The international knowledge base for new care models relevant to primary care-led integrated models: a realist synthesis

Accountable care organisations

Of the international models, arguably the key influencer of the new care models was the US accountable care organisation (ACO) approach. ACOs were introduced in 2012, following new legislature for health-care reform in the USA, which directed the Centers for Medicare and Medicaid Services (CMS) Innovation Centre to develop and test new payment and service delivery models through the Patient Protection and Affordable Care Act. 43 ACOs are described as a group of providers that are contractually responsible to a commissioner to provide all of the care for a given population over a defined period of time, and to uphold pre-agreed quality outcomes within a given ‘capitated’ budget or ‘fee-for-service’ agreements. 44 The programme theory for ACOs is that if providers have an opportunity to share in financial rewards (or face penalties), then ACOs will reduce fragmentation and duplication in medical care. It is suggested that the improved communication and co-ordination across providers, physicians and patients will improve the quality of services and reduce spending. 45 The Pioneer ACO programme started in 2012, with 32 pioneer sites chosen by the CMS to participate, and ended in 2016 with eight organisations. 45 Forty-four next-generation ACOs have been selected by the CMS and will begin the programme in 2018. 45

Gesundes Kinzigtal

Gesundes Kinzigtal (meaning ‘healthy Kinzigtal’ – the town in south-west Germany where the model was introduced) is an accountable network of physicians, partnering with a health-care management company, driven by the needs of multimorbid patients requiring co-ordinated care. 46 Long-term contracts were put into place in 2006 with two German non-profit sickness funds to integrate health and care services for their insured populations. The integrated system covers a population of 60,000, of which Gesundes Kinzigtal is in charge of the health-care budget of 31,000 ‘assureds’. 40 It has a responsibility to deliver regionally integrated health and care services, covering all age groups and care settings, with the explicit triple aims of improving the health of the population, enhancing the patient care experience and reducing the per capita costs of care. To date, improved population health outcomes,40 potential cost-effectiveness,47 more appropriate access (decline in overuse, underuse and misuse of health care) and an increase in health-care quality have been reported. In early 2017, the lessons from Gesundes Kinzigtal were applied to the launch of a new integrated care model in a city in northern Germany, which was designed to improve access to health services in two of Hamburg’s disadvantaged suburbs: Billstedt and Horn. We discuss the underpinning evidence around access in integrated systems in Chapter 4.

Alzira

The Spanish ‘Alzira’ model (named after the town where the model was first implemented) is akin to the financially led model of the US ACOs. However, it has a unique feature of private and public contracting, such that the private sector finances the construction and operation of the health-care infrastructure and delivers clinical services through a capitation fee. This reduces the public sector partner role to that of commissioner of health care. 48 The regional government entered into a long-term contract in 1999 for 15 years, and cost-effectiveness benefits were widely reported. 38 Acerete et al. ,49 however, dispute some of the claims, stating that ‘the precise level of cost savings achieved has never been subject to public scrutiny and remains controversial, with a true like-for-like comparison impossible’. 49 The financial innovations of both the ACO and the Alzira models have influenced the vanguards’ contracting and budgeting mechanisms, and we explore this further in Chapter 5.

Nuka

The Nuka System of Care50 is built on the premise that the relationship between the primary care team and the patient/customer-owner is the single most important tool in managing chronic disease, controlling health-care costs and improving the overall wellness of a population. It was developed in the late 1990s by the Southcentral Foundation, a non-profit health-care organisation serving a population of around 60,000 Alaska Native and American Indian people in Southcentral Alaska. 41 The accountability for improved health outcomes in Nuka lies with the partnership of providers and people in the Native Alaskan community, who changed from being the ‘recipients of services’ to the ‘owners’ of their health system, so that they could co-design services and oversee implementation. The Southcentral Foundation’s annual progress report documents outcomes that include increased patient satisfaction and significant reductions in the use of specialist and hospital-based treatment. 51 The key to the success of this model of integrated care is suggested to be the transfer of the partnership ‘power’ to the recipients of services. 51 We explore partnerships for integrated care further in Chapter 6.

Buurtzorg

The Dutch Buurtzorg (neighbourhood care) not-for-profit model was born out of a desire to improve the staff experience of community nurses delivering integrated home care. It aimed to provide an accountable partnership with social innovation, ‘humanity over bureaucracy’, at its core,52 by attempting to shift the care delivery mindset from managerial protocols and administration to clinical autonomy and accountability. 53 Buurtzorg Nederland was founded in 2006, with a small team of professional nurses, and following 10 years of operation, the model has extended to encompass 10,000 nurses in 850 teams in towns and villages all over the Netherlands, and has spread to a further 24 countries. 54,55 In response to criticisms of ‘cherry-picking’ of patients in its reporting of cost-effectiveness, the Dutch Ministry of Health, Welfare and Sport commissioned the consulting firm Klynveld Peat Marwick Goerdeler (KPMG; Amstelveen, the Netherlands) to compare Buurtzorg with other home-care providers. This evaluation showed that, even after controlling for differences in patient characteristics, Buurtzorg remained a low-cost provider of home-care services; the addition of patients’ total costs (nursing home, physician and hospital costs) resulted in Buurtzorg’s total per-patient costs being around average for the Netherlands. 39 Moreover, the quality of care was deemed efficient, as demonstrated by patient satisfaction56 and a high level of staff experience, as shown by Buurtzorg being named the best national employer in 2010–12. 39 Chapters 4–6 provide further support for the inclusion of staff experience as a health-care aim.

Best-fit framework synthesis

We took the innovative approach of using a best-fit framework synthesis as a means of organising and mapping data, as a prequel to a formal realist synthesis. A best-fit framework synthesis offers a rapid tool by which to facilitate data extraction and analysis, and thus maximise the value of the interpretive synthesis in a way that is amenable to the production of practical recommendations. 66 Framework synthesis is considered ‘especially suitable in addressing urgent policy questions where the need for a more fully developed synthesis is balanced by the need for a quick answer’. 67 A best-fit framework synthesis offers a pragmatic variant of framework synthesis. Instead of spending a disproportionate amount of time in developing a definitive conceptual framework with stakeholders, a best-fit framework synthesis uses the best-available conceptual or instrumental frameworks as a starting point for extracting and analysing the data. In doing so, it harnesses ‘the recognised strengths of both framework and thematic synthesis’. 68 Furthermore, in this project, we specifically sought frameworks with ‘practitioner credibility’ with those working within a rapidly developing area of models of care. Framework synthesis was thus selected as a vehicle for the future dissemination of findings. A best-fit framework synthesis was subsequently combined with a synthesis following realist principles, which is particularly suited to exploring what works, when, for whom and in what circumstances.

The realist principles used in this project derive from the need to use methods that acknowledge the inherent complexity within programmes of change and their evaluation. 69 Although the initial development of the realist methodology focused on realist evaluation, using purposely collected and readily available primary data, Pawson64 subsequently identified the need for, and documented, a corresponding process within the emerging science of reviews. The realist synthesis approach has been consolidated and explored more fully, as demonstrated by Rycroft-Malone et al. 70 in 2012. Particularly influential is the trend to combine elements of realist evaluation and synthesis in order to optimise relevance and transferability. The realist approach acknowledges that interventions do not necessarily transfer easily from one setting to another, and offers deeper insights into the contextual factors involved in change. Realist approaches have been demonstrated as having particular utility within the NIHR’s Health Services and Delivery Research (HSDR) programme-funded projects. 63,71,72

Programme theory development and refinement requires undertaking a series of activities. These activities do not necessarily take place sequentially, and are typically iterative. However, we present them serially for transparency and ease of understanding. We undertook the following activities, reinterpreted from Pawson64 and others in a forthcoming book chapter (due for publication in June 2018):65

• formulating specific questions as lines of inquiry73

• exploring a proposed area of research to identify previously published research and, if necessary, refining the research question (the background search)64

• identifying theories as hypothetical explanatory accounts (initial programme theories) of how an intervention works (the search for programme theories)64

• identifying empirical evidence to test and refine the initial programme theory (the search for empirical evidence)64

• responding to new information needs that emerge during testing and refining the initial programme theory (the final search to refine programme theories)64

• documenting the search process in an explicit and transparent manner. 62,74

A realist synthesis is theory driven and seeks to unearth ‘the unseen elements of a programme (the mechanisms) that lead to its success or failure’. 72 Configurations of contexts, mechanisms and outcomes are mapped to show how particular contexts or conditions trigger mechanisms to generate outcomes. Multiple mechanisms may be activated within each intervention, each representing a ‘pathway from resource to reasoning and response’. 75 The reasoning and response may derive from the perspectives of the receivers, the organisers or those delivering the programmes. 76

Programme theory ‘describes the theory built into every programme’,62 being expressed as ‘if we provide these people with these resources it may change their behaviour’. 77 A strength of realist synthesis is its facility to accommodate diverse sources of evidence when ‘reconstructing’ programme theories. It draws from a systematic process that includes stakeholder engagement, an overview of relevant extant theory62 and close examination of primary research. 64 Such an approach faces three challenges: (1) many commentaries may not articulate a programme theory, or at least may not make a particular programme theory explicit; (2) programme theories may be expressed in a variety of forms and formats, making translation and comparison problematic; and (3) the non-standardisation of format makes it difficult to prioritise theories for further exploration.

However, important adaptations from the classic realist approach acknowledge our ready access to an explicit source of programme theory (i.e. the logic models produced by the local MCP teams). Although programmes of change (i.e. New Care Models programmes) may be considered to be ‘theories incarnate’;69 these theories truly ‘take on flesh’ when realised within the structure of logic models. In this review, we developed a theory by analysing the definitive published logic models, along with their associated narratives, as a source of articulated theories about ‘what works’ in MCP models and the conditions that might result in the achievement of programme success.

For the purposes of this review, a MCP was considered to be ‘a new type of integrated provider’ that ‘combines the delivery of primary care and community-based health and care services’ with the result that ‘services previously based in hospitals’ (e.g. outpatient clinics, diagnostic, day surgery, etc.) may be delivered in the community. It may also involve the integration of ‘mental and physical health services and, potentially, social care provision, together with NHS provision’ [adapted from NHSE (2016);11 contains public sector information licensed under the Open Government Licence v3.0].

As this definition implies, successful implementation and transformational change are contingent on diverse factors that must be acknowledged within a review approach that accommodates complexity and contingency. As a consequence, we selected a theory-driven approach to evidence synthesis, underpinned by the realist philosophy of science and causality. 78

Formulating specific lines of inquiry

In many realist reviews, the research team needs to ‘surface’ programme theories from interviews with stakeholders, and from the published literature and more ephemeral contributions, such as blogs and newspapers. Our access to the formal logic model submissions submitted by vanguards as a requirement for applications for funding allowed us to explore a full spread of interventions, as identified by the vanguards, against a backdrop of different contextual characteristics. The selection of vanguard sites holds an inherent advantage in offering a form of ‘intensity sampling’. However, from a methodological point of view, the concept of a ‘vanguard’ holds an inherent risk of selection bias (i.e. the likelihood that sites have been selected according to their readiness for new care models). So, although we sought to explore the implementation of new care models across a diverse sample of care providers, the total sample itself had been preselected. Such caution is particularly important when looking to apply review findings to roll out and sustainability for ‘failed’ vanguard applications and for those sites that could not aspire to a vanguard application. Nevertheless, existing MCP vanguards represent significant diversity of both contexts and target populations, as acknowledged in our choice of an approach that emphasises ‘what works for whom under what circumstances’,69 and offer the potential to identify context-specific issues and constraints at a particular level, even where a general pattern of successful implementation appears to prevail. Under these conditions, it becomes particularly important to implement a series of validation checks and balances within the strategy for data synthesis.

Agreeing the focus of the review

To agree the focus of the review, we presented our initial programme theory components and a basic overview of constituent if–then statements at a 1-day stakeholder meeting. Our stakeholder event included the following people:

• an academic, who was the chief investigator of another HSDR programme-funded New Care Models project (reference number 15/77/10)

• the accountable officer/chief clinical officer of a CCG, which was a MCP vanguard site

• the chief officer of a CCG, which was a MCP vanguard site

• a communications consultant and fellow

• the Director of Primary Care in a CCG

• an evaluation lead in NHSE

• an executive director of a MCP vanguard site

• a regional manager of the New Care Models programme in NHSE

• two patient and public representatives

• the Director of Strategy of a community health-care foundation trust.

We presented our initial programme theorising as ‘propositions about how new care models are intended to work’, and invited participants to comment on these ideas and refine, extend and prioritise them.

Contribution of the project advisory group

A detailed assessment of the area under study identified a lack of a single appropriate framework. In June 2016, the project advisory group was briefed on the aims, objectives, scope and approach of the study, with a specific focus on the best-fit framework synthesis approach, sharing with them a tentative framework with which the review team intended to support the realist synthesis. The discussion proved to be extremely fruitful, as the review team were made aware of a forthcoming framework from NHSE.

Working with the advisory group allowed us to test the proposed approach and programme theories with key stakeholder groups. It also made us aware of relevant work within the New Care Models programme and the vanguard sites. For example, the review team secured early sight of an implementation matrix being developed by NHSE, which sets out the key characteristics and expectations of new care models.

Creation of a metaframework

Subsequently, we identified five existing frameworks to cover the various aspects relevant to the MCP model of care. These were derived from three relevant guidance documents for MCPs (a, b and e) and two integrated care frameworks (c and d):

1. the eight enablers of transformation in The Forward View into Action: New Care Models: Update and Initial Support89

2. MCP Framework: Working Document 86

3. Framework on Integrated, People-Centred Health Services 90

4. System Leadership: Lessons and Learning from AQuA’s Integrated Care Discovery Communities 91

5. The Multispecialty Community Provider (MCP) Emerging Care Model and Contract Framework. 11

Finally, a sixth quadruple aim framework was identified to capture the intended outcomes, namely, patient experience; the health of the population; health-care costs; and staff experience. 17 The chosen frameworks each contributed unique elements, and so we followed the approach, previously reported by Carroll et al. ,68 of merging the six frameworks into a single metaframework. Essentially, the best-fit framework provided an additional analytical lens for the data, separate from the realist terminology and principles, and offered resonance with the world view of the stakeholders. By combining the six frameworks within a single metaframework, we constructed both a conceptual map and a vehicle for the organisation and extraction of data (Figure 1).

FIGURE 1.

Best-fit framework.

Identifying and searching for abstract theories

Once the candidate programme theories were finalised, the next step was to seek to make connections between these lower-level service-based theories and higher-level, more abstract theories. This allowed us to develop a series of hypotheses that would be testable against empirical studies identified from systematic searching. More importantly, the level at which such theories operate would make it possible to produce transferable lessons about how, and in what circumstances, new care models produce their intended outcomes.

