Embedding in practice: the case accounts

Introduction

This and the following chapter present the research activities and findings of workstream 3. Informed by the findings of preceding workstreams, this involved in-depth qualitative research with four purposively selected ‘cases’ of embedded research, with the objective of understanding the contribution such initiatives make to the co-production and translation of research into practice. The reporting here and in Chapter 5 (and with methods further elaborated in Report Supplementary Material 3) makes use of the reporting guidelines of COnsolidated criteria for REporting Qualitative research (COREQ)169 to set out what was done. In this opening part of the analysis, we focus on ‘telling the story’ of each initiative, drawing attention to their key features, backstories and trajectories; Chapter 5 develops the analysis across the cases, looking for repeated patterns that speak to similarity and differences in the themes surfaced in the earlier workstreams. Thus, this chapter focuses more on description, whereas Chapter 5 takes a more analytic slant; the findings are separated in this way to ensure that the overall account does not become unmanageable, and to allow a clear focus on both the particularities of individual cases and the analytic insights that can be gleaned by looking across the cases.

As shown in Chapter 3, embedded research initiatives can vary considerably in terms of intent, structure and processes. This in-depth work was designed to investigate how such factors, together with other unanticipated issues, might shape how the embedded initiatives were formed, how they contributed to the co-production of knowledge and how they supported the translation and use of knowledge in the organisation and delivery of care.

The specific research objectives addressed in workstream 3 were as follows:

• purposively select four exemplar embedded research initiatives reflecting anticipated typographical differences (e.g. in their affiliations, project characteristics, relational networks and funding)
• describe the embedded researchers’ career histories, motivations and networks of research (university) and practice (NHS) partners, including changes over time in relationships, tasks and activities, and pattern of interactions
• understand how embedded researchers, and members of their networks, mediate different forms of knowledge and cultural and social boundaries to promote co-production
• appraise the contribution of embedded researchers and their networks to knowledge co-production, including detailed examples in selected NHS settings.

The conceptual framing for this work drew on a growing body of research (introduced in Chapter 1 and explored in depth in Chapters 2 and 3) that examines how co-production activities can help mediate and overcome a variety of well-documented social, cultural, organisational and epistemic boundaries. For example, there is strong evidence that research producers and research users are often decoupled or disconnected because of institutionalised knowledge boundaries.170 These include differences in the syntactic nature of knowledge (words or terminology), the semantic nature of knowledge (meanings and assumptions) and the pragmatic nature of knowledge (its purposes and intent).

It has been shown that such differences can significantly retard the translation and application of research knowledge in health and care settings,171 but that a number of interventions and strategies can mediate these factors and boundaries, from knowledge brokers to co-design methodologies. Embedded research initiatives represent one prominent strategy for mediating such barriers. Although recognising that knowledge co-production comes in many guises (see Chapter 2) and that embedded research initiatives themselves can take many forms (see Chapter 3), this part of the project was designed to explore in depth the mechanisms employed and the dynamics that unfold.

Overview of study methods

A detailed account of the research methodology for workstream 3, the case selection process, exact methods and the extent of the data gathered can be found in Appendix 3. Here, we simply give an overview of the research strategy to contextualise and frame the subsequent findings.

Sampling and selection

Workstream 3 involved qualitative comparative case studies of four purposively selected examples of embedded research, although, as we show, one of these comprised additional subcases (multiple embedded researchers). Qualitative case study research affords depth, providing descriptive and explanatory understanding of any given case. Cases can be selected for a range of reasons: from the discovery or exploration of unique situations, the examination of comparative aspects or search for replicative logics, through to more conventional forms of representative sampling.172–174

Sampling here aimed to identify and select examples of embedded initiatives that were likely to co-produce knowledge, while also taking into account key differences in their configuration that might explain how and why co-production varies (see Chapter 2). Selection was informed by the learning from workstreams 1 and 2, especially the national review of current initiatives, the emerging understanding of knowledge co-production (see Chapter 2) and the thematic summary of the design considerations and operational features of embedded research initiatives (see Chapter 3).

Through a robust and independent review process with additional moderation (detailed in Appendix 3), four cases were identified, reflecting differences in approach, outcomes, setting, scale, employment, maturity and location, as set out in Table 6. These sites were selected to explore anticipated variations in approach, outcomes, settings, scale, employment, maturity and location, with these considerations being informed by the design choices and operational features developed in workstream 2.

