or service user need’. This was supported by the ranking of the three most justifiable reasons, which comprised these two reasons and a third: ‘if a service is not being used very much’. The third instance of a high level of consensus was in disagreement with the statement that ‘“if a service is high in cost” is a good reason to move or take away services’.

There was a high degree of consensus around two bad reasons: ‘there is a shortage of staff with the right expertise’ and ‘political pressures at the local level’. This was again supported by the ‘top three’ rankings of bad reasons, which were ‘to make financial savings’, ‘political pressures at the local level’ and ‘to meet the interests of organisations (e.g. hospitals) or individuals (e.g. professionals)’.

After three rounds there was a high degree of consensus that four considerations do actually drive decisions in practice:

1. to make financial savings

2. meeting national policies and targets

3. decision-makers (e.g. commissioners) are in favour

4. the service is not considered to be good value for money.

There was also a high degree of consensus that ‘the public are in favour’ is not a reason for services actually to be moved or taken away in practice. These findings were supported by the ‘top three’ rankings, which were ‘to make financial savings’, ‘decision-makers (e.g. commissioners) are in favour’ and ‘political pressures at the local level’.

Overall, the ratings and rankings indicate a match between bad reasons identified by participants and those that participants felt were most likely to be drivers of decommissioning in practice. In other words, the consensus was that, in practice, decisions to move or take away services were made for bad reasons.

Analysis of overall responses on approaches to engagement

The rankings of statements related to engagement were largely supported by the open-text responses (e.g. ‘I would agree with the rankings’, ‘seems reasonable’, ‘fair distribution’, ‘cannot argue with any of that’). The same was broadly true for the ranking of involvement methods albeit with some exceptions:

There seems to be broad agreement that bringing patients together and encouraging participation and discussion in decision-making is most effective in influencing decision-making.

Patient/service user/carer/member of the public

Idealism with the top three then no effective robust processes to complain about changes in services in the bottom three. Very worrying stats.

Member of the public/carer/service user/patient

It’s interesting that online engagement received so few votes – might this reflect the demographic involved in the Delphi study? Do you have any representation from younger people (who I suspect are more interested in social media as a vehicle for involvement)?

Researcher working in the field of user and public involvement

These seem sensible arrangements with attention to patient panels & focus groups.

Patient/carer/representing a community/third-sector organisation

This supports the above: people want to be active decision-makers, not passively consulted after a decision has already been made.

Patient/representing a community/third-sector organisation

However, in other sections of the survey there were some specific concerns expressed about the public as an influence over decommissioning decisions. The following quotations were included in response to the suggestion that public opinion shapes decisions in practice:

Public are frequently misinformed by newspapers but not sure they have that much influence as they can’t stop it in practice, in my experience.

Patient

Public will normally be emotionally motivated, for example, would prefer a veterans service over a young people/homeless service.

Carer

‘The public’ generally speaking have no idea what services encompass and are unlikely to make such a contribution to the process.

Patient/representing a community/third-sector organisation

Reflections on the methodology

The decontextualising of some questions in the third round (e.g. ‘there are no good reasons to move services’) led to some confusion, with two respondents indicating lack of clarity over the aim of the question. Some questions involved a potential double negative (e.g. ‘I disagree that this is a bad reason’), which one respondent reflected may have affected responses. We recognise this limitation, although sense-checking against open-text comments would appear to show good levels of understanding of the task as set. One respondent referred multiple times to the small sample group for the Delphi survey and others expressed a general desire for more information.

All of my responses would be subject to more information and reliant on specific circumstances, so this is very generalistic.

Member of the public

It’s difficult to answer some of the above without knowing what the alternative to move/take away is – the alternative might be nothing (i.e. the service stays as it is), it might be work to improve the service and bring it back to standard. In some of the above cases, giving services time to improve or to change what they do is more desirable as a first line response than the more radical option than moving it or taking it away.

Patient

What is the international evidence and expert opinion regarding best practice in decommissioning health-care services?

