New Game,Old Rules? Mechanisms and Consequences of Legitimation in Boundary Spanning Activities

Research setting

This paper draws on a qualitative longitudinal case study conducted in a large-scale collaborative partnership (‘Collaboration’) established in 2008 between a university and UK National Health Service (NHS) organizations. This partnership illustrates a recent international policy trend towards the creation of system-level collaborative entities in which researchers work closely with other stakeholder groups, such as clinicians and managers. Such collaborative initiatives seek to create an environment in which research findings can be more readily applied in day-to-day clinical practice, increasing the effectiveness and efficiency of healthcare service provision (Evans & Scarbrough, 2014).

This particular case was selected because the Collaboration was one of the few similar organizations in the country deploying designated boundary spanning roles to support knowledge translation between the producers and users of research. It also granted us a broad access to collect research material prospectively and longitudinally. Most importantly, the Collaboration represented a theoretical case suitable for exploring our research questions: from an ‘organization-as-field’ point of view (Emirbayer & Johnson, 2008, p. 22), it can be conceptualized as a joint ‘field-in-the making’ emerging at the interface of the established domains of science, medicine, management and policy (Kislov, 2014).

The Collaboration recruited eight designated non-clinical facilitators (‘boundary spanners’), whose broadly defined remit included a two-way transfer of knowledge between the university and multiple primary medical care organizations (‘general practices’) and supporting the latter in applying research evidence in their day-to-day practice. Designed by the Collaboration’s management academics, this newly created full-time role was adapted from the Knowledge Transfer Partnership (KTP) scheme which facilitates the links between the academic research base and industry in order to identify research-informed solutions to specific problems faced by practitioners (Acworth, 2008). Prior to joining the Collaboration, two facilitators worked in the NHS as nurses, three were employed in various managerial roles in the public and private sector, and the other three had recently graduated from university and had no previous full-time work experience. Embedded in cross-boundary multiprofessional project teams, the facilitators had hot desks both in the university and the NHS. They received guidance from (and were accountable to) management academics, clinical academics and programme managers, all of whom contributed to the development of this novel role.

The Collaboration’s boundary spanning activities were aimed at the creation of new co-produced programmes of research implementation, ultimately resulting in improved patient care in the participating general practices. In the UK, general practices are the main organizational actor in the field of general medical practice, a subfield within the field of medicine with a number of unique characteristics that need to be taken into account for the subsequent analysis (McDonald, 2009; McDonald, Checkland, Harrison, & Coleman, 2009). First, general practitioners (GPs), endowed by a high amount of field-specific cultural capital, are the most powerful professional group in this field, which contrasts with their relatively low status in the broader field of medicine. Second, most of the GPs, rather than being salaried public sector employees, enter into a contract with the national commissioning body, which means that GPs’ economic capital is invested in their practice and that the latter needs to be economically viable if they are to remain in the ‘game’. Finally, general practices are organized into clinical commissioning groups (CCGs) whose governing bodies are responsible for commissioning the majority of health services in the country (Checkland et al., 2013). ³

Data collection and analysis

This study emerged from a broader research programme which was conducted in 2009–11 and explored the internal boundaries of the Collaboration (Kislov, 2014). Changes in the patterns of boundary interactions that became apparent relatively early in the course of fieldwork inspired a longitudinal extension of the initial programme with an explicit Bourdieusian focus on the legitimation of boundary spanning roles and practices. In total, 88 research participants were purposefully sampled both from the Collaboration’s core team and the fields of applied health research and general medical practice. The eight facilitators were naturally seen as key informants, and all of them were interviewed in the course of this research. At the same time, we significantly expanded the sample by including representatives of multiple groups who were involved in the intrinsically collective boundary spanning activities (Currie & White, 2012; Kislov et al., 2016a) and/or were in a position to contest or disapprove the legitimacy of the Collaboration’s boundary spanning roles and practices.

Semi-structured interviews (30 to 95 minutes in duration) served as the main method of data collection and were conducted (face-to-face or by phone) in two rounds (2009–10 and 2012–13) (Table 2). Almost all of the Collaboration’s facilitators and managers remained in their posts for the second round of data collection and were interviewed twice. However, as different general practices participated in the Collaboration’s projects in 2009–10 and 2012–13, this part of the sample significantly differed between the two rounds, remaining comparable in terms of the professional and organizational groups represented. In addition, a focus group was conducted with all facilitators at the end of 2013 to discuss the development of legitimacy over time. The first author also regularly engaged in informal conversations with research participants and observed meetings in which they participated, which provided opportunities for clarification and triangulation, significantly enriching the analysis.

