Centralisation of specialist cancer surgery services in two areas of England: the RESPECT-21 mixed-methods evaluation

Major system change in the context of specialist health-care services

Major system change (MSC) is an international issue of growing importance and relevance in health care. A review of the literature defined MSC (or large-system transformation) as a ‘coordinated, system-wide change affecting multiple organisations and care providers, with the goal of making significant improvements in efficiency of health care delivery, the quality of patient care, and population-level patient outcomes’. 1 There are long-standing recommendations in the English NHS and internationally to reorganise specialist services into integrated networks of services in which aspects of specialist care are delivered by a reduced number of larger units, treating a higher volume of patients and hosting a full range of experienced specialists and equipment to support care delivery. 2–6 It is argued that such changes (commonly termed ‘centralisation’) may improve care delivery and patient outcomes, and associations between higher volumes and better outcomes have been demonstrated in some clinical settings. For example, recent research7 indicates that centralising acute stroke services into ‘hub-and-spoke’ systems (i.e. a centralised model of care), in which a smaller number of high-volume units (i.e. hubs) provide specialist hyper-acute care and a larger number of units (i.e. spokes) provide ongoing acute rehabilitation closer to home, is associated with significantly better provision of evidence-based clinical interventions and significantly better clinical outcomes, including patient mortality. However, the strength of this relationship varies between specialties. 8

Recent guidance indicates that centralising specialist services will remain a priority in the English NHS in the future. 9–12 However, although this is a growing field of research, relatively little is known about the processes by which services are centralised, the impact of changes on patients and staff, the cost of implementing change13 and which factors influence implementation. 14–18

Centralisation of specialist cancer surgery services

This study evaluated centralisations addressing four cancer pathways that include complex surgery: (1) prostate cancer, (2) renal cancer, (3) bladder cancer and (4) oesophago-gastric cancer. In the UK, there are over 85,000 new cases of these cancers every year (i.e. prostate cancer, > 48,000 cases;19 bladder cancer, > 10,000 cases;20 renal cancer, > 13,000 cases;21 and oesophago-gastric cancer, > 15,000 cases22–24). Prostate cancer is the second leading cause of cancer deaths in men,25 and 5-year survival rates are around 50–60% for bladder and renal cancers,20,21 16% for gastric cancer and 12% for oesophageal cancer. 22

There are long-standing recommendations to centralise specialist services,2–4 citing the potential to reduce variations in access, increase patient volumes and improve patient outcomes by increasing the likelihood of patients receiving care in hospitals that have a full range of experienced specialists and equipment to support provision of care.

Higher volumes in specialist cancer surgery are associated with better outcomes for oesophago-gastric cancers22 and urological cancers. 26 Research indicates that there is limited evidence of the cost impact of centralising cancer services,27 and limited evidence on patient, public and professional preferences in relation to centralisations of this kind. 28,29 Research indicates that centralisation of cancer services is likely to place increased travel demands on patients and families, and may limit some people’s access to quality care. 30 A review of research evidence indicates that patients are willing to travel further for care for a number of reasons, including for specialist care, if a hospital has a good reputation, if a condition is serious or urgent or if the patient is of a higher socioeconomic status. In contrast, older patients and frequent users of services are less willing to travel further. 31 A recent study suggests that cancer patients are willing to make more frequent, but not longer, journeys to services if it means that they will receive care that is slightly more effective or associated with fewer side effects. 32

London Cancer and Greater Manchester Cancer

This research focused on two integrated cancer systems in the NHS in England: London Cancer and Greater Manchester Cancer. London Cancer (London, UK) covers north-central London, north-east London and west Essex (population 3.2 million). At time of writing, this area was covered by the North Central London Cancer Alliance (London, UK) and the North East London Cancer Alliance (London, UK). Greater Manchester Cancer (Manchester, UK) covers Greater Manchester and east Cheshire (population 3.1 million). At time of writing, this area was covered by the Greater Manchester Cancer Alliance (Manchester, UK). 33,34

London Cancer: context for changes

In London Cancer, when changes were being planned, potential cancer patients were referred to their local cancer centre for diagnosis and either remained there or were referred to a specialist centre (Figure 1a).

FIGURE 1.

Organisation of specialist cancer surgery (a) before planned changes; and (b) after planned changes. MDT, multidisciplinary team. Adapted with permission from Fulop et al. 18 This is an open access article under the terms of the Creative Commons Attribution 4.0 International License (CC BY 4.0), which permits use, distribution and reproduction in any medium, provided the original work is properly cited (see https://creativecommons.org/licenses/by/4.0/). The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Adapted with permission from Fulop et al. 18 This is an open access article under the terms of the Creative Commons Attribution 4.0 International License (CC BY 4.0), which permits use, distribution and reproduction in any medium, provided the original work is properly cited (see https://creativecommons.org/licenses/by/4.0/). The Creative Commons Public Domain Dedication waiver (https://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

The care received by patients varied across centres providing cancer care, including specialist centres. For example, prostate and bladder patients could receive robotic surgery in only certain specialist centres; the majority of renal surgical patients underwent surgery in local non-specialist centres (performed by a specialist or general urologist), rather than in specialist centres (potentially limiting the surgical and other therapeutic options afforded these patients); and oesophago-gastric and urological cancer patients were not guaranteed to see a tumour-specific surgical specialist out of hours or at weekends. Variations existed in the protocols used for referral to specialist centres. Across specialist centres, patient volumes were substantially lower than recommended, and there were variations in access to technology (e.g. robotic surgery), innovative techniques and opportunities to participate in research. At the time, all surgeons provided all types of radical surgery within their specialty (e.g. urologists offered all specialist surgery for bladder, prostate and kidney) and there was limited opportunity for greater surgical ‘subspecialisation’ in specific techniques (e.g. robotic surgery).

Greater Manchester Cancer: context for changes

In Greater Manchester, at the time of planning the changes, patients were referred to a local cancer centre and, depending on diagnosis, either remained at that service for staging or palliative care or were referred to a specialist centre for specialist surgery, chemotherapy and/or radiotherapy (see Figure 1a). Specialist centres were located across the Greater Manchester region and took patients referred from nearby hospitals. Certain aspects of urological care (e.g. robotic surgery) were provided by The Christie Hospital (Manchester, UK). Similar to London Cancer, there was substantial variation in patient volumes across specialist centres.

Changes proposed by London Cancer and Greater Manchester Cancer

In both areas, work started in 2011/12 to create integrated cancer systems. It was proposed that specialist surgical services for these cancers should be centralised into hub-and-spoke systems, with a reduced number of specialist centres providing specialist surgery and local units providing most other aspects of pre- and post-surgical cancer care closer to patients’ homes (see Figure 1b).

Patient pathways to specialist cancer surgery were to be standardised, with the aim of reducing variations in access to care. Within specialist centres, it was anticipated that increased patient volume would permit greater specialisation of staff and greater experience and expertise across teams. In addition, specialist services would offer a full range of surgical technologies (e.g. robotics) and equal access to innovative techniques (e.g. less invasive procedures).

It was planned that local units would continue to provide much cancer care closer to home, including diagnosis, ongoing radiotherapy and chemotherapy. In addition, it was anticipated that local units would benefit from closer involvement of specialist centre staff [e.g. through joint multidisciplinary teams (MDTs)] and specialists providing training and delivering some outpatient care, thereby improving quality of care across the whole system. Both centralisations emphasised the importance of continuity of care (e.g. in terms of dedicated keyworkers to co-ordinate patient care and provide relevant information). 33,34 Table 1 provides an overview of the proposed changes in terms of the number of specialist centres for each type of cancer.

Timeline/progress of changes studied

Implementation of the London Cancer centralisations was completed between December 2015 and April 2016 (see Chapter 4). Implementation in Greater Manchester was delayed for a range of reasons (see Chapter 12). Centralisation of Greater Manchester Cancer’s oesophago-gastric cancer surgery services was completed in September 2018, whereas Greater Manchester Cancer’s planned centralisations of specialist surgery for urological cancers (i.e. bladder, prostate and kidney) were not implemented over the course of this study. Consequently, we were able to study the impact and cost-effectiveness of only the London Cancer changes, but we were able to study factors influencing the progress of implementation for both London Cancer and Greater Manchester Cancer. We provide details of our updated study design under Aims and objectives and Overview of the research, and in Chapter 2.

Over the course of this study, the National Cancer Vanguard was operational. The National Cancer Vanguard was a partnership that included London Cancer and Greater Manchester Cancer. 36,37 In 2016, incorporating the learning from the National Cancer Vanguard, the English NHS introduced cancer alliances (of which there are now a total of 21 covering the whole of the English NHS) ‘to bring together local senior clinical and managerial leaders representing the whole cancer patient pathway across a specific geography’. 37 In 2019, the geography covered by London Cancer reverted to two separate cancer alliances (i.e. the North Central London Cancer Alliance and the North East London Cancer Alliances) to align better with sustainability and transformation partnership footprints in the region. The split resulted from stakeholder decisions after a self-assessment process initiated by NHS England and Improvement and applied to all alliances nationally. The two alliances have continued to use the same pathway configurations, with single specialist cancer surgical centres serving both alliances.

Framework for understanding major system change

The approaches were combined using a framework that reflected key processes of MSC and how they are inter-related (Figure 2). This framework drew on several established conceptual frameworks, describing different aspects of the planning, implementation and outcomes of change,1,14,38–44 and was developed originally to support the evaluation of MSC in acute stroke services, with the potential for application in other contexts. 14 The framework covered (1) stakeholder preferences (note that this was an addition to the original framework), (2) reaching a decision to change, (3) developing and agreeing the new service model, (4) implementing the new model, (5) adherence to the new model throughout the system, (6) impact on provision of care and (7) impact on outcomes (including clinical outcomes, patient experience and cost-effectiveness) (note that the order of these factors should not be taken to imply a linear relationship between them).

FIGURE 2.

Framework for analysing MSC, adapted for the RESPECT-21 (REorganising SPECialisT cancer surgery for the 21st century) study. DCE, discrete choice experiment; HES, Hospital Episode Statistics; NCRAS, National Cancer Registration and Analysis Service; ONS, Office for National Statistics; RQ, research question. Adapted with permission from Fulop et al. 18 This is an open access article under the terms of the Creative Commons Attribution 4.0 International License (CC BY 4.0), which permits use, distribution and reproduction in any medium, provided the original work is properly cited (see https://creativecommons.org/licenses/by/4.0/). The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Adapted with permission from Fulop et al. 18 This is an open access article under the terms of the Creative Commons Attribution 4.0 International License (CC BY 4.0), which permits use, distribution and reproduction in any medium, provided the original work is properly cited (see https://creativecommons.org/licenses/by/4.0/). The Creative Commons Public Domain Dedication waiver (https://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

There are important differences between the context in which this framework was developed and the context in which it was applied in this study. Stroke is a health-care event that requires immediate response, whereas specialist cancer surgical services operate at a different pace, affording greater opportunities for care-planning and engagement with the patient and family regarding treatment options. These contextual differences between stroke and specialist cancer surgery potentially introduced different considerations into the decision to change, model selection and implementation approaches (all with potential implications for progress of change).

Stakeholder preferences for centralisation: London Cancer, Greater Manchester Cancer and national control (research question 1)

The centralisations had the potential to significantly change how care was organised and delivered, with implications for patient travel times, choice of treatments and, potentially, outcomes. To examine the acceptability of such changes to patients, the public and professionals, we conducted a discrete choice experiment (DCE). 45–47 The DCE examined stakeholder preferences for centralisation, the relative importance of attributes of surgical services and how preferences vary between stakeholder groups. The DCE was designed in line with international best practice guidelines (detailed methods are presented in Chapter 3 and Vallejo-Torres et al. 48).

Implementation and sustainability of change: London Cancer and Greater Manchester Cancer (research questions 2 and 3)

We used qualitative methods to understand how the London Cancer and Greater Manchester Cancer changes were planned and implemented, and to understand the progress and impact of changes (including factors contributing to non-implementation). We analysed key documents (e.g. project plans, meeting minutes and local press) to develop a timeline of which processes were carried out and when, in the planning and implementation of change. Documentary analysis was also used to supplement and extend understanding of findings emerging from other qualitative components. We interviewed a range of stakeholders related to the London Cancer and Greater Manchester Cancer changes to understand relevant perspectives on how and why change happened, and the implications of change for different groups. We used purposive sampling combined with snowball sampling to ensure a good range of perspectives. Figure 3 presents our sampling for qualitative work, which focused at the governance level (including programme teams, pathway boards and wider context, e.g. patient representative groups, payer organisations and NHS England) and the service level (including specialist centres, local units that had provided specialist surgery pre centralisation and local units that had not provided specialist surgery pre centralisation). Clinical staff interviewed included surgeons, specialist nurses, oncologists and allied health professionals (e.g. therapists, dieticians and radiologists).

FIGURE 3.

Sampling for qualitative research.

Reflecting previous research on evaluating MSC14,18 and our framework for analysing MSC (see Figure 2), our interviews were guided by semistructured topic guides that focused on such topics as drivers for change and factors influencing key stages of change (e.g. agreeing the case for change, selecting the service model, planning and implementing changes, and their impact on quality of care) (see Appendix 8). We conducted non-participant observations of events to obtain a direct understanding of key processes in action. Observations focused on governance and implementation of change (e.g. oversight and planning meetings, including the overarching London Cancer and Greater Manchester Cancer boards) and organisation and delivery of care post centralisation (e.g. MDT meetings).

To interpret these data, we used a comparative case study approach. 49,50 Our analysis was guided by our framework for understanding MSC, which we developed with reference to several other key conceptual frameworks for understanding implementation and outcomes of change. 51 In addition, to understand specific aspects of MSC, we drew on relevant literatures on networks and interorganisational collaboration (see Chapter 5), the influence of history (see Chapter 6), subtractive loss (see Chapter 7) and organisational context (see Chapter 12). In each case, the literature is introduced and discussed in the relevant chapter.

We organised interview transcripts and observation field notes with NVivo version 12 (QSR International, Warrington, UK) and Microsoft Excel^® (Version 2205; Microsoft Corporation, Redmond, WA, USA) software. Initial analysis and category-building was led by the London and Manchester qualitative researchers. The analysis was developed with a subgroup of co-investigators who have qualitative expertise and the whole research team contributed to the interpretation of findings. Ongoing iterative and thematic analysis of all data was undertaken, following established procedures of constant comparative analysis. 52

Table 3 presents the qualitative data collected and analysed. Summaries of interview data are disaggregated by organisation level (i.e. governance and service levels) and specifying the types of change services underwent (or anticipated undergoing) through centralisation.

In conducting 212 interviews (including follow-ups), we interviewed a total of 176 stakeholders (Greater Manchester Cancer, n = 76; London Cancer, n = 100). Of those interviewed, 105 were clinicians working in oesophago-gastric and urological services (Greater Manchester Cancer, n = 44; London Cancer, n = 61), including consultant surgeons, consultant oncologists and pathologists, specialist nurses and allied health professionals (e.g. dieticians, occupational therapists, physiotherapists and radiologists).

Impact on clinical processes and clinical outcomes: London Cancer only (research question 4)

The proposals for change in London Cancer and Greater Manchester Cancer identified the potential for significant improvements in care and outcomes. Therefore, we analysed a range of national data sets that captured clinical outcomes and delivery of clinical interventions. We assembled data from the National Cancer Registration and Analysis Service (NCRAS) data linked to Hospital Episode Statistics (HES) and Office for National Statistics (ONS) mortality data to analyse the impact of selected cancer surgery service centralisations on a range of outcomes [e.g. mortality, re-admission, length of stay (LOS)] and impact on care process measures (e.g. surgical complications, surgical technique). We used a difference-in-differences approach, evaluating the changes in the outcomes over time following centralisation in London Cancer, accounting for changes seen during the same time period in the rest of England (see Chapter 9 for further details of the data, measures and methods used).

Impact on patient experience (research question 5)

Patient experience was a priority for London Cancer and Greater Manchester Cancer change planners. We originally planned to study the impact on key aspects of patient experience as part of our quantitative analysis, drawing on National Cancer Patient Experience Survey (NCPES) data. However, owing to issues with this data set (e.g. it was not possible to distinguish patients who had surgery from other types of management, nor disaggregate by specific cancer types addressed by the London Cancer changes) we were unable to analyse quantitatively the impact of London Cancer changes on patient experience. In lieu of quantitative data, we analysed qualitative interviews to establish London Cancer staff perceptions of how MSC influenced patient experience.

Implementation costs and cost-effectiveness: London Cancer only (research question 6)

As noted in Chapter 1, little is known about the cost-effectiveness of MSC, and the implementation costs of MSC are rarely evaluated and even more rarely incorporated into incremental cost-effectiveness analyses. To address these gaps, we studied the cost of implementing the London Cancer changes (see Chapter 8), analysing key supports of change (e.g. meetings, events, clinical and managerial staff time, and programme team costs) and costs of new services (e.g. staffing, space and technology). To assess overall value for money of the London Cancer changes, we conducted a cost-effectiveness analysis (see Chapter 10). We analysed clinical processes and patient outcome data, alongside national and local cost data, incorporating implementation costs, to generate an incremental cost per quality-adjusted life-year (QALY) gained in the reconfigured services in the London Cancer region compared with the equivalent scenario without these specific changes, based on the difference-in-differences analysis framework used in Chapter 9.

How lessons might apply in other contexts: London Cancer, Greater Manchester Cancer and national control (research question 7)

We anticipated that lessons from this research would be of interest and use to people planning MSC in both cancer and ‘non-cancer’ specialist settings. However, we also recognised that stakeholders in different contexts would have valuable perspectives on how lessons may be adapted to enhance their applicability to other contexts. Therefore, we conducted a workshop both for people involved in planning centralisations of specialist cancer services elsewhere and for those involved in planning centralisation of non-cancer specialist services (see Chapter 13). Through this workshop we explored factors influencing applicability of the lessons to different settings and developed lessons that could be of use in these settings.

Stakeholder preferences for centralisation: London Cancer, Greater Manchester Cancer and national control (research question 1)

Recruitment to the DCE was arranged by the research team and Quality Health (Chesterfield, UK) (which, at the time, administered the NCPES). The DCE questionnaire included a cover letter and an information sheet that included study details, what participating would entail and information on data management and storage. We recruited the three stakeholder groups as follows.

Implementation and sustainability of change: London Cancer and Greater Manchester Cancer (research questions 2 and 3)

Potential interviewees were identified using documentary evidence and ‘snowball’ sampling, and were contacted via e-mail or telephone. Interviewees were given at least 48 hours in which to consider the study information and interviews were conducted only after fully informed and written consent had been obtained. Interviews lasted approximately 50 minutes and were audio-recorded and professionally transcribed. Non-participant observations were conducted with fully informed consent from the chairperson and members. All documents analysed were either in the public domain or obtained from local stakeholders.

Data collection for other research components

Our approach to data requests for the quantitative and cost-effectiveness analyses is presented in the relevant findings chapters (see Chapters 8, 9 and 10), and our approaches to recruitment and data collection for our stakeholder workshop are presented in Chapter 13.

Presenting qualitative data

When presenting interview quotations, we use anonymised participant identifiers. The identifiers for each level of our sample are presented in Table 3. For each quotation, we also present a short statement of the individual’s role (e.g. urological cancer surgeon, clinical nurse specialist) and geographic location in which they were based. For document quotations, we state the document sources. For quotations from non-participant observations and documents we state the event and date on which it took place (e.g. project board meeting, 25 December 2010).

What is already known?

• Centralising specialist cancer surgical services is expected to reduce variations in quality of care and improve patient outcomes.

• One disadvantage of centralisation is that it leads to an increase in travel demands on patients and families.

• Aligning system changes with stakeholders’ preferences might increase the likelihood of successful implementation.

What does this chapter add?

• Patients’, health professionals’ and the public’s preferences were particularly influenced by the risk of complications, the risk of death and the access to specialist MDTs, whereas travel time was considered the least important factor.

• Individual preferences were found to be consistent with the major goals of centralising cancer surgery services.

Sampling and recruitment

Discrete choice experiment responses were obtained from three groups: (1) cancer patients (target sample size n = 200), (2) members of the public (n = 100) and (3) health professionals involved in the treatment of patients with cancer (n = 100). Data were collected by hard-copy postal questionnaires (which were sent to patients) and online surveys (which were made available to the public, patients and health professionals). The sample was recruited through a number of routes (see Chapter 2 for more details).