From the summary of previous reviews presented in Chapter 4, it is clear that the international evidence on health-care decommissioning is skewed towards clinical interventions and, in particular, the replacement or withdrawal of medicines, and this applies to England and the wider UK. The most commonly used term in the recent health-care literature – ‘disinvestment’ – is synonymous with this form of decommissioning, and with a suite of tools (e.g. low-value lists, HTA) designed with such activity in mind. Most definitions of disinvestment incorporate principles of cost-effectiveness and opportunity cost and are therefore embedded in a health economics paradigm. 58,79 Furthermore, the literature in this area is primarily (although by no means exclusively) concerned with the use of evaluative evidence to promote population health and to improve clinical decision-making. However, this is a relatively fast-moving field and recent contributions have begun to reflect critically on the limitations of the disinvestment ‘movement’, and in particular to argue the need to develop strategies for overcoming obstacles to disinvestment practices. 60 Our study adds to this developing appreciation of the determinants of decommissioning of clinical interventions, and how these interact in practice with tools and formal resource allocation frameworks (including PBMA, HTA, low-value treatment lists and so on).

Across the world, the study of disinvestment is thus in a nascent phase and in its current form does not extend far beyond the removal and replacement of clinical interventions. However, this somewhat narrow focus is in contrast to the range and scope of decommissioning observed in the research carried out for this report. For example, five of our seven narrative vignettes, three of our four case studies and 89% of the decommissioning reported by our 56 CCG survey respondents (as well as the majority of examples put forward by participants in work package 4) all fall outside the purview of the extant disinvestment literature and of the tools and frameworks advanced within it. Our study therefore redresses the relative inattention paid to other forms of decommissioning such as service/organisational relocation and closure.

The review of reviews, and in particular the evidence synthesis provided by Niven et al. ,60 suggests a shift in the literature from an initial focus predominantly on problem identification and decision-making towards a greater awareness of decommissioning programme design and implementation mechanisms and strategies. Our study is one of the first to take a longitudinal approach incorporating the stages of decommissioning, and to reflect on how each of these is subject to influencing factors. In this we have drawn on additional literature such as Fulop et al. ,67 who examine the implementation of reconfiguration programmes and find that the nature of the programme of change and ‘local politics’ are more significant in shaping outcomes than the strength of the evidence, and that public resistance is greater when reconfiguration is perceived to include hospital downgrading. Similarly, Greenwood et al. 68 identify organisational determinants of abandonment (albeit in non-health settings) and Johns et al. 66 identify an ‘infrastructural inertia’ that inhibits abandonment of established services. To this, Younkin80 adds the notion of ‘de-commitment’ as a prerequisite of abandonment of practices and supports the conceptualisation of decommissioning as a dynamic, multistage process rather than a discrete decision point. Although drawn from research in a range of countries, each of these insights into more diverse forms of decommissioning (and/or related activity) has shaped our appreciation of the factors shaping the phenomena observed in our study.

Despite these insights, there remains an absence of published evidence with regard to ‘good practice’ in the implementation of the full range of decommissioning types. We therefore conducted a Delphi panel of expert opinion designed to combine perspectives on a broad range of decommissioning types, and to go beyond the normative case for decommissioning (i.e. what should the drivers be?) to reflect on what drives decommissioning in practice. This was international in scope, albeit with strong UK representation.

Our findings indicate that the considerations that experts from a number of health-care systems believe ought to drive decommissioning (cost-effectiveness, quality and patient safety, clinical effectiveness) are different from those that they believe do so in practice (e.g. cost/budgetary pressures, political imperatives). This disjuncture suggests the importance of taking a critical approach to research inquiry and avoiding decommissioning strategies that rely purely on gathering evidence to support decision-making and/or exhorting individuals to change their practice.