Preliminary discussions with key gatekeepers signposted us to important individuals and groups involved with each initiative. We also used a ‘partner identification tool’ (see Appendix 1, Figure 13) to identify additional stakeholders for each initiative. Ultimately, we sought to identify those practitioners, managers, commissioners and patients who, in different ways, supported, worked with and interacted with the embedded researcher. Appendix 2, Table 19, offers an overview of the participants from each site. After appropriate ethics and research governance permissions [see the project web page: www.journalslibrary.nihr.ac.uk/programmes/hsdr/165221/#/ (accessed 18 February 2021)], these people were contacted by e-mail and invited to participate in the study. The invitation to participate, including details of the project and use of data, and the consent form can be found on the project web page [www.journalslibrary.nihr.ac.uk/programmes/hsdr/165221/#/ (accessed 18 February 2021)].

Data collection

We used a range of qualitative research methods to collect and analyse data, including semistructured interviews, focused observation and reflective diaries. In practical terms, these were organised with each participating case through a series of site visits over a 12-month period with the intention of studying changes over time. For the more geographically remote sites, on-site data-gathering took place over two discrete periods, each of 2–3 days’ duration, supplemented by multiple and various physically distanced interactions (telephone and e-mail). One site was more local to the research team, making multiple day-long visits possible. In total, each site received around 7–10 days of observation.

The initial site visit involved introducing the research to key study partners, carrying out the initial interviews with the embedded researcher and key partners, observing aspects of day-to-day practice and observing one or more planned meetings. On subsequent visits, we conducted follow-up interviews with the embedded researcher and additional partners (identified through the use of the partner identification tool; see Partner identification tool), and we carried out additional observations and shadowing of embedded research activities.

Between visits, additional short telephone interviews were carried out with each embedded researcher to discuss ongoing and planned embedded research activities. The research methods are detailed in the following paragraphs and elaborated on in Appendix 3, with major data extracts in Report Supplementary Material 3.

We recorded 46 formal interviews with 31 participants across the four case study sites. Interviews were usually one to one; some were completed face to face, others over the telephone (see Appendix 3, Table 19). Participants included embedded researchers and members of their wider network, identified through use of the partner identification tool and a broader ‘snowballing’ approach. We also completed extensive informal interviews as part of the periods of observation. The observations took place at three case study sites (Evansville had already concluded), and included observing a variety of specific events and more generalised shadowing of the embedded researchers in their roles (specific details of which can be found in Report Supplementary Material 4). Further details on each of these modes of data gathering now follow.

Interviews

Semistructured qualitative interviews were the primary method of data collection for workstream 3 and followed two topic guides, the first for embedded researchers (see Appendix 4, Table 21), the second for other stakeholders (see Appendix 5, Table 22). These were designed to explore participants’ views and experiences of their embedded research initiative, focusing on key stages and activities in the development and function of each initiative, investigating the contribution of key people, meetings or activities, and discussing participants’ views about the types of knowledge or evidence co-produced and how it becomes translated into practice. Interviews were then coded (see Appendix 6, Table 23).

Partner identification tool

Data collection also made use of less common techniques, derived from stakeholder analysis and social network analysis,175 to facilitate discussion with individual embedded researchers about the key contacts or people they engaged with in their activities.

This partner identification tool (see Appendix 1) prompted participants first to identify people they engaged with on a given issue or activity, to think about the quality or frequency of each relationship, and to consider ways in which these people might be categorised. The use of this tool in workstream 3 served two key purposes: to assist the team in identifying potential participants related to the work of each embedded researcher, and to understand each participant’s reflections or views of these relationships.

Observations

Each site visit included focused observations of the embedded researchers (except at Evansville, which had ended as an active initiative). The primary purpose of these observations was to deepen our insight into the embedded researchers’ activities ‘at work’, including the types of activities carried out and the relationships developed. They included observations of project meetings, staff consultations, training workshops and research seminars. We also shadowed the individual embedded researchers on their day-to-day activities.

Diaries

Given that site visits could take place several months apart, the study design made use of reflective diaries (voice-recorded) to enable embedded researchers to capture reflections on their work. This provided data in its own right and informed subsequent interviews. We produced a short guide, outlining the purpose of the diary method. It asked the embedded researchers to describe their activities and tasks, the types of knowledge utilised, the relationships involved, and any challenges or issues faced. A flexible, semistructured approach was adopted from the outset.