Attributes and attribute levels

Analysis of the information from planning documents, covering development, planning and implementation of the changes33,34,55,56 and responses from the questionnaire,57 identified the following six attributes as those that are most likely to be important to respondents and likely to change as a result of centralising specialist cancer surgical services: (1) travel time to hospital, (2) risk of serious complications from surgery, (3) risk of death within 30 days of surgery, (4) number of operations the centre carries out each year, (5) access to a specialist MDT and (6) availability of specialist surgeon cover after the operation (Figure 4). The levels of each attribute were based on planning documents (as above) and input from the RESPECT-21 RSG. Descriptions were developed for each of the attributes to help participants understand the nature of each attribute that they were being asked to consider (note that the complete questionnaire is available in supplementary material for Vallejo-Torres et al. 48).

FIGURE 4.

Discrete choice experiment survey design and format. (a) Attributes and levels used in the DCE; and (b) example of a DCE choice set. Adapted with permission from Vallejo-Torres et al. 48 This is an open access article under the terms of the Creative Commons Attribution 4.0 International License (CC BY 4.0), which permits use, distribution and reproduction in any medium, provided the original work is properly cited (see https://creativecommons.org/licenses/by/4.0/).

Adapted with permission from Vallejo-Torres et al. 48 This is an open access article under the terms of the Creative Commons Attribution 4.0 International License (CC BY 4.0), which permits use, distribution and reproduction in any medium, provided the original work is properly cited (see https://creativecommons.org/licenses/by/4.0/).

Questionnaire design

Respondents were asked to choose their preferred option from a series of pairwise choices (i.e. in which of two fictitious centres would they prefer to have surgery). Similarly, health professionals were asked in which centre they would prefer their patients to have surgery. Each centre was described by a unique combination of different levels of the attributes (see Figure 4 for an example of a DCE question).

In addition, the questionnaire included an initial question that asked respondents to rank the six attributes according to their overall importance, from 1 (most important) to 6 (least important). Information on demographics, socioeconomic status and cancer-related experience was also collected.

Data analysis

Responses to the ranking questions are presented graphically (Figure 5) and we measured inter-rater agreement using kappa statistics. 58

The DCE data were analysed using a conditional logit regression model in which the outcome was centre preference (i.e. A or B) and the variables in the equation were the individual attributes. We ran the model on the whole sample, as well as stratifying participants by the three groups. We tested for differences in preferences between the groups using chi-squared tests.

We included the travel time attribute as a continuous variable, taking the higher-end value of each interval (i.e. 30, 60, 90 and 120 minutes). This specification allowed marginal rates of substitution with respect to this variable to be computed.

In addition, we used the regression analysis results to calculate the predicted probabilities of choosing cancer surgical services with attribute levels corresponding to the goals of centralisation, compared with a non-centralised service. Specifically, we compared the probability that a respondent would choose a hypothetical non-centralised service [which was defined as 30 minutes’ travel time, 10 operations carried per year at the centre (for both attributes these were the lowest levels included in the study), no access to a specialist MDT, specialist surgeon cover during normal hours only, a 5% risk of complication and a 1.5% risk of death] against various different centralised service scenarios. In each case, travel time was fixed at 120 minutes and the number of operations at the centre was increased to 100 operations per year (i.e. the highest levels included in the study). In addition, the following characteristics were added individually and then jointly: (1) access to a specialist MDT, (2) access to specialist surgeon cover 24 hours per day, 7 days per week (24/7), (3) risk of complications reduced to 1% and (4) risk of death reduced to 0.5%.

All analyses were undertaken using the software package Stata^® version 12.0 (StataCorp LP, College Station, TX, USA).

Respondents’ characteristics

From July to November 2016, we obtained 444 responses (patients, n = 206; health professionals, n = 111; members of the public, n = 127). DCE questions were completed in full by 199 patients, 109 health professionals and 125 members of the public. Our analysis was a complete-case analysis using only these respondents’ answers. Table 4 provides a summary of demographic characteristics by group.

Simple attribute ranking

A total of 328 respondents (patients, n = 119; members of the public, n = 113; health professionals, n = 96) provided full responses to the ranking question. Figure 5 shows graphically the responses for each of the three groups separately. The kappa statistic overall was 0.1166. For each subgroup, the kappa statistics were 0.0765, 0.1268 and 0.1501 for health professionals, patients and the general public, respectively, representing ‘slight’ agreement among rankers in each case. 59

FIGURE 5.

Attribute rankings by (a) patients (n = 119); (b) members of the public (n = 113); and (c) health professionals (n = 96). Adapted with permission from Vallejo-Torres et al. 48 This is an open access article under the terms of the Creative Commons Attribution 4.0 International License (CC BY 4.0), which permits use, distribution and reproduction in any medium, provided the original work is properly cited (see https://creativecommons.org/licenses/by/4.0/).

Adapted with permission from Vallejo-Torres et al. 48 This is an open access article under the terms of the Creative Commons Attribution 4.0 International License (CC BY 4.0), which permits use, distribution and reproduction in any medium, provided the original work is properly cited (see https://creativecommons.org/licenses/by/4.0/).

When using this method of ranking, risk of death and risk of complications were ranked highly in each sample, and travel time was consistently considered to be the least important factor by each group. However, we did observe some differences across groups. For example, patients appeared to consider the availability of a specialist MDT highly important, whereas health professionals considered the availability of a specialist surgeon 24/7 more important.

Discrete choice experiment analysis

We found no statistically significant differences in the effects of the attributes between groups, except for the risk of complications, which had a slightly larger impact in the public sample compared with the patient sample. Therefore, we focused on the model conducted on the whole sample (Table 5).

We observed that, as expected, individuals preferred to have surgery in a centre requiring shorter travel time, where the risk of complications and the risk of death were lower, the number of operations carried out each year was larger, and there was access to a specialist MDT and specialist surgeon cover 24/7. We found that participants were willing to travel 75 minutes longer to reduce the risk of complications by 1% and over 5 hours longer to reduce their risk of death after surgery by 1%. Participants’ willingness to travel increases by 5 more minutes for every additional surgery carried out by the centre each year, and by approximately 4 and 3 hours to have access to a specialist MDT and access to specialist surgeon cover, respectively.

The probability that respondents would choose a centre with attribute levels corresponding to a centralised service compared with a non-centralised service is presented in Figure 6. Compared with a centre requiring 30 minutes’ travel time that carries out 10 operations per year (i.e. a generic non-centralised service), respondents are less likely to choose a centre that carries out 100 operations a year, but for which the travel time increases to 120 minutes, holding the rest of the attributes constant (i.e. ‘worst-case scenario’). However, the probability that respondents would choose the centralised service increases if the centre also achieves the goals with respect to each of the other attributes. The probability that respondents would choose the centralised service is 72% if the centre provides access to specialist surgeon cover 24/7, 74% if there is access to a specialist MDT, 76% if the risk of complications is reduced from 5% to 1% and 76% if the risk of death is reduced from 1.5% to 0.5%. If the centralised service achieves all of these changes in the attributes, at the expense of increasing travel time from 30 minutes to 120 minutes, defined as the ‘best-case scenario’ in Figure 6, then the probability that respondents would prefer to have surgery in the centralised service reaches 92%.

FIGURE 6.

Predicted probabilities of choosing centralised cancer surgery services. Adapted with permission from Vallejo-Torres et al. 48 This is an open access article under the terms of the Creative Commons Attribution 4.0 International License (CC BY 4.0), which permits use, distribution and reproduction in any medium, provided the original work is properly cited (see https://creativecommons.org/licenses/by/4.0/).

Adapted with permission from Vallejo-Torres et al. 48 This is an open access article under the terms of the Creative Commons Attribution 4.0 International License (CC BY 4.0), which permits use, distribution and reproduction in any medium, provided the original work is properly cited (see https://creativecommons.org/licenses/by/4.0/).

Principal findings

In this study, we explored patients’, health professionals’ and the public’s preferences for centralising specialist cancer surgery services using a DCE. We found that, consistently across the three groups, respondents’ preferences behaved as expected. Individuals preferred attributes with better values (e.g. shorter travel times; lower risk of death and of complication; and access to more specialised centres, teams and surgeons). We also found that preferences were particularly influenced by the risk of complications, the risk of death and the access to a specialist MDT. Travel time was considered the least important factor. Preferences were, therefore, found to be consistent with the goals of centralisation.

The estimated probability that participants would choose to have surgery in the best-case scenario, with a centre meeting the changes corresponding to the aims of centralisation, was very high, estimated at 92%. However, it is important to note that the impact on mortality and complications in this best-case scenario might not be achieved. Furthermore, if centralisation of cancer surgery services implied an increase in travel time to attend a hospital that carries out a larger number of operations in a year, but that does not offer the additional benefits associated with centralisation, then participants would prefer to have surgery in a non-centralised centre.

Strengths and weaknesses

The analysis applied a DCE that allowed the evaluation of both attributes service respondents would prefer to receive and the trade-offs that respondents are willing to make between attributes.

The analysis was undertaken with responses from over 400 participants, including patients with each type of cancer under evaluation, health professionals and members of the general public. Content validity of the DCE was obtained by grounding the attributes and levels of the changes expected from centralisation based on careful reviews of planning documents and on responses from initial questionnaires to identify the most important factors. The questionnaire was carefully tested and revised during piloting.

We acknowledge several limitations. DCEs elicit hypothetical choices and, therefore, might lack external validity if individuals do not make the same choices in real-life situations. The representativeness of the sample responding to the questionnaire might be limited, as the generalisability of the findings depends on individuals elsewhere having similar preferences. Although the selection of attributes included in the DCE was carefully considered, we acknowledge that there might be other factors affected by centralisation not included in our analysis that may also be considered important for individuals. Similarly, we explored patients’, health professionals’ and the public’s views, but the views of other groups might also be important in the planning, implementation and delivery of MSCs, such as hospital managers and health-care decision-makers.

Another limitation is that in this study we have analysed preferences for specialist cancer surgery services in general, and these preferences might vary by different types of cancer.

Comparison with other studies

This study provides, to our knowledge, the first evaluation of individual preferences with respect to changes associated with the centralisation of cancer surgical services. Previous studies have explored the impact of travel on cancer patients’ experiences of treatment, finding a paucity of research in this area and inconclusive evidence. 53

Implications

There are several implications of our study. First, planners who are redesigning services might consider and measure the impact of the reorganisation on the factors identified as being important in this study. Health policy in England has focused on improving access to services; however, our findings highlight that, in the context of centralising specialist surgery services for cancer, people are willing to trade travel time for better outcomes and quality of care. For centralisation to be judged favourably by patients, the public and health professionals, compared with a non-centralised model, it needs to demonstrate improvements in outcomes (e.g. complications and mortality) and/or delivery (e.g. in terms of postoperative surgeon care and specialist MDT input). In addition, although travel time was identified as the least important factor, the DCE analysis showed that this factor still plays a role in people’s preferences for care and, therefore, plans for transport support and parking facilities for patients and their families would also improve individuals’ views towards centralisation.

What is already known?

• MSC has multiple, sometimes conflicting, goals and involves implementing change processes across a number of organisations.

• Evidence of networks has demonstrated their capability to attempt to address ‘wicked problems’ that are characteristic of MSC in complex, although not uncontested, ways.

What does this chapter add?

• This chapter provides a new understanding of MSC by discussing the strategies used by London Cancer to facilitate MSC in a health-care context in the absence of system-wide authority.

• MSC was a contested process in London Cancer. Actors across the network, including clinicians and patients, questioned the rationale for the changes, the clinical evidence behind them and the ways in which the changes were made.

• A core central team composed of network leaders, managers and clinical–manager hybrid roles was able to drive changes forward by developing different forms of engagement with provider organisations, distributing leadership across vertical and horizontal layers, and maintaining constancy in central leadership over time.

Design

The study built on previous research exploring network leadership and evidence on the strategies used by networks to implement MSCs in health care. In this chapter, we focus on the role of the network in implementing the planned changes. 66

Data collection and sampling

We conducted a qualitative study of the centralisation of specialist oesophago-gastric, prostate, bladder and renal cancer surgery in London. The qualitative study focused on 10 sites. We combined documentary evidence (n = 100 documents), non-participant observations (134 hours) and interviews with stakeholders (n = 81) (sampling is shown in Table 6).

Data analysis

Interview transcripts, observation notes and documentary evidence were analysed using thematic analysis. 67

Central network leadership drove the changes forward

The role of chief medical officer at a local academic health science partnership was established to oversee the design, planning and implementation of the changes. The chief medical officer was a clinician by background, but her clinical specialty was not involved in the centralisation. The chief medical officer was also based at an ‘independent’ organisation, in the sense that it was not a part of any of the provider organisations in the network. A network board (i.e. an independent skills-based board formed of experts external to London and chaired by a former cancer patient) was created to make clinically led recommendations for the model of care. The network board oversaw a bidding process in which provider organisations stated how they would host services as specialist centres and how they would work with the other providers in the network. Where prior consensus was not achieved and competing bids were submitted by provider organisations, the proposals were reviewed externally. These recommendations were agreed by the chief executives and medical directors of the network provider organisations. The chief medical officer and chairperson of the board were perceived by many managers and clinicians across the network as providing strong and objective leadership, with a clear vision and mandate to implement the changes outlined in the model of care.

The relative independence of the chief medical officer role and the board was seen by some members of the network as a factor that allowed the central leadership of the network to be seen as ‘neutral’. However, other actors in the network associated central leadership figures with dominant provider organisations, that is those organisations that obtained most of the specialist cancer workload as a result of the reconfiguration. Constancy in network leadership over time was also perceived to have enabled the implementation of changes, even in the light of the profound organisational restructuring of the health-care system during and after the 2012 reforms.

The central leadership team drew from existing evidence on the potential benefits of the centralising specialist cancer surgery and previous experiences of centralisation as a way to justify the need for the changes. Data were not always readily available and some interviewees spent a considerable amount of time searching for and collating data, and developing new sources when these were not available. There were some discussions about the quality, veracity and inclusivity of the data used to guide decisions on the reconfiguration. Some local surgeons expressed doubts about how the data were used.

Network managers supported leaders

The chief medical officer and the chairperson of the board played a central leadership role, but staff members in managerial and clinical roles across other layers of the network also played important roles. The network managers played an instrumental role in supporting leaders, mediating relationships across sites and facilitating the day-to-day requirements of the changes. The board appointed clinical leads to chair pathway boards and design the integrated cancer care pathways. 68

The network core team designated, arranged training for and supported leaders from each of these pathway boards and these leaders became a core leadership team that remained beyond the implementation of the changes. Leaders were selected for their skills in leading teams, engaging with a wide range of stakeholders and building relationships. Leaders needed to act in hybrid roles, that is having clinical knowledge, as well as managerial skills. An early planning document stated: ‘as we interview pathway directors, we will carefully screen for those with the leadership competencies to spread our culture quickly through the community’. 69 The screening process was based on the use of competency models and role-play simulations. 69 The appointed leaders took part in personal leadership development sessions.

Engagement across provider organisations

The network encouraged pathway leads to engage with a wide range of stakeholders within the network and ‘develop relationships with colleagues across the care pathway’. 69 These relationships were considered central aspects of their role objectives and the leadership development programme. Leaders used their own styles to create change. Successful implementation of changes was associated with leaders who were viewed as trustworthy, as having a clear vision and showing dedication, and as good at building relationships.

Planning the changes entailed the inclusion of representatives from all provider organisations. There was an expectation that if all organisations were involved during early stages, then the implementation of the changes would be smoother, as all organisations would have a sense of ownership of the new pathways, building momentum to drive the changes forward and sustain the changes over time.

This expectation of engagement, however, was practised in a context that was interpreted by some as infused with a spirit of competition for future surgical activity. There was also a concern that specialist centres would be ‘taking over’ the network and absorbing all of the specialist care, leaving local centres without the ability to retain surgical expertise and recruit new members of staff.

Engagement of other stakeholders

Commissioners had a legal obligation to lead a public engagement process on the centralisation. 70 In England, following the 2012 reforms to the NHS, it was commissioners [i.e. Clinical Commissioning Groups (CCGs)] that had to make the decision of whether or not to centralise services (i.e. the network board could only make recommendations to commissioners). Although commissioners were kept informed in the early stages of planning, they only became more involved in 2013 when commissioning structures consolidated. Commissioners were involved in developing the reporting and governance mechanisms for the implementation of changes (in the form of gateway reviews, a process by which programmes and projects are examined at key decision points in their life cycle) and were involved in carrying out two phases of an engagement process (towards the end of 2013 and in 2014), which comprised meetings with health-care professionals, drop-in sessions for members of the public and workshops. The purpose of this engagement process was to discuss the proposals for changes with a wide range of stakeholders to consider all available options before moving forward with implementation.

The engagement aimed to bring together a wide range of stakeholders to ‘gain their views and experience of current services and hear their aspirations for the health services they would receive in the future’ (contains public sector information licensed under the Open Government Licence v3.0). 70 This engagement process included considering recommendations made by the board as well as additional options not recommended by this board (i.e. independent assessments). Patients, families and members of the public were asked about these changes through public meetings.

Some patient groups used the engagement process to express their concerns about the centralisation plans. These patient groups felt that the needs of patients would not be considered a priority in a centralised model of care; for example, some patients would be forced to travel for longer periods of time to undergo specialist surgery. Patient representatives who had worked with members of the network and were in favour of the changes responded to these patients’ opposition by publicly highlighting the potentially positive effects of the changes on patient care, experience and health outcomes. Travel was recognised as a burden, but not a barrier to care.

Principal findings

Our analysis of the implementation of the centralisation of oesophago-gastric and urological cancer surgical services across the network combined existing knowledge regarding the processes that facilitate MSC and the characteristics and strategies used by network leaders and managers to agree and create change. Our findings pointed to the role played by central network leadership figures and network managers in creating, supporting and maintaining consensus to drive the changes. These changes were implemented after major reforms in 2012, in a context with substantially reduced system-wide, or ‘top-down’, authority. The network relied on key leadership figures, as well as distributed, or more horizontal, forms of leadership across the network to bring together a diverse group of stakeholders and drive change. As in other types of health-care networks, an emphasis on the ‘political neutrality’ of central network leaders and the board was employed as a strategy to position themselves outside the competitive health-care landscape and drive the changes forward. 66

Constancy in the people who enacted leadership roles allowed the network to drive the changes forward, even in the face of the organisational restructuring of the health-care system. Senior clinical leaders who acted as pathway leads occupied hybrid leadership roles71 in which they assumed management responsibilities while still retaining their clinical role. This granted senior clinical leaders credibility within the network because of their clinical knowledge71 (also referred to as ‘reputational framing’66) and allowed them to use their managerial skills to bridge organisational boundaries and bring together representatives from multiple professional groups. 72

Clinical leadership of the changes was prioritised and actively fostered, as clinicians were strategically recruited and trained as leaders, developing their hybrid roles. Leadership training was based on inclusive models of participation in the changes and gave clinicians the skills to create ‘buy-in’ across sites and bring stakeholders on board. The work of these clinical leaders depended heavily on the support provided by programme-level network managers who acted as facilitators, connections or bridges across organisations and professional groups.

Early engagement of a wide range of stakeholders led to the creation of local champions across different layers and sectors of the networks, building up the pressure to drive change. Fitzgerald et al. 73 have discussed the role played by this cumulative effect of distributed leadership on organisational change and service improvement, whereby leadership enacted across multiple tiers creates a driving force to move events in a particular direction and sustain these changes through time. The changes were fully implemented, but many clinicians and managers continued to express disagreement with the rationale for the centralisation, the evidence used to justify the changes and the ways in which the changes were made (i.e. how sites were selected to act as specialist centres), even several years after implementation.

Consistent with other studies of MSC,62,74 we found that PPI served various purposes, from a genuine interest in ensuring that the needs and interests of patients were represented in the planning and implementation of the changes to more instrumental forms of involvement (i.e. to minimise resistance or act as champions). In 2013, engagement with commissioners became stronger after the 2012 reforms were implemented, when clear reporting and governance mechanisms were established to ensure that the provider network complied with the model of care and public consultation outcomes.

Strengths and weaknesses

The retrospective nature of some of the interviews meant they could have been influenced by recall bias. To reduce the risk of bias, we used documentary evidence to complement interviewees’ narration of past events. We made an effort to maintain an inclusive sampling strategy, but we might have missed relevant individuals. Our study analysed the implementation of MSC in a specific health-care area and in an urban setting; however, additional work is required to explore the role of provider networks in MSC in other specialties and contexts.