The four factors that drew the highest level of consensus as being important in shaping whether or not decommissioning is implemented as planned were found to be strength of executive leadership, strength of clinical leadership, clarity of rationale/case for change and quality of partnership working with relevant agencies. When asked to identify good-practice principles, the six statements for which there was most support were:

1. Engage and involve clinical leaders from an early stage.

2. Establish a clear rationale and narrative for change.

3. Ensure clear and thorough project planning and governance.

4. Secure high-level political support (national and local) at an early stage.

5. Base decisions on evidence of what works.

6. Adopt a whole-systems perspective from the beginning.

How and to what extent are organisations currently decommissioning services?

Subsequent work packages enabled us to explore the phenomenon of decommissioning in the specific context of the English NHS. This is the most recent study to gauge rates of local decommissioning activity in the English NHS, and the first dedicated solely to exploring this area of activity. Hitherto, the literature has focused on the work of NICE or else has been confined to a relatively small number of case studies. 65,81 Evidence related to wider activity rates at the meso level in England is thus somewhat limited. However, reviews appear to show a shift in activity levels, with later reviews (e.g. Niven et al. 60) apparently uncovering a greater volume of examples than earlier reviews (e.g. Gallego et al. 56) (notwithstanding differences in approach to inclusion). Despite this, the published evidence in England, as elsewhere, suggests that decommissioning at the local level remains relatively rare, especially for reasons other than safety. 60 Furthermore, our mapping interviews indicate that while there are sources of assistance for CCGs in carrying out commissioning and service reconfiguration, very little of this is tailored specifically to meet the challenges of decommissioning. The most active national body in this area, NICE, produces ‘do not do’ recommendations through their appraisal and guidance processes. However, this covers a relatively small proportion of the existing interventions in the NHS, and NICE has yet to launch a programme specifically designed to identify disinvestment opportunities. 81

Our sample of 27% CCG survey responses contained notably higher levels of reported decommissioning than both these previous reviews and a similar survey published in 2011. 82 What is more, much of this activity goes beyond the ‘low-hanging fruit’ described in the study of Daniels et al. 83 to include programmes of significant scale and scope. In support of this work, our survey suggests that CCGs were employing ‘public engagement’, ‘stakeholder workshops’ and ‘needs assessment’ most commonly, and ‘stakeholder mapping’, ‘PBMA’ and ‘scenario planning’ to a lesser extent. The most cited drivers of decommissioning were ‘cost-effectiveness’ and ‘patient experience’, with ‘alignment with strategic priorities’, ‘clinical effectiveness’ and ‘cost/budgetary savings’ also cited by a minority of respondents. These data would appear to suggest that decommissioning activities are both more prevalent and more ambitious than they have previously been, and that cost-related criteria are among those currently driving this change.

What factors and processes facilitate the successful implementation of decisions to decommission services?

The CCG survey findings suggest that the key determinants of successful implementation include:

• executive leadership/management

• quality of evidence base

• clinical engagement and support (both as enabler and inhibitor); and

• quality of implementation plans/planning.

Open-text comments from the survey suggest a perceived lack of guidance to support decision-making, limited local sources of support for decommissioning, and difficulties deriving from NHS restructuring and associated staff turnover and loss of expertise. We have noted that the first Delphi panel identified areas of consensus with regard to good practice. As helpful as these are, they remain somewhat abstract, and the aim of our case studies was to delve deeper into the decommissioning experience in a range of settings.

In case study 1, the review and reorganisation of intensive paediatric burn care, implementation was not achieved in the lifetime of the project and doubt was expressed over future prospects. It became clear that the national political context inhibited progress in a number of respects. First, the specialty area was not felt to be sufficiently high profile in media or government spheres and was therefore considered unlikely to command very high resources in support of change. These factors marked it as different from, for example, the reconfiguration of paediatric services recounted in one of the narrative vignettes, despite both being nationally led specialty reconfigurations. Somewhat paradoxically, the drive for NHS efficiencies was felt to impede implementation of the burn care review, as no large-scale savings were projected to flow from what was likely to be an expensive implementation process. To compound this, the mechanisms for implementation of the decommissioning components of the changes were both limited and subject to challenge, which in turn had led to something of a standstill. Taken in isolation, this experience would suggest the following implications for future national decommissioning:

• Reviews and reorganisations are more likely to succeed to implementation phase when the potential gains in efficiencies promise to make worthwhile the short-term political and financial costs of change.