Despite our best efforts to facilitate diary completion, the researchers, in general, found the diaries somewhat difficult to engage with, owing to the high level of demands of their day-to-day work. Engagement was partial and somewhat superficial, and we recognised that the data from the diaries were likely to be of rather limited use. What this lost from our study was some of the immediacy and fine-grained detail of the day-to-day activities, the tensions and emotions from living embedded work; we therefore needed to rely more on retrospective reflections gathered through interviews, on-site conversations and e-mail communication.

Data analysis

We designed a co-production workshop with the aim of testing preliminary ideas and themes to develop a shared understanding of the in-depth case study findings. The workshop was attended by embedded researchers from three of the four cases and members of their wider teams; other colleagues from the embedded network (and some who attended the first workshop) also attended. Workshop 2 was facilitated by Kaleidoscope Health and Care colleagues (see Acknowledgements). The event was thought-provoking and engaging, and we gathered extensive feedback. These outputs were particularly helpful for the workstream-3 analysis and also fed into the work of workstream 4.

The initial (descriptive) account for each case is presented in the following section. Further details of the analytic strategy, together with significant data extracts, can be found in Report Supplementary Material 3. Through descriptive overviews of each embedded initiative, we sought to show how and why embedded researchers developed their roles, what activities they undertook to co-produce and translate knowledge, what types of relationships they developed and the boundaries they negotiated, and what types of impact they made on translating knowledge into everyday practice. For each case, we looked at the:

• backstory to the initiative, including when it was established and the organisational context, such as employment positions
• rationale and specifications for establishing the initiative – its strategic reason
• features of the embedded researcher, such as their professional background and experience
• key relationships the embedded researcher developed or used in their work, including those in university and NHS contexts
• PPI built in to the initiative
• embedded activities that reflected the particular rationale and specification
• achievements and challenges that characterised the initiative.

In the next section, we outline the findings case by case under each of these headings, before exploring in more detail the diversity and range of experience for each of the descriptive categories.

Findings 1: case-by-case overview

This section provides a brief narrative account of each case, summarised in Table 7 using the seven basic features listed in the previous section. (The subsequent section will illustrate in greater depth the full diversity of experience across the cases.) The four cases have been anonymised and given pseudonyms: Bridgetown, Coxheath, Porter and Evansville.

Bridgetown

Bridgetown is a hospital-based initiative in Scotland that was formed through collaboration between an NHS trust and its local university. At the time of the study, the embedded researcher (Anna) had a joint appointment with both organisations and, in broad terms, was engaged in providing academic and professional input to a portfolio of service-facing research and improvement activities. She had a particular interest in implementation science and knowledge mobilisation theory. As Anna’s role was in its infancy, her time was spent mapping out the research culture and building relationships. Her early findings suggested a commonplace fear among front-line staff relating to academia and research; however, she was already beginning to receive positive feedback regarding her role, from the NHS trust in particular.

Coxheath

Coxheath is an acute hospital-based embedded initiative located in the South East, formed through a collaboration between an NHS trust and a local university. The initiative was well established (in excess of 8 years) and focused primarily on the role and activities of one embedded researcher, Jane, who had a joint appointment with both the trust and the university. Her role encompassed various work related to practice development, implementation and evaluation. Jane had established a broad network with multiple stakeholders, inside clinical practice, in academia and elsewhere: her collaborative relationships were at both a national and an international level. The impact of Jane’s work – and as such, the initiative – was far-reaching; however, Jane felt that her role had not yet been used as effectively at a strategic level as it might have been.

Porter

Porter represented a relatively distinctive case study. Rather than involving a single researcher, it comprised five embedded researchers working across four initiatives or areas of activity in an NHS trust specialising in community and mental health services. Together, these initiatives represented an embedded research portfolio that was nominally led by Carol, who, as the trust’s head of research and development (R&D), was responsible for some aspects of the embedded research initiative and acted as the trust’s primary link with the local universities and research ecosystem.

The five embedded researchers worked on different initiatives across the trust’s clinical divisions, typically with direct line management within the relevant division. Some of these roles could perhaps be interpreted more as traditional clinical research or quality improvement posts, rather than as embedded researchers seeking to co-produce knowledge. However, the participants and the oversight lead (Carol) did construe them as embedded researchers. The researchers faced individual challenges in their roles, but there was a shared sense of achievement relating to the sustainability of the initiative.

