Introduction

Context: rethinking how research influences practice

Managers and clinicians in the NHS, as in all health-care systems, are under growing pressure to improve and redesign services in a way that both optimises health outcomes and controls costs. The research community has great potential to contribute to this process, by producing and disseminating evidence on a variety of aspects of service organisation and care delivery. And yet, the disconnect between, on the one side, the theory and empirical evidence underpinning how best to design and deliver high-quality health services and, on the other side, what actually happens in practice has challenged policy-makers, academics and practitioners for several decades.

How people respond to this challenge depends on whether they perceive the problem to be how academic knowledge is conveyed from researchers to practitioners (‘knowledge transfer’), or the fundamental nature of knowledge and how it is produced (‘knowledge co-production’).1–4

Knowledge transfer

When framed as a knowledge transfer problem, researchers are seen as having expert knowledge that needs to be transmitted to decision-makers in the health service in an accessible and timely fashion.5 Knowledge is perceived as a relatively tangible, bounded and moveable ‘product’, whereas the decision-making process is likely to be seen as time-limited, linear and rational. Research evidence, regarded as the most valid and reproducible form of knowledge, is ‘pushed’ from the research community, through guidelines or evidence summaries, or ‘pulled’ by practitioners who are (or should be) well informed about the benefits of using research evidence.

The emergence of sophisticated informatics and communication technologies in recent years, and their use by the National Institute for Health and Care Excellence (NICE) and others, has helped to reinforce a view that the knowledge transfer model is the most appropriate way of closing the so-called ‘know–do gap’.6 This may be a reasonable assumption in situations in which scientific knowledge is fairly unambiguous, easy to interpret and largely uncontested, such as is often the case (relatively speaking) for much of the clinical evidence underpinning the practice of evidence-based medicine. The transfer approach is more troublesome in the field of health services improvement or service redesign, where the issues that research attempts to address are more complex and the nature of the social science evidence is more contested.7

Knowledge co-production

This recognition of the limitations of the knowledge transfer model has resulted in a reframing of the integration of research and practice, as requiring different approaches to the nature of evidence, its production and use, and the complexities of the challenges faced by care systems.3,8

Rather than being separate processes, the production and utilisation of research evidence are seen to merge as complex, iterative and situated social processes.9–12 That is, knowledge is created, understood, adapted, used and reiterated in the context in which it is needed, and through partnership between different actors.

Furthermore, decision-making by practitioners is regarded more as a situated, social and evolving process than as a one-off, rationally determined event.6,11 Rather than research evidence needing to be fed into this in a linear and methodical way from outside, the emphasis becomes one of the integration of empirical evidence with other forms of knowledge in situ, including practical knowledge about how to improve and redesign services, and user knowledge about the experience of care services.

This integration is seen as a situated social process based on productive ongoing relationships, effective systems and conducive organisational cultures and contexts.13–16 Increasingly, the literature in this area embraces a ‘complex adaptive systems’ way of understanding the world, as a means of understanding how actors have to overcome social and epistemic boundaries to helpfully facilitate knowledge processes.17–19

Such a reframing suggests that the relationship between research producers and research users should focus on the ‘co-production’ of knowledge, rather than merely its transfer.2,3,20–23 Co-production models of knowledge mobilisation are based on the assumption that knowledge created by researchers often needs to be adapted in some way if it is to be useful. Indeed, for knowledge to have influence, all participants need to be involved in its creation and use.24 Researchers, using the scientific method as their predominant way of knowing, are not seen to have a monopoly on expert knowledge. Instead, they need to be willing to negotiate actively their forms of knowledge with others (‘a meeting of experts’, e.g. experts in research evidence and experts in improving health services), to recognise and act on power differentials in who dictates how knowledge is defined, and to adopt a more pluralistic orientation to knowledge to achieve change.25,26

Broadly defined as joint working between people who or groups that have traditionally been separated into categories of user and producer, the term ‘co-production’ came to prominence in the 1970s27 and has gained much currency in public service discourse since.28,29 When applied in health services research, it is increasingly used as a term to describe the co-production of research-informed knowledge through the engagement of policy-makers and practitioners with researchers.30–33