• Executive authority and power of implementation granted to an identifiable and credible body, which in turn is granted autonomy and political support from relevant national system actors, are prerequisites of successful implementation.

• There are merits to adopting an approach with a simple and coherent case for change supported with evidence that opinion-formers accept as legitimate.

Of course, these propositions are highly aspirational and in practice may generate other sources of difficulty. For example, executive authority and discretion can often be challenged or undermined and higher-profile service areas are more likely to be subject to opposition from interest groups, the media and organised patient and public opinion. Indeed, our narrative vignettes confirm that even very well-resourced decommissioning processes that satisfy the above requirements can suffer from delay and, in some cases, discontinuation.

In our case study of the closure of an EOL service, implementation was achieved and, as such, this was unusual within our sample. The CCG was able to propose and manage the closure of the EOL home support service over a relatively short implementation period, following a 6-month review of the contract with the provider. The most important factors in achieving this outcome appear to have been a relatively powerless patient group and a provider organisation that was resigned to the fate assigned to it following the CCG review. Outcomes of the decommissioning were subsequently questioned by some of those involved and doubts were raised over the availability of alternative patient pathways during service changes and the challenge of post-discharge patient monitoring. Unfortunately, we were unable to explore these stages within our case study.

Case study 3 focused on the decommissioning activity of an APC. Leaders of the APC adopted an approach that largely precluded substantive, planned decommissioning. Instead, they had chosen to focus on building mechanisms and processes of decision-making and governance that might later be employed in pursuit of decommissioning, but that were for the time being directed towards promoting evidence-based prescribing (rather than de-prescribing) throughout the patch. These mechanisms and processes included a shared formulary, auditing of compliance with APC guidelines, initiatives to align prescribing across care sectors and development of economies of scale in decision support. Decommissioning had been undertaken but had remained somewhat peripheral and there was recognition, albeit tacit in some cases, that the risks generated outweighed the promised benefits. This position seems rational when considered in the light of the decommissioning failure reported elsewhere, while also perhaps at odds with a wider financial context of acute fiscal constraint.

Case study 4 was a major CCG-led review and planned reorganisation of local primary and acute care services. Successful implementation was not facilitated during the period of research, despite a lengthy and generously resourced process of preparation and stakeholder engagement. The CCG was able to propose and plan the reorganisation but implementation faltered as tensions grew between the provider organisations involved. This example offers a cautionary tale against the assumption that following procedural good practice can ensure delivery of decommissioning outcomes, especially where complexity makes prediction of effects extremely difficult.

In combination, these case studies suggest that the good-practice principles generated in our Delphi study have some limitations. Within the case study sample we found examples of good practice apparently being followed but success nevertheless remaining elusive (e.g. case study 4). Conversely, many of the good-practice recommendations were not observed in case study 2 but implementation was nevertheless achieved. This would suggest either that these recommendations are erroneous or, more likely, that their application is highly dependent on local contingencies. Their merit may reside primarily in their function as general principles of approach and we should perhaps avoid the temptation to codify and apply them mechanistically across contexts.

The case studies also highlight how the role played by evidence and actors in identifying and framing a need for decommissioning depends in turn on factors such as the scope of the project and the number of actors implicated in or affected by the decommissioning, as well as the resources that these can mobilise to oppose plans. The proposed closure of the EOL service generated a muted response in which the patient voice was not loudly heard, and the provider organisation elected not to challenge the decision. This apparently unusual set of circumstances may have negated the requirement for Delphi panel recommendations such as securing high-level political support, engaging service users and stakeholders, and adopting a whole-systems perspective. By contrast, case study 4, drawing on frameworks employed in reconfiguration programmes elsewhere in the NHS, employed expensive and somewhat elaborate processes of option appraisal, consultation and engagement. Despite this, at the time of writing, acceptance of the ‘solution’ of decommissioning remained elusive. Furthermore, the reconfiguration involved considerable complexity in the implementation phase, in contrast to expedient removal of the EOL service, which was ultimately implemented via the cessation of a contract between two parties. The case studies also provide insight into how evidence thresholds required to build the case for decommissioning vary, in terms of both volume and rigour, and this has important implications, how high-quality evidence to inform decommissioning changes is often hard to generate or access, and how professional consensus is therefore frequently employed as a proxy. However, this consensus can be hard won, as illustrated by the three case studies that had been unsuccessful in implementing their plans. The case studies also show how the mobilisation of evidence is often inextricably interwoven with the politics of decommissioning.