Evansville

This initiative was located in a large acute hospital trust and was funded for 3 years from April 2014; it has now ended and, for the purposes of this study, represented a mature and concluded initiative. Over its course, the initiative was characterised by a number of different configurations as the embedded team worked on different applied research projects, each requiring a slightly different working model. As the funding was not extended for this initiative, sustainability and legacy were relevant concerns. Despite the lack of ongoing funding, there was still a sense of enduring impacts relating to the research capacity-building undertaken, with ongoing developmental research projects and relationships.

Findings 2: comparative case descriptions

The chapter now works through a comparative analysis of each of the main descriptive dimensions outlined in Findings 1: case-by-case overview. The purpose here is to look closer at each case to develop analytic and explanatory understanding about how and why each case differs, and what implications these differences have for the co-production and translation of knowledge. We take as our organising framework the features listed in column 1 of Table 7. A full analytical description of each case study can also be found in Report Supplementary Material 3.

Backstory

Here we focus on understanding the history of the initiatives, taking into account the broader context from which they developed, such as funding arrangements, primary location and current (active) status (Table 8). Clearly, the background of each initiative is closely linked to its intended purpose and specifications, as discussed subsequently, but the history is also relevant to its trajectory, priorities and activities, making its inclusion important.

An important background issue relates to the history of collaboration between different partners, especially university and NHS organisations or leaders. These vary from highly developed institutional partnerships, often in the form of clinical academic departments, to less developed collaborations often relying more on personal connections between key individuals in each organisation. Linked to this, it could be suggested that the wider environment of applied health research, in particular translational health research, also conditions the development of each initiative, for example when there are connections with other professional research and training pathways or NIHR infrastructure.

A further consideration is the extent to which the initiative was led or shaped by one partner organisation more than another, or if it represented a ‘marriage of equals’. Certainly, one initiative (Porter) had been championed by a research-active leader based in an NHS hospital trust who then nurtured connections with university partners to further develop the initiative and its specifications.

Each initiative’s contextual background was reflected in notable practical differences in configurations, such as contractual arrangements, funding, line management and accountability. Some embedded researchers were employed solely by the trust or university with honorary status in the second institution, and these arrangements seemed to matter for subsequent relationships.

Each initiative told a different story regarding its funding. For some, the original funding was ongoing; in others, the source of funding had changed over time; and in one (Evansville), it had ended. Porter offered insight into some of these challenges and the associated negotiation, sometimes ongoing, involved. Two of its researchers, Rachel and Victoria, saw significant changes in funding throughout the course of their roles, which unsurprisingly created anxiety and uncertainty for them. For many of the researchers, contractual arrangements and sources of funding were split across different organisations; line management and accountability were therefore sometimes shared and came with competing ideals in relation to the purpose of the initiative. Researchers often had to juggle expectations in response.

The study found that each initiative had been devised, designed and established at a particular time to address a set of specific and locally significant issues that had been identified by one or two strategic partners. In all cases, however, the focus and significance of issues evolved over time, influenced by changing organisational priorities and expectations and by the experiences and expertise of all parties. As such, the more developed or ‘mature’ initiatives – those with a longer history – often had a clearer, more evolved sense of purpose. Bridgetown was a prominent example of this, whereby the experiences of the initial embedded researcher, Jill, had provided lessons for the current, ongoing initiative and her successor, Anna. Anna’s role was specifically designed to reduce the risk of the challenges Jill had faced, relating to the overall objectives, degree of impact and practical issues such as workload.

In terms of promoting the development of embedded research, each case had one or two ‘champions’ who strongly advocated the need and potential of the embedded approach. For example, the chief nurse at Coxheath (at the time of role construction) engaged with constructing a dual role and collaborative working across the university and her local hospital. In Porter, all five embedded researchers described how one or two senior people had helped get their roles off the ground and sustained them, alongside the background support of Carol (head of R&D at the trust). Embedded initiatives appear to need an advocate or champion operating at a relatively high or strategic level, who can ‘see the big picture’ and influence organisational and interorganisational decision-making, leverage necessary resources and legitimise the risk of investment.

Rationale and design specifications

To understand the rationale and specifications of each initiative, we examined their aims and intended purposes as linked to their particular background (Table 9). This analysis was informed by the findings of workstream 1, specifically the differentiation of initiatives according to their approach towards co-production (i.e. from the more conventional to the more radical; see Chapter 2). We also observed how the intent and purpose of each initiative changed over time.