OPEN IN VIEWER

Table 2.

Research participants.

	Interviewed in both rounds	Interviewed in the first round only	Interviewed in the second round only	Total
Facilitators/boundary spanners	7	—	1	8
Programme managers (Collaboration)	2	—	—	2
Practice managers (primary care)	—	7	6	13
Managers (professional networks)	—	2	—	2
Specialist nurses	1	4	3	8
Practice nurses	1	2	12	15
General practitioners (GPs)	1	7	13	21
Applied health researchers	—	3	1	4
Management academics	—	3	—	3
Clinical academics	—	3	—	3
Primary care trust (PCT) executives	—	4	—	4
Clinical/care coordinators	—	—	3	3
Healthcare assistants	—	—	2	2
Total	12	35	41	88

The interviews and the focus group were digitally recorded and transcribed verbatim; transcripts were coded and analysed with the aid of NVivo. The first stage of data analysis was predominantly inductive, involving a series of emergent descriptive codes and following a narrative analytical strategy that aimed at the construction of a detailed story from the raw data (Langley, 1999). The second stage of analysis involved a shift towards an ‘interpretative/theoretical case’ (Pettigrew, 1990, p. 280), linking the emerging narrative with the Bourdieusian framework and related theoretical debates in the literature. This involved the recoding of the two datasets, with most of the codes derived from the work of Bourdieu (e.g. embodied cultural capital; field boundaries; social trajectory). Matrix analysis was deployed to compare the coded material across different rounds of data collection (Table 3) and subgroups of boundary spanners (Table 4). This resulted in developing three main themes described in detail in the following empirical sections: (1) changes in the distribution of economic, cultural and social capital mobilized by boundary spanners; (2) implications of these changes for the relationships between the intersecting fields; and (3) effects on the social trajectories of boundary spanners.

OPEN IN VIEWER

Table 3.

Evolution of capital in the process of legitimation.

	2009–10	2012–13
Economic capital	- Buying out physicians’ and nurses’ time- Financially remunerating the practices for participation	- No direct financial incentivization- Indirect incentivization through linking project outcomes with the pay-for-performance system
Embodied cultural capital	- Previous work experience, often from non-related fields- Quick acquisition of sufficient knowledge about (but not competence in) clinical conditions- Emerging contextual awareness about the field of general medical practice	- Project management knowledge and skills applicable to a wide range of projects- Ability to tailor the message to the expectations of different groups- High level of contextual awareness
Objectified cultural capital	- Evidence derived from international and national guidelines and systematic reviews- Emerging local evidence of improvement achieved by ongoing projects	- Evidence from the previous projects run by the Collaboration (explicitly linked to local priorities and interests)- Tools, guides and websites produced by the Collaboration (and evidence of their impact on performance)
Institutionalized cultural capital	- Lack of clinical qualifications and an ‘academic’ job title seen as a hindrance (internally more than externally)- The ‘outsider position’ signals ‘no threat’ to clinicians’ professional autonomy	- Collaboration’s reputation in the field is growing- Its association with the university and ‘research’ is seen by some as a hindrance but legitimizes the evaluation-related activities
Social capital	- Connections with opinion leaders (mostly from the field of secondary care)- A broad network of contacts across different healthcare sectors- Boundary spanners acting as ‘bridges’ between different general practices	- Connections with CCGs – dominant actors in the field of general medical practice- Decreased reliance on clinical opinion leaders (especially those outside the field of general medical practice)

OPEN IN VIEWER

Table 4.

Variation in the evolution of capital among boundary spanners.

	2009–10	2012–13
Boundary spanners with prior nursing experience	- Generalist clinical knowledge is useful for understanding the clinical aspects although not always directly applicable to specialist clinical areas- Contextual knowledge of the NHS and clinical qualifications are seen as a distinct advantage- Perceived lack of project management skills	- As the boundary spanning role shifts towards project management, clinical knowledge is seen as less important- Acquisition of project management skills
Boundary spanners with prior experience in management	- Project management skills are seen as an advantage- Intense acquisition of contextual knowledge of the NHS and (disease-specific) basic clinical knowledge in the first 12–18 months- Cases where absence of clinical qualifications becomes an obstacle for dealing with clinicians are rare	- Project management skills are valued even more as the boundary spanning role becomes more managerial- Disease-specific clinical knowledge is seen as less important
Boundary spanners with no prior work experience	- Lack of NHS/managerial experience is initially seen as a major drawback within the Collaboration (less so externally)- Support has to be provided by the more experienced members of the Collaboration’s project teams- Intense acquisition of contextual knowledge of the NHS, (disease-specific) basic clinical knowledge and managerial skills	- Similar to boundary spanners with prior experience in management