What are the perspectives and experiences of citizens, patient/service user representatives, carers, third-sector organisations and local community groups in relation to decommissioning?

The conclusions we have drawn so far reflect something of a management bias, drawing disproportionately as they do on the views and experiences of health-system actors, as well as academics and national bodies with an interest in this field. The fourth work package enabled us to directly address the perspectives of citizens, patient/service user representatives, carers, third-sector organisations and local community groups through focus group discussions and a second Delphi survey, both of which recruited from within these groups.

Findings from work package 4 suggest some commonalities with the views of participants in work package 1. First of all, there was support for the principle that decommissioning in some circumstances is legitimate (although the same consensus was not found in relation to moving services). Both Delphi panels also considered decommissioning, in practice, to be at least partly a function of financial pressures and the need to make cost-savings. The majority of participants in work package 4, however, felt this to be an unjustified basis upon which to make decommissioning decisions. Their rejection of cost pressures as a basis for decommissioning was therefore emphatic.

Results from across each of the work packages suggest a perception that patients and the public are rarely influential in decommissioning decision-making. For example, Delphi study 1 participants strongly disagreed with the statement ‘patient and public views do actually in practice inform decisions to carry out decommissioning’ and Delphi study 2 participants strongly disagreed with the statement ‘the public are in favour’ as a reason why services are actually moved or taken away in practice. This apparent consensus with regard to the relative lack of influence of patients and the public was supported by the CCG survey finding that ‘patient and community groups’ and ‘public/citizens’ are among the least-involved stakeholders in each stage of the decommissioning process. Opinions diverged, however, on the normative question of engagement: while the first Delphi panel did not rank patient and public engagement high on the list of good-practice principles, the second Delphi panel expressed an appetite for substantive engagement and co-production of decisions.

These findings are of crucial importance to understanding attitudes to decommissioning. They appear to suggest that patients and the public are rarely influential in decommissioning and that decommissioning is informed, in part, by considerations that these groups are unlikely to consider to be justified. Given this confluence of factors, it was not surprising to hear frustration and mistrust in decision-making voiced by participants in work package 4. Few had faith that calculations of patient benefit or cost-effectiveness (for example) trumped financial pressures and political considerations in the selection of targets for removal.

Allied to this was resentment expressed at the lack of apparent awareness in decision-making of the lived experience of service users. It was felt that little or no value is placed on the relationships that make up service provision and the damage done to these in the process of service removal or replacement, or of the benefits of service continuity as a counterweight to the projected benefits of service replacement. Third-sector providers involved in the focus groups voiced their distress at having to communicate service removal decisions to vulnerable user groups, and their own feelings of uncertainty when under threat of decommissioning. Others participants emphasised the harms to carers and service users of extending travel times resulting from service changes.

Decommissioning is highly context dependent

Despite generating consensus around areas of good practice, we conclude that a formula for decommissioning is not only unavailable but also impossible given its dependence on local context and the apparent unpredictability of outcomes. Furthermore, it is likely that managerial approaches adopted elsewhere in health will not always be sufficient in the face of the additional dynamics associated with decommissioning. We emphasise the need for greater levels of honesty with regard to the rationale for removal and for co-production with affected parties at each stage of the process with the possible exception of decommissioning where low levels of controversy are projected.