Across the four cases, there were significant and stark differences in how each embedded initiative was specified in terms of purpose. This revealed important differences in how ‘embedded research’ as a broader concept or construct was seen by the wider health-care and health research communities. Initiatives were intended to address markedly different issues, such as:

• capacity and capability development (organisational or professional) (e.g. Bridgetown, Coxheath)
• professional development for nursing and allied health professional (AHP) staff [e.g. Bridgetown, Porter (Victoria and Rachel)]
• research development (design, planning and management) (e.g. Coxheath, Evansville)
• research translation and use (dissemination and mobilisation) [e.g. Coxheath, Bridgetown, Porter (Victoria)]
• applied (co-produced) research projects (e.g. Coxheath)
• service improvement projects (e.g. Porter, Evansville)
• service evaluation [e.g. Porter (Katrina and Bev)].

In all cases, there was no single or sustained purpose; rather, each initiative addressed multiple purposes that matured and developed over time. The experiences and interests of the embedded researcher(s) could have an important bearing on how these aims and purposes evolved, especially as they could bring new ideas to the initiative or mediate between the strategic priorities of multiple organisations. We found, for example, that the Coxheath initiative was shaped by the embedded researcher’s passion and commitment to professional development and co-creation. There was, however, arguably less scope for the researcher to influence the initiative’s rationale and direction when working as part of a team or with less organisational or strategic involvement. For example, some researchers at Porter, particularly those with bounded roles in service improvement, appeared to have very little influence over the purpose or direction of their role.

Although the explicit rationale and design specification of each initiative differed (through job specifications, descriptions and other documentation), there were also evident similarities in, for example, the translation of evidence into practice; empowering of front-line staff; transformation of the research culture; and, ultimately, the aim of improvements to service delivery.

Importantly, there was no single or shared understanding of the purpose of embedded research initiatives, a finding that echoed the conclusions of workstreams 1 and 2 (see Chapters 2 and 3). In each of the case studies, the embedding was designed and evolved against the backdrop of a particular institutional context to address certain strategic or operational priorities for research, research use, staff development and service improvement. However, the precise configuration or specification of these purposes varied according to the influence of key partners in the background and design, and the experiences of key stakeholders in the operation and maturation of the initiative. Significantly, embedded research could mean different things to different groups, and one of the roles of the embedded researcher could be to mediate and reconcile these differences in their own work.

The embedded researcher

As the previous sections suggest, embedded research initiatives are, in many ways, defined or differentiated by the embedded researcher(s), who, in many ways, provide the tangible focus and actionable work of the initiative (Table 10). We investigated each researcher in terms of their professional background, career and history; motivations and aspirations for undertaking the role; perceived skills, strengths and qualities; and understanding of the purpose and significance of embedded research.

Embedded researchers in the four cases held a variety of formal titles. Anna from Bridgetown was an AHP research lead and university lecturer – clinical academic. Jane, at Coxheath, was a university professor of practice development, research and innovation and an NHS trust research centre co-director. At Porter, Rachel was an NHS trust senior research associate and university honorary fellow; Bev and Katrina were service evaluation and research associates at their NHS trust; Karla was an NHS trust service evaluation analyst; and Victoria was a university associate professor of communication in mental health, an NHS research consultant and an NHS and associate research lead. Finally, Bella from Evansville was an embedded researcher team member and anthropologist.

The embedded researchers thus varied in a number of key areas or dimensions that provided important points of comparison and shaped how their initiatives functioned. These included the following:

• clinical or non-clinical background, affording different levels of understanding of clinical practice, culture and care organisations or health (or other academic) research environments and priorities
• specialism, in terms of their area of professional (e.g. nursing, AHP) or research expertise (e.g. clinical, social science, translational)
• level and type of qualification, especially whether or not the embedded research was trained to doctoral level in an area relevant to applied health or translational research
• motivation, relating to their personal interests and commitment to embedded research and knowledge co-production, especially the perceived benefits to organisational and professional partners.

There appeared to be a marked difference between ‘traditional’ academic researchers in embedded roles and embedded researchers with a professional clinical background who had subsequently trained to doctoral level in an area of applied health research. The latter background gave the researcher a dual and combined understanding of two relatively distinct areas of practice, with relatively high levels of credibility in both domains (through professional and doctoral qualifications). Therefore, they seemed more qualified and legitimate to work across the boundaries of research and practice than those who might have primary qualifications in only one area.