To identify the abstract theories, we pursued references from within the relevant literature, in particular analyses, commentaries and opinion pieces. This required looking beyond the actual programme theory to identify more specific mechanisms that might be activated by a MCP component (e.g. governance and control, local identity and place, collective ownership). In doing this, we sought to answer such questions as ‘how might we typify this intervention?’, ‘what core underlying idea is at work here?’ and ‘what other interventions are underpinned by similar ideas?’. Several key ideas underpinned more than one of the prioritised programme theories:

1. demand management

2. shared accountability

3. community ownership and resilience.

Abstract theories also emerged from the search for empirical evidence, as described below (see Table 3). Programme theory-related searches do not distinguish between empirical and theoretical literature, so it is efficient to look for examples of each when reviewing identified items. Our systematic approach, utilising Publish and Perish 5.23.9 software [(Harzing.com, Hendon, UK) see Data extraction, quality assessment and synthesis for the realist reviews] represents an improved and more auditable strategy than that used in comparable projects, such as screening the first five pages of Google Scholar (Google Inc., Mountain View, CA, USA) results. 63

Population health

A systematic review154 of the substitution of physicians by nurses in primary care found inconclusive evidence on the effects of nurse-led care on quality-of-life measures for patients; the authors note that interpretation was difficult because of heterogeneous outcome reporting, valuation of resources and the small number of studies. The review154 did find nurse-led care to be associated with lower overall mortality, particularly when providing routine care and non-urgent visits. However, earlier reviews154,156 did not find mortality rates between physician and nurse-led care to differ, but did find demonstrable improvements in clinical parameters of chronic conditions by nurses compared with physicians. Similarly, interventions in which pharmacists supported GPs to optimise prescribing or patient medication use were reported to show a limited effect or no effect on outcomes related to quality of life; however, improvements in the management of chronic conditions and the attainment of health goals were observed. 158

Expanded primary care roles have been found to demonstrate an acceptable level of patient safety or better. Nurses have been reported to show no difference in appropriate diagnosis and prescribing in primary care compared with physicians,156 whereas pharmacist interventions in general practice clinics have been shown to improve the quality of prescribing and medication appropriateness. 158 The introduction of a self-referral pathway for physiotherapy has been reported to be clinically effective compared with usual care, and has been deemed to be safe. 162 The evidence about the appropriateness of care provided by physician associates is weak; however, where evidence exists, it is reported positively. 164,181 In general, studies reporting on the transfer of care from secondary to primary care have reported that the quality of care achieves an acceptable standard. 71,173,174 The safety of diagnostic services transfer is equivocal; a literature-mapping exercise175 found evidence that the quality (including diagnostic accuracy and the appropriateness of test ordering) and safety of diagnostic service provision in the community is mixed.

Cost-effectiveness

Evidence on the effect that expanded primary care roles have on health-care service use is underdeveloped, with most evidence reported in case studies including anecdotal claims. Reviews looking at the upskilling of non-physician clinicians, such as nurse practitioners154,156 and pharmacists,156 and the contribution of physician associates in primary care,181 report a dearth of economic data. Furthermore, much of what does exist is based on direct costs only (e.g. resource use, follow-up consultations, length of consultations and salary costs), with inconsistent evaluation approaches used across studies. 154

Single studies and case studies have painted a more positive picture. A cost-effectiveness analysis of the introduction of a self-referral pathway for physiotherapy reported that the service was cost-effective as a result of a reduction in the average number of patients attending and the wasted appointments by patients who do not initiate or actively commit to treatment. 161 The use of physician associates in UK settings has been found to have lower consultation costs than GPs, albeit with limitations, as costs considered only the time spent in consultation and no other factors, such as support time from GPs. 164 Case studies reporting on pharmacists working in general practice have shown some reduction in primary and secondary health-care usage when pharmacists work in general practice,32 and cost-savings through reduced locum costs. 182 Case study reports of paramedics working in primary care provide a more mixed picture. 163 Although a lower conveyance rate than for other paramedics has been reported, other accounts conclude that the cost-effectiveness remains uncertain, given the supervision time required and longer appointment times. 163 Many authors warn that the provision of care in the community may not be cheaper than outpatient services, and, in some cases, services may actually lead to increased costs. 174,183,184

Few economic evaluations exist regarding the transfer of services. Pawson et al. 71 conclude that there is little solid evidence that GPwSIs have reduced the outpatient workload, in regard to anecdotal claims of this reported in the literature. Furthermore, the evidence on cost-savings from the use of GPwSIs is mixed and based on inconclusive data. 71 This view is supported by a scoping review by Winpenny et al. ,174 who confirm inconclusive cost-effectiveness findings, with several studies suggesting that interventions might increase the cost of care provision. Reviews exploring the use of GPwSIs report that the higher cost of GPwSIs can be justified only if the increased cost can be offset against the benefit in improved access to health. 173,174

Similarly, shifted outpatient models are likely to be more expensive than hospital-based outpatient clinics,174 with the ‘drag and drop’ model that replicates the hospital model in different settings being an inefficient use of time and resources. 176 Few formal evaluations have been reported on more recent innovations, such as attaching a specialist to primary care, so that education is at the core of the model;174,179 however, currently, such services may appear to be costly. 174 Early evaluations of a specialist paediatric outreach clinic (monthly joint clinics between GPs and hospital-based general paediatricians) and MDT meetings held in GP practices are showing promising results. 185 Outpatient referrals have been reduced by avoiding appointments altogether through MDT discussions and improved care co-ordination, or by shifting the appointment from the hospital to the GP practice. 185 Furthermore, the joint GP consultant clinics have reduced the follow-up rate, which is attributed to the presence of the GP in the consultation, allowing a joint plan to be agreed, and the awareness of easy connectivity for further paediatric input, should it be needed. 185

Improved connectivity to specialist advice through the use of e-mail or telephone has shown a substantial opportunity to reduce the number of patients who are seen in outpatient clinics,174,186 support the GP’s management of the patient in primary care and identify potentially unnecessary referrals. 186 Similarly, where consultants have supported staff to work in extended roles through training sessions and ongoing support (telephone and by e-mail), a decrease in referrals has been reported, returning to pre-pilot levels after the pilot had finished. 187 Specialist advice can also help to guide the referral to the most appropriate specialist for patients who are not suitable for community-based care, assisting the patient in getting to the right professional first time, saving duplicate appointments. 186

Patient experience

In general, expanded primary care roles are associated with higher patient satisfaction for nurses,154 physician associates,164 self-referral physiotherapy services32,161 and pharmacists working in general practice. 182 A systematic review focusing on pharmacist services provided in general practice clinics found limited or no effect on outcomes relating to patient satisfaction; however, the included studies focused on interventions aimed at optimising prescribing and/or medication use, and may therefore not represent a direct substitution of the care received by GPs. 158 Patient experience in relation to the transfer of services from secondary care to primary care is consistently reported as being positive for receiving care closer to home,179,187 improved access and convenience,178 and a better overall experience (e.g. less intimidating than the hospital). 178,179 There may, however, be some instances that require a trade-off between the convenience of community-based care and the patient experience for some patient populations; for example, families with small children who enjoy the excitement and exclusivity that the children’s hospital provides, and where the absence of medical technologies decreases confidence in the quality of care provided. 178 This may be particularly noticed during symptomatic illness and the need for reassurance of a medical environment with immediate access to investigations. 178

Patient acceptance is generally high for nurses157 and physician associates164,181 used as substitutes for doctors; however, both alternatives report conditions for which patients would prefer to see a doctor. 157,181 Case study experience from pharmacists working in primary care in Cornwall has shown how pharmacists have been accepted with some patients, who now specifically book directly to see a pharmacist for medication issues. 182

Nurses more frequently provide advice and information to patients and can improve access to health-care services and treatments. 156 The introduction of a self-referral pathway for physiotherapy has also been reported to improve access by reducing waiting times,161,162 as has the introduction of GPwSIs. 71,173 However, evaluations of both services acknowledge that it is possible that times increase with longevity of the service,71 or expansion of the service without substantial resources. 161

Experience from the GP Access Fund pilots has found that patients of hub practices have benefited more from extended-access initiatives than those in other practices. 32 The introduction of specialist paediatric community care in London was reported to increase attendance because of decreases in ‘did-not-attend’ rates;185 however, a controlled evaluation of a relocated paediatric outpatient clinic in Birmingham found that higher attendance was associated only with shorter travel distance. 177 Children living in the most deprived areas were more likely to miss appointments. 177

Staff and provider experience

Both staff undertaking new roles and their colleagues have reported increased satisfaction. Case study accounts from a paramedic practitioner pilot found that paramedics were enthusiastic about career progression and development;163 however, it should be noted that paramedic practitioners offer a new way of working that is characterised by the development of a career framework. 188 Physicians and nurses also report consistently high levels of satisfaction in response to pharmacists159 and physician associates166,181 working in primary care, and GPs using consultant e-mail advice highly value the ease of access and efficiency of the response. 186 Physician associates have also proved to be a flexible workforce resource to cover the work of nurses in the case of absence. 189

The expansion of the primary care workforce might also help to relieve the pressure of GPs and free up time as shown in the case of pharmacists,182 nurses,32 paramedics32,163 and direct referral to a physiotherapist via trained care navigators. 32 Although the use of physician associates has been reported to have the potential to increase physician or practice productivity, it carries the associated risk that they will increase the workload of some staff, given the need for supervision and prescribing support in the UK. 181

Unintended consequences

The main reported unintended consequences of R1 are the loss of continuity, increased inequity, the burdening of the wider workforce, increased demand and cost, and workforce challenges.

If continuity of care is defined as ‘seeing the same health care provider every-time’,155 then the evidence suggests that practice-size expansion to include enhanced and/or new roles could potentially disrupt relational continuity. 156 Although some patients may value the new forms of access that larger-scale primary care can bring, others have expressed concerns over losing an ongoing, trusted relationship with their own GP and their own practice. Continuity of care is likely to be more important for elderly patients, those with long-term conditions and those with multiple morbidities, whereas patients with episodic or minor problems value access more highly. 168,169 Analysis of the 2009/10 English General Practitioner Patient Survey168 has, however, highlighted ‘the difficulty for practices in offering speedy access and continuity of care simultaneously’. Rosen et al. 190 warn that the benefits of larger-scale primary care need to maintain the expert generalist skills of general practice. New roles in primary care might best concentrate on areas where workload is high, such as acute ‘same-day’ workload in general practice, so that doctors can focus on complexity in their caseload. This has been demonstrated by physician associates181 and pharmacists. 182 Experience from a Child Health GP Hub system in west London found that, for economies of scale in the reductions in secondary care usage to be achieved, hubs must operate at a sufficiently large capacity of at least 15,000–20,000 people in the practice population and to engage enough professionals,185 while avoiding detrimental access to primary care. Imison et al. 184 point to evidence demonstrating that access to primary care is more sensitive to distance than the use of outpatient or acute services; therefore, more accessible outpatient services would insufficiently compensate for primary care access. A hub-and-spoke model, where the polyclinic acts as a central resource base in a co-ordinated network of practices,184 is proposed as a better strategy to provide equitable access. The subsequent experience of GP Access Fund pilots suggests that practices operating as a hub can offer more extended-access benefits than independent practices. 32

The opportunity for existing general practice staff to enhance their skill set is appreciated. However, without the ability to change their original role, the additional activity of the enhanced role makes it difficult to balance workload,32 in some cases leading to staff using their own time,187 which is ultimately unsustainable. 172 The appeal of an enhanced role may lead staff to pursue a specialist career and leave their original role,187 or lead to a depletion of their source professional role;163 in both cases, succession-planning is required, such that enhanced roles are within an overall career ladder. 163 Experience from Canada found that physician and nurse practitioners were reluctant to fully integrate the pharmacist’s role during the first year, when the service was referred to as a pilot project and the funding to support the service was therefore not permanent, as a result of concerns that they may become dependent on a new service that might not remain available the following year. 159

Introducing new services has the potential to induce demand,71 or there is potential for the new service to address a previously unmet need, if, for example, incentives encourage procedures that had not previously been available. 174 New or transferred services have the tendency to become supplementary rather than a substitute service. 151,163,172,174 This can create a situation in which the wrong case mix is addressed; staff may misuse new services and refer patients who ordinarily would have been managed by GPs;71,187 or patients seen in the new community-based services might then need to be referred to hospital174 or seen again in primary care. 151 In addition to increased demand through supplementary services, money might also be needed to backfill traditional posts if necessary. 187

Other reasons for the increase in service delivery costs include time spent supervising (including signing prescriptions for non-prescribing staff)154,163,164,166,181 or supporting staff;174,191 longer appointment lengths for new roles;156,164,192 more frequent patient contact;149,192 lower case mix;71,181 fewer patients seen than traditional services;174 loss of economies of scale;174 increased cost of staff, as they take on tasks that might traditionally be delegated;174,176 set-up costs, such as indemnity costs for new posts;32,163 and Care Quality Commission registration for hubs and federations. 32 These, along with administrative support and resources to ensure that protocols can be reliably followed, are aspects of cost that must be factored into any service confirmation. 174,187

Experience from the UK has shown that some of these new ways of working face challenges in their demands for the new and extended roles required by the workforce. A critical shortage of ANPs, coupled with a short time frame of the GP Access Fund pilots to train ANPs, meant that they faced challenges with recruitment and, subsequently, implementation. 32 Similarly, the low numbers of currently qualified physician associates in the UK is a limitation to the scope of use. 163,189 The capacity of some workforce groups may also be reduced as they face competing priorities, such as consultant 7-day working. 163,189

The pay of staff also needs to be considered; if it is too low in comparison with other specialties or primary care physicians, then recruitment of staff might also be difficult,155 and, conversely, where it is too high or ‘virtually the same as a GP’, this may affect commissioners’ decisions about the potential benefits. 172

Patient factors

Improved utilisation of community-based health services (outcome) is dependent on the location of such services (resource); location is tailored to the characteristics of the population (context) with respect to travel distance, deprivation and ethnicity177 and how the care provided in community settings ‘fits into their lives’, meaning that there is consideration of patients’ or carers’ place, space and time and their emotional connection (‘at homeness’) with the location (reasoning). 178

Provided that patients are made aware of services (resource) involving new roles or ways of working in primary care (context), such as practice-based pharmacy and paramedic practitioners,163 and that they have sufficient time to experience this care or envisage it181,185 to build confidence (reasoning) to use this service in the future,185 they will accept the care from an alternative clinician (outcome). 182

Staff factors

Where new roles or ways of working are introduced (context), resources such as clinical champions of innovation,159,174,176,177,179,182 regulatory frameworks,156,163 training and development,156,179 clearly defined roles and responsibilities,156,176,179 an explicit case for change176 and dedicated change-management staff32,156,191 assisting with regular communication and engagement,32,191 can be facilitative. In turn, these resources can motivate professionals (reasoning) to embrace change through a better understanding of the new scope of their role, practice and organisation163,176 and an enhanced perception of the credibility, legitimacy and sustainability of new posts. 159,163,174,176 This results in enthusiasm and willingness to adopt new ways of working, including clinical buy-in,191 and combats resistance to change (outcome). 179

Embedding the new and expanded roles within primary care settings (context) requires time and/or funding (resources) to develop the role163 and upskill, as relevant, through training and education. 155,191 For example, hospital specialists can train clinicians in primary care to manage complex patients in the community. 179 Funding will also provide diagnostics in the community where necessary. 187 As for adopting new roles, the ability to embed new roles and the resources of funding and training provide primary care professionals with both the experience149 and the confidence (reasoning)179,186 to diagnose or treat more complex patients themselves in primary care, without onward referral to a hospital (outcome). 179,187 Similarly, notwithstanding limited evidence on care navigation in community-based settings, receptionists who are trained in care navigation (resource) and have increased awareness and knowledge of alternative services (reasoning), could facilitate a direct referral to new services, avoiding a GP consultation (outcome). This has been demonstrated in practice through the use of care navigators in the implementation of a ‘PhysioFirst’ service,32 and through the inclusion of consultations with pharmacists for telephone triage of same-day appointments. 182

Building staff confidence and empowerment can also lead to greater job satisfaction. Job satisfaction can be improved further (outcome) if regulations of the roles allow for greater autonomy and enhancement of the scope of the role (resource), so that professionals are empowered in their position (reasoning). This has been demonstrated by pharmacists working in a general practice who can prescribe,182 and also, conversely, through the inability of physician associates to currently prescribe as per the exclusion of physician associates from the regulatory framework for health-care professionals. 189

System factors

In the earlier section, Staff factors, we reported a dependency on staff willingness to adopt new ways of working and subsequent training to empower them to take on new tasks. In this section, we move on to consider the task-shifting phase. ‘Task shifting requires both willingness to give up tasks on physicians side as well as the ability and capacity to perform these tasks on non-physician health professionals side’. 155 A systematic review on the substitution of physicians by nurses in primary care found GPs unwilling to give up tasks, as demonstrated by a low level of clinical autonomy in nurses. 154 Similarly, Pawson et al. ,71 in their realist review on demand management, note ‘the lack of success of General Practitioners with Specialist Interests (GPwSIs) initiatives due to reluctance to let go by consultants’.