Implications

Our study of the role of the network in the centralisation of specialist cancer surgery has shed light on the strategies that may be used by networks of provider organisations to implement MSC in health-care contexts in the absence of a system-wide authority to lead the system. Our study extends previous frameworks developed for the study of MSC, which pointed to the need for a combination of top-down and bottom-up leadership to implement MSC,1,61 by describing the role networks can play in facilitating MSC and the processes of negotiation involved in the implementation of such changes.

In the case of our study, MSC in specialist cancer was a contested process with actors across the network, including clinicians and patients, questioning the rationale for the changes, the clinical evidence behind it and the ways in which the changes were made. A core central team composed of network leaders, managers and clinical–manager hybrid roles was able to drive the changes forward by developing different forms of engagement with provider organisations, distributing leadership across vertical and horizontal layers and maintaining constancy in central leadership over time.

What is already known?

• Models of MSC, such as centralisation of specialist care, require aggregation of effort by multiple organisations to plan and implement the change.

• One way to implement MSC is through the creation of new, or the use of existing, provider networks.

• Interorganisational collaboration has been identified as one of the key mechanisms enabling care delivery across provider networks.

What does this chapter add?

• This chapter sheds light on factors that enable and complicate interorganisational collaboration, which is a key component of health-care delivery in most countries.

• Provider organisations in London Cancer negotiated power relations to establish shared goals, reached consensus in relation to maintaining patient-centred care, maintained central figures who could create and sustain collaboration, and promoted distributed forms of leadership.

• The implementation of interorganisational collaboration was still under transformation during our analysis. Future research will need to explore the sustainability of these collaborative relationships and identify the factors that might prompt changes in approaches to collaboration used in networks of provider organisations.

• We present a revised framework for understanding interorganisational collaboration that can be relevant for networks of provider organisations in other contexts.

Design and conceptual framework

The analysis of interorganisational collaboration in this chapter was guided by a conceptual framework informed by previous work developed by D’Amour et al. 83 on collaborative relationships in health-care contexts and was refined by drawing from the field of cultural politics, as well as the empirical findings from our study presented in this chapter.

Data collection and sampling

Data collection for the study took place between September 2015 and April 2019 and focused on 10 sites (including specialist and local centres). We combined an analysis of documentary evidence (n = 100 documents), non-participant observations (totalling 163 hours) and interviews with stakeholders (n = 117) involved in the centralisation of cancer surgery (see Chapter 2 and Table 3).

Data analysis

Interview transcripts, observation notes and documentary evidence were analysed using thematic analysis. 67

Interlocking directorates (collaboration at board level)

Professional affiliation (collaboration between health-care professionals)

Professional affiliation and patient transfer

Principal findings

We identified some factors that contributed to the development of collaborative relationships across the network. These factors included the establishment of shared goals (at least by some organisations), attempts to reach consensus in relation to maintaining patient-centred care, the existence of central figures who could drive the centralisation and the promotion of distributed forms of leadership. Processes for enabling interorganisational collaboration, such as pathway-level meetings, specialist MDTs, joint clinical roles and discipline-specific meetings, were developed over time (experiencing some early ‘teething issues’), then consolidating into routine practice. However, some areas were still under development toward the end of our study.

These differences point to the need to visualise study provider networks as dynamic entities, made up of relationships that are dependent on historical factors (in this case, previous connections between members of the same professional group) and existing infrastructure, but also recognising the potential for these relationships to mutate into new types of collaboration. In an analysis of integrated care networks, Mitterlechner90 found that the relationships and governance of these networks changed repeatedly through repetitive sequences of collaborative inquiry. These sequences allowed the network to address problems in experimental and innovative ways. 90 In the case of the network in our case study, we were also able to see that collaborative relationships changed in relation to the negotiation of power relations between organisations, whereby new powerful actors in the form of specialist centres emerged under the role of ‘system leaders’ to drive the centralisation forward. 91,92

We identified key areas that require further development to guarantee an active degree of collaboration across the 10 indicators. These areas involved developing opportunities for mutual acquaintanceship across all professional groups; the active sharing of knowledge, expertise and good practice across the network; the fostering of trust; and the creation of information exchange infrastructures fit for collaborative purposes. 93 As our study ended, staff members across different hierarchies of the network indicated that active work was under way to address challenges in information exchange and the sharing of expertise, but other areas, such as lack of trust, mutual acquaintanceship and connectivity, had not been addressed yet.

We found that it was not enough for provider organisations to maintain shared goals, it was also important to address different viewpoints on how these goals should be achieved. Although all provider organisations aimed to deliver patient-centred care and improve outcomes, not all agreed that the centralisation model proposed for London Cancer was the best way to achieve these aims. In the case of those who agreed with the centralisation, not all believed that the sites that were selected to act as specialist centres were the best to deliver specialist cancer surgery.

Fotler et al. 94 have proposed the concept of ‘incremental interorganisational relations’, arguing that organisations tend to establish collaborative relationships that require less commitment and have lower risks first, and then move to riskier and more resource-intensive relationships. The organisations included in our study reached consensus in relation to delivering the care that was best for patients, but required additional time and the development of other collaboration mechanisms to engage with the centralisation. We would add that the process of incremental interorganisational collaboration identified by Fotler et al. 94 is also dependent on the constant negotiation of power relations and reinforcement of the status quo. 16,64,95

Another factor that came up in different ways in our study was the presence of hierarchies within the network. Some provider organisations (mainly Specialist Centres) were seen as more powerful than others, as they were able to influence decision-making processes in relation to care delivery (i.e. when patient transfers were made and how). Networks have traditionally been portrayed as devoid of hierarchies, privileging horizontal forms of governance over vertical ones. 77 However, recent work in the management literature has highlighted the prevalence of power imbalances in ‘collaborative governance’ as well. 96,97

A horizontal lens (i.e. looking at relationships across professional groups) allowed us to identify different gradients of collaboration, identifying some professional groups in which collaboration was active and others in which it was still under development. This horizontal focus also entailed looking at the sharing of staff across sites, as the evidence in the literature is mixed in relation to the benefits of having staff members move across organisations. 98–100 In the case of our study, staff members identified joint clinical roles as a mechanism that enabled interorganisational collaboration. However, toward the end of the study, some of these roles had started to disappear.

Strengths and weaknesses

Our study had several limitations. The retrospective nature of some of the interviews meant that they could have been influenced by recall bias, as a significant number of the data analysed for the chapter was collected shortly after the implementation of the centralisation. We made an effort to include the views of a large group of stakeholders and maintain an inclusive sampling strategy, but we might have missed relevant individuals. Our study analysed the development of interorganisational collaboration in a specific health-care area and in an urban setting. Additional work is required to explore collaboration in other specialties and contexts. Our analysis was based on the conceptual framework developed by D’Amour et al. ,83 but other conceptual frameworks might shed light on aspects of collaboration we did not explore.

Implications

We have explored the processes, challenges and strategies used to create and maintain interorganisational collaboration between professionals in a provider network where services were centralised. The provider organisations in the network we studied negotiated power relations to establish shared goals, reached consensus in relation to maintaining patient-centred care, maintained central figures who could create and sustain collaboration, and promoted distributed forms of leadership. These were dynamic processes still under transformation during our study. Future research will need to explore the sustainability of these collaborative relationships and identify the factors that might prompt changes in approaches to collaboration used in networks of provider organisations.

What is already known?

• ‘Attending to history’ is likely to enhance the implementation of MSC and may influence the process in a number of ways.

• Multiple stakeholder voices within change processes could thwart leaders’ ability to transform services in response to historical evidence.

What does this chapter add?

• This chapter shows how ‘attending to history’ can be influential in MSC through an in-depth analysis, demonstrating how history can influence other components of change. Change leaders in Greater Manchester recognised that having a change process within the context of competition, led by any one single group (i.e. commissioners or providers), with poor stakeholder engagement and processes amenable to challenges, contributed to the failure of previous reconfiguration attempts.

• This chapter illustrates the multifaceted nature of ‘history’. The history of failed attempts to reconfigure oesophago-gastric surgery was plain to see, but also evident was more granular detail, for example the history of relationships between individuals. Change leaders responded to all of the various facets of history in their attempt to achieve change.

• This chapter demonstrates the importance of maintaining awareness of how previous change attempts have affected the willingness of local stakeholders to engage in change.

• This chapter provides a description of the impact of the availability of numerous personal accounts of historical change attempts.

Background and setting

The need for reconfiguration of oesophago-gastric specialist cancer surgery services was recognised in Greater Manchester; however, agreement about surgical sites had never been achieved,108 despite attempts to reconfigure services over the preceding decade and a half (Table 7).

There was some reduction in the number of hospitals undertaking oesophago-gastric cancer surgery before formal service reconfiguration, from eight hospitals in the early 2000s to three hospitals at the outset of the reconfiguration attempt studied here (i.e. 2015). The expectation was that this would be reduced to one site, based on clinical guidance. 108

In 2015, another attempt to reconfigure oesophago-gastric cancer surgery services in Greater Manchester commenced. In September 2018, the reconfiguration was completed, consisting of a single surgical centre, a Greater Manchester-wide specialist emergency on-call rota, three ‘sector’ MDTs and a centre MDT.

Sample

Non-participant observations (160 hours) took place at relevant meetings. Documentary evidence (≈ 300 documents) was gathered from online resources, meeting papers and involved stakeholders. Interviewees were purposively sampled, reflecting the range of boards and groups set up to oversee the planning of the new services and the range of professionals involved. Forty-six interviews were conducted.

Participant recruitment and data generation

Recruitment and data generation occurred between September 2015 and December 2018 (see Chapter 2 and Table 3). A semistructured interview schedule was developed to reflect the different stages in planning and implementation of changes (see Appendix 8).

Analysis

A thematic analysis110 of the interview transcripts, observation notes and documentary evidence was undertaken, using a deductive approach guided by the Best et al. 1 ‘simple rules’ framework and the work of Turner et al. 61 As the importance of ‘attending to history’ emerged, an inductive approach was used to explore how interviewees described this.

Awareness of history and availability of historical accounts

There was strong awareness of history among interviewees and this was captured by an individual who, when asked what the main challenge of the reconfiguration process was, stated ‘well history, I guess’ (GM03, manager). In 2013, documentary accounts of past attempts to reconfigure oesophago-gastric services were available, for example reports of enquiries by Monitor (which, at the time, was the sector regulator for health services in England). However, what seemed most important was the proliferation of personal accounts, that is interviewees were keen to talk about change attempts. Differences of opinion were expressed about how knowledge of history could be useful. Some interviewees considered a broad understanding of history more important than knowing the ‘nitty-gritty’:

I understand the general background; when it comes to specific services, less so, I’m not really hugely familiar with the history around OG [oesophago-gastric], I don’t mean to be.

GM31, manager

However, many interviewees who were involved in planning changes spoke about the need to really understand the history of failed reconfiguration attempts:

[It was] important to understand why previous procurement processes had not been successful . . . initial stages were just examining the history, talking to people, understanding exactly what had got in the way of procurement in the past.

GM03, manager

Competition

A history of competition between provider trusts (organisations) in Greater Manchester was perceived to have contributed to the lack of success in reconfiguring oesophago-gastric services. A number of factors were suggested to have engendered the competitive atmosphere. The large number of trusts in a small geographical area was mentioned:

Lots of big trusts, three teaching hospitals, all in a metropolitan area, all not wanting to give anything up.

GM04, surgeon

In 2003, legislation111 enabling the establishment of foundation trusts (i.e. NHS bodies with a degree of autonomy as decision-making powers were devolved from central government) was cited as intensifying competition:

Seeking of foundation trust status put all trusts in active competition.

GM13, surgeon

Although not unique to Greater Manchester, interviewees perceived that Greater Manchester was different from other areas in that historically hospital staff, both managers and clinical, ‘stayed’ in Greater Manchester, rather than moving to other geographical areas. This was thought to contribute to competitive thinking/loyalty to a single organisation:

GM [Greater Manchester] doesn’t seem to have the churn [of personnel] that other places, therefore their memories are long . . . managers who are around for a long time . . . are very wedded to their organisation.

GM24, manager

In response to this competitive environment, those interviewees planning change put firm emphasis on the oesophago-gastric cancer surgery service as Greater Manchester-wide, involving all providers working together. Minutes from the first meeting of the Oesophago-gastric Implementation Board, which was set up to oversee the detailed design and implementation of the new service, stated:

Whilst the new service specification references a sole provider for [oesophago-gastric surgery], there will be a single service and it is important that organisations work together . . . all involved organisations should be proud of what they will achieve together.

Minutes, Oesophago-gastric Implementation Board, 17 March 2017

A manager on the board commented:

This is a GM [Greater Manchester]-wide service . . . and that’s a slightly different way of thinking . . . This is a single service for GM, it’s for the benefit of the entirety of GM; do you want to be part of the solution or part of the problem?

GM20(2), manager

In the following sections we turn to how ‘attending to history’ influenced approaches to other rules in the Best et al. 1 framework.

Designated and distributed leadership

Designated leaders are formally in charge of a programme of work and distributed leaders are people who share responsibility for implementing a programme, and all of these people are change leaders. 1 We focus on designated leadership. Since 2013, specialised services for oesophago-gastric cancer surgery were commissioned by NHS England. In 2016, health and social care funding was devolved to Greater Manchester and the Greater Manchester Health and Social Care Partnership (GMHSCP) was established. A memorandum of understanding between GMHSCP and NHS England included the commissioning of specialised services. Therefore, the GMHSCP had the role of designated leader and funded a Greater Manchester NHS consultancy (i.e. the NHS Transformation Unit) to facilitate the creation of a single service for specialised oesophago-gastric cancer surgery services. Although governance arrangements were enabled by the devolution process, the history of failed attempts to reconfigure services led purely by providers or commissioners influenced the development of a different leadership structure:

. . . using [the Transformation Unit] to almost be an independent, so neither a provider nor a commissioner but with expertise in transformation.

GM10, manager

The Transformation Unit designed an eight-step transformation process. This process involved service providers, patients and the public putting together clinical standards, after which the Transformation Unit worked with providers to develop service access frameworks and explore models of care. From this, the Transformation Unit designed the final model of care and service specification, allowing commissioners to then make decisions about where services should be provided.

The Transformation Unit was keen to emphasise that the process was engaging with provider sites appropriately, as previous change attempts had been halted because of fear that providers had excessive influence on the placement of services. Having engaged with providers through the oesophago-gastric pathway board (i.e. a multiprofessional group overseeing the oesophago-gastric patient pathway from a clinical perspective) about clinical standards for the service, the Transformation Unit wrote to the board to advise that the Transformation Unit would take the clinical standards and use them to develop the service specification. A member of the oesophago-gastric pathway board commented that this occurred so that ‘I could not contaminate the process [of developing the service specification]’ (GM10, manager).

The independent role of the Transformation Unit was viewed positively:

They’ve [acted] in a very independent honest broker kind of way. They administer meetings, support. . . . the implementation group . . . where there’s organisational difficulties they support the board in overcoming those.

GM10, manager

The design of this transformation process was heavily informed by the history of challenges to previous reconfiguration attempts (see Table 7). This history was ‘an undercurrent’ ensuring that ‘we’ve worked on the premise that the decisions will be challenged because they have been in the past’ (GM01, manager). The transformation process was designed by the Transformation Unit to be collaborative, with the goal of avoiding challenges.

Engagement with stakeholders

Although the ‘simple rules’1 refer to engagement with physicians, subsequent work61,112 has indicated that engagement with a wide range of stakeholders is crucial. Change leaders in Greater Manchester understood that the history of attempts to change had affected how stakeholders engaged with the current attempt (an example of history as a barrier to change). One manager commented ‘a process that’s failed so many times . . . there is an issue associated with credibility’ (GM16, manager). By maintaining awareness of these attitudes, change leaders were able to work to encourage engagement. This was achieved partly through acknowledging past experience {‘I said, “We recognise you’ve been marched up the hill, but we’re making a personal commitment [to change]”’ (GM11, manager)} and partly through designing a change process [‘that was so obviously different that it didn’t look like trying to do a failed process”’ (GM16, manager)]. In addition, change leaders talked about working individually with reluctant stakeholders:

Lead clinicians . . . had to be persuaded, conversations took place to persuade people that we need to make change and this time we are actually going to do it.

GM12, manager

Understanding the history of the relationships between stakeholders was important in engaging stakeholders effectively. A manager commented:

One of the reasons that these services have never been sorted is that surgeons are big personalities . . . they have a lot of history, they’ve known each other for a long time, they’ve trained each other.

GM11, manager

These relationships were taken into account, for example a manager spoke about the seating plan at an engagement event:

. . . certain personalities together are not going to come up with a consensus because whatever one says somebody will say the opposite . . . the way [Transformation Unit] stage-managed the tables around who was sat where was helpful.

GM12, manager

Lack of early engagement was also understood to result in active opposition to change. This understanding led to a deliberately inclusive approach, with the belief that if commissioners, service providers and users worked together from the start, then change would happen. A surgeon stated:

. . . there’s been every attempt made to keep this open [with] clinician involvement from the outset, service user involvement from the outset.

GM06, surgeon

A manager representing commissioners described using principles of ‘co-design and getting ownership and buy-in to an overarching process’ (GM16, manager). Another manager explained:

. . . working collaboratively with providers who have an understanding first-hand of the issues of running services day to day. You tap into their knowledge and expertise, that was very important . . . engaging with the clinicians from the start.

GM03, manager

Establish feedback loops

A further check to reduce the possibility of challenge was the establishment of feedback loops, with external feedback used from the outset to assure both clinical decisions (e.g. clinical standards) and the transformation process itself. Change leaders talked about ‘clinical assurance’ and ‘process assurance’ and previous reconfiguration attempts had been challenged on both fronts.

For clinical assurance, an External Clinical Assurance Panel was set up, consisting of clinical experts from outside Greater Manchester and patient representatives. This panel commented on the outputs of every step of the planning process:

[They] played an important role in terms of providing that external assurance . . . very helpful in terms of challenging us on the standards and . . . the service access framework . . . they were very insightful.

GM03, manager

For process assurance, those planning the reconfigurations ensured that relevant regulatory bodies were informed about the work. For example, previous reconfiguration attempts had been referred to Monitor and so change leaders:

. . . involved Monitor early on and invited them to events . . . they got all the paperwork and we maintained regular discussion with them . . . we shared our process as it was developing and asked for any comments . . . to make sure that the potential for challenge was minimised.

GM03, manager

In addition, change leaders talked about keeping senior stakeholders in provider trusts informed so that challenges could not be made on the basis that individuals or institutions did not know what was happening, as had occurred previously:

Regularly met with the chief executives, so that they were well aware of what was going on, no surprises. I used to write a briefing . . . every 4 to 6 weeks just to keep everybody up to speed, again with the aim of minimising challenge.

GM03, manager

The ‘eight-step’ transformation process was designed so that feedback was received at each stage and the decisions that were made were ‘locked down’ so that they could not be revoked:

Kind of a stepwise iteration process so there’s various points along the pathway where the commissioners lock down what’s been agreed already, the idea being that you can’t go back hopefully and question the process, the nuts and bolts, the logistics, the administration of the process, as has happened before.

GM06, surgeon

Principal findings

It has been argued that ‘attending to’ history is important in MSC, but there is relatively little empirical evidence to support this and as far as we are aware there has been little unpacking of what the term means. 1 Drawing on Suddaby and Foster,107 the Best et al. 1 review takes a history-as-fact ‘objective, positivist view of history’107 and, therefore, conceptualises change as difficult, focused at organisational level and resulting in new structures or operations.

Through attending to history, those leading the reconfiguration of oesophago-gastric cancer surgery services in Greater Manchester recognised that having a change process within the context of competition, led by any one single group (i.e. commissioners or providers), with poor stakeholder engagement and processes amenable to challenge, contributed to the failure of previous reconfiguration attempts (see Table 7). The change process analysed here took these issues into account and utilised an independent broker for the process (i.e. the Transformation Unit) that maintained an awareness of how previous change attempts had affected the willingness of local stakeholders to engage in change. Although the Transformation Unit did not explicitly describe their approach as ‘attending to history’, it is clear that it was taken into account in the process described here.

Strengths and weaknesses

Strengths of the study include this being a rare opportunity to study the planning of a major service reconfiguration contemporaneously, enhancing recall of events. A broad range of participants was included (see Chapter 2 and Table 3). In terms of weaknesses of the study, there was no opportunity to study implementation after the very early stages to reflect on how ‘attending to history’ may have affected the developing service.

Comparison with other studies

Some of the issues encountered during the change process, both past failed attempts and the one studied (i.e. oesophago-gastric cancer surgery services in Greater Manchester), were related to professional power, as noted by various authors, including Addicott and Ferlie,113 who also studied cancer services. Relational power between professionals and managers identified by Alford114 was particularly evident in this study.