Porter was a particularly interesting example in terms of diversity. Here, one initiative housed a mixture of professional backgrounds. For example, one researcher had a psychology background and a PhD in family therapy, and another had degree-level qualifications and experience in pharmaceutical research. These individuals’ backgrounds held a relevance both for their own embedded roles and for the initiative more broadly: the former influencing a broader, interdisciplinary network and contributing to organisational and strategic priorities, whereas the latter carried out work largely bounded within a specific clinical context.

Perhaps most interestingly, despite the significant diversity between researchers, we found similarities in the embedded researchers’ motivations in undertaking the role. There was a strong shared sense of the potential for research to provide tangible and applied ‘real-world’ change in the design and delivery of health services, and to improve patients’ experience of this care. This motivation was also relevant for those supporting the initiatives, particularly initiatives based in clinical practice.

Relationships

Key to the role of the embedded researcher is the ability to develop, use and broker relationships within and across the research and practice boundaries. We therefore investigated each embedded researcher’s patterns or networks of relationships, and how they understood these networks as informing and supporting their activities. To do this, we used the partner identification tool (see Partner identification tool and Appendix 1) in conjunction with qualitative interviews. The tool helped participants think about, identify, appraise (in terms of frequency or quality) and categorise (as NHS, university, other) their relationships.

The study found that interpersonal relationships, and the strength of these relationships, were fundamental to how the initiative was configured and functioned. Key contacts (Table 11) were seen as fundamental for each initiative’s success and sustainability, and also played an important role in securing and maintaining funding. As noted previously, such ‘champions’ offered sponsorship, endorsement and a degree of protection for the researchers and for their initiatives more broadly. These individual champions spanned clinical settings and university, although examples of the former were more widely noted. Champions tended to be those in senior management roles with associated status and power, with the ability to ‘open doors’, fight the researcher’s corner and ensure that they had ‘a seat at the table’.

The diversity and scope of the researchers’ networks and intrinsic relationships differed greatly. These networks can be seen most clearly in the completed partner identification tools for each embedded researcher depicted in Figures 5–11 [note that Bev and Katrina at Porter (see Figure 8) completed their partner identification tool together].

FIGURE 5 Embedded researcher key partners: Bridgetown – Anna. (Figure not reproduced here; see the original article.)

FIGURE 6 Embedded researcher key partners: Coxheath – Jane. HEE, Health Education England; HR, human resources. (Figure not reproduced here; see the original article.)

FIGURE 7 Embedded researcher key partners: Porter – Rachel. (Figure not reproduced here; see the original article.)

FIGURE 8 Embedded researcher key partners: Porter – Bev and Katrina. (Figure not reproduced here; see the original article.)

FIGURE 9 Embedded researcher key partners: Porter – Karla. QI, quality improvement. (Figure not reproduced here; see the original article.)

FIGURE 10 Embedded researcher key partners: Porter – Victoria. CAMHS, child and adolescent mental health services; QI, quality improvement. (Figure not reproduced here; see the original article.)

FIGURE 11 Embedded researcher key partners: Evansville – Bella. (Figure not reproduced here; see the original article.)

As the completed tools show, all embedded researchers had developed relationships across university and NHS organisational boundaries, but it was also clear that some relationships were weighted towards one more than the other, often depending on their background and contractual location or home. Given the focus and intent of all the embedded researcher initiatives (largely, research-informed service improvement), relationships within the NHS (especially NHS trusts) were seen as especially important. Moreover, some of the researchers at Porter (Bev and Katrina in particular) had very few academic relationships and infrequent contact with academia because of the ‘service evaluation’ nature of their work. As a result, university contacts were informal and based largely on involvement with previous treatment trials in their clinical setting. In comparison, but also at Porter, Victoria had strong, formal (contractual) relationships with both practice and academia. Again, the intent and purpose of her role was relevant: it was much more flexible, diverse and targeted at a broader organisational and strategic level.

In addition, some embedded researchers had large local, regional, national and international networks, involving a range of different stakeholders and collaborators. These could be drawn on to enrich the work of the embedded research (in terms of new or different ideas) and also to spread learning beyond the local initiative.

Overall, our research found (albeit to varying degrees and reflecting their particular backgrounds and contexts) that the embedded researchers acted as relational conduits or brokers within and between different communities to support the (co-)production, dissemination and utilisation of research to address particular strategic and operational purposes (see Embedded activities).