Where new ways of working are introduced (context), resources such as the negotiated division of labour,71 clearly defined roles and responsibilities,154 protocols that define practice boundaries155 and communication and engagement with the wider workforce179 can facilitate task-shifting (outcome). These resources, in turn, create an agreed and accepted case mix of patients to be treated in the community. 71,190 Preconceptions and attitudes of what professional roles entail are eliminated,154,179 enabling professionals to better understand and appreciate the role of each team player,155,159,193 which helps to diminish a fear of losing control or professional insecurities160,163,193 (reasoning).

With the introduction of new roles (context), resources such as co-location,159,193 electronic medical records193 and meetings193 can be facilitative. These resources, in turn, enable professionals to develop collaborative relationships and improve communication, trust and respect,159,185,190 resulting in improved delivery of community services (outcome). For instance, improved trust (context) will determine the level of support that staff in new roles receive (resource) as consultants accept the new service (reasoning),71 improving the safety and efficacy of the community-based services (outcome). 174

Mechanisms summary

From a patient perspective, improving the utilisation of community-based services can be enhanced if service locations are tailored to patients’ emotional connections, as well as developing patient confidence through greater awareness and experience.

From a staff perspective, overcoming resistance to change and the willingness to adopt new ways of working, along with maintaining new ways of working, are key considerations. In order to create a culture receptive to change, structured processes (e.g. regulatory frameworks, training and development programmes) need to be incorporated to develop staff confidence in the new way of working. Once staff have been encouraged to adopt new ways of working, investment is needed to embed such roles, so that primary care professionals have both the experience and the confidence to maintain the new way of working. Staff can further be empowered through autonomy and enhancement of the scope of the role, increasing staff resilience.

From a wider staff perspective, the introduction of new ways of working needs to be done in a structured systematic way, so that all stakeholders have confidence in the new way of working. Collaborative relationships and the promotion of new services will encourage their use.

The evidence base

The evidence base underpinning R2 is characterised by limited empirical evidence, although there have been earlier reviews of available evidence; there are a number of commentaries, drawing on experiential evidence (case studies and informal evaluations). Details are provided in Appendix 6. The review is based on 12 reviews, 3 qualitative studies, 7 experimental studies, 8 official publications and 18 commentaries. Support for this component is therefore relatively weak. Limitations of the evidence base mainly relate to the heterogeneity of contracting models and variable reporting, resulting in a moderate level of uncertainty around the conclusions.

There has been substantial debate and commentary on contractual models within UK and international literature, but limited empirical studies that test and evaluate the theory, perhaps indicative of the challenges involved in measuring impact. Owing to the features of NHS funding and its flows, much of the international evidence base needs to be considered for its applicability to a NHS context. It has even been suggested that, given the significant variation, the labels used may be unhelpful, implying a level of consistency not visible in reality. 201 The evidence base, although limited, will help to articulate the theory and underlying assumptions of the MCP model. When appropriate, gaps in the knowledge base are highlighted, suggesting potential priorities for further research and evaluation.

Population health

Reviews197,208 of contracting arrangements in health and social care present a relatively mixed picture of the impact on population health outcomes, with those that had ‘well-developed and comprehensive pooling arrangements across a range of sources, creating large health and social care budgets’208 demonstrating positive impacts. The reviews found limited studies exploring the impact of contracts on the quality of care, and considerable variation in the measures used. The findings suggested that pooled resources might offer providers the ability to create a more holistic package of care that meets patients’ needs; however, this carries an associated risk that capitation might act as a perverse incentive, encouraging providers to sacrifice quality to reduce cost. The opportunity hinges on the availability of information on health needs, service utilisation and quality indicators. 197

Learning from legacy initiatives, such as the Better Care Fund, suggests that expectations on population health outcomes may be ambitious, failing to take into account the time taken to influence changes in behaviours and to demonstrate an impact on morbidity or mortality;209 a qualitative analysis of Better Care Fund implementations suggests that a time scale of ≥ 5 years is realistic. This same analysis notes how several local health economies chose to scale back ambitions, focusing on specific population groups to achieve short-term goals. Elsewhere, it is proposed that financial reform in health care could have a negligible or negative impact on health inequalities,210 although the evidence is limited.

Three Cochrane reviews204–206 have explored the effectiveness of financial incentives, which highlights the need for more robust evidence on the impact of financial incentives on patient outcomes. Although incentives seem to have some effect on the behaviours of health-care professionals, this may not necessarily translate into improvements in patient outcomes. 205 Capitation, fee for service and pay for performance were found by Yuan et al. 206 as likely to have little or no impact on health outcomes, leading to their conclusion that decision-makers should consider the selection of performance measures and targets, the availability of funding and the means to test the payment level needed to influence behaviour change. 206

Several MCP vanguards have expressed a commitment to develop new outcome frameworks for primary care, replacing the QOF. The QOF was established to drive up quality in primary care, but has since been criticised for (1) its failure to address multimorbidity;211 (2) encouraging an overly mechanistic approach to care, focusing on biomedical surveillance and neglecting self-management;212,213 (3) a focus on measurable, rather than truly improved, health outcomes;214 and (4) imposing barriers to case-finding, particularly in deprived practices, in which prevalence is higher. 211 The focus on single diseases211 may have inhibited a more holistic approach for comorbid patients. However, it has been argued215 that, although guidelines could be created for common comorbidities such as depression, the task would be overwhelming, because of the multitude of potential conditions and combinations.

A recent review216 suggests some lessons, relating to the QOF, which may usefully inform new outcome frameworks. The review found a limited effect on outcomes, including glycaemic control and high blood pressure, suggesting that impacts may be transient and may not lead to improvements in population-level morbidity and mortality. Attribution is problematic – typically many influences, such as housing, environmental factors and deprivation, will have contributed to these outcomes. 118,197,209,216,217 It has been argued that the long-term contracts being developed by MCPs may offer an opportunity to invest in interventions with partners to address these wider determinants. 11

Cost-effectiveness

The current focus on savings and reduced utilisation does not seem to be borne out in the literature. Findings regarding the cost-effectiveness of new contractual arrangements in earlier legacy initiatives have been mixed. A review208 of 34 schemes found only three that reported significant reductions in utilisation; 11 had no significant effect; and, for the remaining studies, results were mixed. Few studies have assessed the long-term impacts, making it difficult to draw conclusions on the sustainability of any reductions in service utilisation. 208 Notwithstanding several evaluations of integrated care contracts, such as Torbay and Wye Valley, the methodologies make it difficult to compare findings accurately. 197 It is possible that integration may help to identify unmet need, which may initially limit savings and tip resources towards acute-based care;208 in addition, the current climate of widespread provider deficit holds a risk that ‘crisis management’ may over-ride prevention. 197 Even when costs are reduced, it can be difficult to translate these into actual savings, without reducing provision or closing beds. 197

A commentary, by Humphries,218 argues that the Better Care Fund, designed to enable integration through pooling resources, ‘created risks as well as opportunities’, because of implicit assumptions ‘based on optimism rather than evidence’ about the savings that could be achieved. The ambition to deliver £511M of savings in 2015–16 was not attained, as a result of increased emergency admissions and delayed transfers of care. 209 Plans to reduce delayed transfers of care by 293,000 days in total actually resulted in an increase of 185,000 days and additional costs of £146M – approximately 75% of areas were unable to achieve the targeted reductions in delayed transfers of care. Similar numbers of local areas experienced increases in emergency admissions, leading to costs of £311M over the plans. Qualitative interviews suggest that the initial administration of the fund created a level of bureaucracy that affected capacity and existing integration schemes. 209 Similar comments about the administrative overhead have been made, by Allen and Petsoulas,219 about the Commissioning for Quality and Innovation scheme. The assumption that pooling resources will help to identify and address barriers to improvement fails to take account of the inevitable redundancy within any system. 217

Looking further afield, to understand what may be learnt from international systems of care, commentaries on the early ACOs in the USA point to mixed results on costs and health outcomes. 119,126,220 A difference-in-differences analysis221 suggests that the success of the ACO model may derive from the shift to lower-cost facilities and the emphasis on outcomes, rather than activity, based on learning from the Massachusetts-based initiative incorporating capitation and pay for performance. These changes are associated with reduced utilisation in some population groups. At this stage, given the maturity level of ACOs, it may be too early to see, where reductions in utilisation have been demonstrated, if these can be sustained over time. 126 Questions remain regarding the expectations of cost-savings from shifting care closer to home191,222 within a NHS context. An evaluation of the CMS Group Practice Demonstration Programme223 found significant reductions (US$500 per person per year) for the sickest patients (people receiving both Medicare and Medicaid support). The authors go on to speculate that ACOs may have a role in ‘slowing the rate of increase in health care spending and in bringing about some improvements in the quality of care’; however, they also point out that ‘if ACOs are to bring benefits then they will need a range of capabilities in order to manage cost and quality and to be able to implement what is known about the characteristics of successful integrated systems’. 223

Capitation and pay for performance were found to be ‘generally effective’, based on the findings from a Cochrane review. 204 In an outpatient setting, pay-for-performance incentives were associated with improvements in quality in outpatient settings, particularly in the use of tests and treatments for long-term conditions, but little or no difference to health outcomes or utilisation of services. 206 Learning from the QOF suggests that the effectiveness has been variable. The QOF has been associated with a small curb in the increase in emergency admissions216 for those conditions that are included in the framework. However, it is difficult to attribute any improvements to the QOF, given the multiple factors that may have contributed. A key criticism of the QOF is that the framework incentivises activity rather than quality. 204,224

Lessons from earlier initiatives suggest that potential savings could be limited by the time and set-up costs to design and implement new contracts, outcome frameworks and performance management systems. 201,225 There may be costs associated with data collection and management, actuarial analysis and training. It has been suggested that this work can take up to 2 years. 201,226

Patient experience

With the emphasis on outcome-based commissioning within MCP contracts, it could be argued that the use of financial incentives is geared more at improving productivity than quality. However, little evidence exists to support the notion that financial incentives will drive up quality without some investment of organisational development,205 particularly focusing on collaboration, communication and teamwork. 118 The review found little evidence on the impact of financial incentives on patient experience.

This is indicative of the challenge of defining and measuring outcomes that are meaningful from both a patient and service perspective118 and is a recurrent theme in the literature: ‘. . . complying to arrangements for patient empowerment and patient centredness will be challenging as these terms are not self-defining and there are a wide variety of approaches to achieving these goals, depending on how they are understood.’118

Staff and provider experience

The evidence base appears to focus on the experience of provider organisations rather than individual staff, suggesting a potential research gap.

For providers, there may be greater flexibility for investment within a pathway and over the lifetime of a contract, to manage risks and gains. 227 It has been argued that this creates an environment for innovation118 and enables a focus on the wider determinants of health.

The context in which MCPs will be operating presents challenges to commissioners and providers, in particular, political pressure to deliver change within time frames that have been described as idealistic; Burns and Pauly220 suggest that a time frame of 5–7 years is more realistic. Anecdotally, MCP teams are noting the time involved in building relationships and structures. In a conference presentation on new care models, Hopson226 noted that the logistics of getting out to talk to stakeholders (e.g. practices, local authorities) presented significant challenges to vanguard teams.

Unintended consequences

Elsewhere in this report, the issue of defining ‘place’ or ‘community’ is noted. Although many of the MCP vanguards refer to ‘natural communities’, the notion of ‘community’ is fluid and subjective (further described in Chapter 6). As with any boundary, there may be consequences from where contractual lines are drawn. For patients living near the boundaries, there may be impacts on patient choice and on the liability of the MCP, as suggested in the following commentary:

If a patient opts to seek care from a different provider outside the area, the local ACO would remain accountable for that individual’s care and bear all the costs. In addition, the ACO or other provider that receives this out-of-network patient would be remunerated depending on the level of success.

Reproduced from Ahmed et al. 119

Uncertainty exists regarding the tensions and trade-offs resulting from the shift of care from hospital to community settings, with speculation that organisations may act in their own interests rather than those of the overall system. 118 Conflicting policies may exacerbate this, as pointed out in the National Audit Office analysis of integrated care. 209 Although some areas (such as Salford and Northumberland)200 are making progress on pooling budgets, hurdles remain in pooling across health and social care, making it difficult to align incentives,208 and in transferring savings and risks across sectors. Those areas that are making progress are building on a history of collaborative working. 200 In the USA, there has been speculation228 that the ACO model, in its emphasis on consolidation and co-ordination, may inhibit those market forces/competition that are necessary in driving cost and quality improvement.