During the development of the service specification and model for the oesophago-gastric cancer surgery service, professional power was limited to one particular stage of the process until clinical standards had been agreed (see Table 7). This deliberate separation of clinical involvement and the ‘service specification and model’ process could be viewed as an attempt to limit the impact of professional power. The one-to-one conversations between the Transformation Unit and various key stakeholders was also evidence that professional power was recognised, and attempts were being made to mitigate its negative impact on the process.

Taking a history-as-power perspective107 highlights that the focal point of change is not organisational design, but the power structure of the various stakeholder coalitions and the power differences that are ‘solidified’ through history. This study clearly shows how change was ‘characterised by long periods of relative inertia maintained by countervailing political pressures’107 and that change finally occurred when a different approach was taken. The history-as-power perspective enables insight into how a lack of professional agreement was the stumbling block to previous plans. This appeared to be the view of those at the Greater Manchester level attempting to progress this reconfiguration, although not expressed explicitly as such by our interviewees. The aim appeared to be to get an agreement rather than to push a specific model of reconfiguration, with the Transformation Unit facilitating that process as a neutral broker, with no ‘interest’ in pushing a specific model. The focus was more on assuring the process and making it resistant to challenge from any specific stakeholder group (clinical or otherwise).

We presented the leadership issues in this study from a distributed/delegated perspective, but other perspectives view leadership itself as an ‘extension of managerialism’115 and as something that is increasingly a part of public service reform in the UK. This study shows clear attempts, through attending to history, to ‘alleviate tensions by drawing them [stakeholders] together into a unifying discourse of a leading vision for their services in which they, collectively, play a major role’. 115 It is argued that this perspective also supports a form of ‘strong’ leadership that enforces radical change, rather than negotiating compromise or undertaking incremental change. 64 Therefore, attending to history (when history is viewed as fact), and the leadership approaches that are then employed, can be viewed as a form of managerialism. We found no empirical evidence that this was a concern of those involved, but our study did not continue beyond implementation itself when such reflective views might have been identified.

In terms of stakeholder involvement, our findings show that among change leaders, based on history, there was a belief that wider involvement gave more chance of ‘success’ (i.e. of reconfiguration being agreed and implemented rather than being enforced). It could be argued that the use of the Transformation Unit as an external agent to enable the change process to move forward is an example of the manipulation of stakeholder consultation to serve ‘powerful interests’, as described by Fraser et al. 16 Fraser et al. 16 described the way in which management consultants were able to control how problems were understood, and which solutions were adopted, which can be perceived as similar to the way in which the Transformation Unit developed and implemented its eight-step process. However, in this study, we identified positive views from stakeholders about this process that might be explained by the fact that the Transformation Unit had ‘attended to history’ in developing the process, and that this was clearly stated at the outset. In addition, the experience and history of the failed change attempt processes made stakeholders more positively inclined to this different ‘independent broker’ way of working. The Transformation Unit approach can be viewed as history-as-sensemaking, with the efforts made to get people on board with a change process that has failed before, taking into account how history affected the willingness of Greater Manchester stakeholders to engage.

The ‘stage-managing’ of events to keep ‘big personalities’ apart, which was, again, something developed as a result of knowledge of previous change attempts (i.e. history), could be regarded as exercising ‘top-down’ managerial power. The ‘stage-managing’ of events could also be viewed, however, as a pragmatic approach to move toward a decision, facilitated by those independent of (although commissioned and funded by) senior system managers. This history-as-power perspective provides insight by suggesting that it enables the overcoming of ‘the constraints of history through retrospection, critical reflection and creative visioning’107 and was arguably the approach taken and enabled by the Transformation Unit.

The availability of both personal and documentary historical accounts is important when attending to history,1 and it might be that personal accounts lead to a need to view history-as-power, whereas documentary accounts are less contested and are viewed from a history-as-fact perspective. In this study, there was a wide awareness of the history of change attempts in oesophago-gastric cancer surgery services in terms of both power and facts. It may be that this was particularly acute because of the plethora of personal historical accounts that were evident (i.e. almost everybody could ‘tell a story’ about previous change attempts and whether or not they had been involved themselves, although these accounts always represented their personal interpretation of events). This ease of access to information perhaps contributed to an atmosphere in which it was difficult not to attend to history. It may be that in situations in which historical accounts are mainly documentary and not a topic of discussion that there is less likelihood that history will be so widely known or to such a level of detail and, therefore, acted on.

This study highlights how history intersects with the other rules in the Best et al. 1 framework. Through ‘attending to history’, using a range of perspectives, change leaders considered other rules, in this case aspects of leadership, stakeholder engagement and feedback loops, recognising history as both power and sense-making, in addition to a history-as-fact approach. Through historically informed leadership arrangements and stakeholder engagement, an approach more nuanced to the context within which it was being undertaken was developed, and one that attempted to take into account the power of the various groups involved (i.e. recognising subtleties and interstakeholder dynamics). The separation of clinical and service issues into different phases of the process minimised the influence of professional (clinical) power, which was viewed as having been a barrier to change in the past.

Implications

Although there were changes in the context of the reconfiguration studied here, in terms of the devolution of funding to Greater Manchester, which arguably encouraged the development of pan-Greater Manchester services, our data indicate that these changes alone would not have been enough to ensure successful reconfiguration, given the competitive nature of the system and the history of previous change attempts. It was the combination of the context becoming arguably more supportive of reconfiguration, along with the recognition of, and response to, history, which led to the oesophago-gastric surgery changes being implemented. This chapter provides further learning about MSC and how learning from history can be exploited to enable successful change.

What is already known?

• Centralisation and other forms of MSC can be stressful for the staff involved.

• We are aware of a limited focus in the literature on the emotional aspects of MSC, but research has shown that these reorganisations can cause stress caused by insecurity, increased workload, fatigue, bullying and feelings of loss or grief.

• Experiences of organisational change are known to provoke strong emotions when a part of the work environment is removed or ceases to exist.

What does this chapter add?

• MSC involved processes (e.g. bidding for specialised status) that provoked feelings of loss and personal failure.

• The movement of financial and workforce resources to specialist sites can destabilise the ‘ecosystems’ in local teams and create issues with maintaining and recruiting skilled staff.

• Changes can cause loss of motivation and reward in daily work for staff at sites that have lost activity.

Study design

This is a qualitative study that draws on interviews with key stakeholders in the centralisation process, non-participant observations and document reviews. The study was primarily focused on, but not limited to, interviews drawn from participants in sites that lost surgical activity and also participants in the central leadership team.

Immediate subtractive change: loss of activity, skill and continuity

Long-term subtractive change: loss of staff, trainees and autonomy

Staff at local sites perceived that their trusts suffered from losses after centralisation that resulted from the devaluing of their organisational status and activities. This made continuation of routine clinical activity challenging and there were concerns about decreasing standards of patient care.

Emotional repercussions of subtractive change: loss of self-image, status, autonomy and motivation

Support and coping strategies offered

We identified various different strategies offered during the implementation phase, usually by the central leadership team and other managerial roles. Anticipation or expression of loss from staff at local sites was often countered by an offer of instrumental support, for example offering joint contracts, collaborative interventions or educational opportunities. Emotional support was mentioned far less often, particularly in relation to short-term loss experiences, and normally in the context of supporting other members of the implementation team.

Principal findings

• In the short term, staff experienced loss of activity, skill use and interaction with familiar team members.

• Over time, staff perceived that loss resulted in shrinking, de-skilling and destabilising local sites and teams.

• Both individual staff members and their host organisations felt devalued, and people experienced loss of status and motivation.

• Leaders mitigated losses through joint contracting, surgical skill development opportunities and trainee rotation.

• Relatively little emotional support was offered, and emotional reactions to the centralisation were often characterised as resistance to be overcome through persuasion and appeals to the success of the new system.

Strengths and weaknesses

It is important to see these findings in context. The majority of interviewees (including those in local units) felt that the reorganisation was positive and that conducting a higher volume of specialist surgeries at a designated centre was the correct thing to do for patient benefit. Interviews were retrospective and subject to recall bias. To reduce this risk, we used documentary evidence to complement interviewees’ narration of past events. Attitudes towards loss may also change further over time, and our data are limited to a 3-year window after the implementation of the change. Despite our inclusive sampling strategy, key individuals who were affected by loss may not have been represented.

Comparison with other studies

This analysis builds on previous studies of major system and organisational change by reinforcing the conceptualisation of change as a stressor117,121 and, particularly, a loss. 129,130 Our findings draw parallels with the emotional and politics aspects of other organisational health-care loss, such as decommissioning. 134,135

We concur with studies91 that attribute responsibility to leaders of MSC to engage in emotional work and to mediate resistance to MSC with efforts to create shared individual values. Leaders who promise a successful new system could be onto a good strategy. Hakak136 suggests that individuals are helped by perceiving the new system as prestigious. However, our study suggests that this is potentially insufficient when coping with MSC processes such as abrupt changes in team composition and staffing shortages. 125 Participants agreed with the values associated with centralisation, but expressed resistance and disengagement from leadership through the unmitigated stress of losses sustained interpersonally, as well as individually.

We also echo concerns articulated by Fraser et al. 16 about leaders’ use of clinical arguments or evidence of ‘success’ in persuading stakeholders to drive home MSC. These approaches are at odds with the ‘collective’ leadership styles advocated by stakeholders in health-care leadership, such as the NHS Leadership Academy (Leeds, UK), giving power to clinicians and patients. 137 However, these types of models ignore the impact of lost power and the importance of being able to provide relief and coping strategies for MSC through tangible organisational processes, such as effective collaborative trainee schemes. These types of models also highlight that leaders who prioritise system processes (e.g. surgical centralisation) over indirect aspects of centralisation (e.g. joint contracts and trainee rotation) may undermine successful implementation.

Implications

• Leaders of MSC should prioritise planned indirect changes, such as training and skills development.

• Centralised service design should consider social interactions and team dynamics, support staff after losing bids and address key workforce issues in local sites.

• Social or personal interventions would benefit staff at local sites, as well as encouraging coalitions to ask for more tangible support measures. 129

Conclusion

Stress incurred by system change cannot be prevented; however, leaders of MSC should consider changes from the perspective of individual staff members and what their role will be in relation to the organisation before, during and after large-scale reorganisations of this kind. Finally, in the new era of emergent, emotionally attuned and relational leadership, leaders need to reconsider the narrative of ‘overcoming resistance’, considering how this may be supported by providing adequate resources to mitigate stress and loss. 138

What is already known?

• The implementation of major reorganisations of clinical services can carry substantial cost, partly as financial expenditure and partly as staff time, spent within working hours and as discretionary out-of-hours effort.

• Economic evaluations of MSC tend to fall outside health technology assessment and within service evaluation, where the evaluation of costs and consequences is often neglected.

What does this chapter add?

• The London Cancer changes cost £7.2M to plan, design and implement (adjusted 2017–18 prices).

• The highest costs were for robots, which might not apply in other reconfigurations. The total adjusted costs were £3.2M when robot costs were excluded.

• Cross-disciplinary collaborations involving health economists, qualitative researchers, clinicians, managers, patients and the public facilitated this work and are recommended for future similar work.

• The framework we used to design the data collection can support different stakeholders, including service planners, researchers and policy-makers, in collecting and analysing implementation costs, as these costs are often considered too complex to measure or are excluded as sunk costs.

People’s time

Using meeting minutes alone to calculate staff time was insufficient, as staff also spent substantial time outside meetings. Therefore, we created a list of 19 key actors based on the qualitative work and additional conversations with key central figures, and included a weighted portion of their salary and on-costs on top of the time spent by other staff in board meetings. These key actors spent an estimated 12.5% (11 people), 25% (five people), 40% (one person) or 50% (two people) of their time on the four specified cancer reconfigurations. Time spent in board meetings by these specific people was excluded to avoid double counting.

On top of the above, other informal discussions and planning tasks took place among other staff. This would have constituted substantial time for clinicians and managers but was not possible to quantify, and so represented missing information.

Costs to sites

We initially considered obtaining financial documentation from all 14 sites in the region to confirm expenditure, but conversations with key management and clinical staff suggested no direct external expenditure during the time frame, except at the new prostate/bladder and renal specialist centres. There was some suggestion that, as a result of the reconfiguration, one new oesophago-gastric centre purchased new equipment for surgeons who were now operating at that site, but specific estimates were unavailable.

Some specialist renal, bladder and prostate surgeries are increasingly carried out using robotic techniques, and replacement robots were purchased by two sites during the timeline of the reconfiguration study.

Component A: options appraisal, bidding and external review processes

As part of the appraisal process, a number of analyses were performed at NELCSU and by external consultants. These analyses included a complex business case (which included capital works across different hospital sites), programme management support, and competition (i.e. market) and transport analyses. The total cost was estimated at £1,850,000, encompassing the whole London Cancer programme (i.e. covering eight cancers, not just those we studied) and cardiac changes (see Chapter 1).

Regarding bidding preparations at prospective specialist sites, we obtained no information beyond some key actors writing bids as part of their role. These costs are, therefore, included within key actors.

Component B: stakeholder engagement

There were two engagement phases, October–December 2013 and May–June 2014. These phases were led by NHS England and CCGs, and included workshops attended by clinicians and the public, and planning and engagement meetings with NHS staff. Provider staff time at workshops totalled 475 person-hours (£23,248) in the first phase (i.e. urology only) and 220 person-hours (£10,768) in the second phase (i.e. all eight cancers). A further 520 person-hours (£25,451) was spent in ongoing meetings covering the joint cancer and cardiac changes, and £23,713 was spent by the London Cancer Board on room hire, catering, etc., for events (all eight cancers).

There were no minutes available for engagement events and, therefore, information came from memories and calendar invitations from current and former Transforming Cancer Services Team and NELCSU staff and mentions in various documentation, including archived news items. London Cancer Board direct expenditure figures came from a report discussed at a 2014 London Cancer Board meeting. We could not exclude key actors’ time here as event attendee lists were not available. No clinic sessions were cancelled for these events and almost all occurred outside working hours, particularly when patients and the public or their representatives were invited. No distinction was made in this analysis between staff time spent during working hours and during leisure time.

Component C: planning, monitoring and board meetings

Staff excluding key actors spent an estimated 1459 person-hours (£71,309) on board meetings (four cancers of interest), including during the options appraisal period (see component A in Figure 9), the engagement period (see component B in Figure 9), planning and monitoring (see component C in Figure 9) and for auditing and monitoring performance (see component E in Figure 9). Expenditure on internal change support for planning and monitoring totalled £100,000 (eight cancers and cardiac).

Component D: making the change

Owing to increased patient volumes at the specialist sites, there were some new hires and some sharing or movement of surgeons between sites, but these changes were not included in the analysis. Only one-off costs of new roles created specifically for doing the design, planning and implementation were included in the analysis.

Component E: audit and performance monitoring

Time spent on implementing audit and monitoring systems in board meetings was included under component C (see Figure 9). The time included covered the four cancers of interest only.

Total implementation cost

The cost of implementation was £6.9M by expenditure year or £7.2M in 2017/18 costs (Table 9). The cost of implementation included some costs that could potentially be attributable to the cardiac reconfiguration or to the other four cancer pathways, although it was not clear how much could be apportioned to these other areas, as certain joint events and activities would have happened regardless. Sensitivity analysis removing half of these shared costs for the activities in which there was overlap with cardiac reconfigurations reduced the total to £6.2M in 2017/18 costs, mostly as a result of halving the component A (see Figure 9) costs. In other reconfigurations, it is possible that such a large equipment cost might not be required, and excluding the robot costs gave an adjusted total cost of £3.2M.

Principal findings

We conducted a bottom-up costing exercise of the total cost of implementing the London Cancer reconfigurations with a total adjusted cost of £7.2M over 4 years.

Strengths and weaknesses

The costing exercise was conducted contemporaneously alongside the reconfigurations, which aided in certain aspects of the data collection. However, there remained a number of challenges in collecting the required information for the relevant cost components. Some of the costs collected relied on indirect sources and some estimates of time spent were made from memory by NHS and other sources. Deciphering which costs could be attributed specifically to the London Cancer configurations and which were related to other activities going on at the same time was also challenging. Over half the cost was due to the purchase of equipment for robotic surgery, which may or may not be required for other regions that may want to undergo a similar reconfiguration. The inclusion of costs for robotic surgery also implicitly raises questions regarding the relative cost-effectiveness of robotic and traditional surgery, and, to the best of our knowledge, there is no current evidence on this. In both renal146 and prostate cancers,147 there is some evidence that better clinical outcomes can be obtained with robotic surgery, and reductions in LOS in renal, prostate and bladder cancers have been detected in the analysis presented in Chapter 9. Use of the robot to avoid open surgery with prolonged LOS could be a contributing factor to this, and the purchase of the replacement robot was partly justified by the high case volume created through the service reconfiguration.

Comparison with other studies

One of the aims of this analysis was to incorporate the cost of implementing the reconfigurations into the economic evaluation reported in Chapter 10. Economic evaluations of reconfigurations are often neglected, falling outside the standard health technology framework used to guide the methodology for economic evaluation. 148 Systematic reviews of economic evaluations of implementation strategies have found a limited number of economic evaluations, which are generally rated as being of poor quality, and only one study was related to reconfigurations. 149,150 The cost of reconfiguring stroke services in London has been estimated at £9M, covering capital, equipment and premises refurbishment,151 although the methods used to obtain this estimate were not as extensive as what has been used here. Additional national government funding was also allocated to London to support it with the reconfiguration of stroke care, but this was not the case with the cancer reconfigurations.

Implications

Regardless of the inclusion or exclusion of the cost of robots, the cost of implementation is not insignificant, and its analysis requires planning and significant additional work to collect the required data. The implications of including the costs of implementing the reconfiguration alongside the direct costs of treatment and the effectiveness of the reconfigurations are explored as part of the full economic evaluation reported in Chapter 10.

What is already known?

• Some studies have shown that centralising surgeries for some cancers can result in lower mortality rates and shorter LOS, but may have negative impacts on access to specific treatment modalities.

What does this chapter add?

• Centralisation of specialist cancer surgery in London Cancer was associated with fewer surgeons doing more operations.

• Centralisation of specialist cancer surgery in London Cancer was associated with a significant decrease in LOS. In the case of renal cancer, we found evidence that patients were more likely to receive less invasive treatment, suggesting a broadening of the range of treatment modalities offered.

• We found no evidence of impact on mortality or re-admissions, although this may be because the underlying risk of these outcomes was already low.

Data

We obtained patient-level data from the NCRAS database for all patients diagnosed with one of the four cancers between January 2012 and December 2017. The four cancers were defined using International Classification of Diseases, Tenth Revision, codes (see Appendix 5, Table 21). The NCRAS database (requested via Public Health England’s Office for Data) was linked at the patient level to the HES database of all hospital admissions in the English NHS, which contains information about the type of operation patients underwent and the date of the procedure. Our analysis includes people who had surgery for their cancer, which was defined using Office of Population Censuses and Surveys (OPCS) codes (see Appendix 5, Table 22). We omitted patients treated at private hospitals, treated in hospitals that were not in England or who had their surgery in Greater Manchester (we did not exclude private patients who were treated in NHS hospitals). The HES data were linked to mortality data supplied by the ONS using an anonymised unique patient identifier.

We obtained data from British Association of Urological Surgeons Radical Prostatectomy, Nephrectomy and Cystectomy audits for patients who had these surgeries between January 2014 and December 2017. However, we identified that not all hospitals participated in these audits and, in particular, for prostatectomies and cystectomies, these data were incomplete for the period prior to the reconfiguration for London Cancer and the rest of England. Therefore, we were unable to analyse these data sets (see Appendix 4, Table 20). After continued attempts, we were unable to obtain any data from the National Oesophago-Gastric Cancer Audit and so could not use this database for our analysis (see Appendix 4, Table 20). We obtained data from the NCPES for patients who had cancer-related treatment by trust between April 2016 and June 2018. It was not possible to distinguish patients who had surgery from patients undergoing other types of management for cancer. Data were available at the aggregate level for all urological cancers combined and for all stomach-related cancers combined, but not for individual cancers. We applied to access cancer-specific data under a special license, but these data were not available for the period prior to the reconfiguration for London Cancer and the rest of England. Therefore, we were unable to analyse this data set (see Appendix 4, Table 20).