Economic capital

Our data clearly indicate the decreasing role of direct financial incentivization in the process of legitimation. In the beginning, some of the Collaboration’ projects provided remuneration to general practices for their participation and/or bought out clinicians’ time which could then be used for project work. Interestingly, the clinicians held somewhat ambiguous views about the Collaboration’s incentivization policies, perhaps due to the intrinsic propensity of economic capital to remain misrecognized (Bourdieu, 1986). Most clinical respondents claimed that their engagement was driven by the need to improve knowledge of a certain clinical area and the quality of patient care rather than by financial incentives, saying that the money ‘didn’t enter any of the team members’ heads’ (PM1-R1). ⁴ At the same time, they often admitted that economic incentivization was an important factor for ‘some other people’ working in the same practice and increased the practices’ motivation and capacity to do project work with the Collaboration’s boundary spanners:

For me, [the funding issue] wasn’t [important], I must admit. For the GP it was, because we had to get a locum in to cover him. (PM5-R1)

We worked more than we’ve been paid for, but it certainly comes in handy . . . (PM2-R1)

Money certainly helps. It does certainly make people more motivated. . . (GP2-R1)

Over time, however, direct financial incentivization of practices started to be seen as requiring too many resources, making it unsustainable in the long run. In addition, it turned out to be insufficient on its own for recruiting the practices and engaging them in joint work with the Collaboration’s boundary spanners. In the later stages, the Collaboration stopped providing monetary remuneration to project participants, relying instead on indirect economic incentivization through linking the project activities with the pay-for-performance system adopted in the primary care sector:

I personally don’t think we should [pay people to take part in the project] because it’s work that they should be doing anyway. . . . As a team we have definitely taken a different approach to that. We don’t pay people like we did at the beginning. (BS7-FG)

There are [pay-for-performance] indicators around improving your register. So that was the main driver for the project. (BS8-R2)

Making these links explicit was often used by the boundary spanners, especially in the second half of the Collaboration, as a way of engaging the general practices in the Collaboration’s projects:

We can demonstrate some of the benefits [to the practices] . . . ‘Every patient you add on to your register is however many [pay-for-performance] points and this amount of money’ kind of thing . . . (M2-R2)

Cultural capital

The boundary spanning roles created at the outset of the Collaboration were explicitly non-clinical, although two of the boundary spanners had a background in nursing. This meant that most boundary spanners did not possess sufficient clinical knowledge, skills and expertise (embodied cultural capital) or clinical qualifications (institutionalized cultural capital) and had to build their legitimacy with clinicians in other ways. First, the boundary spanners quickly acquired some knowledge about the clinical conditions they were working with, which often required an extensive reading of research papers and clinical guidelines until they felt that this knowledge was ‘good enough’ to allow them to implement their duties:

After a time you realize that you can’t know everything. I’m not a clinician, I’m not going to know everything, but I can get to the point where I know enough and understand enough to hold a conversation without stepping on anybody’s toes, and then liaising or speaking to people who know an awful lot more than I do. (BS2-R2)

Second, they developed a deep understanding of the NHS context in which they were operating. One of the managers described this growing contextual awareness as:

Understanding the context, understanding the NHS as a whole and understanding how the different groups of stakeholders that you work with within the NHS all fit together and what their motivations are and what are . . . the levers you can use to engage them. (M2-R2)

This knowledge allowed the boundary spanners to tailor their message differently for different groups operating within the field:

At a CCG level it’s more about delivery . . . so we adapt to that level and we tap into what we know they want . . . We know they want an increase in prevalence . . . so when we go and talk to them, we will emphasize that . . . Whereas in a feedback meeting [at a general practice], if that’s not what we think makes them tick, then we won’t focus on that. We’ll focus more on clinical management. (BS3-R2)

Finally, boundary spanners emphasized their position outside the medical field as a way of differentiating their role from that of clinicians and preventing the latter from perceiving boundary spanning activities as an infringement of their professional autonomy:

It can be an advantage, this outsider position: ‘I have no clinical knowledge at all; I don’t touch your expertise there. We are working in two very different arenas and let’s find a way to work together.’ We’re establishing almost the same goal and then saying, ‘I’m not a risk to your credibility and your status.’ (BS5-R1)