Approaches to decommissioning should be modified according to type

Our definition of decommissioning encompasses reduction and replacement in, as well as removal of, services. It also includes medicines, discrete services and whole-provider organisations. Our study points to the differences between these forms of decommissioning, as well as factors held in common. In relation to medicines we believe that the relatively embryonic field of replacement and withdrawal would be well served by learning lessons from the longer-standing field of evidence-based medicine. This includes the need to mobilise and incorporate the experiential evidence of practitioners and patients, on the grounds that end-user intelligence is required if replacement and withdrawal is to be implemented without unintended consequences. 101,102 In this, professional stewardship of limited resources will be key. 103 Other, more substantive forms of decommissioning at the organisational or interorganisational level require greater attention to factors such as context, scale and scope (along with interests and systems). Inclusion of decommissioning in health-care improvement stands an increased chance of success if it is based on a more sophisticated appreciation of each of these aspects and how they cohere and conflict in local service settings.

Terminology is important

The literature on removal and replacement employs multiple terms and there is a danger of this becoming something of an obsession for the research community. The proliferation of terms has aided theoretical development but has not increased clarity or accessibility. Debates over the merits of candidate terms such as ‘de-adoption’, ‘de-implementation’ and ‘undiffusion’ will no doubt continue. Our aim in employing the term decommissioning is merely to denote the phenomenon of study and we revert to the terms suggested by our participants (including ‘stopping’, ‘moving’ and ‘replacing’) when disseminating our work outside research settings.

Successful decommissioning is a long-term pursuit

Our study suggests that successful decommissioning is likely to be the end point of a process in which the case for change has been shared and argued widely. This ‘de-commitment’ is important to the building of acceptance of both problem and solution in relation to decommissioning. 80 Where this was not achieved, for example in the case of our APC, the preference of those involved was for more passive approaches to decommissioning. Our case studies also reaffirm the importance of developing long-term engagement strategies and addressing issues of trust and public legitimacy before embarking on substantive decommissioning activities. 104 This in turn suggests that reliance on a combination of evidence and exhortation is likely to be insufficient in making the case for decommissioning, and that a more root-and-branch analysis of the politics and processes involved is required.

Delineating decommissioning decision-making

In our case studies and vignettes there is evidence of some confusion over roles and responsibilities for decision-making and it is clear that the challenges of decommissioning are exacerbated by lack of clarity and authority in this regard. There is a further requirement for clarity over the specific terms of decision-making. For example, do decision options include reduction in investment or full withdrawal? Is substitution or partial reinvestment built into the decision-making parameters? Once these questions are answered, more informed choices over the use of decision-support tools and frameworks can be made.

Substantive decommissioning requires attention to programme design

The design of effective decommissioning requires evidence to support the causal efficacy ascribed to the programme in meeting its aims, as well as a grasp of the technical and political feasibility of putting the programme in place. 105 This study suggests that local commissioning processes and frameworks may not be sufficient in themselves to deliver effective decommissioning. While decommissioning and commissioning may be seen in normative terms as elements of the same processes (i.e. needs assessment, service review, contracting and procurement and so on), in practice our study suggests that received NHS models of commissioning good practice may be insufficient to meet the challenges of decommissioning. Key to this are the additional implementation challenges involved in dislodging entrenched services and practices. Our study suggests that the more ambitious a programme of change is, the more likely it is that this will require active participation from the organisations implicated in the changes, because, as we have seen, resistant provider organisations can call upon an array of strategies to delay or derail changes. The requirement for some degree of service self-sacrifice (or ‘arranging their own funeral’, as our interviewee described it) sets decommissioning apart from other commissioning activities.