In addition, concerns have been expressed that the lead provider/contractor model, advocated in the fully and partially integrated MCP contracts, could have negative consequences, creating further distance between commissioners and providers and affecting clinical governance and patient safety. 118 The actions of individuals are also unpredictable; a review, by Billings and de Weger,118 points to earlier analysis of the effectiveness of incentives: ‘. . . although agents will often change their behaviour in response to sanction/reward mechanisms, this is not always in desired ways’. Outcome frameworks may risk focusing on process outcomes, rather than those that are more challenging to define. The implications for providers should also be considered – qualitative measures may involve new data collections and could incur more cost. 118

Learning from the ACO models in the USA, it has been noted that current savings typically derive from the first cohort, set up in 2012, demonstrating the time needed to deliver efficiencies/improved productivity. 229 Similarly, it seems unlikely that improvements to the quality or experience of care can be delivered in a matter of months. 221 This may translate into an optimism bias,201 whereby commissioners may develop contractual forms that do not have the support of providers and local communities. The challenge of designing a population-based capitated contract, in which subcontracted providers may hail from both the public and private sectors, should not be underestimated; ‘gains’ may be seen as profit for shareholders or reinvestment for improvements. 200 New guidance230 from NHS Improvement suggests that adjustments will be considered for agreed periods, to enable integrated working to become embeded.

There is a risk that incentives may reward inappropriate behaviours; for example, capitation could restrict access for some patients to some services;208 however, commissioners have the opportunity to set thresholds to mitigate this. The focus on reducing hospital utilisation may also have an impact on patient choice, if packages of care are focused around this aim to the exclusion of patient preferences. 208 In particular, it can be difficult to define outcomes relating to the experience of care.

The different contractual forms described in the MCP framework11 offer different combinations of risk and opportunity. Reviews of contracting arrangements have found that alliances (the vehicle for GP practices to join partially integrated MCPs) benefit from incentives to collaborate, improved relationships, active involvement of commissioners,201 information-sharing, reduced capital costs and better co-ordination. 225 However, some evidence suggests a high failure rate in alliances118 because of the conflict between competition and collaboration:

Alliance partnerships differ in the amount of pressure the firms feel to cooperate vs compete. These competitive forces are affected by contracts existing between the parties, the amount of product and service overlap, as well as the amount of trust in the relationship.

Reproduced from Billings and de Weger118

The lead provider model offers commissioners the opportunity to address pathways via one contract, as opposed to the multiple contracts currently in place. This raises questions on how group accountability for outcomes will be managed. It has been argued that incentives might result in unintended behaviours and outcomes. Incentivising case-finding could initially increase costs as unmet needs are identified. Professionals may be incentivised to focus their efforts on outcomes that are rewarded, thus neglecting other aspects of care. 205 Furthermore, there is an associated risk of ‘gaming’ unless robust information systems underpin performance management. 197,208 The availability of information is associated with the power balance between commissioner and provider; information asymmetry is a risk227 when commissioners are ‘price-takers rather than price-setters’,208 leading to inefficiencies in the contract.

As the theory behind longer-term, place-based and outcome-based contracts suggests that outcomes can be improved, recent examples – such as Hinchingbrooke231 and the UnitingCare contract in Cambridgeshire and Peterborough200,232–234 – have demonstrated the difficulty of pricing services previously procured on block contracts. The UnitingCare contract – a 5-year £725M contract awarded in 2014 for acute and mental health services for older people and community services across Cambridgeshire – failed, resulting in at least £16M in termination costs shared by the parent companies of the provider and the CCG. 200 The viability of the procurement, which stipulated a 10% reduction in costs over the course of the contract, was questioned. 233 In addition, the financial risk and value-added tax (VAT) implications were not sufficiently addressed. 232,233 The context of an existing unsatisfactory contract combined with local ambition to deliver to tight time scales influenced decision-making behaviours to prematurely sign the contract. 233 Potential implications may relate to circumstances in which commissioners prefer larger providers or partnerships, which offer greater financial resilience200 over smaller providers.

Staff factors

It is widely recognised that the contracting arrangements alone are unlikely to deliver change without an explicit commitment to organisational development. 200,209 A culture of trust and collaboration (context), which takes time to develop, is needed to underpin the new ways of working (outcome) required of new contracting arrangements. A key theme to emerge from the literature is the mutually reinforcing relationship between structural interventions, for example, organisational forms (context) and governance structures (resource), and behavioural change [e.g. staff motivation (reasoning)] from staff engagement and development (resource), leading to a shift from siloed working (reasoning). 235 In addition to these cultural factors, specialist capacity and capability (resource), with clear roles and responsibilities in contract management and procurement,233,234 are underlined as critical to enabling new contracts. Investment to stimulate engagement and organisational development (resource) is considered imperative to improving the value of health care221 (outcome).

At an individual practitioner level, it has been argued that extrinsic incentives (resource) ‘may crowd out or reduce intrinsic motivation’204 (reasoning). 182,192,205 This reflects the logic of the value-based health-care movement203 that value should be defined by the customer (patient/local, community/clinicians). This might be facilitated by the co-production of incentives with clinicians and patients (resource) to focus performance around measures that are meaningful within local communities118 (outcome). There is also an argument for considering the appropriate ceiling216 and ‘strength’204 of incentives:

For example, a weak incentive to perform a valued behaviour may be more effective than a strong incentive to perform a behaviour that is not seen as important by the physician.

Reproduced from Flodgren et al. 204

System factors

Co-production (resource) encourages outcomes that are meaningful to patients, practitioners, local communities, providers and commissioners236 (outcome) which, in turn, depend on robust information11 (resource). Incomplete information, partly attributable to siloed working, was one of the issues identified in the analysis of the UnitingCare contract failure. 232 This approach suggests a focus on value rather than activity203 (context), presenting the challenge of translating a complex system of care into a set of measurable outcomes and indicators. 196 Accountability ‘as active inquiry over time: listening, asking questions and talking, as opposed to taking a snapshot’196,236 has the potential to build the relational behaviours (reasoning; see Chapter 6) that underpin the MCP vision.

Collaborative behaviours (outcome) and sustainable relationships (outcome) are dependent on trust (reasoning). 118,120,200,219,225 Contracts in the NHS depend on a level of flexibility (resource), which is challenging where relations are fragile or adversarial (context); Sanderson et al. 225 note the importance of relationships in building a shared sense of identity and values, which is particularly important in the context of agreeing opportunities for improvement or innovation and in determining what outcomes should be monitored.

Collaboration (reasoning) may be encouraged through a co-productive approach (resource), described in an ethnographic study196 as ‘social accounting [. . .] through the acknowledgement and enactment of multiple accountabilities and relationships’ – in recognition that changes to structure and process are not sufficient to facilitate a culture of shared accountability (outcome).

The effectiveness of incentives is explored by Mehrotra et al. 237 in their application of behavioural economics theory to the use of incentives. Although commissioners find the threat of penalties (resource) to be a useful incentive, not all of them act on the penalites219 (reasoning), citing the risk of impeding provider efforts to improve quality as one reason; thus, incentives may fail to influence individual practitioners’ behaviour118 (outcome). The unpredictability of behaviours also makes it difficult to ensure how measures influence individual practice. 212,238 Findings from The King’s Fund inquiry into the quality of general practice concluded that protected time (resource) is needed alongside incentives, to allow for reflection239,240 (reasoning). The co-production and testing of outcomes and incentives with stakeholders may help to ensure alignment with business plans and interagency strategies118,227 (context).

The readiness of organisations (context) to take on the role of lead provider has been questioned,118 with the suggestion that larger institutions providing secondary care may be in a more favourable position to take on the level of risk, particularly in areas that lack a strong GP federation. The need for investment in technology, information systems, governance, process re-engineering and cultural change (resource), and the resulting need for revenue streams to manage high set-up costs (context), may also favour larger institutions. 220 It has been suggested that implementation costs could outweigh potential savings. 118 Within a primary care context, there is a suggestion that the large multispecialty group practices in the USA are able to improve quality as a result of shared infrastructure; however, this has not been robustly tested. 241 It is likely that GP practices will vary in enthusiasm for existing contracts and readiness to give up existing contracts (reasoning) to participate in the partially or fully integrated model. 200 Readiness is dependent on a host of contextual factors,200 notably the strength of existing relationships across the health and care economy (context). It can be challenging to build collaborative system-wide relationships in a regulatory system built around the performance of separate organisations (context),200 and it is acknowledged that relationships will evolve over years rather than months. NHS Improvement and NHSE are jointly developing the Integrated Support and Assurance Process to support new care models. 230

The programme theory articulates the notion that new contracting and payment arrangements will incentivise shared accountability. However, it is clear from the evidence that consideration needs to be given as to how to encourage inclusivity (reasoning) through the co-production of outcomes (resource) and engagement in data collection196 (resource). An ethos of co-production combined with a long-term perspective (context) goes some way towards the vision of the culture of stewardship described by Gray242 (outcome).

Population health

Although there is some evidence of the impact of a higher level of participation (through co-production, volunteering, empowerment through self-management and social prescribing) on health and well-being outcomes, there are issues around heterogeneity, the quality of reporting, the lack of sufficient control comparison and short follow-up periods. 264,265 Evaluations have reported findings of varying quality, citing improvements in lifestyle changes, social isolation, reduced symptoms, health literacy, improved support for carers and improved independence. 33,132,265–272 From a service perspective, it has been suggested that co-production can help improve the quality of care. 273 Boyle and Harris274 suggest that participation in public services could help service users to develop a social network, building social cohesion, which may be key in supporting them to manage their condition(s). In a case study of an arts intervention for multisensory impairment, Vogelpoel and Jarrold270 hypothesise that interventions such as social-prescribing can help to address the needs of high-risk groups that are particularly hard to reach through traditional channels. However, the impact of poorly managed participation does not seem to be reported.

A key characteristic of the MCP model is the focus on social action. There is some evidence to suggest that involvement in public services, through volunteering or co-production, is associated with higher levels of health and well-being and can also have an impact on social outcomes within local communities. 256,275 A Cochrane systematic review276 examining community-driven interventions targeting racial and ethnic minority populations reported mixed results. Interventions led by community health workers were found to lead to changes in health status, but results were not consistent. Wider health and social care system interventions led to small changes in measures of health behaviour and health status.

These findings, although sometimes based on anecdotal evidence, suggest a need for a longer follow-up period, to understand how levels of engagement and health outcomes might change over time and the contextual factors that are likely to influence behaviours and, thus, the resilience of local communities.

Cost-effectiveness

The MCP focus on participation is based on a theory that service users who are supported and engaged will be better informed to make appropriate use of services and, in turn, better able to contribute to a mutually supportive network. However, there are considerable challenges in translating this to an impact on the cost-effectiveness of services, notably, the lack of robust measures and the difficulties of relating outcome measures to costs. 265,272,277 Few studies report costs, and even when costs are reported, it is often difficult to determine the cost of individual elements of services and interventions. 264 This is particularly challenging in services involving volunteers, whose unpaid contributions, training and support may be excluded from analysis. 132,264,269 Evidence suggests that higher engagement may lead to earlier intervention, which could help to prevent downstream costs. 272,274 A broader range of metrics may be more appropriate in assessing the impact of greater participation, including on admissions; workload in primary and community care; and the service-user and carer experience. 247

Several case studies suggest potential impacts on cost-effectiveness; for example, Boyle and Harris274 cite the Expert Patient programme, which reported a 7% reduction in GP consultations and 16% reduction in accident and emergency (A&E) visits. Partnership-working seems to be a key factor; large-scale self-management initiatives, involving national or regional authorities, were found to reduce admissions, deaths and time off work. 278 The Brighton and Hove social-prescribing pilot evaluation279 estimated potential savings of £1.36M per year of GP time. Several authors,33,280 however, caution that programmes often understate supplier-induced demand and overstate cost reductions, failing to account for set-up and operational costs appropriately:

The idea of self-management as a demand management strategy is also based on an assumption that utilisation behaviour is patient-led, when some aspects of utilisation (such as clinical attendance) are also a function of professional behaviours and may not be affected by changes in patients or carers.

Reproduced from Panagioti et al. 280

Dayson et al. 132 also point to the issue of scale, suggesting that, for some interventions, cost-effectiveness depends on a large critical mass of service users. There is also the potential to deliver improved cost-effectiveness through partnership-working with public health services; a recent study281 found a median return on investment (ROI) of public health interventions of ≈14.1. Health promotion interventions, including falls prevention and smoking cessation, delivered a median ROI of 2.2; health-care interventions, including disease management, offered a median ROI of 5.14. There may be potential for partnership-working across public services to address wider determinants of health; interventions were found to deliver a fivefold ROI. Partnership-working with the third sector may need specific consideration in terms of financial sustainability – the uncertainty around budgets, typical of the public sector, can have very far-reaching ramifications for small third-sector groups and for individual volunteers. 132

A case study282 reports on an instance of community engagement following a consultation to close a community hospital as a result of the recruitment and retention issues in primary care. The community, opposed to the closure, rallied to support GP recruitment, producing promotional materials to attract applicants to the area.

Patient experience

Reviews suggest that service users’ experience of participation in health care271 seems to be generally positive, reporting benefits relating to physical and emotional health and well-being, self-confidence, self-esteem, social relationships and their individual sense of empowerment. 267 However, the experience can prove to be negative for some service users. 267 Co-production is viewed by some service users as an opportunity to move ‘beyond token involvement and consultation towards more equitable power relations’ through ‘genuine collaboration’. 283

A greater role for third-sector and community groups may offer service users access to specialist knowledge, for example, on dementia,282 which could prove to be a valuable resource. 284 Where service users are actively engaged, they may also begin to contribute their knowledge; this is illustrated by del Castillo et al. ,282 in which patients with rare diseases collectively shared knowledge to support each other. Increased involvement of the third sector can also create and mobilise a body of volunteers, who can contribute to a more holistic care and support package, connecting service users to provide further support, complementing that provided through public services. 132,270 Farenden et al. 279 suggest that this collaboration adds value by spanning the gaps between services. This mutually supportive relationship is demonstrated in a Hospital to Home service,282 which recruited volunteers to support early discharge, reporting a statistically significant positive impact on patient experience, mood, anxiety levels, nutrition and hydration, and releasing nurses’ time to care. Similar findings are echoed by Maughan et al. ,33 who also cite ‘improved socialisation and occupational outcomes’.

Where service users are referred to social-prescribing programmes, the evidence suggests a mixed picture. Attendance varies considerably, with one review, which reported on two studies, reporting a 58% attendance rate and a 100% attendance rate, respectively. 265 An evaluation of a pilot reported that 101 out of 393 referred patients were unable to complete the package of support offered. 279 There are various possible reasons for low uptake suggested in various studies and evaluations, including inappropriate referrals, readiness to engage and logistical difficulties. High levels of satisfaction have been reported;265,279 an evaluation of a pilot service reported that service users felt that their needs were better understood. 279 A further theme to emerge from the research and evaluation is the sense of mutual support that service users can give and receive through initiatives such as social-prescribing. Vogelpoel and Jarrold270 describe the development of ‘supportive relationships […] where skills, practical advice and mutual support were shared’ between participants of an arts intervention for people with sensory impairment. Over time, this grew into a ‘genuine sense of community’, stemming from ‘a willingness to extend informal support to one another’.