The use of objectified cultural capital (documents, tools and evidence) by boundary spanners also changed over time. While national-level guidelines and statistics were heavily relied upon as a source of legitimacy in the beginning, the focus shifted in later stages towards relying upon the context-specific evidence produced by projects in the local health economy. Where electronic tools or improvement guides were produced by the Collaboration teams, a conscious effort was made to present the evidence of their effectiveness on financial savings and/or patient outcomes:

So you just have to find that thing that really plays on them and influences them. Sometimes it can be the financial outcome of it; sometimes it can be the patient outcome of it. (BS1-R1)

Starting to use the audit tool in the second phase of work was actually a really powerful way to change practice because we were able to present concrete evidence to people. (BS6-R2)

Social capital

Unsurprisingly, the importance of building relational networks was seen as a major enabler of legitimacy. The focus of these networks, however, differed across the two rounds of data collection. In the early stages, connections with clinical opinion leaders, some of whom were formally affiliated with the Collaboration, were used by the boundary spanners both as a source of contacts and as a symbolic form of capital conferring legitimacy by virtue of their perceived reputation and influence in the field of general medical practice:

The Kidney Tsar is a great reference source, who gives us a lot of credibility when we’re giving information [to the practices]. We can pass it on to a third party and they say, ‘Where’ve you found this from?’ and we can say, basically, ‘This is where we’ve heard this from.’ It’s a very, very credible source to have. (BS6-R1)

The second-round data show a relative decrease in the importance of clinical opinion leaders as a source of boundary spanners’ legitimacy. First, many of the opinion leaders were hospital doctors, which increasingly rendered their reputations, knowledge and contacts less relevant for the boundary spanning activities which were becoming more focused on the field of general medical practice:

When it comes to practice-level delivery . . . they couldn’t give two hoots if [the hospital doctor’s] name is at the top of it or not; it doesn’t really have an impact then. (M1-BS2)

Second, as the boundary spanners were increasingly accumulating other forms of social and cultural capital, they relied less on their connections with influential clinical opinion leaders:

It was [the clinical lead’s] contacts, liaisons, links, clinical colleagues who were really important initially, and now she kind of takes a step away from that and we use her . . . for advice about new things emerging rather than for a contact into different places . . . [Boundary spanners] do all that work rather than using [her] contact here, [her] contact there, because you have much more knowledge . . . (BS2-FG)

Finally, while the authority of clinical opinion leaders was instrumental for the initial engagement with practices, it proved insufficient for recruitment on a large scale and, in particular, for the spread of interventions from the practices taking part in the piloting stage of the Collaboration’s projects to a wider group of primary care organizations. In the later stage of the Collaboration, relationships with CCGs – local commissioning bodies – were seen as the most important form of social capital:

Now it’s gone beyond just having that one person, like one GP, and in the projects that we’re running now . . . we’re invariably dealing with the CCG . . . (BS7-R2)

The CCG level – that’s where our credibility comes; being able to pull things together to demonstrate that it’s not just about these individual practices. (BS3-R2)

Power relationships between the fields

The impact of the boundary spanners’ evolving capital structure on power relationships between the Collaboration and its primary care partners was complex. In many cases, boundary spanners were perceived as influential change agents, helping clinicians modify their everyday clinical work by applying clinical guidelines or service improvement approaches:

[The boundary spanner] could bring the knowledge of how to manage the patients. . . and through the meetings that knowledge was transferred to us which is now in our practice and working there. (PM6-R1)

[The boundary spanner]’s role was. . . reminding us about where directing us, where we needed to be . . . We were glad when it was over, in some respect. Because . . . we felt under a lot of pressure to perform with it. (PM2-R1)

The work that the general practices had to do as part of the Collaboration’s projects almost always fell within their normal clinical remit, relied on voluntary participation, was agreed with participants prior to the beginning of the projects and was financially rewarded either directly (through cash payments from the Collaboration) or indirectly (through meeting the practices’ performance targets). However, as the last quote shows, the requirement to perform more work was often met with at least some degree of resistance. In some cases, where the expectations, needs and opinions of local powerful players were not taken into account, the Collaboration failed to gain legitimacy with and exercise influence over primary care partners:

We hit a brick wall in terms of our credibility when we went out and said, ‘This is what we are doing here.’ And our credibility in some places, where that didn’t align, plummeted . . . Where that happened, we’re not credible at all because we didn’t deliver what they expected, and their expectations weren’t managed at all. (BS3-FG)