However, the increasing popularity of the principle of co-production in public services comes on the basis of promising, but far from complete, evidence.29,34–36 The literature highlights a range of dilemmas and challenges that need to be explored more fully; for example, is it desirable or feasible to bring researchers and practitioners more closely together, or might the logistical challenges and potentially greater costs than traditional approaches outweigh the benefits?32,37,38 What are the challenges of reconciling divergent or even incommensurate epistemologies?39 Should greater attention be paid to the political and social dimensions of co-production (e.g. the different interests, power and expectations of the parties)?30,31,39,40 Might co-production lead to a narrowing of focus towards problem-solving dimensions of research use,41 instead of encouraging important broader perspectives?23 Should the boundaries between researchers and practitioners be firmly drawn, or are there advantages in flexibility and in the blurring of boundaries?9,21 In what ways can researchers and service users get involved in the co-production of service redesign or improvement?42–44 And can health-care research that is co-produced in one location be translated effectively to other settings?32

The challenges posed by these questions have been exposed by the growing range of new initiatives that seek greater interactivity over research (and, by extension, knowledge co-production in some form) that have emerged in the NHS over the past two decades,17,45–47 as well as in the in the health systems of Canada,48,49 Australia50 and the USA.51,52

Embedding researchers to encourage co-production

Many different terms and models have been used to describe approaches to research use that seek greater engagement and sustained interactivity, including knowledge brokers,53–60 NHS management fellows,12 Health Foundation improvement science fellows, National Institute for Health Research (NIHR) knowledge mobilisation research fellows, and researchers-in-residence.61–65 What many of these models have in common is a desire to ‘embed’ researchers in service settings, for significant periods of time, to secure the sorts of sustained interaction thought necessary to promote knowledge co-production. Yet much of the literature is unclear about the relationship between knowledge co-production as a generic concept, ‘brokerage’ (the process by which knowledge is shared across boundaries) and ‘embeddedness’ (whereby researchers are, to a variable extent, located within practitioner teams).

Moreover, other aspects of the schemes that have emerged are highly variable and usually not well described. For example, schemes may vary markedly on the individuals involved (e.g. researchers, health professionals, managers, service users), the degree of close interaction or embeddedness of individuals (e.g. physical locations, contractual arrangements), the nature and degree of co-production sought (calling into question complex issues of epistemology and power) and the types of activities that take place (e.g. the balance between the brokering in of external knowledge and the co-creation of new knowledge in situ). These different models are emerging largely independent of each other, with little evidence of shared learning and only a very few examples of formal evaluation. What has been published mostly comprises descriptions of isolated initiatives, but little in the way of deep analysis and interpretation, and even less in terms of practical guidance.

Both the literature and current practice, therefore, highlight a rich research agenda around embedding researchers as a means to knowledge co-production and, hence, better research use. This includes a clear need to develop a better theoretical and empirical basis for such schemes in the NHS, and the need for evidence-informed practical guidance to support implementation of embedded co-production in practice.32,38 This project addresses these gaps.

Research aims and objectives

The Embedded project aimed to increase the influence of health services research on decisions about the improvement and redesign of NHS services by:

• developing insights and understanding about the nature, challenges and effectiveness of co-production initiatives in which researchers are embedded in service settings
• producing practical guidance on the design and implementation of embedded models of co-production for managers and clinicians in the NHS, their academic partners and people who use services.

In addressing these aims, the project focused on the following specific objectives:

• to review the theoretical and empirical health services, management and organisational literature relevant to embedded research initiatives and knowledge co-production, and identify the relationship(s) between the two (workstream 1)
• to gather examples of embedded models in operation around the UK’s health services and public health sectors, focusing on examples where embeddedness and co-production co-exist, and to describe the features of these models, including their history, context, participants, scale, scope and content (workstream 2)
• to undertake in-depth case studies in four of the examples identified, to understand their mechanisms, effectiveness and challenges (workstream 3)
• to provide resources aimed at assisting in the recruitment of embedded researchers, alongside recommendations and guidance for their training and development, customisable for the different ways in which embedded co-production may be framed and specified, to allow those interested in developing and using such approaches to understand the design choices they face (workstream 4).