Outcomes

Clinical members of the research team identified the primary and secondary outcome measures for each type of cancer. These outcome measures are listed in Table 10, along with the data sources for each measure. The LOS measures were based on the number of nights in hospital for the index surgery. The re-admission measures were based on the further hospitalisations within a given time period when the primary diagnosis for the re-admission was the specific cancer. We calculated the number of procedures per surgeon per month by first calculating the number of surgeons doing at least one operation on patients with each type of cancer in London Cancer and the rest of England in every month before, during and after centralisation. We then divided the total number of surgeries performed during each period by this number. Waiting time from diagnosis to treatment was defined as the start date of the first cancer treatment minus the date when the cancer treatment period was defined to start, and, according to national guidelines,153 this should occur within 31 days. Waiting time from referral to treatment was defined as the start date of the first cancer treatment minus the date when a decision was made to refer the patient to secondary care with suspected cancer, and, according to national guidelines,153 this should occur within 62 days. Mortality was based on deaths from any cause and at any place (i.e. hospital or otherwise) at a predetermined time point after surgery. For renal cancer, the probability that patients with stage 1 cancer had partial nephrectomy was defined as the probability of having partial nephrectomy compared with any type of nephrectomy, using the categorisation of OPCS codes in Appendix 5, Table 23. The probability of non-invasive treatment was defined as the probability of non-invasive treatments compared with active surveillance, nephrectomy, partial nephrectomy or non-invasive treatment, using the OPCS codes in Appendix 5, Table 23. For oesophago-gastric cancer, these two measures were similarly defined (see Appendix 5, Table 24).

As explained above, several of the data sets we planned to use were either not available to use or the data they provided were not sufficient for us to be able to investigate between-region difference-in-differences. The measures and data that we could not analyse are detailed in the final version of the study protocol154 and are listed in Appendix 4, Table 20.

Explanatory variables

As noted above, we obtained patient-level data from the NCRAS database for all patients diagnosed with one of the four cancers between January 2012 and December 2017. To estimate between-region difference in differences required us to assign each patient who had surgery to a time period before, during or after the centralisation in London Cancer. ‘Before’ referred to the period before the reconfiguration was agreed and the service infrastructure changes were completed, ‘during’ referred to the transition period when the reconfiguration was officially signed off and first patients started to be transferred to the centralised centres, and ‘after’ referred to the period when all patients were expected to receive care under the new system. The timeline was different for each cancer type (Table 11).

To identity between-region differences, we identified patients who had surgery at one of the eight NHS trusts within the London Cancer Network (i.e. Barts Health NHS Trust; Barking, Havering and Redbridge University Hospitals NHS Trust; Homerton University Hospital NHS Trust; Royal Free London NHS Trust; North Middlesex University Hospital NHS Trust; The Princess Alexandra Hospital NHS Trust; University College London Hospitals NHS Trust; and Whittington Health NHS Trust). In the rest of England comparator, we included patients treated at other trusts within London. We did not include patients treated within Greater Manchester given the planned centralisations, nor did we include patients treated outside England.

We adjusted for differences in case mix in the regressions, controlling for age (as a linear term), sex, ethnicity (i.e. white, not white), Index of Multiple Deprivation quintile of place of residence, cancer tumour stage at diagnosis (T1, T2, T3, T4, TX, not known), combined Gleason grade for prostate cancer (low: less than 7; moderate: 7; high: more than 7; missing), tumour grade for the other three cancers [G1 (well differentiated), G2 (moderately differentiated), G3 (poorly differentiated), G4 (undifferentiated) or GX (grade of differentiation is not appropriate or cannot be assessed)], Charlson Comorbidity Index and number of diagnosed cancers (linear term).

Statistical methods

We evaluated whether or not the centralisations in London Cancer had an impact on the outcomes using regression analysis of difference in differences between regions to compare the changes over time in London Cancer with the change over time in the rest of England. The analysis was carried out at the patient level. The regression equation was:

where y is the outcome, i indicates an individual patient, j indicates location (i.e. London Cancer, the rest of England), t indicates whether the surgery took place before, during or after centralisation, α is a constant term, u are location fixed effects (i.e. London Cancer, the rest of England) and v are time fixed effects (i.e. before, during, after centralisation). D¹ is a variable taking the value 1 if the provider trust in which the patient had surgery was in London Cancer and 0 otherwise, D² is a variable that equals 1 if the observation belongs to the time period after the reconfiguration and 0 otherwise, and D³ equals 1 if the observation belongs to the time period during the reconfiguration and 0 otherwise. x are patient-level variables that might be associated with the outcomes (e.g. age, sex, ethnicity, deprivation, stage of cancer, grade of cancer, comorbidity and number of diagnosed cancers). We controlled for these variables as they might affect case selection for treatment at specialist centres and also be correlated with the outcomes of interest. We are interested in the sign and statistical significance of the coefficient δ₁, which quantifies the changes in case mix-adjusted outcomes over time in London Cancer, controlling for the changes over time in the rest of England. Most of the outcome measures were binary (yes/no) at the patient level, and we used logistic regressions to evaluate difference in differences. We calculated marginal effects (i.e. difference in differences in probabilities), making pairwise comparisons of the predicted outcomes estimated at the mean values of the case mix-adjusting covariates (x) for the before period for London Cancer and the rest of England combined. We used a similar approach for LOS, but used parametric survival models assuming a log-normal survival distribution to account for the time-to-event (i.e. from admission to discharge) nature of the variable. We ran models with and without the case mix-adjusting covariates (x). For the number of procedures per surgeon per month and waiting times variables we did not include the case mix-adjusting covariates (x), as these do not depend on patient-level factors.

We undertook pre-trends tests to examine whether or not the case mix-adjusted primary outcomes had a different linear trend in London Cancer compared with the rest of England before the centralisations. We reran the models on all patients who had surgery before the centralisation and included a linear time trend for month. We added an interaction term between London Cancer and the linear time trend, and tested the individual significance of the interaction term. In every case, the significance was non-significant (p > 0.05), except for LOS > 3 days in the prostate cancer analysis.

Principal findings

Although there were differences between the four cancers studied, our between-region difference-in-differences analysis suggested that centralisation of specialist cancer surgery in London Cancer was associated with a significant increase in the number of operations per surgeon per month, with fewer surgeons doing more operations. It was also associated with a significant decrease in length of hospital stay. In the case of renal cancer, we found evidence that patients were more likely to receive less invasive treatment, suggesting a broadening of the range of treatment modalities offered. We found no evidence of impact on mortality or re-admissions, although this may be because the underlying risk of these outcomes was already extremely low. For some cancers, waiting times worsened.

Strengths and weaknesses

The main strengths of our study are that we used a large national data set, containing detailed information on outcomes and patient characteristics, and the robust quasi-experimental framework, and these allowed us to control for trends in the rest of England and other factors that could affect outcomes during the same period. There are, however, several weaknesses. The main weakness was the lack of available data, which meant that we were unable to investigate the full range of outcomes that we originally intended. In particular, we were unable to analyse the effects of the centralisations on surgery-related complications and other measures or on morbidity. It might be expected that such complications would decline with greater concentration of procedures among fewer surgeons, although there is a potential risk of under-reporting of complications in surgeon-reported audit data. 155 We were also unable to investigate the impact on patient experience because of issues around data access and likely small numbers of respondents with the specific cancers studied. Second, our analysis relies on accurate coding of surgical procedures within trusts. For example, the analysis of the type of surgery used among patients with renal cancer requires that hospital coding staff assign the correct procedure codes to individual cases. Third, the analysis may have been underpowered to detect statistical differences associated with MSC, especially in terms of the mortality measures, in which the probability of mortality was very low. Fourth, we also acknowledge that our findings may not be generalisable to other parts of the UK. Services in London may be different from those in other parts of the country, for example in terms of travel times and distances, number of centres and the range of facilities and specialist expertise available in them.

Comparisons with other studies

Some studies have shown that centralising surgeries for some cancers can result in lower mortality rates and shorter LOS in hospital. 22,26 We found no evidence of impacts on mortality, although, as explained, the underlying mortality rates were already very low, with potentially little scope for improvement. Some studies have shown that centralisation may have a negative impact on access to different treatment modalities. 152 We found that, in the case of renal cancer, centralisation appears to have increased the range of treatment modalities offered; however, for oesophago-gastric cancer, there appears to have been no change.

Implications

Further work would be beneficial to understand the impact of the centralisations on complications and, importantly, on patient experience, as we were unable to evaluate these impacts in the present study.

Our findings imply that centralisation was associated consistently with reduced LOS, which suggests that further research to investigate the value for money of centralising specialist cancer surgery services is warranted.

What is already known?

• Chapter 9 reports improvements in clinical outcomes for the three urological cancers regarding reduced LOS. The results for oesophago-gastric cancer were not suggestive of clear improvements.

• The overall cost of implementing the London Cancer changes was £7.2M across the four cancers (see Chapter 8).

What does this chapter add?

• There was a medium to high probability that the London Cancer region changes led to more cost-effective treatment provision in prostate (79%) cancer specialist surgery, and a medium probability of the same for oesophago-gastric (62%) and bladder (49%) cancer specialist surgery, compared with the rest of England, excluding Greater Manchester, at a standard cost-effectiveness threshold of £30,000 per QALY gained.

• There was a low probability (12%) of the London Cancer reconfigurations being cost-effective for the renal reconfigurations at the same cost-effectiveness threshold.

• These analyses incorporated information on estimated QALYs and treatment pathway costs from routine data, plus the implementation cost in the ‘London Cancer after’ group, split across the four cost-effectiveness analyses, according to relative cancer incidence.

Overview

The cost-effectiveness analysis consists of four decision-analytic models, one per cancer, calculating costs (from a NHS payer perspective) and outcomes, summarised as the NMB for the reconfigured services in the London Cancer area compared with services delivered elsewhere in the rest of England, excluding the Greater Manchester Cancer region. Patients were classified into ‘before’, ‘during’ or ‘after’ groups based on their surgery date (see Chapter 9 and Table 11). Patients whose surgery date fell in the ‘during’ period were excluded in the base-case cost-effectiveness analysis and, therefore, four scenarios were considered per cancer:

1. patients receiving surgery before the reconfiguration in London Cancer area

2. patients receiving surgery after the reconfiguration in London Cancer area

3. patients receiving surgery before the reconfiguration in the rest of England

4. patients receiving surgery after the reconfiguration in the rest of England.

Each model calculated costs and QALYs over a 30-day decision tree (a 90-day decision tree for prostate cancer), plus a further 10-year Markov model, for 1000 hypothetical patients with the same initial disease status and demographics as those in the linked patient-level data (see Chapter 9).

The analyses were performed using Stata version 16 and Microsoft Excel.

Data set

The analysis used the same linked patient-level data, cohorts and timelines as the analysis described in Chapter 9, namely patient-level data from NCRAS, linked at patient level to HES and ONS, from Public Health England’s Office for Data Release and including Cancer Registry data; HES inpatient, outpatient, and accident and emergency (A&E) data; and ONS mortality data.

Model structure

The design was similar to that used in previous work on stroke. 151 The design involved modelling patient pathways using a short-term decision tree followed by a longer-term state transition Markov model. The structure is illustrated in Figure 10.

FIGURE 10.

Illustrations of the model structures for each of the four cancers. The length of decision tree was 90 days for prostate, and 30 days for bladder, renal and oesophago-gastric.

The prostate model was a 90-day decision tree followed by a 10-year Markov model with 6-monthly cycles. The renal, bladder and oesophago-gastric models were 30-day decision trees, followed by 10-year Markov models with 6-monthly cycles. Patients entered the decision tree on the date of their specialist surgery for that cancer, and were subsequently classified at the end of the decision tree as falling into one of three health states: (1) ‘healthy’, (2) ‘not healthy’ or (3) ‘dead’.

Patients were classified as ‘healthy’ in the prostate cancer model if their LOS for the index surgery was < 3 days and they were not re-admitted within 90 days. Prostate cancer patients who either had a LOS of > 3 days at their index surgery or who were re-admitted for prostate cancer within 90 days, or both, were classified as ‘not healthy’ at 90 days. Patients who had died by 90 days after surgery were classified as ‘dead’. Patients who were alive (i.e. ‘healthy’ or ‘not healthy’) at the end of the 90-day decision tree moved into a two-state Markov, with the two states being alive or dead (see Figure 10).

In the bladder, renal and oesophago-gastric cancers, patients were ‘healthy’ at the end of the 30-day decision tree if they were not re-admitted to hospital for that cancer within 30 days and were ‘not healthy’ if they were re-admitted within 30 days. Patients who had died by 30 days after surgery were classified as ‘dead’ at the end of the decision tree. Patients who remained alive at the end of the decision tree moved into the two-state Markov model, as above, for prostate.

Costs

The mean hospital treatment pathway cost per patient was calculated using HES data from the linked patient-level data sets. Unit costs were obtained from NHS reference costs 2010/11–2017/18156 and were applied to hospital events, including inpatient, outpatient and A&E attendances. Unit costs for inpatient stays were converted to an average cost per bed-day to capture the cost impact of reductions in LOS. The published reference costs for 2018/19 were not used, as the format in which these costs were reported changed between 2017/18 and 2018/19 so that bed-day costs could no longer easily be calculated for 2018/19 reference costs.

Inpatient unit costs were categorised by Secondary Uses Service Healthcare Resource Group (corresponding to currency code in NHS reference costs156) and class (ordinary admission, day case admission, regular day or night attender), with the cost per bed-day generated by dividing the full consultant episode cost by reported average LOS. Outpatient unit costs were categorised by treatment specialty (corresponding to service code in NHS reference costs156), using published average costs that had been weighted according to national proportions of consultant- and non-consultant-led attendances. A&E unit costs were categorised by A&E department type. The latest available NHS reference cost156 was applied based on the categories above and adjusted to the 2018/19 financial year using the new Health Services Index using Consumer Price Index (Health), and the previous Hospital and Community Health Service indices for part-adjustment of older prices. 143,144

Per-patient alive treatment pathway costs for the Markov model cycles were calculated for a 6-month period by summing outpatient decision tree costs, then dividing by the number of days in the decision tree and multiplying by 183 days. A 6-month cycle length was chosen as this was considered short enough to capture changes in patients’ costs and utilities, and long enough to not extend the computation time for the probabilistic sensitivity analyses beyond what was feasible. In accordance with recommendations in the relevant National Institute for Health and Care Excellence (NICE) guidelines for each cancer157–161 only outpatient costs were included in the alive Markov cycle costs, as patients would not continue to have the same high level of resource use over the following 10 years as they had in the decision tree. This approximation aimed for a middle path, as all patients were grouped together in the ‘alive’ group in these analyses. Patients could have higher follow-up costs that were curtailed because of death, or lower follow-up costs that were stopped when they were discharged. We compared these cost estimates to values found in the published literature157–161 for similar patient groups and the cycle costs used in our model were similar to these. For example, in BOXIT (Bladder COX-2 Inhibition Trial), Cox et al. 158 provided estimates with a mean of £1385 (2017 prices) over 6 months when considering follow-up treatment over 3 years following surgery.

The one-off costs for death in the 10-year model were calculated using literature values reported by Round et al. ,162 adjusted to 2018–19 prices, as described above, using reported prostate cancer values for the three urological cancers and reported colorectal values for oesophago-gastric cancer.

For each state (i.e. alive/dead), these costs were calculated for each of the six scenarios (i.e. before/during/after and London Cancer/the rest of England) or four scenarios in oesophago-gastric cancer. The unweighted means of the six or four scenarios were applied for each cancer, as it was not expected that differences in state costs between scenarios would persist over the longer time frame, and any differences in costs between the scenarios were likely to be skewed by small patient and event numbers.

Quality-adjusted life-years and utilities

The outcomes used in the cost-effectiveness analysis were QALYs. Patient-level health-related quality-of-life data were not available for this cohort, as they are not collected routinely. Instead, assumptions were made regarding patients’ health states based on treatment events observed in the linked data set, and utility scores from published studies in these cancers were applied on this basis. Searches were performed on 26 August 2020 for utility scores, focusing on NICE technology appraisal documentation163 and the Tufts University database,164 and then by snowballing from those starting points to find other relevant work.

In the literature, we found no reported utility scores that corresponded exactly to the scenarios defined here (i.e. based on the main statistical primary outcomes, as described above) and, therefore, ‘healthy’ patients were approximated to ‘pre-progression’ patients in published studies,160,165–177 and ‘not healthy’ patients to ‘post-progression’ patients. Published values had been obtained in a variety of ways, including utility scores calculated from patient-completed questionnaires [i.e. EuroQol-5 Dimensions (EQ-5D),178 European Organisation for Research and Treatment of Cancer Core Quality of Life Questionnaire (QLQ-C30)179 and Short Form questionnaire-12 items],180 as well as estimates of utilities by clinical oncology experts when patient-reported outcome measures information was unavailable. When there was a choice, we used values calculated from EuroQol-5 Dimensions, three-level version (EQ-5D-3L) responses, as NICE recommends that QALYs are calculated using utility scores generated by the EQ-5D163 and the EQ-5D-3L was most common, and attempted to include patient populations that had been offered standard of care in trials. The means of the utility scores used in the model are reported in Appendix 6, Table 52. Standard errors reported in the literature varied. The uncertainty in utility scores informing this analysis ranged from not reported (i.e. only the mean reported) in oesophago-gastric cancer to utilities with ranges of 0.5–1.0 in bladder cancer and, therefore, a median standard error of 0.1 was used for all utilities. The utility scores of the amalgamated ‘alive’ patients in the Markov models were the weighted means of these two utility scores for each cancer, according to the overall relative proportions of healthy/not healthy patients at the end of the decision tree. The dead health state in all cases carried a utility score of zero, and we assumed that patients who died during the decision tree time horizon had zero utility for the whole 30- or 90-day period.

Statistical analysis

Principal findings

In 2016, compared with the rest of England scenario, and assuming a population of 2077 patients, the reconfigurations of prostate cancer surgery services in London Cancer potentially led to additional costs of approximately £1.1M over 10 years, with a potential increase of 173 QALYs (see Figures 11 and 12). The improvements in quality of life in patients with prostate cancer, compared with the rest of England, were driven by comparatively shorter LOS and fewer re-admissions after surgery, as detailed in Chapter 9. The relative increase in costs was driven by the implementation costs and possible small relative increases in outpatient costs in London Cancer, compared with the rest of England, after surgery. The evidence suggested that for oesophago-gastric reconfigurations, the differences in costs and QALYS were small and that there was an approximately 60% chance that the reconfigurations were cost-effective at any threshold above around £20,000 per QALY gained (i.e. the CEAC in Figure 18 levelled off at around 60%). For bladder reconfigurations, the CEAC remained at around 50% for all thresholds (see Figures 13 and 14). In addition, there was a very low probability of the renal changes having been cost-effective, driven by the larger reduction in QALYs (see Figures 15 and 16).

It is possible that the London Cancer reconfigurations would not have occurred in this way if the group of cancers had not been reconfigured together at the same time. The delivery of health-care services in the NHS is a highly networked and collaborative activity, as we have seen, particularly, during the COVID-19 pandemic, and as discussed in Chapter 4 and related published work from the RESPECT-21 study team. 60 With the reconfigurations taken together, they were calculated to have resulted in a 10-year additional cost of £2.3M to London Cancer compared with the rest of England, with 102 more QALYs (see Table 14) in a cohort of 3413 patients.

Strengths and weaknesses

We have attempted to control for other contemporaneous changes that might have taken place by using the rest of England in the difference-in-differences analysis in the absence of a randomised design. Future work could consider using the ‘rest of London’ as the control group, if sufficient patient numbers were involved. Randomised controlled trials are the gold-standard to minimise bias; however, randomised controlled trials are rarely possible in service change evaluations and, therefore, we have used the most appropriate non-randomised method and adjusted for available confounding variables. 184 In terms of the regression and survival models, we adjusted for available baseline patient and disease characteristics recommended by the clinical authors, but it is likely that there are further confounding variables and other factors that are important to patients and health-care professionals that we could not include. We also note that this analysis included only patients who received surgery and so wider differences in resource use due to different mixes of treatments offered would not be reflected here.

Patient-reported health-related quality-of-life data were not routinely collected in this context and so estimates of utility scores were obtained from published sources160,166–177 and applied to patients in the model. It is possible that not all the impacts on clinical outcomes were fully reflected in the QALY values because of the lack of patient-level utility data in the linked data set.

There were some differences in methods used between the analysis conducted in this chapter and the analysis in Chapter 9, and these differences were necessary because of the structure of the health economics models. For example, decision tree proportions were calculated over three groups and, therefore, ordered logistic regression was used here instead of logistic regression (see Chapter 9). Use of three categories may have exacerbated issues related to small sample sizes and event numbers.