Decommissioning programme implementation

Key considerations at the implementation phase include how the extent of ambiguity and conflict surrounding a policy can mediate and constrain approaches to implementation. 106 For example, the implementation of decommissioning in the NHS is often characterised by high levels of conflict as well as ambiguity over the different priorities and criteria that should inform and guide decisions, leading to implementation slow-down and/or failure. 89 We have seen that services can be reduced or removed through the application of eligibility criteria, practice guidelines and other forms of service restriction and scale-back. Alternatively, they may be withdrawn as a result of in-house service closure, external contract termination, service reconfiguration and formulary delisting. As above, our study suggests that full implementation of decommissioning requires the active support (e.g. through temporary double-running of services) of the affected provider, which suggests the need for their co-option into the process.

Management and community engagement in decommissioning

This study expands and refines understanding of the processes related to decommissioning across a range of health-care settings and contexts. Overall, we conclude that no decommissioning blueprint exists and that outcomes, especially of more ambitious programmes, are hard to predict. Approaches to planning and management should reflect the factors found to shape the outcomes of decommissioning, including change management and implementation, evidence and information, and relationships and politics. Those planning decommissioning will need to account for history, culture and quality of interprofessional relationships. In a highly politicised and contested landscape, skills in negotiation, sense-making and leadership are paramount. Furthermore, the contentiousness of decommissioning suggests that an explicit and proactive media strategy is more likely to be required, as is the political acumen to negotiate the inevitable influence of power and interests. 107

Without community support, it is often difficult to achieve planned changes. Evidence-based models for working with communities are required to support difficult decisions of this kind. Our findings suggest that established approaches to engagement and involvement require rethinking. In particular, in cases in which decommissioning is driven in part or in full by financial pressure, we find that attempts to downplay cost-savings run the risk of increasing cynicism and challenge from stakeholders who are often fully aware of the financial ‘big picture’.

Sampling

In most aspects of the study we have sought to recruit participants who have experienced or are currently involved with some form of decommissioning. For work package 1 this included national bodies with a possible role in decommissioning (mapping exercise), previous leaders of decommissioning processes (decommissioning stories), and ‘experts’ in decommissioning from research, policy and practice backgrounds (Delphi study). In work package 3 this included individuals involved in current and/or recent decommissioning processes (case studies). In work package 4 this included citizens, patient/service user representatives, carers, third-sector organisations and local community groups, many of whom will have had direct experience, in these capacities, of decommissioning. This is a deliberate sampling strategy intended to draw on the perspectives and experiences of those with a ‘take on’, as well as possibly a ‘stake in’, decommissioning. Results will therefore be different from what might have been expected if we had either deliberately sought to target those with no involvement in decommissioning or pursued a sampling strategy in which we were blinded to this information. Our results are therefore not intended to be automatically generalisable to, or typical of, wider managerial or stakeholder populations.

Clinical Commissioning Group survey response rate

By contrast, work package 3 targeted all CCGs in England. Although the response rate reached over one-quarter (27%), the potential for self-selection bias has led us to be cautious in our approach to extrapolation. In Chapter 3, we acknowledged the possibility of response bias in our survey of CCGs, with responding CCGs perhaps more likely to be actively engaged in decommissioning than non-responders. We also noted that, with the exception of an over-representation of Midlands CCGs (perhaps reflecting the location of the lead research institution), the sample of responding organisations shows no obvious atypical characteristics. Although we consider that a response rate of 27% enables us to infer some general patterns across the wider population of CCGs, we acknowledge that a larger sample size would have reduced the chances of sampling error and that we could have mitigated for gaps in the national data with, for example, a detailed mapping of decommissioning in one region or locality.

Public and service user involvement

As noted in Chapter 3 of this report, our aim was for the perspectives of patient and service user representatives and members of the public to form an integral element of the study. Although this was partially achieved in work packages 1 and 3, this was not to the intended degree of depth and breadth. This is a product of the limited engagement and participation activities enacted in the case study processes, but which nevertheless means that incorporation of these perspectives into our case study analysis has been hindered. In more general terms, we have sought to mitigate these deficiencies. The research team applied for and secured NIHR HSDR funds to build an additional work package into the project designed to strengthen our understanding of decommissioning from a service user, patient and public perspective.