Overall study design

In this section, we set out the overall research strategy governing the four workstreams. Detailed descriptions of the research methods used for the three research-based workstreams (workstreams 1–3) are provided in Chapters 2–5. The activities underpinning workstream 4 (engagement, dissemination and influencing) were ongoing throughout the project, and are described here briefly, with specific outputs emerging from this work covered in Chapters 6 and 7. All appropriate research ethics and research governance permissions were obtained, with approval letters reproduced on the project web page [www.journalslibrary.nihr.ac.uk/programmes/hsdr/165221/#/ (accessed 18 February 2021)].

Workstream 1 used two separate narrative literature reviews to identify and describe the principles and practices of knowledge co-production and of embedded research initiatives. The results were used to develop practical frameworks for understanding knowledge co-production and embedded research; these are described in Chapters 2 and 3, respectively.

Workstream 2 identified and described the breadth and scope of embedded researcher initiatives operating in health settings across the UK and how these were designed to enable knowledge co-production. The results were used to further develop the frameworks of knowledge co-production and embedded research from workstream 1 and to identify candidate case studies for further in-depth examination in workstream 3 (see Chapters 4 and 5). Taken together, the results of workstreams 1 and 2 were used to inform development of an initial framework for planning embedded researcher initiatives that enable knowledge co-production (elaborated on in Chapter 6).

Workstream 3 built on the work undertaken in the first two workstreams by conducting four in-depth case studies to develop a deep understanding of how embedded models of knowledge co-production actually work in practice. The cases and methods of data gathering are detailed in Chapter 4, and the cross-case analysis is displayed in Chapter 5. Drawing on these insights, the implications for scheme design and an enhanced practical framework are set out in Chapter 6.

Workstream 4 used a range of approaches to engage stakeholders with the findings of the project, including developing, testing and disseminating practical guidance for managers, clinicians and academics. This work drew on all the emerging insights and outputs from workstreams 1–3. The processes deployed are described in the next section, and the practical outputs, tools and resources that emerged from that process are set out in Chapters 6 and 7.

Engaging and influencing (workstream 4 activities)

Throughout the project, considerable attention was given towards engaging with and influencing those already embarked on developing embedded research initiatives and those who might be interested in doing so. We aimed to engage with stakeholders to help shape our programme of work and to guide us in the production of useful materials, tools and resources for that community. This work addressed the objectives of workstream 4, namely that, as a consequence of undertaking the project, we wanted to see the following:

• organisations already engaged in embedding researchers making fuller use of the evidence-based guidance, person specifications, training templates and training resources that we produced to optimise the effectiveness of their work
• organisations that have not yet considered embedding and/or co-production models stimulated to explore their potential use in improving decisions that affect service delivery
• future work commissioned to explore the effectiveness, cost and value of embedded research initiatives, focusing on those having the greatest potential to improve services.

Approach to engaging and influencing

The fundamental premise of this research is that knowledge needs to be produced in and through relationships with those who are going to use it (the rationale that underpins both embedded research and knowledge co-production). This, therefore, was also the approach that we took to knowledge creation in this research project. Hence, engaging and influencing key stakeholders were core elements of the proposal.

The principles underpinning our approach were aligned to the theory and practice of co-production models of knowledge mobilisation and to the principles of participatory research.17,46,66 To that end, the influencing plan was co-produced by practitioners, service users and academics, and used approaches that focus on social interaction, as well as more traditional academic approaches to dissemination.

Audiences and actions

We identified three target audiences that we considered most likely to benefit from the new insights and the associated tools and resources. The primary audience was NHS and local government leaders, both managers and clinicians, whose provider and commissioning decisions could potentially be improved by making better use of health services research evidence.

The secondary target audience comprised the applied research and associated implementation communities, such as members of NIHR Applied Research Collaborations (ARCs) [formerly known as Collaborations for Leadership in Applied Health Research and Care (CLAHRCs)] and Academic Health Science Networks (AHSNs). The twofold aim here was both to encourage actors in these initiatives to explore the benefits and risks of engaging in embedded models of knowledge co-production and embedded research, and to contribute theoretical and empirical knowledge to the field.

The third target audience comprised front-line staff and service users; their interaction with embedded researchers are at the heart of the model. Helping those who want to access and use applied health research to see more clearly how the structural arrangements for the creation of that research affect its use is key.