Rather than imposing something on people – it’s not about some secondary care clinician who happens to be a clinical lead for [the Collaboration] who’s got this idea and then goes out and tries to impose that on various PCTs – it’s more what [the partners] are interested in. (M2-R2)

As the last quote indicates, the authority of the Collaboration and its boundary spanners to enact change in primary care was contingent on how well their activities were aligned with the interests of primary care stakeholders. In the early stages, the recruitment of the practices and gaining access to them was managed on a practice-by practice basis, often involving ‘ego massaging’ (BS2-R1) of locally influential clinical opinion leaders, with an understanding that ‘you have to make yourself useful’ (BS5-R1) to general practices. In the later stages, the boundary spanners were able to negotiate involvement of larger groups of practices through demonstrating the ‘dirigible outcomes’ of projects (BS2-FG) to commissioners and persuading them that participation in the Collaboration could help them meet their financial and patient care targets.

Exclusion of applied health research from boundary spanning

It was expected that the Collaboration would span the boundaries between several key domains, including applied health research, implementation, health policy and clinical practice. Bridging the boundary between researchers and practitioners was seen to be of utmost importance. However, the accumulation of capital viewed as legitimate in the field of general medical practice was not matched by a corresponding increase in boundary spanners’ legitimacy with applied health researchers. Boundary spanners were seen by the latter as ‘ill-equipped to be trying to propagate best practice’, both due to the lack of research skills and because they ‘hadn’t got enough of a relationship’ (HR2-R1) with clinicians to gain access to the data that the researchers were interested in obtaining. Boundary spanners, in turn, felt disadvantaged by their lack of scientific knowledge and credentials:

I haven’t got the academic background and I’m always aware of that, and sometimes it makes me feel uncomfortable when I’m sat around a lot of researchers because I can’t talk on your level. (BS4-R1)

It is therefore unsurprising that the intermediary position of boundary spanners at the interface of applied health research and general medical practice shifted towards the latter:

There’s been very little work with [facilitators] . . . from the perspective of my research. (HR2-R1)

There’s clearly some academic research work going on, but I’m not sure what it is, and I’m not quite sure how relevant it is to us as a primary care trust and working with our GPs . . . The [facilitation] stuff is the important stuff. It’s the research stuff that feels less important and distant to us. (PCTE1-R1)

At the same time, there remained some ambiguity in the boundary spanners’ relationship with research. On the one hand, there was little interaction with applied health researchers, and a perceived connection between the Collaboration and the scientific field was claimed to have a negative effect on the boundary spanners’ legitimacy with the field of general medical practice:

Sometimes there’s a bit of misunderstanding, that [primary care staff] maybe think that we do research . . . It can really put practices off then, can’t it really? (BS7-FG)

On the other hand, boundary spanners claimed to be using peer-reviewed papers and the national guidelines to inform the implementation projects in primary care (especially in the initial phase of the Collaboration). In addition, they admitted that the research-informed evaluation work built into most of the projects was perceived by healthcare organizations as an enabler of the boundary spanners’ legitimacy:

The CCG . . . see us as being much more forthright and more credible [than a consulting firm] regarding evaluation work. (BS2-FG)

Variation in initial capital endowments and internal legitimacy

In the first round of data collection, boundary spanners’ core responsibilities included a wide range of frontline activities involving coordination, negotiation and facilitation:

[The boundary spanners] transferred knowledge about problems that other practices were having, and how they’d got over that; they transferred knowledge about things we were doing to other practices. (PM1-R1)

What am I doing? It’s a lot of networking, stakeholder engagement, chasing people up and making things happen, negotiating, finding solutions, convincing people, and a lot of data analysis. (BS5-R1)

While remaining within the broad remit described above, in the first half of the Collaboration’s lifetime internal legitimacy of boundary spanners within the Collaboration was affected by their initial capital endowments determined by their previous experiences in other fields. Facilitators with a nursing background experienced a distinct advantage of possessing both institutionalized (clinical qualifications) and embodied (general clinical and contextual knowledge) cultural capital specific to the field of general medical practice. This enabled them to ‘open doors earlier and quicker’ (M1-R2), not least because ‘having certain letters behind your name helps in dealing with clinicians’ (BS5-R1), but also meant that they were not free from ‘the cynicisms and the preconceived ideas’ (M2-R2) about the NHS. By contrast, those facilitators who had no work experience prior to assuming the boundary spanning role were significantly disadvantaged by their lack of capital and widely perceived within the Collaboration as ‘pretty green and pretty junior and inexperienced’ (CA3-R1):