Most significantly, relationships with senior actors, within either or both the NHS and university sectors, were seen as particularly important for the success of each initiative. In broadly similar ways (but varying according to the level and experience of the embedded researchers), these relationships provided:

• clarity and assurance about the embedded role and initiative (intent and purpose)
• ongoing mentorship, support and guidance (for academic and/or clinical development)
• sponsorship or buy-in for embedded research initiatives with third-party groups or organisations, by ‘opening doors’ and ‘providing air cover’
• additional support and assurance around contractual and funding matters.

Researchers, then, were embedded, not just in service-related settings, but in dense and diverse networks across multiple institutions. Creating, maintaining and navigating these networks was clearly a major part of their day-to-day activities.

Patient and public involvement

The level of engagement with formal PPI groups in the embedded initiatives was variable, as were the structures and mechanisms that underpinned this involvement (Table 12).

This finding only illuminates further the level of diversity (in intent, structures and processes) across the case studies, especially in terms of stakeholder involvement. Often this relates to the nature of the embedded researcher and the distinction between an explicit embedded researcher team (e.g. Evansville) and a single embedded researcher working on various, individual projects (e.g. Coxheath). As an example of the latter, as noted in Table 12, PPI at Bridgetown was not relevant for Anna’s broad embedded role, but was relevant for external projects [e.g. through INVOLVE at the NIHR: see www.invo.org.uk (accessed 3 March 2021)] and also in terms of her capacity-building work, whereby Anna pointed clinical staff towards more effective PPI in their own work.

It is important to note that experiences of PPI were shared only sparingly during the interviews and observation periods, although specific questioning by e-mail did elicit useful additional data. An analysis of the relational dimension for the PPI activity set out in Table 12 is offered in Chapter 5.

Embedded activities

The embedded researchers took part in a variety of activities and tasks commensurate with the prescribed intent and specification of their initiatives; so some activities in those initiatives oriented towards capacity development differed from those more concerned with service improvement. We observed striking differences in relation to the diversity and variability of these activities between initiatives, as outlined in Table 13.

Detailed examples of the spectrum of activities in each of the case study sites are given in Report Supplementary Material 3, but, in brief, these comprised different mixes of the following:

• capacity development – supporting and reviewing funding applications (e.g. master’s degrees or PhDs); methods training; assisting with individual projects (research design and implementation); mentorship, support and encouragement; relationship-building; signposting
• improvement – service evaluation (information collation, report writing and dissemination), quality improvement (e.g. individual project support and supporting broad organisational agenda), supporting clinical colleagues with involvement in improvement work
• research development and use – writing for publication (academic and practice audiences); grant applications; presentations (e.g. at conferences, events); networking and building opportunities for collaboration; implementation and dissemination.

As with other aspects, the embedded activities often evolved over time. In some cases, this change of focus was less visible or perhaps even intended. For example, as the research team at Evansville became established and relationships with front-line staff grew, the researchers found that their involvement in capacity and capability activities also developed and became more prominent. This had implications for their workload (as their role and work related to ongoing health services projects). The success of this ‘extra’ work left a void in some areas when the initiative ended.

Activities associated with more ‘traditional’ and ‘original’ research were also diverse, again relating to the overarching purpose of the initiative, but also the individual role. For some, contact with research in this sense was in building site capacity for randomised controlled trials. For others, they took part in funding applications, publications and various dissemination activities, offering a further example of contrast between roles in different initiatives.

The level of activity relating to strategic involvement and influence was also an important observation when examining the embedded researcher’s role: their strategic ‘voice’ either directly or indirectly was variable. For some, attending senior management team meetings was an explicit part of their role, for others their ‘attendance’ was indirectly through line managers and key strategic-level contacts. Some researchers undertook no activities outside their clinical specialty or department. In the main, the potential to engage embedded researchers in strategic activities was not always thought to have been fully realised.

Achievements and challenges

Embedded researchers and identified stakeholders were asked to reflect on the achievements and challenges of the initiative. There were stark differences in responses both within and between case study sites, suggesting differences in the perceptions, expectations and assumptions of participants. In other words, it was not easy to list achievements that were universally recognised or agreed on; rather, each local stakeholder group saw particular benefits and achievements, but also challenges, with each initiative (Table 14).