The Embedded project team had extensive personal and professional networks in the field of NHS service management and knowledge mobilisation, in the UK and internationally. We used the team’s wider local and national networks in local health economies, ARCs/CLAHRCs, AHSNs and higher-education institutions to ensure that the outputs of the project had maximum impact.

A range of actions were devised to engage and influence these audiences.

Workshops

We hosted two participatory workshops during the study, each of which enabled us to co-produce practical outputs (see Chapters 6 and 7) and raise the profile of the project with target audiences.

Workshop 1 co-produced a framework of design options for embedded researcher initiatives and visual representations of the findings from the literature reviews (workstream 1) and the scoping exercise (workstream 2). Details of the design, attendees, process and outcomes are laid out in Chapter 6.

Workshop 2 co-produced a shared understanding and a clear narrative of the case study findings. Details of the design, attendees, process and outcomes are covered in Chapter 4.

A third planned workshop was precluded from taking place because of the COVID-19 crisis. This would have co-produced practical guidance and tools and resources for the design and implementation of embedded models. These goals were met instead by tapping into professional networks and through social media interactions, and by using data gathered from workshop 2.

Tools and resources

Working in partnership with the relevant stakeholders, we co-produced the following resources for use by health service and academic organisations to encourage and support the practical implementation of the learning from workstreams 1–3:

• an animation to describe the role of embedded researchers in an engaging and accessible way
• detailed guidance on the design and implementation of coherent embedded co-production models of knowledge mobilisation, presented in the form of webinars and interactive virtual workshops
• job descriptions and person specifications to support recruitment to new embedded research posts
• a description of the knowledge, skills and attitudes required of embedded researchers, and the different career pathways that could be pursued
• guidance for the training and support of embedded researchers, together with resources that could be used by co-production knowledge mobilisers
• an outline of how embedded researchers might prepare for their role and how organisations could create a conducive environment for them to thrive.

Additional information on each of these outputs is presented in Chapters 6 and 7, where we integrate the research findings from workstreams 1–3 with the engagement activities of workstream 4. The recruitment resource pack detailed in Chapter 7 can be found on the project web page [www.journalslibrary.nihr.ac.uk/programmes/hsdr/165221/#/ (accessed 18 February 2021)]. Practical outputs are also documented on the project’s Embedded Research website.67

Network

In the original application, we proposed identifying and supporting embedded researcher learning sets that we were aware had been established in a number of locations across the country. As part of our scoping work, we discovered that these networks were often short lived or only partially active. When we discussed the desirability of greater peer support at our first workshop, the participants expressed considerable enthusiasm to establish a self-managed and lightly facilitated network to allow researchers to support and learn from each other.

At the request of the participants, and with their express permission, a network was established, first using a Google Group (Google Inc., Mountain View, CA, USA) and then a WhatsApp (Facebook, Inc., Menlo Park, CA, USA) group. We hoped that the group would provide a peer support resource for those involved in embedded research and a rich opportunity for the research team to get members’ input into our practical frameworks, emerging learning and influencing activities. Observations made through this network influenced the materials and tools created (presented in Chapters 6 and 7, and showcased on the project website).67

Publications and presentations

To engage public sector leaders, senior clinicians and service managers, accessible audience-appropriate articles were prepared for the Health Service Journal and Local Government Chronicle. The emerging findings were also presented at national and international conferences and seminars, with more detailed reporting in academic journals (see Acknowledgements for a full list of publications and presentations). Through these means, we accumulated resources for the project website67 and prompted interest in and conversations about the emerging programme of work.

Social media

Working with our communications and influencing partner, Kaleidoscope Health and Care (London, UK), we established a website, Embedded Research,67 published regular blogs and set up an Embedded project Twitter account (@_embedded) (Twitter, Inc., San Francisco, CA, USA). In addition, we produced a series of webinars exploring key issues relating to embedded research (www.embeddedresearch.org.uk/resources/webinars/) and we designed an animation to popularise the key messages for audiences not yet familiar with embedded research.67 The website and associated resources are described more fully in Chapters 6 and 7.