That was a bit of a stumbling block at the beginning, because they knew nothing about the NHS, nothing about the culture, or anything . . . They’ve had to learn, quickly, how the NHS works. (BS7-R1)

In addition to having ‘a lot to pick up in a very short space of time’ (BS6-R1), novice non-clinical facilitators had to be supported by the more experienced academic and managerial staff:

Because, obviously, [the clinical academic] is a working clinician and she was kind of the other part of me with these people, then that kind of filled the full . . . So I always had the best possible clinical input that I didn’t provide. (BS3-R1)

With the exception of rare cases where non-clinical boundary spanners were dismissed by clinicians due to the lack of clinical knowledge and credentials, in general these strategies (along with those described earlier) proved successful in establishing their external legitimacy with the NHS despite their capital deficiencies:

Medical background was our territory. . . .What [the boundary spanners] were telling us is: what we should be looking [at] next, what we should be doing next. And it was our clinicians’ job to reach those targets. So they didn’t need to have medical knowledge. (GP4-R1)

Shift from boundary spanning towards project management

With an accumulation of managerial knowledge, contextual awareness and relational networks, initial differences between boundary spanners became less prominent, with a corresponding increase in the status of novice non-clinical boundary spanners within the Collaboration:

. . .Now we see our guys are very skilled at working with primary care and they are able to relate to them a lot better than they were three or four years ago. . . .When you actually get in the meeting now, [non-clinical facilitators] can soon come over as very credible and they can soon build an effective relationship. . . They’ve become like opinion leaders in their own right. (M1-R2)

As their authority grew, boundary spanners started to be more involved in managing multiple projects, with a corresponding prioritization of skills related to project management over disease-specific knowledge:

With the [previous project]. . . I was reading high-level clinical journals. . . With the [current project] it’s completely different. I don’t know anywhere near as much detail. . . it’s much more about the process where my key skills lie. . . in providing advice to help facilitation, almost being like a project manager, structurally organizing things, developing systems. . . much more so than having to know the clinical ins and outs of disease management. (BS2-R2)

The ongoing shift from an intermediary boundary spanning role towards a more traditional project management function was further reinforced by the secondment to the Collaboration’s project teams of several part-time clinicians, who took over some of the boundary spanning functions (Kislov et al., 2016a):

It’s been a natural process that as we’ve become more experienced in delivering and leading the project, we’ve become more project management focused. And then the specialists we’ve brought in to support the project have become more of the actual boundary spanners. (BS6-R2)

In an interesting reversal of fortune, clinical knowledge and credentials of boundary spanners with a nursing background now mattered less, with the managerial skills of non-clinical boundary spanners being favoured in the new distribution of capital within the Collaboration:

When I first started I know what made me credible is the fact that my background is in nursing . . . but I don’t think it’s as much as it was at the beginning. (BS7-R2)

It might be worse to bring clinical staff in, because they don’t have the more core business skills really. . . But there’s an advantage to bringing in people with some . . . core project management skills. (M1-R2)

Evolution of capital

Similarly to previous observations (Levina & Vaast, 2005), the boundary spanners in this study both accumulated and related different forms of capital in the process of joint working with the field of general medical practice. However, as our findings show, legitimation not only increases the volume of the overall capital mobilized but also involves significant changes in the relative weights, mutual convertibility and original sources of its specific forms, leading to the transformation of capital structure over time.

First, we demonstrate the changing role and nature of economic capital mobilized in the process of legitimation. Despite the acknowledged importance of economic factors in general practices as independent businesses (McDonald, 2009), boundary spanners increasingly relied on cultural and social capital rather than on the direct use of economic capital. The latter, however, was often mobilized indirectly, whereby linking the projects and their outcomes to the nationally mandated metrics led to a delayed but substantial financial benefit for the participating general practices. This shows how the linking of disparate forms of objectified cultural capital, such as project data, audit tools and performance measurement systems, can lead to their (partially disguised) conversion into economic capital (Bourdieu, 1986). At the same time, increasing reliance on performance targets can be explained by their legitimating function as a form of symbolic capital in its own right. These have become internalized by the field of general medical practice partly due to a strong tradition of healthcare standardization within the dominant discourse of evidence-based medicine (Cheraghi-Sohi & Calnan, 2013), and partly because performance measurement is deployed for benchmarking individual general practices notorious for their competitiveness (McDonald et al., 2009).