Staff experience

The evidence base suggests that health-care workers have mixed experiences of initiatives such as social-prescribing, with some evaluations reporting relatively high levels of satisfaction,247,265,279 whereas others have reported a reluctance among primary care staff to get involved in non-medical aspects of care. 275 Potential reasons for this are explored in the literature, with suggestions that a medically oriented model of care (as opposed to a socially oriented model), as well as confidence in third-sector organisations, may influence perceptions of the value of social prescribing. 275

Conversely, volunteers and third-sector workers felt that they were able to spend more time with service users and could therefore offer a more intense level of help and support than those offered in primary care. 275 Pilot evaluations suggest that volunteer retention can remain relatively high when part of a formal structure. 279 In the Brighton and Hove pilot,279 volunteers gave an average of 7 hours per week during the first 12 months, totalling 5824 hours.

It has been suggested that a contract enabling collaboration across public sector and third-sector organisations could offer a balance of scale and infrastructure, with an expert understanding of communities, thus helping to create the conditions for a culture of improvement. 284 Some of the services that have been developed within the NHS have developed a ‘broker’ role, contracting with an umbrella third-sector organisation,132,269,279 offering a pragmatic option to commissioners and providers to manage relationships with multiple groups and organisations.

Unintended consequences

There are challenges in sustaining community involvement over time, and barriers such as the technical terminology used routinely in the NHS can act as a barrier to involvement. 285 There is some evidence of uneven representation or engagement across demographic groups, particularly from ethnic minorities and those with low levels of literacy. 256,286,287 Similar findings have emerged from initiatives aimed at individual service users, for example, social-prescribing. 132 This presents the risk of widening, rather than resolving, health inequalities. A universal community engagement intervention (covering both relatively advantaged and disadvantaged groups) risks increasing health inequalities. 264,266 Inequalities may also be widened when social networks benefit members of one particular group, but exclude others outside the group from sharing those resources. 264 As well as difficulties in recruiting patients and carers, it may also be difficult to retain them through the different stages of projects; for example, patients who are no longer using the service may decline further participation. 271

There is some debate regarding the extent to which service users may wish to engage in decisions about their care or about local services, with clear differences in expectations affecting willingness and capacity to engage in self-management and co-design. 261,288 A case study from the UK reports some reluctance to participate, with some patients stating that ‘it was the clinician’s role to manage their care’. 288 This is echoed in a recent review:

Not all patients have the desire or capacity to be active participants in coproducing their healthcare service. Sometimes patients are sick, and need a health professional to relieve a burden rather than enable self-care.

Batalden et al. 261

Howieson and Fenwick289 use the example of a patient’s right to choose their general practice to highlight that patient choice is generally not exercised. They also argue that the assumption that the public – with whom the NHS would be in collaboration – would be reasonable and open-minded and would have (and take) a holistic view of the health service and understand and appreciate the best way to allocate resources for health provision is very much open to debate.

Greater engagement may be experienced as burdensome by some, with the physical and mental health benefits of community engagement diluted by the stresses and strains associated with involvement for some participants. 267 Sorting helpful from available information is complex; diminishing the value of professional expertise by transferring care responsibility to patients and families could result in bad health outcomes. 261

Implementing community engagement among disadvantaged communities has highlighted complex challenges, such as poor health system infrastructure and service delivery, and poor staffing and resources. 266 Furthermore, these complex challenges often result in unmet community needs, causing community partners and study participants to feel dispirited, and compromise the potential for community engagement. 266 Clarke et al. 271 report that a lack of progress in community engagement projects can lead to frustration for both staff and patients. Programmes are likely to be considered ‘tokenistic’ in nature when suggestions are not acted upon,267 and compromise the trust and relationship-building purpose of community-engagement initiatives, resulting in poor individual health outcomes. 266

The role of social movements and third-sector organisations and groups in health care is growing. However, tensions can emerge from working across very different cultures and governance structures. The agile and dynamic nature of social movements can clash with the more structured approach of publicly funded services, and del Castillo et al. 282 recommend a considered approach with realistic expectations. Voluntary and community organisations, although able to mobilise volunteers, are limited by resource constraints. In the example of one social-prescribing pilot, it became clear that NHS-led time scales for roll-out across practices had not factored in time to cascade training and guidance around the availability of volunteers and the capacity of individual practices. 279

Where there is willingness to participate, there is potential for confusion regarding this new more participatory role for service users, as reported in a 2014 systematic review,290 which highlighted contradictory information and conflicting instructions about individuals’ needs, priorities and management strategies.

Self-management and social-prescribing programmes are based on implicit assumptions that improving health literacy and self-confidence will lead to long-term behaviour change, increasing independence and reducing service utilisation. However, as pointed out in a NIHR-funded systematic review,280 few trials and evaluations measure outcomes beyond 1 year, so the extent of ongoing support needed remains uncertain. This leads to the question of ‘whether or not the reductions in utilisation achieved are significantly greater than the service input required to maintain gains in knowledge, skills and confidence’. 280

Furthermore, Batalden et al. 261 highlighted experience from shared decision-making, demonstrating that, even after training, many patients and professionals in Scotland inconsistently apply newfound skills and orientation.

Service-user participation can help to identify unmet needs and gaps in service provision. In the short term, this may result in higher costs. 247 It is debatable as to the extent to which voluntary and community services have the capacity to offer help to people who do not meet thresholds for formal care, but who may benefit from social contact and low-level practical help. Conversely, service gaps were also thought to contribute to the inappropriate referral of service users with highly complex needs to a social-prescribing pilot: ‘some clients fall through the gaps’. 279 Service evaluations report on the impact of funding cuts and the uncertainty of future funding on the capacity of voluntary and community services, highlighting the risk of delays in agreeing contracts. 279 The impact on volunteers and service users can be significant, with longer waiting lists for support, reduced or closed services and low retention of volunteers being reported. 279 There is a risk that service users, who may be vulnerable, are left feeling ‘abandoned’291 and that other services, such as general practice, are placed under additional pressure. 279 In a competitive market, staff may be more likely to move on to new roles in the event of continued uncertainty regarding future funding. 279

There is also the potential to introduce unnecessary duplication and waste if roles and responsibilities are not explicitly clear – for example, various referral routes to social-prescribing support. 132

Service users

Patient mutuality and co-ownership in health-care services (context) require patients to be appropriately trained [e.g. in health literacy287,292 and self-management261 skills (resources)], improving their self-efficacy and confidence to manage their care292 (reasoning). More widely, low awareness of long-term conditions (context) could be improved by a stronger narrative (resource), informing ‘patient help-seeking behaviour and public perceptions of need’. 278 This can help individual service users, who are interested in getting more actively involved in decisions about their care, but is dependent on an understanding of the service user’s context (e.g. carer/family support)275 and a shared understanding291 (reasoning), which can help to develop confidence and personal resilience (reasoning):

Self-management was conceptualised by health care professionals as incorporating both a biomedical model of compliance and individual responsibility. Lay people understood self-management in wider terms, reflecting biomedical, psychological and social domains and different expectations of responsibility.

Sadler et al. 291

The willingness of service users to participate in social-prescribing programmes265,275 depends on an individual’s sense of confidence265,270 (reasoning), the relevance of the activities provided (reasoning),265 the individual’s literacy (context) and logistical issues, such as travel270 (resource). Facilitated support (resource) has been shown to be more effective in assisting people with wider social needs than simple signposting, and in one case study, maintaining follow-up (resource) ‘enabled participants to implement important choice-making and autonomy building processes’ (reasoning). 270 Greater confidence (reasoning) can help service users to make connections with their peers and to explore options for support (outcome). 270

Where community engagement is introduced (context), research should be conducted264,266 (resource) to build a clear understanding of the community and its resources293,294 (reasoning), which would then enable appropriate community resources to be mobilised264,266 (outcome). Coulter294 emphasises the importance of starting with a very clear sense of the local community, centred on the ability to describe target stakeholders and the value of engagement.

Through the successful mobilisation of appropriate community resources (context), the programme can be carefully designed, with community involvement and validation where possible264,266 (resource). Services can therefore be tailored to reflect community representation, not tokenism, such as through the cultural adaptation of programmes264,266,294 (resource); community resources, such as community leaders, outreach workers and lay workers,263,266,293 are used to facilitate and promote opportunities to relevant participants287 (resources). This increased awareness of authentic and credible programmes can enable people to identify with a community264,293 (reasoning), helping to remove barriers to participation and to empower citizens to become involved266,287,293 (outcomes).

The design of community engagement programmes (context) should include clearly defined roles and responsibilities,271 and training and support. 250,264,278,287,291,294 Methods of participation that address concerns and support requirements,287 facilitated or delivered by community representatives,264 can be facilitative. In turn, these resources can build the confidence of service users so that they become more aware of their contribution273 and expectations283,287 (reasoning), thus encouraging involvement263,267,287 (outcome).

To encourage service-user participation, incentives, such as financial rewards or recognition,264,273,283,287,294 support with practical issues such as time and transport287 (resource) or meetings to suit community members’ needs can help to engender a sense of trust and respect (reasoning) between patients and health professionals287,295 (resources). Methods for participation should be carefully selected so that they recognise patient abilities261,267 and give consideration to practical and logistical problems. 271,294 Recognising that service users differ in their preferences for involvement in decision-making, there should be an offer of alternative forms of participation (resource). 263,264,267,287 Carers, in particular, may need to be consulted, as there is some evidence to suggest that programmes of self-management and social-prescribing can assume a certain level of capacity among carers and can also cause confusion regarding responsibilities. 291 Delivery should be culturally relevant264,266 and goals should be realistic. 263,273 Communications should be clear and transparent (resource),264,294 using bidirectional communication when necessary,295 avoiding overloading service users,263,267 and making involvement an accessible and enjoyable experience264 (reasoning), thus encouraging commitment and retention266,283 (outcome):

. . . individuals are less likely to find community engagement a positive experience where consultation is the main method employed by professionals and no real power to effect change is ceded to community members.

Reproduced from Attree et al. 267

Staff factors

Where leaders have identified participation in decision-making and co-production of services as a priority (context), a clear case for change (resource)273,294 championed by confident leaders with the energy to innovate and motivate (reasoning)296 (resource) has been shown to facilitate cultural change (outcome):263,289

Some of the key challenges in co-production within health care are the different education, culture, philosophy and interaction models of different professional groups.

Reproduced from Philips and Morgan263

Practically, support for staff (resource) can encourage new ways of working (outcome) – for example, training using patient stories,261,263,265 time to develop and utilise skills,261,263,264,271,287,289,294 supportive tools and infrastructure,261,273,287,289 feedback33 and appropriate guidance274 – can motivate staff (reasoning) through understanding the case for new ways of working. 249,273 This also applies where volunteers are engaged in the delivery of services; evaluations demonstrate the importance of mentoring support, feedback and information-sharing (resource), to help volunteers to feel like a part of the team (reasoning). 33,275,279 A change in mindset to one that recognises service users as equal partners261,263,274,289 and diminishes the fear of losing status and control273,289 can help to develop confidence in ‘sharing power and accepting user expertise’287 (reasoning). Underpinning this is the challenge to build a sense of mutual trust266,275 (reasoning) across different cultures:

Co-production calls, instead, for a slower pace of sharing and producing knowledge and for open-ended, experimental forms of learning, critical thinking, and working together in health care and research.

Reproduced from Filipe et al. 283

Given that some local health economies may experience a high turnover of staff, there needs to be an ongoing programme of training and development to embed and sustain more participative decision-making. 278

Relevant policy and governance structures (resource) will need to support new ways of working, articulating the shared purpose, values and principles that emerge from joint working across the local health and care system. 249,261,263,271,273,286,292 On a practical level, this will help to clarify, for example, the referral criteria for social-prescribing support. 33,270 This can provide staff with clarity of purpose and expectations (reasoning), and provides assurance for system leaders through regular reviews of progress and impact. 263,274,283 This can encourage sustainable partnerships and collectively agreed, realistic outcomes (outcome).

System factors

In a MCP context, a co-production approach is intended to achieve a more participative relationship with individual service users and community groups, contributing to a system-wide impact. 250,261 Within public sector organisations, working with service users to understand their experiences (resource) has been shown to create a sense of urgency among staff to act, and served to legitimise (reasoning) the realisation of quality improvements to higher management bodies (reasoning). 297 To support greater levels of participation by service users in decisions about their own care and about services (outcome), a shift in the role of staff from ‘fixers to facilitators’ is required. 274 This shift balances power towards service users (reasoning) through front-line professionals who are trained (resource) and subsequently empowered (reasoning) to take on these new roles. 250 Management of expectations will be an important consideration in bringing together people with different motivations, cultural backgrounds and prior experience, coupled with realistic goals (resource):

. . . co-productive experiments are best seen as generative processes that are less about delivering predictable impacts and outputs and more about developing new communities, interactions, practices, and different modes of knowledge and value production.

Reproduced from Ryder268

Poortinga256 references the five capital model, which emphasises the importance of health services working with other public services and community groups to address the wider determinants of health:

. . . healthy neighbourhoods have a balanced combination of human capital (e.g. skills and education), social capital (e.g. social networks), built capital (e.g. access to amenities), natural capital (e.g. access to green space), and economic capital (e.g. income) resources.

Reproduced from Poortinga256

Social movements may be a route to encouraging open innovation to address specific community needs,282 but there is a risk that the bureaucracy inherent within the public sector can stifle smaller and more agile groups, which suggests a need for strong relationships to manage potential tensions in relation to priorities, governance and culture. 282 A useful early step may be to map community ‘asset’ or resources (resource),33,269 potentially via an umbrella third-sector organisation acting as a broker. 269,279

Although trust is often mentioned as important to building and maintaining strong relationships, there is also an argument that ‘distrust is not necessarily detrimental to community involvement and wellbeing’, helping to engender greater ‘political involvement’ or ‘critical trust’, encouraging individuals to get involved in decisions about local public services. 256

When working across multiagency sectors (context), time and space for learning,261,263,264,274,283,294 role definition,283,291 shared planning and values,261,284,294 external facilitators271,295 and administrative support264,295(resources) can be important enablers. Opportunities for face-to-face contact (resource), to build social connections, can help to develop trust (reasoning) across diverse services and groups;275 for example, across primary care and community groups, this can help to build the legitimacy of groups that may be unfamiliar to staff. Training (resource) may be needed at multiple levels of the local health-care system, including at the board level. 284 This can support the development of collaborative relationships and trust261,263,283 through a shared understanding and recognition of each other’s expertise and values. 250,261,283 External support (resource) can help to resolve tensions and conflicts,295 allowing critical inquiry and the confrontation of ideas283 (reasoning). This can help to create an environment of improved power-sharing250,266,271,274 and collective decision-making264 (outcome).