Case study progress

Our study is inevitably constrained by the phenomena we observed within it. Although instructive, the modest progress against aims of the case studies means that some stages of our hypothetical decommissioning model – notably implementation and review – are given less attention than others. We were unable, for example, to analyse how organisational close-down periods were managed (i.e. the time from an announcement until the final day)108 and still less to measure the outcomes and impacts of decommissioning processes. In the single case study in which close-down was enacted (the EOL service), we were unable to re-establish contact with case study sites after key respondents moved on to new posts.

Case study interviews

Although we had intended to conduct a total of 90 interviews across the four case studies, the eventual number was 59. There are a number of reasons for this shortfall. First, we had intended to interview higher numbers of patient and public representatives in case studies. A greater part of the gap is accounted for by a deliberate decision to shift fieldwork in the direction of non-participant observation, as this began to yield important insights (e.g. in case studies 1 and 4), periods of stasis in case studies 1, 3 and 4, which meant less follow-up interviewing was warranted, and key staff moving on from their posts and becoming unavailable for interview during case study 4. We consider the level of overall data to be commensurate with that required for the case studies, notwithstanding the lack of follow-up in case study 4. The addition of unexpectedly fertile opportunities for observation mitigates gaps in the interview data set.

Realist synthesis of the evidence contexts, mechanisms and outcomes in decommissioning

To further enhance understanding of the theory and practice of decommissioning, we recommend that future evidence synthesis takes a realist approach, incorporating insights generated in studies such as this one. 109 In particular, as the volume of empirical case studies increases, future research should identify context–mechanism–outcome configurations and pathways to successful implementation and outcomes. This would provide development of the empirical findings presented here and a complement to the methods employed in evidence reviews to date.

Conceptual and empirical development of loss aversion

We believe that there is a need for theoretical and empirical development of a broader concept of loss than that advanced in prospect theory. 18 It is unclear whether the multiple layers of impact and resistance, including at levels of organisation, institution, politics and community, can be incorporated into an expanded concept of loss and loss aversion or these require separate theoretical constructs reflecting the distinct dynamics at play.

Exploration of the role of financial pressures in limit-setting decisions

The importance of financial constraint runs through all aspects of the study and the experiences of those leading and affected by decommissioning processes. Although it is far from clear what the proper place of affordability considerations should be, our study clearly shows a fissure between the two Delphi panels with regard to the acceptability of financial constraint as a factor in decommissioning. We advocate further investigation, for example using large-scale and deliberative data collection techniques, of the place of financial considerations in health-care limit-setting. There is often talk of the need for a national conversation on limit-setting in the NHS, and we believe that this would be a fruitful way to initiate such a conversation. Our study suggests the further need in future work to disaggregate the perspectives of patients/service users, carers, citizens and third-sector providers so that the subtle differences of emphasis identified in work package 4 can be explored.

Research into community engagement in decision-making

Research is required into the specifics of engaging communities in decommissioning. This would address questions such as:

• How have communities participated in plans for decommissioning of local service provision?

• What dynamics are at play in these processes?

• How might future engagement processes build greater credibility into such processes?

Given the frequency of organisational change in health care, the scale of current reorganisations, the costs involved and the level of community mobilisation against changes, this research is strategically important and urgently needed.

Exploration of decommissioning outcomes

A key limitation of our study is the lack of data on outcomes of decommissioning, either in the CCG survey or the case studies. This would be a valuable focus for future studies, including, for example, through follow-up of case studies included in this study. In case study 3 the decommissioning objectives were largely achieved, but only through follow-up would we be able to determine whether the decommissioning ultimately increased or reduced equality of outcomes. Areas to explore in future research therefore include the impact on patients of the changes as well as budget impact and benefits accrued from released resources. Prior phases of organisational decommissioning processes also require further research, especially linking to organisational closure and how such processes can be managed in ways that minimise harm to employees and service users.

Representatives of agencies with a role in decommissioning (telephone interviews)

Decommissioning narratives (telephone interviews)

Case study interviews round 1 (telephone and face to face)

Case study interviews round 2 (telephone and face to face)