Our analysis used the economic evaluation methodology of quantifying cost using opportunity cost and assuming a threshold for cost-effectiveness. Other techniques, such as programme budgeting and marginal analysis, and maximising outcomes within a fixed budget, could have been used, as sometimes decision-makers find these techniques more relevant;151 however, these were beyond the scope of this analysis. The analysis was restricted to a payer perspective, as we had treatment pathway resource use from hospital records only and not from primary care or other community services. This perspective was also used in calculating the implementation costs that were included in this analysis. The impact of this will vary depending on the cancer and the relative importance of other costs. Participants from one site, which was not a specialist site, mentioned increased staff costs for a time after the implementation phase due to increased use of locums. However, insufficient information was available to include these costs.

Comparison with other studies

Cost–utility analysis (i.e. calculating the incremental cost per QALY gained) is not common in economic evaluations of service reconfiguration. 27,148,185 Greving et al. 186 developed a decision model to evaluate the centralisation of ovarian cancer services and found that services providing semispecialised hospitals, compared with general hospitals, were cost-effective at €7135 per QALY gained, but tertiary hospital care was not (at €102,642/QALY). The majority of evaluations focus on the relationship between hospital patient volume and clinical outcomes, with some evidence that concentrating procedures in high-volume hospitals is related to improved clinical outcomes. 187 The methodology of many of these studies, however, means that caution should be exercised when interpreting these results. 185

Implications

The reconfigurations in prostate, bladder and oesophago-gastric cancer in the London Cancer region appeared to have a medium probability of being cost-effective compared with the rest of England. This could reassure health leaders that large-scale reconfigurations can be appropriate in terms of cost-effectiveness, although including the implementation costs, as was done here, can have an important impact. However, it is not clear if an individual cancer pathway’s reconfigurations could be implemented alone, especially as urology cancer pathways overlap clinically and, therefore, it is likely that the results of the four analyses need to be considered together. There was no evidence of a negative impact on patients, as QALY changes were all insignificant, although some had negative mean values.

What is already known?

• Previous research suggests that analysing MSC in terms of ‘implementation outcomes’ may support better understanding of its impact on clinical outcomes.

• The London Cancer changes had a range of effects on clinical outcomes and cost-effectiveness.

What does this chapter add?

• Some of the main priorities of MSC in London Cancer were to improve clinical outcomes, improve patient experience and increase participation in research.

• For urological cancers, patient volume in specialist centres increased and LOS reduced significantly, suggesting benefits of treatment in a specialist unit. The impact on some measures (e.g. mortality and survival) was difficult to evidence, partly because of there being limited room to improve but also because of gaps in available data and an absence of processes to monitor performance. Post implementation, very few patients were treated in non-specialist units, suggesting high fidelity to the referral pathway.

• National targets for time to diagnosis and surgery were a priority in London Cancer. Challenges with both transfer into specialist services and transfer on to local units potentially reflected ongoing ‘bedding in’ and adaptations of the new system.

• The challenges identified in achieving and evidencing impact on key outcomes reflect lessons from previous literature on implementing MSC and evaluating change more generally. These challenges relate to selecting meaningful measures, based on data of suitable quality and completeness, and having processes in place to monitor and manage performance.

Understanding the outcomes of major system change

Major system change is a complex process: asking ‘did it work?’ is not enough; we must also ask ‘how did it work, and why?’. In addressing the latter, it can be helpful to distinguish between ‘intervention outcomes’41,44,51 (i.e. the impact on delivery of care and clinical outcomes) and ‘implementation outcomes’41,44,51,188 [i.e. the extent to which an intervention was implemented, including the extent of adoption (is the intervention used in all participating units?), fidelity (is it followed reliably across the whole system?) and sustainability (does it continue to be used?)]. Analysing such implementation outcomes of MSC may facilitate understanding of the intervention outcomes it achieves. 41,44,51 As outlined in Chapters 5 and 7, there were wider outcomes of change, including how providers worked together as a network and how actors within the network experienced loss as a result of change.

Previous research1,38,41,42,44,51,60 suggests that outcomes achieved by MSC are influenced both by the service model implemented (e.g. number of units providing care, service specifications and referral pathways) and the implementation approaches employed (e.g. leadership approaches, accreditation against standards and external support to facilitate change).

Looking beyond London Cancer’s intervention outcomes

In Chapters 9 and 10, we described the impact of MSC on London Cancer’s oesophago-gastric and urological specialist surgical services in terms of delivery of clinical interventions, patient outcomes and the cost-effectiveness of the centralised services. Here, we draw on qualitative data to address the following questions:

• To what extent were London Cancer’s objectives for oesophago-gastric and urological cancer services (including those analysed in Chapters 9 and 10) achieved?

• Were there any unintended (positive or negative) consequences of the changes?

• Which factors contributed to these outcomes?

Design

This was a mixed-methods single-case study, which focussed on the effects of MSC on specialist surgery for urological and oesophago-gastric cancers in London Cancer.

Sample

We analysed interviews with 99 London Cancer stakeholders (involving system leaders, programme team members, senior hospital managers, service staff and voluntary sector staff), documents (including project plans, case for change documents and meeting minutes) and meeting observations (see Chapter 2 and Table 3). We also drew on quantitative data on where patients underwent surgery, which had been analysed for Chapter 9.

Analysis

The analysis developed in several stages. We produced a timeline and narrative of implementation (led by CV, with input from AIGR, NJF and VJW). Discussions with the quantitative team (SM, RH, MMM and CSC) and clinical collaborators highlighted several important effects of change that quantitative analysis could not address. To assess fidelity to the patient transfer pathway, we calculated the number of patients undergoing surgery in specialist centres over time. Angus IG Ramsay led the final analysis of the data, drawing on a framework designed to analyse MSC in terms of the dynamic relationships between its key components,38,41,42,44,51 combined with the main objectives of MSC set out in London Cancer planning documentation and stakeholder interviews, with a focus on identifying both intended and unintended consequences of change.

The extent to which London Cancer influenced outcomes

London Cancer aimed to improve specialist cancer surgical services in many ways:

. . . improved 1-year survival for patients within London Cancer; improvement in patients self-reported experience of the care they receive; and increased participation in clinical trials to 33% of all patients.

London Cancer memorandum of understanding189

(contains public sector information licensed under the Open Government Licence v3.0)

Shortly before change was implemented, London Cancer’s objectives were revised to cover (1) early diagnosis, (2) supporting local improvement initiatives and (3) supporting whole pathway improvement. 190 These adaptations can be seen as reflecting appreciation of the potential for system-wide improvement approaches to improve patient outcomes, experiences and research participation.

London Cancer identified a number of system-level benefits that might be seen as examples of ‘implementation outcomes’, including:

• standardised flow of patients into high-volume centres for surgery (while other aspects of care remained at local centres)

• access to full range of appropriate care, delivered by specialists (including innovative non-surgical techniques)

• greater subspecialisation of surgeons (i.e. greater expertise in specific cancers and treatments)

• specialist participation in local units (e.g. training and development) to build a culture of collaboration.

Below, we discuss the extent to which London Cancer’s objectives of improving patient outcomes, experiences and research participation were achieved, while also addressing some of the underlying factors.

Patient outcomes

Improving 1-year survival was seen as an important indicator of wider system improvement and outcomes. As discussed in Chapters 9 and 10, MSC in London Cancer resulted in variable effects on outcomes and cost-effectiveness (Figure 19). There were no significant improvements in re-admissions or mortality and, of the four centralisations conducted, there was only a medium chance of prostate, bladder and oesophago-gastric changes being cost-effective, with a low chance of renal MSC being cost-effective. However, for urological cancers, there were significant increases in the volume of patients treated per individual surgeon and significant reductions in LOS. Reduced LOS is likely to indicate the benefits of treatment in a specialist unit, with greater access to 24/7 specialist support and less invasive procedures.

FIGURE 19.

Simplified summary of significant effects on outcomes (note that findings are drawn from Chapters 9 and 10). An upward arrow indicates a significant increase, a downward arrow indicates a significant decrease and a sideways arrow indicates no significant effect. Light blue indicates a desirable result, orange indicates an undesirable result and dark blue indicates no clear direction of change. a, Cost-effectiveness figures present per cent likelihood of centralisations being cost-effective at the £30,000/QALY threshold, including implementation costs.

As discussed in Chapter 9, the reason for the lack of impact on mortality may be statistical in nature, that is, pre-MSC mortality levels were already very low. However, London Cancer’s prioritisation of cancer patient survival in plans reflected priorities identified in national and international guidance. 2–6 Regarding the cost-effectiveness analyses, the cancer reconfigurations were not implemented in isolation, and urology cancer pathways in particular overlap clinically; therefore, it could be argued that the results of the four analyses need to be considered together (rather than separately, as presented in Figure 19).

Another issue was with data availability, which limited the degree to which measures, such as patient experience and quality of life, could be assessed:

To really demonstrate the full impact [ . . . ] that’s going to be difficult because the baseline data didn’t exist. [ . . . ] we’ll have to use the datasets that were available at that time, which do tend more towards the kind of crude mortality type stuff.

Lon29, senior manager, hospital hosting specialist centre

Factors underlying London Cancer’s outcomes

To understand the outcomes of MSC in London Cancer, we explored the extent to which underlying changes – which were anticipated to enable these outcomes – were achieved. Below, we consider the extent to which the service model (i.e. of all eligible patients receiving specialist surgery in a specialist centre, then returning to their local unit for ongoing care) was delivered.

Patient transfer: access to high-volume centres and specialist surgeons

The original model for London Cancer was that (1) all patients eligible for specialist surgery would be treated in a specialist centre (increasing volume of patients treated by specialist surgeons) and (2) these patients would undergo all other aspects of care – both before and after surgery – at their local centre (see Chapter 1 and Figure 1). According to interviewees, this reassured staff at local units that they would continue to play an important role in care. In addition, by covering only a small percentage of overall care delivered, the changes did not have to undergo formal consultation.

Figure 20 shows that the number of patients undergoing specialist surgery in urology and oesophago-gastric specialist centres increased over time, with very few patients treated in local units post implementation. This suggests good fidelity to the new referral pathway, contributing to increased surgeon volumes and access to specialist treatment. However, it was harder to gauge the proportion of eligible patients receiving specialist surgery (i.e. whether or not all eligible patients were treated in a specialist centre). There was no established process to assess this, but surgeons suggested that it was unlikely that local units would be performing such surgery:

I suspect we are getting pretty much all of the partial nephrectomy work. I don’t think that people are having open partial nephrectomies done peripherally, because that is a difficult operation, so I think that for the partial nephrectomy work we’re getting a lot. There are still local hospitals that are doing radical nephrectomies. [ . . . ] But, you see, now what happens is that their trusts won’t really get reimbursed for doing that, in theory that will be what happens, and they don’t then have the funding to have cancer nurses, to have specialist radiologists, to have oncology, so I suspect that with time those numbers will decrease.

Consultant urological surgeon, London Cancer kidney specialist centre

FIGURE 20.

Overview of specialist operations per year, disaggregated by specialist/non-specialist centre status. (a) Specialist prostate cancer operations by year; (b) specialist bladder cancer operations by year; (c) specialist renal cancer operations by year; and (d) specialist oesophago-gastric cancer operations by year.

London Cancer’s oesophago-gastric services did not significantly increase their surgeon volume. A member of the oesophago-gastric pathway board explained that this was mainly as a result of fewer patients being suitable for radical surgery because of improvements in imaging and detection of invasive and metastatic disease. By identifying patients who would not benefit from surgical resection, essentially avoiding unwarranted treatments, there were no longer sufficient numbers of patients requiring specialist surgery to achieve anticipated surgeon volumes:

If we looked at the [pre-centralisation] trends the re-section rates are falling, and we knew that they would continue to fall basically because of better imaging. So we find more metastases so that those patients who had previously been operated [ . . . ] would now not be operated on. [ . . . ] The total number done for the year ending this April at [specialist centre] was 97 cases. So this is short of the 120 that we predicted. And we have six surgeons, so the problem is going to be how this is going to work out.

Lon19fu, consultant oesophago-gastric surgeon, oesophago-gastric specialist centre

At the same time, staff reported other benefits of centralisation in terms of surgical presence, with the new system ensuring 24/7 availability of specialists:

Having all the specialised surgeons in one place means there’s always a pelvic cancer surgeon on call here at [specialist centre], the whole time, which is separate from the general on call system. [ . . . ] if there were two centres, that wouldn’t have been possible.

Lon46, consultant urological surgeon, pathway board member

Unintended outcomes

Principal findings

London Cancer made clear progress in delivering its intended outcomes, with complex changes being implemented successfully in terms of where patients were treated, but requiring time to ‘bed in’. Post implementation, the number of patients undergoing specialist surgery in urology and oesophago-gastric specialist centres increased, with very few patients treated in local units. This suggested good fidelity to the new referral pathways, contributing to increased surgeon volumes and access to specialist treatment.

London Cancer’s main objective was to improve 1-year survival. Our quantitative analyses of clinical outcomes (see Chapter 9) found that patient mortality left little room for improvement, whereas other measures (e.g. patient experience or quality of life) could not be evidenced strongly because of limitations in available data, especially before centralisation. Reductions in LOS may have resulted from treatment in high-volume specialist units, reflecting the anticipated benefits of the London Cancer changes.

Underlying factors, such as patient flow through the system, suggested that the original model did not sufficiently acknowledge some patients’ preference for continuing care with their surgical team. Information flow emerged as an important gap, with patient data not routinely following patients to their local units for follow-up care. Increased participation from specialist centre staff in local unit activity was broadly not delivered: this engagement had been an important ‘offer’ in the original plans and, therefore, led to a sense of local unit staff feeling a sense of disappointment about unfulfilled promises. This sense of disappointment, in turn, contributed to a sense of ‘us and them’ in local units.

Turning to unintended outcomes, several staff reported the emergence of an ‘us and them’ culture, despite change leaders’ efforts to manage this risk during planning and implementation. At the same time, a ‘culture of leadership’ developed among pathway leads over the course of their selection and training.

Strengths and weaknesses

This analysis presented an important opportunity to look beyond whether or not ‘headline outcomes’ were delivered. By drawing on a large qualitative data set, we were able to consider why and in which ways key outcomes were delivered and to analyse the underlying factors.

As with other qualitative analyses in this study, many of the data were collected retrospectively, making it important to draw on documents contemporaneous to the planning and implementation of MSC. In addition, it was not possible to assess certain aspects of fidelity to the London Cancer system. Instead, we were reliant on our informants’ explanations of how and why patients would not be expected to continue to receive specialist surgery locally.

Comparison with other studies

By studying factors underlying the impact on clinical outcomes, such as patient flow through the system, this analysis builds on previous research showing the importance of examining implementation outcomes (e.g. fidelity) to understand the outcomes of MSC. 41,44,51 Some of the difficulties identified in this analysis reflect the literature on implementing MSC and evaluating change, which notes the importance of developing meaningful performance measures with appropriate baselines. 1,61,193–195 In this case, some measures (e.g. mortality) left little room for services to demonstrate improvement, leaving a situation in which the system could demonstrate increased activity, but not the associated benefits. Evidence on implementing MSC in this context might, therefore, recommend using quantitative and qualitative measures reflecting the benefits of access to innovative technologies or 24/7 specialist cover, especially patient experience and quality of life (e.g. continence). Furthermore, evidence on implementing MSC notes the importance of using (or developing) systems to monitor and manage performance. 1,61 In several cases, we found that processes to monitor key outcomes (e.g. patient experience) were not in place or were developed during the post-implementation period (e.g. introducing increased focus on clinical trial recruitment). Developing such measures and processes may form an important part of the planning process, whereby relevant stakeholders (including patients and professionals) are involved at all stages of the care pathway. Such an approach may help identify relevant measures and processes, but may also increase the degree of local ownership of change proposals. 1,61

Implications

Major system change can result in improvements in outcomes (e.g. LOS) that reflect underlying improvements in care delivery (e.g. access to specialist surgeons). When evidencing impact on key outcomes, it is important to select (qualitative and quantitative) measures that matter to patients, including treatment options, patient experience, functional outcomes and quality of life. These measures should have sufficient ‘room to improve’, with supporting data of sufficient quality at regional and national levels, with processes to analyse progress in achieving outcomes. Such approaches may form a key component of planning and engagement activity, ensuring both appropriateness and system-wide ownership of prioritised measures.

What is already known?

• Many factors influence whether, and in what way, MSC is implemented, and the outcomes achieved. Studying the non-implementation of MSC (i.e. when change does not proceed as planned) may offer insights into how contextual factors may influence change and how implementation approaches play out in different contexts. Although normative analyses of MSC interpret non-implementation as ‘failure’, there may be circumstances in which non-implementation is the appropriate decision.

• Although specialist surgery for urological cancers was centralised in London Cancer in 2016, planned MSCs were not implemented in Greater Manchester Cancer. Studying how and why implementation of these changes progressed differently may enhance understanding of implementation of MSC.

What does this chapter add?

• Greater Manchester Cancer faced several contextual issues. A history of non-implementation reduced clinical support and trust in the process, and several concurrent, linked change programmes increased complexity of local decision-making. Planners did not address clinician concerns about the implications of the Greater Manchester Cancer model (e.g. for benign urology patients and the workforce): this caused loss of trust and ongoing delays, culminating in local urology clinicians resisting the proposals.

• London Cancer faced fewer contextual issues, but still experienced local resistance. London Cancer governance (e.g. obtaining senior management sign-up to the MSC process) enabled system-wide support for proposed changes and this, combined with local clinical ownership of the proposed changes, helped overcome local resistance to MSC proposals.

• Contextual factors are highly influential and may shift over the MSC lifespan, affecting both organisational players and the arenas in which decisions are made. Governance mechanisms to build ongoing system-wide commitment to MSC may enable resilience against local resistance.

Design

To address these questions, we conducted a cross-case analysis,50 focusing on MSC of specialist surgical services for urological cancers in Greater Manchester Cancer and London Cancer, drawing on a previously developed framework describing the relationship between key stages of MSC and its outcomes.

Data and sample

We analysed interviews with clinicians, managers, patient representatives and system leaders in both areas (Greater Manchester Cancer, n = 75; London Cancer, n = 60), and included perspectives from representatives of specialist centres, non-specialist centres and the wider systems (see Chapter 2 and Table 3). Of these interviews, 14 in each area (n = 28) were follow-up interviews, permitting examination of how perspectives changed over time. In addition, we analysed observations of planning and oversight activities (e.g. meetings and events, n = 30) and associated documentation (e.g. plans, meeting minutes, and public reports, n ≈ 100). Details of recruitment and ethics approval are presented in Chapter 2.

Analysis

Over the course of the study, the team developed narratives and timelines of implementation of changes to both oesophago-gastric and urological cancers, based on local documentation, observations and discussions with local clinical leads. Development of the narratives was led by Catherine Perry (for Greater Manchester Cancer) and Cecilia Vindrola-Padros (for London Cancer), supported by the wider qualitative team (including NJF, RJB, AIGR, GB and VJW) in consultation with clinical and patient collaborators. These narratives identified key events influencing progress of MSC in Greater Manchester Cancer and London Cancer. Angus IG Ramsay then led the thematic analysis of interview, observation and documentary data to understand factors contributing to these events, exploring, when possible, the perspectives of stakeholders who might be expected to perceive MSC positively (e.g. change-planners and representatives of specialist centres) and negatively (e.g. stakeholders who would lose specialist activity through MSC). These findings were organised within the team’s MSC framework, reflecting on contextual influences on implementation and non-implementation of change,198–200,205 and discussing the influence of context in terms of the Consolidated Framework For Implementation Research domains of ‘inner’ and ‘outer’ settings. 44,205,206 These findings were discussed within the qualitative group and shared with clinical collaborators for further input as the analysis developed.

Overview

We present our findings as follows. First, we introduce the contextual factors that influenced progress of MSC. Second, we discuss how leadership, planning and implementation approaches influenced differing progress in the two areas. Figure 21 provides a simplified overview of the non-linear relationship between issues that our analysis suggests played an important role in London Cancer’s and Greater Manchester Cancer’s efforts to implement MSC of specialist surgery for urological cancers.

FIGURE 21.

Overview of factors influencing progress of urology MSC in London Cancer and Greater Manchester Cancer.

Context for change

Although Greater Manchester Cancer and London Cancer shared several national and local drivers for centralisation (see Chapters 1, 4 and 6), two contextual factors that differed when comparing MSC of urological cancer surgery were (1) the local history of system reorganisation and (2) the influence of other system and service changes taking place concurrently.