Second, we highlight the mechanisms through which boundary spanners surmount the lack of field-specific embodied and institutionalized cultural capital, i.e. clinical knowledge and credentials. Although we did observe the acquisition of basic disease-specific knowledge by boundary spanners at the beginning, this does not reach the level of competence possessed by clinical professionals. However, boundary spanners can successfully use these deficiencies to their advantage by discursively positioning themselves outside the jurisdiction of clinical professional groups and thus decreasing the perceived threat of boundary spanning projects to clinicians’ professional autonomy. Another strategy involves developing architectural knowledge (Currie & White, 2012), or knowledgeability (Omidvar & Kislov, 2014). This subtype of embodied cultural capital denotes a deep understanding of the ‘rules of the game’ unfolding in the field, its key players and relationships between them, forming an integral part of emerging boundary spanners’ habitus. Boundary spanners’ own capital deficiencies can also be compensated by the ongoing accumulation of social capital which can either be directly converted into the form of cultural capital being sought (for instance, through directly involving clinical professionals in the boundary spanning project) or used as a source of indirect ‘legitimacy by association’ with powerful local agents who have established clinical credibility (Baum & Oliver, 1991, p. 189).

Finally, it is important to trace the changes in the sources of the social and cultural capital mobilized by the boundary spanners over time. While it could be expected that the boundary spanners would acquire and/or mobilize a diverse range of capital assets derived from multiple fields, what we observed was the increasing focus on the cultural and social capital derived from one of the interrelated fields, namely the field of general medical practice, accompanied by the decreasing influence of social connections and objects from other fields. This is exemplified by the switch from national guidelines to the project-specific, locally tailored evidence as the preferred form of objectified cultural capital as well as by the shift from secondary care clinicians to primary care commissioners as the most legitimate co-producers of social capital. This indicates that an influential position in one (sub)field does not easily translate into symbolic capital in another, echoing previous accounts of the decreasing role of clinical opinion leaders beyond the initial stages of change initiatives (Hendy & Barlow, 2012).

Relationships between the fields

Legitimation through acquisition, mobilization and conversion of capital has traditionally been associated with gaining power within a given field (Bourdieu, 1986). Concurring with previous observations that boundary spanning is an inherently political and collective process (Currie & White, 2012; Kislov et al., 2016a), our findings demonstrate that the evolution of boundary spanners’ capital results in complex two-way power play which both shapes and is shaped by several intersecting fields at the interface of which they are operating.

On the one hand, our data demonstrate the growing reputation and influence of the Collaboration and its boundary spanners in general practices, with the increasing number of primary care organizations engaged in its projects. This often involved co-opting clinical agents and drawing on their capital assets. On the other hand, the influential representatives of the field of general medical practice had the power to approve or disapprove the Collaboration’s activities. The selection of clinical conditions for improvement projects, the formulation of their objectives, the recruitment of general practices and the spread of resulting improvement were all contingent on such approval. Overall, the relative gain of power achieved by boundary spanners only becomes possible through acknowledging (and utilizing) the field-specific rules of the game, the power structures and the distribution of capital endorsed by the most influential members of the (established) field they are interacting with.

According to Bourdieu (1977), legitimate authority is a misrecognized, i.e. socially recognized, form of domination. However, neither the Collaboration and its boundary spanners nor primary care clinicians seem to clearly fit the ‘dominant’/‘dominated’ dichotomy (also see Kamoche et al., 2014). The blurring of the distinction between these categories can be explained by the formation of an alliance (Bourdieu, 2000) between the Collaboration’s boundary spanners and the influential agents in the field of general medical practice. This alliance enabled the former to meet their objective of facilitating the application of research evidence in clinical practice. The powerful agents within the field of general medical practice, in turn, gained an opportunity both to increase the volume of capital in their possession and to preserve its structure, potentially acquiring an important bonus in their power struggles with the rival field of secondary care (McDonald et al., 2009). Interestingly, through an increasing alignment with the patterns of capital distribution adopted in primary care, internal legitimacy of boundary spanners within the Collaboration became contingent on their external legitimacy in the field of general medical practice to the detriment of the scientific field (see Drori & Honig, 2013).

As a result, boundary spanning increasingly took place between the Collaboration’s multiprofessional implementation team (the joint field of collaborative practice) and primary care organizations (the field of general medical practice) rather than directly between the latter and applied health researchers (the scientific field). This suggests that, where marked power differentials exist between the intersecting fields, some of these fields can be prioritized in the process of boundary spanning at the expense of others. At the same time, given the low credibility of the boundary spanners in the eyes of applied health researchers, excluding one of the established fields from the emergent field of collaborative practice can also be interpreted as a rather innovative way of responding to the contestation of boundary spanners’ legitimacy. Ironically, the marginalization of a certain group of agents (e.g. researchers) from the emergent collaborative field does not seem to preclude utilization of capital (e.g. evaluation methods) associated with these agents as a way of reinforcing the external legitimacy of boundary spanners and of the emergent field of collaborative practice in general.