Outcomes relating to the quadruple aim

There is some evidence to suggest that integrated care pathways could contribute to improved health outcomes. A disease-management programme in Germany has established disease-specific objectives with defined treatment goals and specific criteria for referrals to secondary care. The programme has an emphasis on patient education and self-management. Patients choose a physician, who co-ordinates their treatment; the framework clearly sets out when specialists should be involved in patient care to avoid gaps in care provision. 303,308 Further afield, the shared guidelines and standard procedures of the Mayo clinic are considered to contribute to lower complications and improved clinical outcomes. 303,309 An end-to-end pathway may enable the focus on prevention common to many of the MCP logic models; one example suggests that integrated pathways may contribute to a reduced incidence of falls, following knee arthroplasty. 302 Although it is difficult to attribute population health outcomes to pathways, theoretically, integrated pathways may help to stimulate an environment that encourages communication, collaboration and peer review, thus leading to improvements in quality and safety. 303 Audit and feedback can strengthen the impact of integrated pathways, identifying what works in improving health outcomes and identifying problems that may indicate potential improvements or amendments to the pathway. 303,306

Integrated pathways, through reducing waste and duplication, could contribute to improved cost-effectiveness and lower unnecessary utilisation. The Nuka system of care, run by the Southcentral Foundation in Alaska, is credited with achieving improved quality for a lower cost, demonstrated by a decrease in unscheduled care alongside improved patient and staff satisfaction. 139 Integrated pathways have been used to enable a shift of care from hospital to community settings, demonstrated by the Veterans Home Administration, which measured fewer hospital bed-days and overall admissions. 303,310,311 Integrated pathways may also lay the foundations to introduce technology to support the delivery of care. Shared health records have a role in driving patient safety and adherence to guidance, supporting smoother discharge and transition between teams or services. 301 Telecare and telehealth can support more home-based care through remote monitoring and, in some contexts, has been found to contribute to reduced bed-days and increased patient satisfaction. 303,312 Opportunities exist to embed technology within pathways; for example, patient deterioration can trigger a responsive multidisciplinary package of care to avoid further decline, based on an agreed care plan facilitated by record-sharing. 303 Some countries have initiated condition-specific approaches, such as disease-management programmes in Germany, Sweden’s chains of care and Scotland’s managed clinical networks, finding the most impact on patients with long-term conditions. All of these approaches involve elements of integrated care, recognising the importance of co-ordinating care and working across systems, and creating pathways across traditional ‘single-disease’ pathways. 303

Mechanisms

A virtuous cycle, akin to the theory of the MCP model, could operate where clinicians and managers work more effectively across boundaries to attain financial and quality goals, which then improve cost-effectiveness, which can support investment in quality initiatives. 302 In such an environment, integrated care pathways can act as a focal point for clarifying and articulating the roles, responsibilities, processes and systems to underpin co-ordinated care, shared decision-making and care-planning. 302,303 This co-ordination can provide the stimulus for information-sharing, supporting patient-centred care. Where the design of integrated care pathways is based on engagement with patients and practitioners, patients may feel empowered and supported to maintain independence for longer. 102,301,303 Such engagement can also be instrumental in enabling access to a more holistic approach to care. 301 From a patient perspective, an integrated pathway can translate into a single point of entry, a single assessment and a personalised care plan, delivered by a MDT (see Chapter 7, M2: multidisciplinary teams provide continuity for patients with long-term conditions/complex needs), which can help to reduce the burden on carers and address unmet needs. 301,303 This co-ordinated standardisation can, in theory, address unwarranted variation, drive up quality and improve equity of access. 302 Patient satisfaction can be strengthened by the routine collection of patient-reported outcomes and regular patient feedback. 303,306

The process of developing integrated care pathways to incorporate shared values and goals can support the integration of teams and influence organisational culture and staff motivation, making it more likely that new processes will be embedded in routine practice. 301,304 Closer working and standard pathways can help to identify those patients with more complex needs who need specialist secondary care, and those who are better managed in the community. 303 Pathways informed by formal needs analysis – for example, Joint Strategic Needs Assessments – and community engagement can help to align the investment of resources within local communities,139,300,305 factoring in the wider determinants of health and significant risk factors specific to local communities. 102 This, in turn, can help to promote holistic and person-centred care and a sense of ‘shared ownership’, shared goals and incentives to address prevention. 139,303

Early engagement of staff will help to ensure that pathways are comprehensive and realistic,299 and can ‘offer an opportunity for professionals to renew their implicit rules for operation and shape collective meanings and objectives’. 302 Affording more opportunities for communication across teams302 can contribute to ‘an environment in which excellence in clinical care will flourish’. 307 Standardisation facilitates monitoring and evaluation;302 supported by transparent reporting, this can be a catalyst for quality improvement when larger teams are better able to exploit performance data for quality improvement, and to address wider issues, such as prevention and health promotion. 303 Benchmarking, using national audits and other tools, can help to demonstrate the impacts of changes to pathways and to highlight issues relating to demand and capacity. A culture of personal professional responsibility and innovation among health-care professionals can be encouraged through more autonomy. 301,304,307 Multidisciplinary training and development, with a focus on communication and trust, can encourage more collaborative working. 299,301,304,307 When individuals understand their contribution and that of others, pathways are more likely to be ‘owned’ by staff and successfully embedded in practice. 139,302 Problems can be identified and resolved collectively,301 thus developing a learning culture and a ‘continuous quality improvement “mind-set” ’. 307 Where a history of co-operation and trust pre-exists, this may be a smoother process; conversely, where relations are more adversarial, behavioural change is likely to require significantly more investment. 139,303 Cultural barriers, such as competitive funding, poor relationships between senior leaders and risk aversion can stifle innovation and motivation. 301 Local barriers and facilitators need to be considered, for example, around data-sharing. 102

Although engagement is to be encouraged, there is evidence to suggest that the design and testing of integrated care pathways may be better managed by small focused teams, which are sufficiently agile to make the decisions needed to progress, but also sufficiently connected to ensure staff commitment and confidence. 301 Pathways alone are unlikely to deliver significant transformation – delivery may be contingent on new or expanded roles within primary and community settings. 139 In areas where there are long-standing recruitment issues, integrated pathways could be hindered by key staff shortages that, without workforce planning and development, could serve to undermine morale. 302 Multidisciplinary working will need to be supported by organisational development, to tackle cultural aspects, contractual incentives to influence behaviours and operational logistics, such as systems to enable shared data,102,301 as demonstrated in Geisenger Health. 303 Information systems, in particular, may help to facilitate timely and efficient information flow, supporting knowledge-sharing, the spread of good practice, consistent standards and communication between generalists and specialists. 299,303 Shared governance can promote care standards, patient safety and professional competencies;301 conversely, instability can severely constrain the implementation of integrated pathways. 102,299,303

Outcomes relating to the quadruple aim

There is evidence to suggest that an integrated approach to care, via MDTs, is associated with improved outcomes for patients with long-term conditions or complex needs;340 however, the majority of studies focus on single conditions – for example, reduced glycated haemoglobin values/improved glycaemic control in diabetes,335,344 blood pressure control,328 treatment compliance,328,344 reduced depression and anxiety324,327,334,352 – and it is unclear if benefits are sustained over time. As with any package of interventions, attribution is a challenge and, typically, MDTs are implemented alongside programmes of self-management, case management and enhanced pathways. 340 It has been suggested that the ‘supportive network of professionals and peers’352 contributes to improved health and well-being. Additional evidence suggests that MDTs can facilitate disease management and encourage self-management,334 reflecting the common features of MDT working, including case management,351 care-planning,351 the promotion of self-management and medication adherence. 325,334,338,353

Conversely, there seems to be little or no impact of MDT-working on patient-reported outcomes, such as activities of daily living,338,353 functioning,322 quality of life322,325 and disability. 325 There may be limited benefits for older people with multimorbidity or frailty. 351 The extensivist model (based on the Care More model in the USA)321 is designed to provide more intensive support for older patients with very complex needs, and is being implemented as part of the MCP model in Fylde Coast. It can be challenging to measure the effectiveness of MDTs, notably the appropriate follow-up period within which to measure impact and the baseline (with a ‘ceiling effect’ of limited impact, whereby primary care was already well organised). 322

Patients with multiple conditions may find it challenging to manage the complexity of their conditions and to prioritise needs, particularly within the context of deprivation or social isolation. 336 It can be difficult to navigate the health and care system, and care packages developed by MDTs could be a critical element in improving support to patients and carers, as demonstrated in Torbay. 323 Patient involvement may be facilitated by certain roles within MDTs; one qualitative study354 reported perceived improvements in care co-ordination and problem-solving, following the inclusion of a nurse practitioner within the MDT. Trust was also greater, possibly resulting from more time spent with patients and more opportunities for communication. The value of a MDT is in the opportunity for interdisciplinary diversity, with multidisciplinary participation emphasised as a key benefit of MDT meetings,337 supporting a range of social functions such as peer support, team-building and bonding,337 which can, in turn, have an impact on health outcomes. 350

Several studies report reductions in service utilisation associated with integrated care via MDTs (hospital admission,326,334,355 A&E attendance,339 readmission341 and length of stay)326 with a potentially higher impact in those ‘with at least moderate disease’. 326 Few cost-effectiveness studies explore the effect of MDTs. Of those studies that do report costs, few have reported savings. 344 One cost-effectiveness study estimated330 a mean cost per QALY of £14,248, arguing that collaborative care is expected to be less costly and more effective.

Mechanisms

The professional composition of MDTs is variable and typically comprises three to eight members, with GPs and nurses most frequently represented. The inclusion of voluntary sector organisations (VSOs) in MDTs has recently increased, often to support self-management and more holistic care. 357 Involving other public services, such as housing, enables a care plan that addresses the wider determinants of health, and supports patient-focused goals to tackle social isolation. 358 Specialist nurse roles, with arguably more capacity than GPs, can help to support patient education329 by spending time with patients to personalise information. This could be of particular benefit to patients who may struggle with more generic approaches, for such reasons as language, learning difficulties and cognitive decline. The ‘drip feed’ of information can support a continuous relationship. 329 For patients with multimorbidities, decisions on the trade-offs between the benefits and harms of treatments can be supported by specialist roles. 329 Conversely, it has been argued that patients with very complex needs might need more support from GPs. 331

In bringing together multiple services, the role of care co-ordinator can be critical in tackling fragmentation,359 supporting relational continuity between the services and the patient; informational continuity between services and with the patient; ensuring accountability to the patient; and acting as the patient’s advocate to ensure patient-centred care. Nurses without advanced qualifications or advanced training have reported feeling ill equipped to take on the care co-ordination role, highlighting the importance of support (information systems, monitoring processes, guidelines), as well as training and development. 347

Space and time for informal communication and shared communication methods are identified as enablers,346 underpinning access to, and the exchange of, essential information and ideas. MDT working can be facilitated by a shared understanding and vision of integrated care,340,346 translated into clear, agreed objectives. 360 Operationally, explicit processes338,345,346 can help to clarify expectations; for example, mandating attendance in job descriptions, regular feedback on team performance, provision of administrative support and sharing of information with other practitioners involved in pathways. An absence of such clarity can have an impact on the implementation of decisions. 337 Organisational development can support the development of shared goals, values and outcome measures. 239

To effectively manage multimorbidity, interventions may offer greater benefit if targeted at specific combinations of common conditions or specific problems facing comorbid patients,332 highlighting the role of dedicated case managers. 327,333 Risk stratification and predictive modelling are often used to support case-finding,361 and several recent studies present alternative methods to predict future utilisation. 319,320 Sharing and access to patient records is critical in MDTs working across organisational boundaries to support clinical decisions and more holistic support. 323,343,345 The use of technology is variable,337 challenging information-sharing within current systems. The experience of the Buurtzorg model suggests that technology can underpin timely communication, operational transparency, knowledge-sharing and quality improvement, through team access to audit and performance data. 343,345

System leadership,346,362 with its attributes of building a common vision and engaging with colleagues and partners to build commitment, may act as a trigger to develop the collaborative culture that is key to MDTs. Maintaining a shared vision that is clearly communicated, alongside an explicit commitment to quality improvement through benchmarking and peer review, are considered to be essential to creating the environment for integrated working. 337

The vanguards seem to recognise that integrated working is unlikely to emerge simply by forming teams. Team development is needed at all levels of the system – not just at the executive level – and investment in organisational development is needed to underpin structural and cultural change. 360,363 Without explicit development, teams can lapse into ‘domain thinking’. 348 Specific training and development can help to define a shared language and strengthen relationships. 348 Development may involve coaching, leadership training, 360-degree feedback, personal development and action-learning sets. 364

Although not a multidisciplinary model, the underlying philosophy of the Buurtzorg model,55 with its emphasis on autonomy and empowerment, may offer valuable learning. Care is delivered by self-organising teams, typically comprising 8 to 12 nurses or health-care assistants. Teams are given the freedom to develop their own models for joint working with other professions. Value-based recruitment has created a virtuous cycle, in which the high levels of patient and staff satisfaction have attracted highly motivated individuals who perpetuate the cycle. The Buurtzorg mission is based on a shared sense of purpose, which ‘appeals to professional pride’, aimed at stimulating a culture of innovation and improvement. Teams are embedded within communities, developing detailed knowledge of their ‘clients’, their families and community resources; this informs problem-solving, and, in turn, helps to build the resilience of patients, families and communities. 343

The mix of professional backgrounds in MDTs can enable ‘participants to transcend their own professional roles and routines, leading to learning and a more collaborative environment’. 345 Support for reflection, innovation, open communication and supportive colleagues can create an environment in which participants feel part of the team, supporting a belief in interprofessional care. 345 On a practical level, a nominated lead to co-ordinate and chair meetings, supported by administrative resource, can help to nurture the team as it develops,345,348 with appropriate personal development. This need for development is emphasised by West et al. ,356 who note that leadership skills are unlikely to emerge from clinical experience alone and will need nurturing through training and support.