Complexity and distractions

Given timings, both London Cancer and Greater Manchester Cancer change processes had to accommodate the effects of macro-level reforms of commissioning and governance of English NHS services (see Chapters 4 and 6).

Greater Manchester experienced a further macro-level influence on local context, as it was the first area to implement the national drive for the devolution of health and social care. 207 Devolution resulted in local programmes to standardise acute and specialised care and to centralise pathology services into two hubs: these reorganisations added to the complexity of planning the Greater Manchester Cancer urology MSC. This complexity was felt to make reaching decisions harder:

There’s so many different bodies in Greater Manchester, and so many different committees and groups, and I’m surprised anything ever gets approved, because every time [ . . . ] you think you’ve got through a hurdle, there’s another one, and another meeting.

GM45, commissioner

Furthermore, the locally developed ‘Healthier Together’ programme (which focused on reorganising complex acute surgery across Greater Manchester),207 represented a further meso-level distraction from Greater Manchester Cancer’s urology MSC.

In contrast, London Cancer may have faced fewer distractions in terms of concurrent/recent change and, indeed, was facilitated by pre-existing systems. For example, the London Cancer programme made use of existing governance structures, as it operated within the local Academic Health Science Network. The main linked changes were reorganisations of other cancer services through the wider London Cancer programme, which may have placed additional burden on programme leadership, but still operated within the same governance structure.

Governance and leadership

In Greater Manchester, surgeons whose services were likely to lose activity suggested that the Greater Manchester Cancer programme governance struggled to manage the balance of power among local provider organisations, resulting in a loss of trust in the overall process:

[Trust] are very strong and what they want is a prostate centre. And because they want a prostate centre the rest of Manchester’s being broken up in order to provide it. But I suspect it won’t be better for the experience.

GM37, consultant urological surgeon

In terms of transparency of a process [ . . . ], there doesn’t seem to be any, there doesn’t seem to be a fair equitable or really patient-centred process; it actually came down to an organisational political decision.

GM49, consultant urological surgeon

London Cancer’s programme governance was designed to manage potential contextual challenges. All local provider organisations signed up to the programme’s terms of reference that delegated substantial authority to the London Cancer Board, ‘to make recommendations and then agree with commissioners the appropriate incentives and any sanctions necessary to drive the prioritised recommendations from Cancer Pathway Boards on behalf of London Cancer’ (contains public sector information licensed under the Open Government Licence v3.0). 189 This agreement was seen as permitting London Cancer leaders to develop MSC proposals with support from all providers (including those who might ultimately lose surgical activity):

We had this mandate for all of the organisations which gave us power [ . . . ] a fairly robust terms of reference which went to the Executive and all the chief executives said, ‘Yes, fine, we’re happy for that to go ahead’. That gave us the right, if you like, to make recommendations to the commissioners about how cancer services should be reconfigured.

Lon02, programme team, London Cancer

In addition, the Gate Review process (see Chapter 4) offered transparency on progress of the changes against established objectives, providing ongoing assurance to senior management of participating organisations and the wider system. Throughout, the London Cancer programme emphasised the overall independence of these processes.

Agreeing service models

Both areas experienced debates about how the system would operate and implications for associated services. In Greater Manchester Cancer, there was much engagement activity to inform and gain clinician support for the changes. The Greater Manchester Transformation Unit (which also managed the successfully implemented oesophago-gastric centralisation; see Chapter 6) conducted events to develop service standards and discuss potential issues emerging from implementing MSC. However, clinicians raised concerns about implications of MSC at these events, particularly about the potential impact on associated services (e.g. for ‘benign’, non-cancer urology patients) both in terms of care delivery and workforce. Leaders of the workshops, whose priority was delivering the centralisation of cancer surgery, felt that these concerns meant that the discussion ‘lost its way’:

. . . the final [workshop] was a bit, I mean it just fell for some reason, it wasn’t as constructive or positive as the previous one. We got bogged down with benign urology and how we would resolve that. It becomes too big I think, the issues. And so that final workshop I think lost its way a little bit.

GM03, Transformation Unit, Greater Manchester Cancer

In contrast, many clinicians believed that their concerns about benign services were central to the questions of if and how urological cancer surgery should be centralised. These clinicians felt that their concerns were not addressed in sufficient depth either at the events or subsequently:

. . . the items that still are grumbling on that we haven’t solved were put into a ‘car park’ and it was simply called the car park because it was too difficult to address and left and despite that a commissioning decision was made.

GM49, consultant urological surgeon, Greater Manchester Cancer

In Greater Manchester Cancer, the proposed model – creating one centre for bladder and kidney cancers and another for prostate cancer – was seen by clinicians as unusual and unjustified. Furthermore, clinicians felt that the model reflected political rather than clinical priorities. In turn, this was seen as driving a loss of trust among consultants, resulting in slow progress of change:

Obviously ones from the centre with more political power [ . . . ] have the ear of people and they’re going, ‘no, we should definitely split up prostates and bladders’. Whereas you look around the world, that’s not been done, because the same skill set does both those operations. [ . . . ] it doesn’t seem to make any sense. And because it doesn’t make sense, people don’t buy into it. And without consultant buy-in as they found out, things move very, very slowly.

GM37, consultant urological surgeon, Greater Manchester Cancer

In London Cancer, the service model also prompted debate. For instance, clinicians questioned the number of specialist centres required and the strength of evidence for centralising kidney cancer surgery. However, substantial engagement (e.g. with staff and through tumour-specific patient representative groups) and provision of data supportive of change (see Chapter 4) helped to build a critical mass of supporters for change across the London Cancer area. Importantly, this critical mass was also achieved in organisations that were to ‘lose’ services, with supportive clinicians and senior managers making the case for change against local resistance (e.g. driven by concerns reported in Chapter 7):

When colleagues are saying, ‘How dare you move what we do here to another trust?’ [ . . . ] they were really very courageous to stand up and put forward the evidence and make the case and say, ‘Actually, this is the right thing to do’, and that’s very powerful when it comes from the trust that is losing a service. [ . . . ] So that leadership and clinical leadership is really an important enabler as is the executive support for the trust. I mean, they need to be backed by senior management of the trust to the trust board to give them that freedom.

Lon05, Academic Health Science Network

Implementation approaches

In both areas, phased implementation was seen as a way to ensure wider stability of the system. In Greater Manchester Cancer, various phased approaches were considered, with patients moving to specialist centres based on (1) which hospital they would be treated at and/or (2) the type or risk level of surgical procedures they were to undergo. Another option was first creating a single service operating across two sites, before moving the centre to a single site. Specialist centre representatives pressed for a staged approach to implementation (which they felt to be more logistically manageable). However, some services that were to ‘lose’ specialist surgery argued that patients should move to the new centres in a single step, reflecting frustration with implementation being drawn out over a longer period:

It doesn’t make sense to move, for us anyway, to move things partially. Either press the button and really bugger it up all at one go rather than buggering it up in little sections.

GM37, consultant urological surgeon

In London Cancer, implementation was staged by site, with patients moving to the specialist centre one region at a time. This staged approach was justified in terms of ensuring appropriate capacity to deliver the new care pathway within the new centre and host hospital, but also across the wider system:

It’s around bed capacity, operating theatre capacity, general clinical capacity and then about transferring the patient care back again when it’s appropriate. [ . . . ] it never would have been the case that they’d move from all centres into one, on the same day.

Lon46, consultant urology surgeon, programme team

Interviewees also described a degree of ‘natural drift’ to the specialist centres before the new system launched, as referring clinicians acknowledged the potential benefit of specialist centre care for certain patients (see Figure 20). Implementation was overseen through regular operational and strategic meetings, which assessed progress against timelines to meet the gateway stages, addressed emerging challenges and enabled ongoing development of aspects of the system (e.g. new referral protocols). Staff noted the importance of independent clinical leadership at service and system levels (e.g. through pathway leads) and senior management support throughout the implementation process.

Progress of implementation

In Greater Manchester Cancer, implementation did not proceed during our study. The scale and complexity of changing urology services were identified as important factors, as was ongoing local resistance to change:

There is goodwill, but it’s just very complicated, I think the complexity of it. What seems like, ‘Oh yeah, just centralise’, the actual delivery of that and the logistics, with people resisting in using clinical or other tactics to stall, and I just get on and just do what I can do and just chip away. I think that, in summary, the barriers would be resource, behaviours, and just the complexity and the number of individuals, and organisations and pathways affected.

GM71, project manager, Greater Manchester Cancer

In late 2019, we learned that 43 of the 53 urology consultants in Greater Manchester Cancer (i.e. both urology cancer surgeons and benign urologists) had written a joint letter to the chief executive of the GMHSCP, stating concerns about implications of the proposed MSC. This letter prompted a meeting ‘to review the grave clinician concerns regarding the proposed changes in Greater Manchester’ (Urology Pathway Board minutes, November 2019). This withdrawal of clinical support – based in organisations that would both lose and gain services – contributed to MSC being paused. Since then, the research team learned of recent voluntary changes to services, enabled by hospital mergers across the Greater Manchester Cancer area. Following these changes, there are now two prostate, three bladder and three renal services across the Greater Manchester Cancer area, all achieving or exceeding volume requirements for safe surgical services.

Implementation of London Cancer’s changes was completed in April 2016. We discuss elsewhere in the report how the changes affected delivery of care, patient outcomes and cost-effectiveness (see Chapters 9 and 10); we also address their influence on interorganisational collaboration (Chapter 5) and other aspects of delivering services (see Chapter 11), and how organisations and individuals experienced loss (in various forms) through the changes (see Chapter 7).

Principal findings

Our analysis suggests that Greater Manchester Cancer faced more challenging circumstances than London Cancer in terms of competing priorities, organisational distractions and history of implementing MSC. However, Greater Manchester Cancer’s approach appeared to have some gaps in terms of transparency and engagement. For example, ‘parking’ clinician concerns about benign urology services, rather than having the (potentially tough) discussions to address them, limited the likelihood of developing a model that was supported by clinicians. A lack of transparency around decision-making (e.g. in terms of the model separating bladder and prostate services) suggested to many that decisions were being guided not by clinical priorities but rather organisational ones; this, in turn, contributed to a loss of trust in the process and growing clinician resistance. Local receptivity to change at organisational (i.e. meso) and individual clinician (i.e. micro) levels is an important component of context for change,44,204,205 and our analysis suggests that some of the approaches employed in Greater Manchester Cancer entrenched some clinicians’ negative perceptions.

Although London Cancer had some experience of successfully centralising specialist cancer surgery and fewer competing initiatives, the changes were still potentially controversial and faced resistance. London Cancer’s approach to governing change was driven toward achieving system-wide receptivity to change at meso and micro levels. For example, system-wide delegation of power to an independent change process permitted development of MSC recommendations that were acceptable to and endorsed by senior managers and clinicians across London Cancer, including within organisations that were to lose specialist surgical activity. Furthermore, by engaging clinicians in the reasoning behind decision-making throughout the process, there was strong front-line support for the proposed changes, and this support, coupled with senior management endorsement, played an important role in responding to local resistance.

Strengths and weaknesses

This analysis represented a rare opportunity to examine implementation and non-implementation of MSC, which could be achieved through only high levels of access (to interviewees, events and documents) enabled by the teams leading MSC in both areas. The cross-case analysis allowed us to assess the degree to which the local contextual challenges were equivalent and to consider how different approaches to leading and managing change might play out in different contexts.

There were some limitations. The bulk of data collection in London Cancer was conducted once key decisions had been agreed or after MSC had been implemented, and these retrospective narratives presented by interviewees may be biased by this knowledge. However, we drew on substantial observation and documentary evidence to mitigate this risk. Timing also limited the Greater Manchester Cancer analysis. As interviews were conducted with the expectation that change would be implemented, questions centred on how MSC would ultimately be delivered, rather than reflecting explicitly on why MSC had not been implemented. Therefore, many of the dynamics contributing to non-implementation had to be inferred from these data and then explored and verified with clinical collaborators.

Comparison with other studies

This analysis builds on previous literature on implementing MSC and other innovations, especially in terms of the role played by context and approaches to leading and governing change. 1 Although much research on how context influences organisational change has tended not to address MSC,205 this analysis suggests that many of the themes identified elsewhere in relation to the influence of context on organisational change apply here.

An important example of this was the inter-related nature of context at macro, meso and micro levels. For instance, national initiatives shaped the meso-level context significantly, including national reforms of NHS governance (which disrupted regional commissioning arrangements in both areas) and the drive for devolution of health and social care (which introduced new layers of governance and prompted several competing programmes of work in Greater Manchester Cancer).

This analysis also illustrated the importance of leadership in enabling change. London Cancer obtained senior management agreement to delegate power to develop recommendations independently and transparently and this delegation enabled a critical mass of senior management and clinical support for recommendations, even in organisations that were to lose specialist surgical services (although, as noted in Chapter 7, some resistance remained in these organisations). This echoes previous research1,60,61 that notes the value of leadership to both drive change and build receptive context at regional, organisational and service levels, and the dynamic relationship between these levels.

Another important process to enable change was stakeholder engagement throughout the change process to ensure that MSC addressed clinical priorities in a realistic way, sustaining system-wide support for proposals. 1,61,197 Gaps in the engagement approach employed in Greater Manchester Cancer – in particular the decision to ‘park’ key clinical concerns and reduced involvement in agreeing the model to go forward to commissioners – suggested that there was insufficient appreciation of the degree to which urological cancer surgery and benign urology were intertwined. These gaps resulted in a service model that most urology specialists in the area did not feel able to support. In addition, these gaps contributed to a perception among staff that the MSC process itself was influenced more by political than by clinical priorities. These concerns link to previous research79 that suggests the importance of achieving and sustaining trust in both the people making the case for change and the evidence used to justify it. These concerns also contribute to a wider debate about how evidence and change processes may be used politically to drive MSC, and the risks associated with not attending sufficiently to concerns raised by clinicians and the public. 15,16,64,208

Implications

Contextual factors, such as history of attempted change, concurrent reorganisations and the local balance of power, are highly influential and may shift over the lifespan of MSC. Certain key aspects of context (e.g. meso- and micro-level receptivity to change) may be influenced, both positively and negatively, by leadership and governance approaches. Governance mechanisms, such as sign-up to change principles and the use of independent oversight processes, may facilitate sustained support from senior managers, while strong clinical leadership may help build commitment from front-line staff. Achieving a critical mass of support for change at meso and micro levels may make MSC more resilient to local challenges.

What is already known?

• The RESPECT-21 study identified potentially useful lessons on stakeholder preferences for MSC, the contribution of leadership and implementation approaches to making change, the impact of MSC on clinical interventions and patient outcomes, and the cost and cost-effectiveness of MSC in this context.

• MSC may play out differently in different (geographic or health-care) contexts. Therefore, there is value in engaging with stakeholders beyond the settings studied (e.g. cancer systems elsewhere in the UK and in non-cancer-specific settings) to explore ways in which lessons might apply in different contexts.

What does this chapter add?

• We conducted an interactive online workshop, which was attended by 32 people, including patients, clinicians, managers, voluntary sector staff and policy-makers from cancer and non-cancer settings. The purpose of this workshop was to share emerging lessons from our findings to explore the ways in which these stakeholders felt that these lessons might (or might not) apply to their own setting. We found that lessons from our research resonated strongly with attendees who raised the following points in relation to the following key themes:

  • With regard to leadership of change, attendees identified challenges related to managing local resistance, political influences and negotiating meaning of evidence.

  • With regard to stakeholder collaboration, attendees discussed the value and challenges inherent in engaging with diverse perspectives, and the importance of prioritising reaching an agreed decision, establishing transparent governance processes and focusing on patient benefit to align priorities.

  • When evaluating MSC and the implications for future work, attendees identified a need to strengthen routine data collection to permit deeper understanding of change and ‘future-proofing’ of evaluation designs. In addition, attendees also urged greater focus on important outcomes (e.g. quality of life and continence) and understanding the lived experiences of patients and carers throughout the care pathway.

Attendees

Table 15 provides an overview of workshop attendees. Our workshop was invitation only to capture a range of perspectives of stakeholders with national and regional leadership roles in relation to cancer and other care settings. Examples of ‘national’ stakeholders included representatives of NHS England and Improvement, and ‘regional’ stakeholders included representatives of Cancer Alliances, health and social care partnerships and senior managers of NHS provider trusts. We categorised stakeholders as ‘non-cancer specific’ if their role encompassed aspects of non-cancer care; however, some of these people may have also held some responsibility for cancer care. An exception to this was when attendees had direct involvement in the changes studied: we classified such people as ‘Cancer specific, Greater Manchester Cancer/London Cancer system’, even if they held a wider regional remit.

We aimed for approximately 30 attendees to permit rich discussion. To identify potential invitees, the research team drew on its own professional networks and worked closely with clinical and patient collaborators.

We sent a ‘save the date’ invitation, which outlined the background to the workshop and the RESPECT-21 study. When inviting people to the event, we emphasised that the event would prioritise learning from attendees, rather than broadcasting study findings.

Sharing evidence before the event

To minimise time spent discussing our findings in the workshop, we shared our key findings in advance of the event so that people would have a good understanding of the research when discussing its implications at the workshop. Our two main approaches to support orientation were (1) our project website and (2) a package of accessible summaries of our key findings.

Timeline

Timing was important: the workshop could take place only once key findings had been established, but it also had to take place sufficiently in advance of the end of the project to permit meaningful analysis and write-up. These interdependencies resulted in a postponement of the workshop when the team realised that additional time was required to finalise some of the analyses. In the run-up to the workshop, we distributed multiple reminders about the event and evidence resources. Appendix 7 provides the key dates and stages of developing this workshop.

Interactive online workshop

The workshop took place online via Zoom (Zoom Video Communications, San Jose, CA, USA) and lasted 2 hours (to minimise attendee fatigue). The event had an external facilitator (Rich Taunt of Kaleidoscope Health and Care, London, UK) and was designed to encourage active discussion among participants. The event structure, including introductory activities, presentation topics, breakout sessions and concluding activities (e.g. evaluation), is summarised in Table 16. Each speaker had relevant expertise, including patient and clinical collaborators, members of the research team and a voluntary sector representative. Most time was devoted to feedback and discussions with attendees to get their views on these themes and how they play out in their respective contexts.

Patient and public involvement and collaboration with other stakeholders

Planning and delivering this workshop was led by two members of the research team (AIGR and PLN), but this was a highly collaborative process, with several members of the research team presenting findings and/or facilitating discussions (NJF, SM, CVP, GB and CSC).

The event could not have happened without important contributions from our patient and clinical team members. Several patient and clinical team members played key roles in planning and delivering our workshop. The workshop was a standing item in our quarterly RSG meetings, which patient and clinical team members attended regularly. As outlined in Collaborating with stakeholders beyond one’s organisation, patient and clinical perspectives were central to the workshop, and we worked closely with our collaborators to develop the talks they gave at the event.

Also valuable was expertise from Rich Taunt, who both facilitated the workshop and advised on event design (e.g. agenda structure and modes of feedback) and practical aspects of delivery (e.g. event timing and ensuring sufficient technical support).

Overview

Here, we present a summary of how our attendees responded to the selected themes and, in particular, their reflections on how these issues had played out in their own context.

Key themes from workshop discussions

Evaluation of event

Although our attendees were highly engaged in the live discussion, our event evaluation had a low response rate (8/32, 25%). The feedback was broadly positive, indicating that lessons had been helpful and would influence future MSC activity:

• Six of eight attendees agreed/strongly agreed that the pre-event information explained study findings.

• Seven of eight attendees agreed/strongly agreed that they could participate as fully as they had wished.

• Half of the attendees agreed that the workshop made them think differently about their own setting.

• Six of eight attendees agreed/strongly agreed that they would draw on lessons from the RESPECT-21 study when implementing change in future.

Strengths and weaknesses

This workshop was an important opportunity to engage with a range of key stakeholders operating at national and regional levels in cancer- and non-cancer-specific settings, including voluntary sector and patient representatives. The event prompted valuable discussions, which identified several ways in which lessons from this research might apply in different settings and how planners and researchers might attempt to work differently in the future. The discussions also contributed to the final conclusions and recommendations for our final report.

Owing to UK pandemic restrictions, our event took place entirely online; this enabled participation from stakeholders across the country, but raised a challenge of striking the right balance between information-sharing and discussion. To minimise ‘Zoom fatigue’, we conducted our meeting in 2 hours and shared the lessons from our research in advance, rather than discussing them in depth at the workshop itself. This made for quite truncated discussions, and one attendee expressed frustration that the main event had focused relatively little on the actual research. Under non-pandemic conditions, we would have held a longer meeting, with greater time devoted to discussing the research findings [see, for example, the event created by the researchers in collaboration with Rich Taunt for research on stroke reconfiguration, URL: www.learningfromstroke.com/ (accessed 1 March 2022)].