Social trajectory of boundary spanners

Newcomers into a field may differ in the volume and structure of their initial capital at the point of entry due to the natural differences between their career trajectories shaped by their previous experiences in other fields (Bourdieu, [1984] 2010; De Clercq & Voronov, 2009b). One might anticipate that different groups of boundary spanners, i.e. those with a nursing background, those with managerial experience and those with no job experience, would differ in their legitimation strategies and social trajectories (Battilana, 2006; Lockett, Currie, Finn, Martin, & Waring, 2014; Lockett et al., 2012). However, despite significant initial differences in the volume and structure of cultural capital in their possession, variation in the legitimation dynamics between individual boundary spanners proved much less pronounced. This may suggest that the peripheral and relatively low-status nature of the boundary spanning position can, at least partly, level out the impact of previous professional and status differences on its incumbents’ subsequent trajectories within the field.

We show that initial capital deficiencies of individual boundary spanners can be compensated not only by gradual accumulation of missing capital assets but also by reliance on capital accumulated by agents they are interacting with. By highlighting the complementarity and exchange of capital in the process of boundary spanning this study enhances our understanding of the mechanisms underpinning its inherently collective nature (Currie & White, 2012; Kislov et al., 2016a). Interestingly, although boundary spanners with a nursing background had a distinct advantage in the form of clinical knowledge and qualifications, i.e. field-specific cultural capital, its legitimating effect was limited to the earlier phases of the Collaboration. Possible explanations may include limited applicability of discipline-specific clinical knowledge to other clinical disciplines (Martin, Currie, & Finn, 2009), the dependence of legitimacy upon continuous engagement in professional practice (Kislov et al., 2016a) and, most importantly, a general trend towards managerialization of boundary spanning over time.

Our findings demonstrate that successful legitimation can effectively transform nominated boundary spanners, i.e. agents utilizing their ambiguous ‘in-between’ position to span boundaries between the intersecting fields, into project managers, i.e. occupants of a traditional role recognized by (and embedded within) one of the fields. This tendency illustrates the transience and instability of nominated boundary spanning roles as a novel mode of organizing and the relatively high risk of its reversal to more traditional modes of management (Chew et al., 2013; Kislov et al., 2016b). Interestingly, despite an associated increase in status and authority, the managers-in-the-making did not generally attempt to contest the existing distribution of capital, internal boundaries and ‘ways of doing things’ within the field of general medical practice. Absence of such strategies, which are often associated with middle-status agents (Bucher et al., 2016; Lockett et al., 2012), can be explained in our case by relative peripherality of boundary spanners in relation to inter-professional struggles within the field, lack of formal ‘consecration’ of their de-facto managerial role (Bourdieu, 2000, p. 243) as well as by the ‘inertia’ of their habitus which was shaped by their previous lower-status position in relation to the field (McDonough & Polzer, 2012).

By describing the radical shift in boundary spanners’ trajectory, our study reveals a highly transformative potential of legitimation, demonstrating that its possible ‘successful’ outcomes are not confined to transitions from an ‘apprentice’ to ‘master’ (Lave & Wenger, 1991) or from a ‘nominated boundary spanner’ to ‘boundary-spanner-in practice’ (Levina & Vaast, 2005) but may include lateral shifts within the field. Furthermore, boundary spanners, perhaps due to the increased propensity of peripherally located lower-status agents to decipher ‘the rules of the game’ and ‘unmask’ power relationships (Bourdieu, 2000; Bucher et al., 2016), can relatively quickly overcome the mismatch between their individual dispositions, which are shaped by their previous engagement in past fields, and the current field (see McDonough & Polzer, 2012). The fact that we observed this shift across all of our boundary spanners allows us to interpret it as a non-random and relatively durable change of their emerging collective habitus under the influence of the forces operating in the most influential field they are interacting with (Bourdieu, 1983, [1984] 2010; Emirbayer & Johnson, 2008).

Postscript to the empirical case

One year after we finished the fieldwork for this study, five boundary spanners were still employed by the Collaboration, all of them now holding formal managerial positions. Three of them embarked on postgraduate courses in healthcare management. These developments support our argument by demonstrating that the capital accumulated by boundary-spanners-turned-managers was objectified as a formal organizational position and further enhanced by educational credentials (Bourdieu, 1977).