Outcomes relating to the quadruple aim

Findings from the UK National Staff Survey suggest that staff engagement is the best overall predictor of NHS organisations’ outcomes (care quality and financial performance; patient satisfaction; staff absenteeism; and health and well-being). 374 The impact of staff experience on the quality and safety of care has been explored, following the Francis inquiry380 and the Berwick report,381 with suggestions that strong team-working and clear aims among staff can contribute to patient outcomes. 382,383 Engagement has also been linked with organisational productivity, quality of patient care, lower mortality rates and better financial performance. 383 In one study, practices demonstrating higher leadership scores were more likely to implement clinical improvements. 368

Leadership development, through training and increasing workforce capabilities, is seen as a particularly critical element of workforce development. Staff on leadership programmes, such as the Leadership Challenge programme, have reported improved organisational impact, including raised awareness of costs and cost-savings, as well as improved patient care. 364 The experience of staff is often linked to patient satisfaction. 350,360,374,384,385 Staff views of leaders are strongly related to patients’ perceptions of the quality of care. Where leadership enables effective team-working, health-care delivery is more effective, and organisations exhibit higher levels of innovation and patient satisfaction. 374 Team working, in particular, has been found to be a critical ingredient in high-quality care, particularly in relation to care for long-term conditions. 365

Higher levels of support for staff from managers has been associated with higher numbers of patients reporting better care. 350,374 Greater engagement has also been linked to safer patient outcomes and significant associations with patient satisfaction, patient mortality and infection rates. 385 When staff experience is positive, for example when they feel that they have ‘meaningful roles’, individual staff outcomes (e.g. job satisfaction) and organisational outcomes (e.g. absenteeism and staff turnover) have tended to improve. 350 Conversely, higher turnover may exist where working environments are perceived as negative, because of bullying, harassment and aggression. 350 A learning organisational culture, exemplified in many high-performing services, has been linked with organisational commitment, job satisfaction, innovation and performance. 366 Linking with the theory components M2 and M4, learning organisations typically demonstrate an effective use of data to reflect and learn from performance at the individual, team and organisation levels. 371

Mechanisms

Organisational development could usefully focus on empowering teams to explore opportunities to improve and innovate. Positive staff experience has been associated with the ability to contribute towards service and quality improvement at work and strong communication. 350 Clarity on the opportunities for involvement and engagement in change has been identified as ‘an important contextual element’. 375 Multidisciplinary learning, potentially working via networks, can help to cement the vision of integrated working365,378 and develop the adaptive skills needed to implement and evaluate service transformation. Leadership and quality improvement skills in primary care have been identified as a priority; team dynamics may also be influenced by value-based recruitment. 365

Leadership development often focuses on processes, targeting more senior or medical roles, thereby missing the opportunity to develop skills more widely. 371 Various groups within health care, notably primary care staff, have not engaged as much with leadership development, possibly because of capacity issues. 365 This points to an opportunity to engage more of the workforce in leadership development. A ‘learning organisation’ culture can be encouraged through processes and systems to support the sharing of knowledge, data and feedback. 364 With leadership support, including time for reflection, teams can be trained and empowered to audit and benchmark their performance, analyse problems and share knowledge. 364–366 For clinicians, autonomy may be a particularly important factor, enabling more collaborative and innovative behaviours. 379,387 Evidence suggests that learning and development should be seen as an ongoing commitment; incremental change to encourage continuous learning and improvement364,375 has been shown to sustain cultural change.

High levels of motivation are associated with higher levels of productivity, and tend to be reflected in positive relationships with patients/service users. 367 Motivation is influenced by levels of engagement and the working environment; continuous high pressure and evidence of discrimination are associated with lower levels of motivation and patient satisfaction. 350,382 A positive environment may be encouraged through clear goals382 and a systematic approach to learning and development. 366 The attributes of senior leaders are an important influence here, with a key role to play in inspiring collaborative working through fair-mindedness, the empowerment of staff, vision and organisational awareness:379,386 ‘capacity building and effective leadership development implies social systems to help build commitments among members of a community’. 386

Integrated care depends on a system perspective, and much has been written about system or collective leadership,362,369,370,374 where there is a ‘notional dispersal and sharing of leadership’. 379 Distributed leadership models, which allocate time and resources to staff engagement, can ‘create a shared sense of energy’,388,389 activating ‘a learner’s sense of psychological safety’. Involving people in projects can help to build the credibility of the change programme, demonstrating the feasibility of change and encouraging uptake, spread and sustainability. 366,375

Interpersonal skills are a critical element in creating ‘the foundation for making good ideas into tangible improvements’;372 this is especially the case with complex change, for which senior clinicians will need to influence others to contribute expertise and ideas. A longer-term view is needed to build the relationships that are needed to support complex change, and this will also help to develop insight to spot opportunities in an environment of ambiguity and uncertainty. 372 A key priority is to develop primary care staff, at all levels, to support quality improvement and innovation,365,376,377 with a clear role for mid-level managers in creating and sustaining change. 368,378

The pace and scale of change involved in new models has the potential to demotivate staff. Although evidence suggests that employing a range of staff can extend the current skill mix and improve working life, an accompanying risk of stress and anxiety highlights the importance of mutual support, which can be encouraged by team development. 365,367 It is important to staff that they feel that their roles make a difference to patient care, which increases their trust in their employer; a sense of shared values and a feeling of belonging can build mutual respect, which is particularly important in potentially stressful contexts. From these perceptions of trust and respect, motivation and satisfaction can increase, demonstrated by lower levels of staff turnover. Similarly, when staff feel like they are under less pressure, training and appraisals seem to increase in frequency. Personal experience is strongly affected by instances of negative behaviours – bullying, harassment, discrimination and aggression – particularly, when it is perceived that it is inadequately addressed, leading to low morale, and potentially increased absence, which affects patient satisfaction. 350,367

Outcomes relating to the quadruple aim

In contrast to other programme theory components, the achievement of transformational change does not map directly to the quadruple aim. Instead, it is best seen as an intermediate outcome by which the medium- to long-term outcomes articulated by the quadruple aim might be achieved. An alternative view might locate transformational change as a subsequent context, within which the achievement of outcomes might be demonstrated through the monitoring of outcomes and the creation of feedback mechanisms.

Mechanisms

Outcomes relating to the quadruple aim

The importance of information to assist with public health is well recognised. At the local level, data are used to target interventions and monitor population health. At the national level, information is used to make informed decisions about prioritisation, planning and commissioning. At the global level, information assists with the burden of disease and tackling global health threats. Routinely collected health data are not only used by public health agencies, but also for academic research and the development of technology. Large population-level data can help to accurately identify social inequities, and target specific groups and geographical and regional challenges, which can be addressed through evidence-based health interventions. 454 The Symphony Project in south Somerset is an example of the use of extensive data sets for the purpose of improving the integration of care. Data on utilisation, costs and patient characteristics have been combined with local intelligence to identify groups that would benefit the most from proactive care. 450 To this end, examples also exist in which segmentation analysis combined with ‘big data’ has divided patient populations into distinct groups, for needs-based targeting of care models and intervention programs. 455

Southampton City CCG is an example of an organisation that has used data to assist with the reduction of unwarranted variation in health care. NHS RightCare provided data that highlighted the key priority areas for the reduction of variation, which could increase the value of health care. The CCG used these data to target areas of concern for spend and quality indicators, which subsequently led to the development of an implementation plan to improve the management of gastrointestinal patients. 451 A UK study suggests that information on variations in costs, activity and outcome could be better exploited, perhaps indicating a need for tools to help understand the causes of variation in health-care delivery. 456 Despite the availability of data, however, data on disease surveillance, cause-specific mortality, intervention coverage and other vital statistics remain underused data sources,454 suggesting missed opportunities in discovery and innovation.

The potential for effective information-sharing to achieve cost-savings remains inconclusive. One systematic review concluded that there is sufficient evidence to suggest a reduction in costs through the use of electronic health records,457 whereas another study suggests that, although such cost-savings are achievable over a long period of time, there is a high degree of variation dependent on the practice setting. 458 The high costs of implementation can also function as a barrier to the uptake of electronic health records. 459 Such costs may also not be recovered. 460 Support for the cost-effectiveness of shared electronic health records remains limited and conflicting. 459

The government-anticipated ‘big data’ and the use of large data sets for the purposes of analysis have been promoted as potentially reducing costs, with an estimated £20B of benefit over a 5-year period for the central government. 449 This estimate was based on the understanding that the cost-effectiveness of services is, to some degree, dependent upon data transparency and effective data-sharing. 449 To this end, linked data have proved to be useful in exploring the utilisation and costs of services. Patient-level linked data have been used to explore the NHS and social care costs of the care pathway associated with elderly patients recovering from a fall in Torbay, Devon, UK. In this case, cost data patterns proved to be helpful in identifying and targeting patients to fall prevention services. 448

Information-sharing can help to improve patient–doctor interaction, including through the use of telephone, e-mail and web-based consultations, with tools such as webGP and askmyGP enabling patients to share information with clinicians to support clinical decision-making. 461 Modality, a super-partnership in Birmingham and Sandwell, and one of the MCP vanguards, operates a remote hub to support triage, self-management and same-day consultations, enabling 65% of health concerns to be dealt with remotely. 461 A similar system, based at south Somerset’s Symphony, one of the PACS vanguards, has established a web platform to enable patients to access their care plan, messaging capabilities (such as Skype™; Microsoft Corporation, Redmond, WA, USA), test results, information about their condition and telehealth functionality. The platform enables a ‘single view of clinical activity between the [integrated] hub and practices’, directly linked with the GP system. From an information governance perspective, such systems can also clarify permissions for sharing information with carers and family members, which can be useful in the event of a crisis or emergency or when the patient is reluctant to engage with technology. 462 Mobile phone applications are increasingly being used463 and may help to support patient empowerment and shared decision-making.

An earlier HSDR programme study452 explored continuity and co-ordination in primary care, including the impact of informational continuity. Up to half of the patients at all five sites studied described instances of information, regarding history, diagnosis and treatment, not being passed between providers. Patients reported correcting health worker omissions and out-of-date information, and having to repeat information to the next practitioner. In one instance, a district general hospital delayed action on an urgent GP referral, resulting in emergency admission of the patient 2 days later. Poor protocols for sharing data and unclear responsibilities were cited as contributing factors. Informational continuity seems to decline across traditional boundaries, with some patients with long-term conditions maintaining their own paper-based histories.

Effective communication, to share and link data, underpins multidisciplinary and collaborative working,174,453 and can range from a basic model through to a shared care record with e-referrals, e-prescribing and e-discharge. Data-sharing, enabled by telemedicine, allows GPs to access advice from specialists; examples include GPs working with dermatologists and ophthalmologists to share data and enable remote specialist consultation, with mixed results. 174 Informational continuity, through shared records across primary and secondary care, seems to show potential in improving decision support outside of specialist clinics. 174

Data-sharing is considered to be one of the key factors driving high-quality care for chronic conditions, with the potential to improve efficiency, drive up quality and reduce duplication. 453,464,465 The chronic illness care model466,467 encourages the use of information systems to facilitate data-sharing and communication across traditional boundaries, helping to translate knowledge into practice. One example demonstrates integrated medical records for frail elderly patients providing read–write access for all health professionals involved in care, as well as patients and carers. 452 Such systems can be built on to develop tools and dashboards to alert MDTs to patients at risk of admission417,452 and identify potential bottlenecks and delays. 417 In north London, an information technology tool was developed and used to support information storage, sharing and analysis to support integrated care programmes. 468 Information-sharing does, however, depend on data quality; GP systems and templates may enable standardised and complete data collection. 453,469

Mechanisms

Facilitated by new ‘disruptive’ technologies enabling large data sets,470 real-time analytics can be a catalyst to drive population-based health care. Future iterations of clinical decision support systems are likely to enable data linking across sectors, enabling primary and secondary care clinicians to access data to inform decisions at key points in the pathway. 461 Currently, information exchange is constrained by such factors as local information governance protocols, incompatible information systems leading to duplication and rescanning, unreliable connectivity and inconsistent clinical coding. 452 Digital strategies should encompass a range of technologies to support data-sharing and co-ordinated multidisciplinary care; shared electronic records, used in combination with predictive analytics; telehealth; and patient-facing tools, such as portals. 469 Consideration needs to be given to infrastructure (e.g. data storage, real-time analytics and integration of distributed data sets);470 capacity and skills;470 and financial sustainability. 454,464 Some form of central administration is needed to ‘monitor, mediate and facilitate data-sharing across multiple stakeholder organisations’454 and to manage protocols, processes and data specifications. 470 At a national level, there is also a role to ensure interoperability between policies, systems, data standards and supporting market development,454,461 in order to avoid unnecessary variation472 and conflicting policies. 469,473

The sustainability of data innovations is dependent on demonstrating benefits for patients, commissioners and practitioners. 472 Data-sharing needs to be supported by cultural changes, such as multidisciplinary working and support for shared decision-making. 174,461 Innovations are also likely to be more acceptable if patients and staff have been engaged, and communication on the legal and ethics implications has been clear. 470

Both patients and health-care professionals will need support in developing the skills and knowledge to benefit from data-sharing. 470 Training centres specialising in big data (large linked data from electronic health records and health data collected by personal wearable devices) and open data (the sharing of data sets) can help services to reap the benefits from data-sharing to address population health needs. 470 In the absence of training, the ‘noise’ of large data sets may risk the misinterpretation and misuse of data. 470

Numerous barriers to technology transformation have been identified, including the absence of standards, significant variation in workflows, inefficiency of health information technology, incompatible legacy systems, incomplete information flows and poor system design. 452–454,472 An absence of incentives for data-sharing, opportunity costs and disagreements around the use of data can also hinder effective information exchange. 454 Poorly defined goals can result in uncertainty about what can be shared and which data need to be included,464 potentially leading to variation in the capability of organisations and services to share and act on data. 464 Relationships based on trust and collaboration will be key to overcoming structural, cultural and ethics barriers. 454

R1

The delivery of new and expanded roles in primary care, including access to specialist care in the community, requires a degree of trust between the different professionals. The provision of relevant training and practical tools in parallel can trigger professional willingness to adopt new ways of working or task-shifting. Once new professional roles and behaviours are effectively embedded, there is the potential to achieve the outcomes of better management of chronic conditions, a reduction in the number of secondary care referrals and cost-savings, following sustained implementation and stabilisation of increased demand (with the costs of training and additional community services provision accounted for).

R2

The opportunity for clinicians and patients to engage with providers and commissioners for accountable place-based contracting and payment systems requires the alignment of personal, professional and organisational values and incentives. Moreover, sufficient time needs to be allocated to learning and development, agreeing outcome frameworks and sharing access to robust high-quality information that includes cost and quality data. This will allow the building of trust, collaboration and shared decision-making for accountability across quadruple aim outcomes. Furthermore, through service users and a diverse group of professionals having the confidence to hold providers and commissioners to account, MCP leaders will be spurred to make informed (re)investments based on clear measures of value and appropriate management of financial risk.

R3

The development of mutually beneficial relationships with communities requires opportunities for equal and reciprocal engagement for all relevant health-care professionals and the local population. The provision of ongoing training, guidance, feedback and practical support for community-based working needs roles, responsibilities and expectations to be clearly articulated. Together, they can empower individuals with the confidence to contribute to decision-making or the sharing of experience and knowledge to inform the priorities for targeted preventative and holistic care. Consequentially, this encouragement of shared community ownership will result in improved health behaviours and increased social participation, and will engender community resilience.

M1–5

The maps further demonstrate the interdependencies between individual theory components at the individual, organisational and system levels. Professional autonomy and empowerment is critical for driving cultural change that is associated with trust and collaboration, particularly for structural developments of MDTs and integrated pathways. Cultural change needs to be stimulated through organisational development and system leadership behaviours that promote collaborative, population-based approaches to health care and aligned processes that support delivery. Shared data, in particular, offer the opportunity to improve the co-ordination and continuity of care at the individual and organisational levels, while MCP-wide learning can be accomplished through training and feedback loops, built into audit and formative evaluation, to support system learning and improvement.