Implications

Workshops of this kind can be useful, as they encourage reflection and development of lessons that are accessible to a wide range of stakeholders. For instance, preparing for the event sharpened our thinking on how findings might apply to other settings.

Themes identified through the RESPECT-21 study resonated with the national and regional stakeholders who participated in this workshop, regardless of whether they were based in cancer- or non-cancer-specific settings.

Research question 1: what are patient, public and professional preferences in relation to these centralisations?

Our DCE established the following points in relation to stakeholder preferences:

• Patients, health professionals and the public had similar preferences.

• Patients’, health professionals’ and the public’s preferences were influenced by the risk of complications, the risk of death and the access to specialist MDTs, whereas travel time was considered the least important factor.

• Individual preferences were found to be consistent with the major goals of centralising cancer surgery services.

Research question 2: what are the key processes in centralising specialist cancer surgery services in London Cancer and Greater Manchester Cancer, and what factors influenced progress of centralisation?

Our analysis of network leadership in delivering change in London Cancer (see Chapter 4) established the following:

• MSC was a contested process in London Cancer: some actors across the network, including clinicians and patients, questioned the rationale for the changes, the clinical evidence behind it and the ways in which the changes were made.

• A core central team composed of network leaders, managers and clinical–manager hybrid roles was able to drive the changes forward by developing different forms of engagement with provider organisations, distributing leadership across vertical and horizontal layers, and maintaining constancy in central leadership over time. An important enabler was leadership training for clinical pathway leads.

Our analysis of implementation of oesophago-gastric centralisation in Greater Manchester Cancer (see Chapter 6) suggested the importance of learning from history:

• Change leaders in Greater Manchester recognised that having a change process within the context of competition, led by any one single group (i.e. commissioners or providers), with poor stakeholder engagement and processes amenable to challenge, contributed to the failure of previous reconfiguration attempts.

• The history of failed attempts to reconfigure oesophago-gastric surgery was plain to see, but also evident was more granular detail, for example the history of relationships between individuals. Change leaders responded to all the various facets of history in their attempt to achieve change.

Our cross-case analysis of centralising specialist surgery for urological cancers in Greater Manchester Cancer and London Cancer (see Chapter 12) suggested the following:

• Greater Manchester Cancer faced several contextual obstacles. A history of non-implementation reduced clinical support and trust, and several concurrent, linked change programmes increased the complexity of local decision-making. Planners did not address clinician concerns about implications of the Greater Manchester Cancer model (e.g. for benign urology patients and the workforce), which caused loss of trust and ongoing delays, culminating in local urology clinicians publicly withdrawing support for proposals.

• London Cancer faced fewer contextual issues, and had a recent history of implementing MSC successfully, but still experienced local resistance. London Cancer governance (e.g. obtaining senior management sign-up to the MSC process) enabled system-wide support for proposed changes and this, combined with local clinical ownership of the proposed changes, helped overcome local resistance to MSC proposals.

Research question 3: what is the impact on staff and health-care provider organisations, including ways of working, skill mix and approaches to collaboration?

Our analysis of network collaboration in London Cancer (see Chapter 5) established the following:

• Provider organisations in London Cancer negotiated power relations to establish shared goals and reached consensus in relation to maintaining patient-centred care.

• Provider organisations maintained central figures who could create and sustain collaboration, and promote distributed forms of leadership. These were dynamic processes still under transformation during our analysis.

Our analysis of loss in London Cancer (see Chapter 7) presented the following lessons:

• MSC involved processes such as bidding for specialised status, which incurred feelings of loss and personal failure.

• The movement of financial and workforce resources to specialist sites destabilised ‘ecosystems’ in local teams and created issues with maintaining and recruiting skilled staff.

• MSC can cause loss of motivation and reward in daily work for staff at sites that have lost activity.

Research question 4: what is the impact of the London Cancer centralisations on provision of care in terms of clinical processes and outcomes?

• Centralisation of specialist cancer surgery in London Cancer was associated with fewer surgeons doing more operations, which research2–6,22,26 suggests is associated with better patient outcomes.

• Centralisation of specialist cancer surgery in London Cancer was associated with a significant decrease in LOS. In the case of renal cancer, we found evidence that patients were more likely to receive less invasive treatment, suggesting a broadening of the range of treatment modalities offered.

• We found no evidence of impact on mortality or re-admissions, although this may be because the underlying risk of these outcomes was already low.

• We were unable to provide evidence on the impact of MSC on key outcomes that are of importance to patients, such as range of interventions offered (including less invasive non-surgical procedures) and patient experience, and certain important functional outcomes, such as continence (see Limitations).

Research question 5: what is the impact of the London Cancer centralisations on patient experience, including choice and continuity of care?

• Owing to issues with the NCPES data set (e.g. it was not possible to distinguish patients who had surgery from other types of management, nor disaggregate by specific cancer types addressed by the London Cancer changes), we were unable to quantitatively analyse the impact of London Cancer changes on patient experience (see Chapter 9).

• Qualitative data indicate that London Cancer staff had varied perceptions of the impact of change on patient experience. Although many staff saw improving patient experience as a priority of the changes, they also reported logistical challenges in collecting experience data.

• Several staff described how patients valued aspects of the centralised system, including organised specialist care at the centres (to the point where some patients indicated a preference to continue receiving care at the centre rather than closer to home) and new information and support resources.

• Other staff described patients’ frustration with aspects of the centralised services, including increased travel to reach the specialist centres, insufficient time for discussions with specialists and disjointedness in the system.

Research question 6: what are the costs and cost-effectiveness of the London Cancer changes?

Our analysis of implementation costs (see Chapter 8) suggested the following:

• The London Cancer changes cost £7.2M to plan, design and implement (in 2017–18 Great British pounds). Costs included activities (e.g. options appraisal, change planning and oversight) that spread across the wider London Cancer programme, incorporating changes to cancer pathways beyond those studied in the RESPECT-21 study.

• The largest proportion of the costs was for equipment (robots), which might not apply in other reconfigurations of urological and oesophago-gastric cancers. The total adjusted costs were £3.2M when robot costs were excluded.

• The framework we developed can potentially support different stakeholders, including service planners, researchers and policy-makers, to collect the required information and analyse implementation costs, which are often considered too complex to measure or are excluded as sunk costs.

Our health economic analysis (see Chapter 10) indicated the following:

• There was a medium to high probability of the London Cancer changes leading to more cost-effective treatment provision in prostate cancer specialist surgery (79%), and a medium probability of the same for oesophago-gastric (62%) and bladder (49%) cancer specialist surgery, compared with services as provided in the rest of England, excluding Greater Manchester, at a standard cost-effectiveness threshold of £30,000 per QALY gained.

• There was low probability (12%) of the London Cancer changes being cost-effective for renal specialist surgery at the same threshold.

Research question 7: how might lessons from centralising specialist cancer surgery services be applied in future centralisations of specialist cancer services and other specialist settings?

We conducted an interactive digital workshop, which was attended by 32 stakeholders, including patients, clinicians, managers, voluntary sector staff and policy-makers from cancer- and non-cancer-specific settings. Lessons from our research resonated strongly with attendees, who raised the following points in relation to the following key themes:

• With regard to leadership of change, attendees identified challenges relating to managing local resistance, political influences and negotiating meaning of evidence.

• With regard to stakeholder collaboration, attendees discussed the value and challenges inherent in engaging with diverse perspectives, and the importance of prioritising reaching an agreed decision, establishing transparent governance processes and focusing on patient benefit to align priorities.

• When evaluating MSC and the implications for future work, attendees identified a need to strengthen routine data collection to permit deeper understanding of change and ‘future-proofing’ of evaluation designs. In addition, attendees urged greater focus on important outcomes (e.g. quality of life and continence) and understanding lived experiences of patients and carers throughout the care pathway.

Stakeholder preferences

• Patients, professionals and the public appear to share priorities for MSC: specifically, stakeholders are willing to accept longer patient travel times for specialist surgery if (but only if) they are associated with significant better care and outcomes (although see Limitations).

What works in terms of care delivery, patient outcomes and cost-effectiveness

• Past research indicates that MSC may be associated with improvements in care and outcomes,22,26 but these effects may vary depending on the health-care setting. 8

• Our study demonstrated that MSC can be implemented for specialist cancer surgery and does influence some aspects of care delivery and outcomes.

• Where change was implemented, there were clear signs of fidelity to the new pathways (e.g. a high proportion of patients receiving surgery in specialist centres).

• There were some clear improvements above and beyond what was seen elsewhere in England, for example in terms of LOS and surgeon volumes. However, we did not find any significant improvement in mortality or re-admissions relative to the national control.

• The centralisation of specialist surgical services for prostate, bladder and oesophago-gastric cancers had a medium probability of being cost-effective, whereas the centralisation of specialist surgical services for renal cancer had a low likelihood of being cost-effective. We note, however, that as it is not clear that the individual reconfigurations could have been implemented in isolation, especially as urology cancer pathways overlap clinically and, therefore, it is likely that the results of the four analyses need to be considered together. Our findings add to a limited evidence base on the cost-effectiveness of MSC, analyses of which are seldom conducted. 149,150 We were also able to estimate costs of implementation, something that is even less frequently conducted, and this had an impact on the cost-effectiveness analyses. 148

• We were unable to provide evidence on the impact of MSC on key outcomes that are of importance to patients, such as range of interventions offered (including less invasive non-surgical procedures) and patient experience, and certain important functional outcomes, such as continence (see Limitations). Strengthening routine data collection at local, regional and national levels, in cancer and other health-care settings, would permit more meaningful analysis of the impact of changes of this kind.

The how and why of implementing major system change

• Provider-led networks can deliver MSC of specialist cancer surgery services. However, several factors are influential, reflecting lessons from research on MSC in other contexts. 1,51,61

• Context may both facilitate change and act to obstruct it. For example, national recommendations and local variations in care and outcomes helped drive change in both areas, and the concurrent reorganisation of cardiac services was felt to have aided London Cancer specialist site selections. In terms of obstructions, both London Cancer and Greater Manchester Cancer had to manage implications of NHS reforms implemented in 2013 and, in addition, Greater Manchester Cancer had to manage several other concurrent change programmes, including devolution of health and social care. Therefore, such contextual processes need to be analysed and addressed.

• Although clinicians may broadly support centralisation of specialist surgery in principle, resistance may emerge in relation to location of specialist services, linkage of specialist sites and implications for the wider system (e.g. workforce and ‘benign’ urology services). Failure to engage meaningfully with clinicians and other stakeholders about their concerns may lead to disengagement from change processes, reflecting the wider importance of stakeholder engagement in MSC1,61,197 and the importance of maintaining trust. 188,213

• Governance mechanisms to build ongoing system-wide commitment to MSC, e.g. transparency offered by the Gate Review process (which assesses system readiness for change at a series of key stages) and terms of reference signed up to by local senior management in London Cancer, may enable resilience against local resistance.

• Clinical leaders, developed through training and supported by administrative facilitation, can play a key role in building a network of distributed leadership and shared objectives across the system. 1,60,61

• Change leaders may enhance their approaches by learning from local history of change to guide their implementation approach. 1,104

• Changes of this kind will commonly have ‘winners and losers’ and managing feelings of loss is a likely task for leaders at every level of local systems.

• MSC is not the only route to delivering high-volume specialist cancer surgery. For example, since our data collection period ended, Greater Manchester’s system achieved high-volume services through voluntary reorganisations enabled by hospital mergers.

• Although we identified important obstacles to change in this study, attendees at our stakeholder workshop noted that the COVID-19 pandemic provided several examples where dramatic service changes were achieved in little time, with relatively minimal resistance. Further reflection on system responses to the pandemic – and the extent to which such achievements might apply to ‘normal’/non-pandemic circumstances – may improve understanding of how established obstacles to change might be engaged with in the future.

• Our learning on how complex change may be delivered at scale in collaboration with multiple stakeholders may be of value in many settings. For example, it may be of particular use to integrated care systems, which are anticipated to become statutory bodies in 2022, with responsibility for leading major changes to health and care services at a system level on the basis of population need.

Strengths

• This research was the result of an active collaboration between multidisciplinary researchers, patients, clinicians and managers, which was central to our approach from drafting the original outline application through to the write-up of this report.

• Our DCE allowed us to address a key question when planning change, that is, ‘what matters more to different stakeholders?’ Recruitment to our DCE was conducted in collaboration with Quality Health, which ran the NCPES, which afforded us a sample of patients that reflected a range of patient characteristics.

• We collected a rich qualitative data set contemporaneous with the planning and implementation of change. Our data set comprised 212 interviews, 185 observations and 873 documents from across the London Cancer and Greater Manchester Cancer areas. These data provided numerous insights on how services developed and why change progressed as it did.

• We were able to analyse several national data sets, which provided substantial data on how surgical services performed before and after the changes, both in London Cancer (the studied region) and in a large national control.

• Our study’s integrated design, organised around our updated framework for analysing MSC, helped to ensure that different research components informed and enriched one another. A key example was close collaboration between qualitative and health economic teams, which facilitated detailed analysis of the implementation costs in London Cancer. 139

Limitations

Our study had several limitations:

• Our DCE was, in part, based on a convenience sample. Although participating cancer patients were recruited to ensure a sample that varied in age, sex, and other characteristics our health-care professionals and, particularly, members of the public were self-selecting. Therefore, care must be taken when inferring wider public attitudes to MSC.

• Only the London Cancer changes were implemented in time to permit analysis of their impact. Having only one setting in which the impact of MSC could be assessed limits the confidence with which we might infer transferability of the findings to other parts of the NHS.

• We were unable to measure several outcomes that are of importance to patients and the systems that serve them, including range of interventions offered, patient experience, and functional outcomes (e.g. patient continence, quality of life) in both the short and the long term. The focus on surgery meant that it was not possible to analyse non-surgical interventions. These important gaps in data limited both our quantitative and our cost-effectiveness analyses.

• The health economic analysis included patients who had received surgery only and so effects relating to different mixes of treatments offered (e.g. radical radiotherapy) could not be assessed.

• Our failure to find significant effects on some key outcomes may result from these outcomes not being sufficiently amenable to change or our analysis being underpowered. For example, post-surgical mortality, although extremely important, was already low, making it statistically challenging to establish a significant effect.

• Our quantitative findings may not be generalisable to other parts of the UK. Services in London may be different from those in other parts of the country, for example in terms of travel times and distances, number of centres and the range of facilities and specialist expertise available in them.

• Owing to a lack of available data, our quantitative analyses were unable to look in depth at the more nuanced changes in care delivery (e.g. the breadth of care options offered to patients).

Patient representatives on the research team

Two patient representatives, Neil Cameron (co-applicant, London Cancer) and David Holden (collaborator, London Cancer), joined our team at the outline application stage. Colin Jackson (collaborator, Greater Manchester Cancer) joined our team for our full application. Veronica Brinton (collaborator, London Cancer) joined our team as we prepared for our project to launch.

Over the course of our project, David Holden and Colin Jackson withdrew from the team, and Neil Cameron very sadly died in 2017. Therefore, we recruited two further patient representatives: John Sandell (London Cancer) and Patrick Fahy (Greater Manchester Cancer).

Application

Patient representatives shaped our project in several ways over the course of our application, discussing the purpose and priorities of the research, contributing to the design of the research component (focusing on patient experience) and shaping our approach to PPI.

At both stages of our application, patient representatives commented on our proposal in terms of focus and research questions. In addition, co-investigator Neil Cameron worked with the team to identify key items from the NCPES for analysis.

In terms of informing our PPI approach, our patient representatives advised us on how best to involve them in the research, for example their preference to receive printed or electronic versions of documents 1 week in advance of meetings or events.

Research management and governance

We held quarterly RSG meetings, which we invited all our patient and clinical collaborators to attend. Patient representatives received payment for meeting attendance and had their travel and subsistence covered. Documents were distributed 1 week in advance in electronic format (or paper format if this was a stated preference). There were formal points on the agenda where PPI representatives were invited to contribute to key decisions on interpretation of findings and dissemination opportunities. The RSG meeting chairperson also encouraged patient representatives to participate in general discussions.

We held SSC meetings, which took place approximately annually. Patient representatives on our research team joined these meetings, as did additional patient and carer representatives on the SSC itself. Again, the SSC chairperson (Professor Lorna McKee) took an active approach to encouraging patient and carer representatives’ contributions.

Another important aspect of research governance was our 6-monthly update to the funder. This update featured a section on PPI, where we reported how patients were being involved in the work. In addition, on several occasions, we involved patient representatives in drafting and approving this section.

Study design

As discussed above, patient representatives commented on our research topic and research questions, and Neil Cameron helped identify the NCPES items for quantitative analysis of patient experience.

Data analysis and interpretation

Our RSG and SSC meetings were important arenas for the discussion of research findings and agreement of next steps. Our patient representatives were actively involved in discussing priorities for research focus and helping to ensure that our research focused sufficiently on the wider implications for patients and carers.

In addition, when sharing research findings with patient groups (see Dissemination), we also discussed our wider research plans (e.g. our qualitative analyses of implementation and experiences of loss). This was a useful way to strengthen our interpretation of findings and to obtain wider patient perspectives on our research.

Dissemination

We actively involved our patient representatives in the dissemination of findings. Our patient team members were invited to act as authors on all our academic journal articles (although, ultimately, the patient representatives decided that they would prefer to be acknowledged rather than be full authors). Team members Veronica Brinton, Patrick Fahy and John Sandell are co-authors of this report and they took a central and valuable role in preparing the Plain English summary.

When we published research findings, we developed two-page accessible summaries that covered our key findings and implications. Our patient representatives reviewed and made important contributions to these summaries, including advice on the clarity and focus of language, and made important contributions to these summaries.

In addition, we attended various patient representative groups, in the London Cancer and Greater Manchester Cancer areas, to share developing findings and to discuss our wider research plans. These were important opportunities to get wider perspectives from patients and public who were less aware of our research.

Throughout our project, we produced quarterly newsletters, through which we sought to engage our stakeholder list with the ongoing development of a research project. Our regular ‘meet the team’ section involved a short interview with team members and we ran several individual interviews with our patient representatives over the course of the study. In addition, as with other documents, we invited our patient representatives to comment on the clarity and accessibility of language in our newsletters.

Finally, our stakeholder workshop aimed to explore how lessons from our research might apply in different contexts. A key theme of the workshop was collaborating beyond health-care organisations, and Veronica Brinton, Patrick Fahy and John Sandell each gave a short talk on their experiences of collaboration and what helps ensure an effective collaboration between health-care services and patients. These talks were seen by attendees as extremely powerful illustrations of the difference that effective and patient-focused leadership can make when planning and implementing change.

Importantly, all of this involvement activity had been costed into our funding application, meaning that throughout the project we could pay our patient representatives for their valuable insights and the time they were committing to our work.

Patient representatives are busy people

• It is important to be flexible in one’s engagement, for example working with people’s communication preferences (e.g. telephone, video call, e-mail or text message) and finding mutually convenient times to meet.

Project leadership matters

• Our patient representatives frequently cited the importance of good chairing to ensure a welcoming and productive environment. The patient representatives commented positively on the inclusive agenda and chairing of RSG meetings, through which they felt empowered to join discussions of all aspects of the work.

Patient involvement requires resources

• Our team were able to dedicate appropriate amounts of time to work with our patient representatives and the research benefited accordingly. The additional capacity afforded by having a project manager on the team was extremely valuable in dealing with the logistics of arranging meetings and payments for patient representatives. In addition, identifying patient involvement activity as an explicit role of some (or all) researchers may support more effective involvement.

Paying patient representatives appropriately makes a difference

• In our evaluation of stroke service reconfiguration, we suggested that payment for other activities (i.e. over and above travel and subsistence) would be valuable. In the current study, we drew on examples of good practice (e.g. via NIHR INVOLVE guidance)216–220 to request additional funds to cover the other activities described above, and we feel that this had an important and positive effect on our collaboration with patient representatives.

What might we do differently in future?

• We enjoyed and valued our collaborations with patient representatives on this project. Of course, working closely with a small number of individuals limits the degree of representativeness, and a greater diversity of representation (e.g. by including carers and a wider range of communities) could only have strengthened our core PPI group. Over the course of our study, we sought to increase diversity, for example by attending patient groups to discuss our research. In our future work, we may usefully explore increasing such engagement, for example by working with patient groups and voluntary sector organisations that focus on minority and hard-to-reach groups, and formalising regular meetings with such groups.