Patient and public involvement

Although progress is being made in involving patients and the public in the provision of their clinical care, in the broader issues of health service improvement and service redesign, patient and public involvement (PPI) has proved considerably more challenging. Experience-based co-design is a rare example of a specific methodology that brings service users and staff together to develop simple solutions to improve patients’ experiences of care.68 However, moving the agenda from highly tangible but tightly bounded initiatives, such as experience-based co-design, to broader issues, such as how patients and the public can assist in ensuring research-based knowledge is used more effectively, has proved to be somewhat problematic.69 The following approach has been gaining traction more recently in some jurisdictions (e.g. the USA and Canada): putting patients and service users at the heart of research teams, using patients as active partners and participants to ensure the relevance and applicability of applied research findings.70 Nonetheless, despite these pockets of innovation, the field of knowledge mobilisation, in general, appears to have not yet been successful in finding a place for the patient voice.3,4,71

We used the literature review and empirical work in this study to examine this challenge, with the aim of exploring and describing in detail examples of good practice in involving patients and the public in embedded knowledge co-production initiatives. As the subsequent chapters will show, embedded research initiatives involve patients to a varied extent and use diverse models. It was notable, however, that embedded scheme specifications rarely considered PPI in any depth, and the PPI that did occur was seen largely at the level of embedded research projects, but not usually at the level of the overarching embedded research initiative.

In addition to examining and conceptualising the role of patients and the public in embedded research, we specifically also wanted to make a practical difference. To achieve this, we wanted to seize the opportunity to involve patients and the public more substantively in knowledge co-production and to understand how to make an effective case for greater user involvement in this field. In our design of the project, we conceptualised patients as the main beneficiaries of a more evidence-informed health system. Indeed, we thought that patients, both existing and potential, should be central because it is primarily they who are disadvantaged when research-based knowledge is not used by health professionals and managers.

To that end, we framed the effective co-production of knowledge as being dependent on a number of different relationships. The primary focus for this research was on how researchers can build effective relationships with practitioners to co-produce knowledge and improve services, but we saw service users as a key element of the context in which these relationships are enacted, as stakeholders and as important motivators for change. The patient and public voice was therefore included from the early design stage of the proposal.

Our patient and public collaborators included a funded consultant (RP) with great experience as a PPI advisor to NIHR and as a member of the NIHR knowledge mobilisation research fellowship selection panel, who took on the explicit responsibility of challenging any tendency for service-centricity in the design and delivery of the proposal. In addition, the service co-applicants and collaborators were chosen for their commitment to PPI and their extensive contacts with patient groups, including user advisory groups that had been established specifically for knowledge mobilisation initiatives run by ARCs/CLAHRCs and AHSNs and within specific knowledge mobilisation roles. We also established a project PPI group comprising three additional lay members, two of whom already worked closely with the lead for workstream 2 and the other with the lead of workstream 4. We return to these issues in Chapter 8 to distil the learning from across workstreams on the role and potential of greater PPI.

Structure of the report

The remainder of this report is structured as follows.

Chapter 2 presents the methods and findings of a focused literature review of knowledge co-production, drawing from writings on health care as well as in more diverse fields. What emerges is a practical framework of different types of knowledge co-production, from relatively orthodox views of collaborative knowledge creation and use, to more radical conceptions of the capacity of co-production to upend existing assumptions and to challenge established patterns of power. This provides an account of the first part of workstream 1.

Chapter 3 presents the methods and findings of a focused literature review of embedded research schemes (the second part of workstream 1) and the scoping review of extant schemes in UK health-care settings (workstream 2). The findings are brought together to provide a framework for understanding the key features of embedded research initiatives.

Chapters 4 and 5 present the findings from our intensive case studies of different types of embedded research initiatives in which knowledge co-production was a feature (workstream 3). Chapter 4 describes the methods of data gathering and provides accessible accounts of each of the initiatives studied (with additional detail on both methods and cases contained in associated supplementary materials). Chapter 5 details the analytic approach to cross-case analysis and thematic analysis, building on the domains identified and elaborated in workstreams 1 and 2.

Chapters 6 and 7 integrate the understanding that emerges from Chapters 2–5 into a series of practical tools and resources aimed at helping the design, analysis and management of embedded knowledge co-production initiatives. Chapter 6 also notes the contributions and outputs derived from the engagement strategy run through workstream 4. Chapter 7 presents the recruitment resource pack in detail.

Finally, Chapter 8 situates and reflects on this work in the wider research literature and contemporary policy and practice concerns. It also considers the role, implications and potential for PPI in such schemes, and offers directions for further research.