Despite the diversity of intentions and processes among the different initiatives, we were able to collate insights into some possible criteria for assessing and measuring success. Key areas where the nature of success might be explored include:

• capacity development – positive staff feedback; clinicians engaging in further research training and seeking out support (e.g. workshops, enquiries, e-mails); clinicians applying for grants, further research education (e.g. master’s degrees, PhDs, scholarships); front-line staff presenting at conferences and other events
• improvement – tangible outputs such as report writing at departmental level (achieving deadlines), meeting the requirements of governance (relating to improvement involvement), interest and engagement with improvement initiatives from front-line staff, changes to practice based on findings
• research development and use – grant applications (number and quality), publications (number and quality), uptake of findings in practice (impact case studies), international collaborations (and uptake of findings outside own organisation, nationally and/or internationally).

Importantly, the degree and nature of achievement of each initiative reflected, in part, their maturity. For example, one embedded researcher had been in post for only 12 months, limiting the scope or reach of their achievements, but also making more noticeable the significance of their impact in such a short time, especially in terms of building new relationships and networks around the role. In comparison, another long-established multiresearcher initiative showed individual role achievements (largely service improvement) and a broader sense of longevity.

Given the challenges of measuring success or achievements, it seemed that, when initiatives were more narrowly defined in their specific activities or goals, stakeholders were more able to comment on their relative success in delivering within an agreed time frame. Similarly, and again noting ambiguity, roles that were less bounded (in terms of focus, activities and scope) were often victims of their own success, and personal capacity became problematic. This was less challenging in roles bound by the requirements and focus of a specific project, service or department. Conversely, those with more ambitious or ambiguous aims, such as capacity-building or cultural change, could be seen and assessed in less concrete ways, with the likelihood of disagreement among stakeholder groups as to the level of success attained.

Concluding remarks

This chapter has presented the methods and data sources used in the four intensive case studies explored in workstream 3. In this initial analysis, we focused on ‘telling the story’ of each initiative, drawing attention to their key features, backstories and trajectories. Some interim conclusions are drawn here from the analysis, but the main discussion, in the light of the other strands of work in the project and the extant literature, and the drawing of conclusions for future scheme development take place in Chapter 8.

The over-riding conclusion from this descriptive analysis is that there is no single model or specification for embedded research, an issue also highlighted by the literature reviews of workstream 1 and the scoping work of workstream 2 (see Chapter 3). Although there may be general understandings of what an embedded initiative might involve or hope to achieve, they are interpreted, translated and enacted in different ways in local contexts. These variations reflect important historical and contextual factors relevant to each case. For example, the particular contextual priorities and pressures facing partner organisations at any given time are likely to shape how the initiative is framed and specified. Equally, the relative investment, funding or commitment to the initiative by different partners also directs the intention and specifications towards certain purposes over others.

Therefore, an embedded research initiative developed in a community hospital with relatively diffuse and limited involvement of university partners is likely to be different from an initiative created through a long-established university–NHS partnership to address specific improvement or capacity development issues. Moreover, as was evident in the trajectory of Bridgetown, where the initiative morphed as the embedded researcher changed (from Jill to Anna), initiatives themselves can shift focus and goals over time. History, context and emerging contingencies, especially in terms of funding and local service priorities, shape the rationale and specifications of any initiative.

Equally, initiatives vary according to the type (and number) of embedded researchers involved. Important variations can be seen in the embedded researchers’ professional backgrounds and their academic and professional qualifications, as well as their contractual and funding arrangements. That said, all the embedded researchers in our cases seemed motivated (albeit in slightly different ways) to close the gap between research and practice, by bringing research expertise closer to or within front-line clinical practice.

Key to each embedded researchers’ activities in closing this gap was the development and use of relationships across university–NHS boundaries. To varying degrees, embedded researchers represented, not just research capacity, but knowledge-brokering capacity. Where they varied was in what knowledge they brokered, whom they brokered between, and how they viewed the nature of knowledge or evidence. All the embedded researchers in these initiatives employed collaborative approaches to knowledge creation, with some embodying the more committed approaches to knowledge co-production explored in Chapter 2.

Success in any given initiative was often hard to define and was, in part, determined by the clarity with which the initiative was specified, for example whether or not it had bounded goals and explicit deliverable outcomes. Achievements could more easily be discerned in the form of successful processes, networks and relationships, and only sometimes in the shape of improved services. Challenges abounded across all the schemes, which suggested that the successful implementation and sustaining of embedded research is difficult. However, many of our participants (embedded researchers and wider stakeholders) spoke with enthusiasm of the potential and benefits of working in these ways.

A fuller discussion of these findings in the context of the literature can be found in Chapter 8. In Chapter 5, we turn to a deeper cross-case analysis of the four initiatives introduced here, to tease out the features and dynamics of embedded research.