Unintended system consequences

The concept of ‘community’, as we discuss in Chapter 6, is fluid and subjective and does not necessarily follow the boundaries set out by MCP designers. As a consequence, the drawing of contractual lines may effect patient choice and the liability of the MCP. 119 There are also organisational boundaries to consider. The shift of care from hospital to community settings is fraught with tensions and trade-offs, with speculation that organisations may act in their own interests rather than in the interests of the overall system,118 and this is not helped by the presence of conflicting policies. 209

A lack of representation or engagement with co-production activities by some groups in a community may result in increased health inequalities, as services are designed without them in mind. 286,287 In any case, any initiative must take care to be consistent in its information-sharing with service users; for example, receiving conflicting instructions about service users’ individual needs, priorities and management strategies in self-management would prevent the successful uptake of self-management initiatives. 290

Defining the continuity of care in community-based settings as ‘seeing the same health care provider every-time’155 poses a problem for access to an expanded primary care, as discussed in Chapter 4. MCP delivery by an enhanced team with more new roles has the potential to disrupt relational continuity156 for those service users who prefer an ongoing, trusted relationship with a specific professional over quick access. This means that an expanded primary care and community system will need to maintain both its connectivity to the local population and the generalism inherent in general practice;190 operate as hubs to offer more extended access;32 and have sufficient and appropriate staffing,185 which pays heed to workload,32,172 staff development163 and the status of innovation. 159 It is notable that those areas seen to be making the most progress are those building on a history of collaborative working;200 their learning is that it is difficult to develop accountable approaches without aligning incentives208 and transferring savings and risks across sectors. In any case, introducing new services has the potential to induce demand71 or incentivise use. 174 There is the potential to misuse new services71,187, or duplicate care at hospital174 or in primary care. 151

The findings from Chapter 5 suggest that the risks of increased costs187 cannot be overstated and include loss of economies of scale;174 set-up costs for new posts32,163 or registration for hubs and federations;32 and the associated administrative support and resource cost to ensure that system protocols are followed. 174,187 There is also the potential that a model focused on consolidation and co-ordination might inhibit those market forces/competition that are necessary to drive cost and quality improvement. 228

Quality of the evidence base

The evidence base identified varies from moderate (for R1 and R3) to relatively weak (for R2). There is a lack of robust, empirical evidence and the findings are therefore somewhat dependent on experiential evidence, informed by case studies and informal evaluations. The ‘intellectual heritage’ of the MCP models – earlier NHS integration programmes and international models of accountable care – offer some transferable findings, although the contexts differ significantly.

Strengths

The project team composition is relatively unique in being embedded in the NHS, with methodological support and mentoring from an academic partner. This model of co-production ensures a connectedness with current policy and practice, while maintaining methodological rigour. A key strength of this partnership is in harnessing existing links to connect with a range of stakeholders at a strategic and operational level. On a practical level, these links have helped in enabling early access to relevant NHS documents. The team is also well placed to translate emerging findings into language that resonates with key audiences and to relate findings to current service priorities. These findings are not limited to MCPs, but can apply to transformation or integrated care more generally.

Expertise in the principles of realist synthesis remains relatively scarce, although the approach is fast emerging as a response to the challenges of evaluating CASs. We were able to harness expertise acquired from prior involvement in other NIHR HSDR evidence syntheses, and from NIHR projects more generally, as well as borrowing heavily from the tools developed by other teams funded by the HSDR programme. Our methodology acknowledges the increasingly hybrid nature of current realist synthesis projects, spanning the published literature and stakeholder input. We believe that our methodology was genuinely innovative, by starting with the analysis of existing logic models, as well as by using a best-fit framework synthesis to produce a conceptual framework, not only for structuring our own analysis, but also as a potential vehicle for the evaluation of future service models. We benefited from pooling the expertise of three information specialists relating to the ‘realist search’ being captured by a book chapter being written concurrently with our own project,65 as well as from existing tools such as Pawson’s time and task template64 and data extraction forms developed by other teams. 70 The team has used the RAMESES training materials480 in expanding our realist synthesis capacity, and RAMESES guidance62 to ensure project outputs and high-quality reporting.

The close involvement of the advisory group has proven to be valuable in articulating and validating programme theories, and validating a clear, consensual basis on which to focus and target the synthesis strategy. The work with the advisory group helped to ensure that the prioritised in-depth realist reviews focused on the aspects likely to add the most value, while ensuring the relatively comprehensive coverage of the breadth of the MCP model, through the series of theory maps. To test and validate findings, the advisory group was expanded to form a broader focus group, involving additional key stakeholders identified during the project. The focus group provided a ‘reality check’ of our findings. A summary of discussions at the focus group and during interviews with advisory group members is included in the review chapters.

Patient and public involvement has been a key component of stakeholder engagement. In developing the review proposal, the team contacted local patient participation groups to invite feedback from current patients and members of the public. As Davies et al. 481 point out, patients and the public constitute a group that has been largely absent from the knowledge mobilisation literature. The project aimed to address this gap through the active involvement of service users. Two PPI representatives were appointed to the advisory group and have provided valuable input throughout, notably in the prioritisation of programme theories for review, reviewing emerging findings and advising on dissemination activities.

Communications and engagement are critical to producing outputs focused on relevant and actionable insights for our various stakeholder audiences. Recognising the importance, the team has been supported by dedicated and specialist communications support and expertise. The skills and experience have been valuable and have enabled stronger relationships with the communications team at NHSE, enabling wider reach to MCP sites and minimising the risk of overburdening them with ill-timed requests and information. The dedicated communications role, with a background in communicating complex research, might be a useful model for future projects.

See Appendix 9 for a summary of how the study has addressed the research objectives.

Limitations

The programme theory was substantially based on the MCP vanguards’ logic models, developed to support the submission of value propositions to bid for transformation funding. However, this approach was dependent on complete logic models being available and other national processes affected the production of the logic models. The team was able to access logic models for 12 of the 14 vanguards and, at this early stage, the format and granularity of the models were variable. The team subsequently used associated documentation available in the public domain, from the vanguard sites (e.g. consultation documents) and NHSE, to supplement the logic models. As the models have evolved significantly during the lifetime of the review, particularly regarding procurement and contracting, the team has endeavoured to keep up to date with emerging policy and guidance. However, given the rapid pace and scale of change, it is likely that the programme theory will continue to evolve subsequent to our analysis.

From a methodological perspective, we acknowledge that conducting a review against a backdrop of continual change, to add to the already-identified complexity associated with diverse contexts, was extremely challenging. Although the new care models are being intensively evaluated, and our networks identified relevant data as soon as these became available, we encountered a degree of prematurity within the specific context of a UK-focused evaluation. We sought to minimise the impact of this in two ways, by (1) exploring transferable lessons from international comparisons and UK legacy models and (2) seeking to identify generic lessons that may be used to explore future care models that seek to achieve large-scale transformational change within CASs. Although vanguard sites offered a ready-made source of what was intended to happen, there was little supporting detail on how this was to be achieved. In particular, individual sites lacked specificity about the perceived and actual intended impacts from the new care models, making inferences about mechanisms extremely problematic. However, we were able to use the composite picture, derived from 12 sites, to mitigate some of these difficulties, and we believe that our analysis was particularly strong in isolating unintended consequences. In particular, the embedded nature of our NHS-based team, continuous stakeholder engagement and telephone interviews with key informants added to the value of our inquiry.

Although the overall review draws on a broad evidence base, this predominantly derives from within the health sector. The team has made a concerted effort to explore social care and third-sector literature, but recognises that there may be valuable learning from other disciplines, for example, lessons from other public services or the private sector in relation to large-scale public-funded procurements. A realist review does not aim for the exhaustive searching of a systematic review and, therefore, it is possible that important papers have been missed. The iterative approach and stakeholder engagement has focused the search on the identification of key sources, but does not eliminate the risk of confirmation bias.

Stakeholder engagement

At our stakeholder event in May 2017 (see Chapter 2) we tested our emerging findings across the programme theory components with stakeholders. We framed our discussion in the context of the observable progress at the MCP sites from the perspective of participants. The discussion content is summarised according to the relevant theory components below; there was of course overlap between the topics.

Dissemination

NHS England has signalled the intention for new care models to continue to evolve,57 with a commitment to developing ACSs, subsequently announced in eight areas, two of which include current MCP vanguards. 59 These ACSs will have responsibility for improving population health, developing horizontally and vertically integrated models of care, learning from the MCP and PACS vanguards. As the vanguards have evolved to fit within their local contexts, it is apparent that they share common characteristics, and any distinguishing characteristics between the two models become more blurred, as acute providers become more involved in MCPs and community providers in PACS. Using the best-fit framework as a structure, and informed by our mapping to the COM-B model (see Figure 21), we provide a quick-view summary of the key findings from our synthesis, extracting the interventions and mechanisms identified in the evidence as significant to the MCP model of care (see Table 10). We describe these conditions as ‘active ingredients’ that, although derived from a programme theory relating to the MCP model, will offer transferable insights to population health-based models of care in general. This table was refined as part of the dissemination activities, which included a presentation to the new care models team but also discussions with the wider Strategy Unit team and NHS colleagues interested in accountable integrated models of care.

Aims and objectives of the project

The project is led by The Strategy Unit, NHS Midlands and Lancashire Commission Support Unit together with the School of Health and Related Research (ScHARR) University of Sheffield. Funded by the NIHR HSDR programme (project reference number 15/77/15), the project is concerned with combining existing knowledge of what works in integrated care with the intentions of the MCPs in developing innovative new care models. It is expected that this work will help MCPs to improve their models.

The aim of this synthesis is to provide decision-makers in health and social care with a practical evidence base relating to the MCP model of care. The synthesis, based on realist principles, will develop a ‘blueprint’, which articulates how and why MCP models work, to inform design of future iterations of the MCP model. 61

Specifically, the objectives of the synthesis are to:61

• articulate the underlying programme theories behind the MCP model of care, by mapping the logic models of the MCP demonstrator sites, prioritising key theories for investigation

• identify sources of theoretical, empirical and practice evidence to test the programme theories

• appraise, extract and analyse evidence, reconciling confirmatory and contradictory evidence

• develop the synthesis, producing a ‘blueprint’ to explain how the mechanisms used in different contexts contribute to outcomes and process variables

• consult with key stakeholders from selected MCP demonstrator sites to validate findings and to test applicability to different contexts

• finalise the synthesis, incorporating stakeholder feedback

• disseminate the findings, preparing a series of practical tools to support knowledge mobilisation.

Terms of reference of the advisory group

This group has been established to provide the project with independent support and advice. Its objectives are to:

• provide stakeholder perspectives on the scope of the synthesis and on emerging findings

• inform the study design, including input into the search strategy

• participate in strategic decisions about the project

• advise on key policy, strategy and communication activities which may inform or impact this project

• advise on the identification of key audiences and networks for dissemination activity

• advise on dissemination formats, routes and activities

• advise on interpretation of findings, to ensure relevance to stakeholders

• support the translation of findings into actionable recommendations and tools for local health economies.

Working methods

It is expected that the advisory group will be convened on up to four occasions during the project, at the following key stages:

1. at the project start, to confirm the scope of the synthesis

2. during the analysis stage, to share and test emerging findings

3. ahead of stakeholder consultation, to consult on the draft conceptual model

4. at the project conclusion, to sign off the final report and dissemination outputs.

Electronic communication may be used between advisory group meetings.

Minutes will be taken at each meeting and will be shared with the NIHR HSDR programme and, potentially, more widely. The advisory group will be supported administratively by Sharon Townsend, Project Officer.

June 2016

Search strategy

Feedback loops

The feedback loops serve to feed back the output of a process, as an input, into the same system. 74 In health-care studies, feedback loops between individuals and their community or environment have been assessed to determine how ‘neighbourhood effects’ or ‘place effects’ affect the individual’s health. 486,487 Feedback loops can generate change or stability in a system, depending on the relationships among different agents;422 positive feedback loops can be facilitative, for example, in the use of PDSA cycles in quality improvement processes. 396 Of our eight theory components, feedback loops are most useful in framing the necessary system learning (M4) through positive feedback reinforcement when intended outcomes are being realised, or taking corrective action following negative feedback from less-desired outcomes. Thus, in a MCP-like system, feedback allows the adaptation476 of the New Care Models programme delivery to align with the constantly shifting NHS landscape, for instance, allowing MCPs to be incorporated into sustainability and transformation plans or to begin journeys as ACSs, as vanguard evaluation findings are fed back.

Connectivity

The manner in which agents in a system connect and interact gives rise to the behaviour of a CAS, highlighting the importance of relationships between the agents over the agents themselves; as Plsek and Greenhalgh418 note, ‘the observable outcomes are more than merely the sum of the parts’. The capacity to connect and leverage social networks is described to be the most distinctive feature of the Buurtzorg model in the Netherlands, in which multilevel networks with intra- and interprofessional group connectivity coexist at local and national settings. 52

This connectivity is a recurrent theme in the MCP programme theory, occurring within components R1, R3, M1, M2, M3 and M5. This includes primary, community and specialist professionals connecting to adopt new roles and ways of working through building or strengthening relationships to co-ordinate care for the benefit of patients (R1). In R3, the inclusion of service users for co-production purposes adds a further dimension of connectivity.

In M1, the responsibility of delivering safe and high-quality care is discharged through collective decision-making with dependency on good communication and working relationships. In M2, interdisciplinary connectivity within a MDT offers shared decision-making and streamlined care for the benefit of the patient, and promotes team-building and peer support for the benefit of professionals.

The importance of connectivity for staff engagement is clear in M3; effective connectivity through sharing and leadership leads to improved staff engagement and workforce morale. Connectivity is also important with a technological interface, as the sharing of data by human agents brings about the intended change of proactive population health (M5).

Edge of chaos

The most productive state for a complex system is proposed to be at the ‘edge of chaos’, to allow for maximum variety and creativity, making it fundamentally unpredictable and, as such, it has to be observed over time. 418,477 A CAS approach seeks to draw out and mobilise the natural creativity of health-care professionals to adapt to circumstances and evolve. This requires system learning from locally owned solutions that are intended to achieve improved quality,424 rather than achieving this through top-down large-scale transformation efforts. Crucially, this requires risk-taking and learning from mistakes or failures (see Chapter 8, Stakeholder engagement), rather than national interventions to limit local risk-taking. There may also be a need to allow tensions to play out, for instance, in R2, the contracting and payment mechanism for MCPs is necessarily at the edge of chaos, using the opposing forces of competition and co-operation to be mutually beneficial,478 to allow local sites to innovate for their accountable local place-based system.

We have abstracted the resource and reasoning aspects of the intangible mechanisms (see Table 12) to six classic middle-range theories, in order to provide a rationale for the underlying social processes to inform future accountable population-based integrated care models. This is by no means an exhaustive search for middle-range theories, but a selection of six classic theories that can apply across system transformation, such as the introduction of a MCP.