Fugitive coproduction: Conceptualising informal community practices in Scotland's hospitals

Within public administration, coproduction is a ubiquitous policy discourse, and increasingly an analytic lens through which public relationships with public services are viewed. This article reports an interpretive qualitative study of community practices around three changing hospitals in the Scottish NHS, comprising semi-structured qualitative interviews with citizens, NHS staff, politicians and journalists, as well as non-participant observation of community and NHS events. Initially focused on community opposition to top-down hospital change, the study identified a surprising range of supportive community actions for their local hospitals, including volunteering, fundraising and innovative co-delivered service models. Building on these examples, the paper presents a model of ‘ fugitive coproduction, ’ where individuals and groups within communities collaborate with local staff in ways which significantly shape the provision of local services, without permission or authorisation from relevant authorities, and in modes that are centrally concerned with immediate perceived need not strategic change. I argue that these forms of public action can make valuable contributions to public services, and that they hold lessons for the wider reform of public administration.


| INTRODUCTION
Coproduction is a 50 year old concept (Ostrom & Ostrom, 1977) which has, since the early 2000s, come to dominate the exploration of relationships between public service organisations and their communities (Bevir, Needham, & Waring, 2019;Nabatchi, Sancino, & Sicilia, 2017). In an effort to tame this substantial and unruly literature, and to aid generalisation, scholars have called for and produced practical, clearly-specified typologies (Nabatchi et al., 2017;Osborne & Strokosch, 2013). This article argues that while these enhance our ability to generalise and learn across cases, such categories also obscure, specifically the presence and consequences of informal and unsanctioned community contributions to public services. I report findings from an empirical study of changing hospitals in the Scottish National Health Service. As is appropriate within an interpretive research design (Haverland & Yanow, 2012), this study progressed inductively from a focus on community 'save our hospital' campaigns, into a broader exploration of the practices-including volunteering, fundraising and campaigningthrough which communities sustain their hospitals. The wider research project explored the range of community responses identified within these case studies (Stewart, 2019). Here, I focus on a subset of practices to offer a conceptual account of 'fugitive coproduction' at the local level.
In this usage, the term 'fugitive' does not denote illegal action (as in fugitives from justice): the word is derived from the Latin root fugere meaning to flee or to wander. I use it to describe practices which are transient and elusive: transient because these practices emerge only in times of need, in order to solve a local problem, and elusive because of they often occur outside official rules and boundaries, even challenging such boundaries. The concept has heritage in Wolin's (2016, p. 111) concept of 'fugitive democracy' as 'a rebellious moment' and recent work by Mattern (2019) on 'fugitive libraries' created by marginalised communities. The understanding of fugitive coproduction that I develop here includes three characteristics: (a) collaboration between community members and local staff, which is (b) significantly productive of the nature of local healthcare provision and crucially, which is (c) unauthorised by and, as a result, 'below the radar' of healthcare authorities. Relatively informal practices-fundraising, volunteering, providing goods and services-are rooted in long-term community dynamics but are often only fleetingly visible. In this article, I interpret these practices as 'fugitive coproduction,' arguing that in the interstices between NHS-led planning and decision-making, they make important contributions to local health services, while also posing challenges for top-down management.

| INSTITUTIONALISING COPRODUCTION
Attributed to the work of Elinor Ostrom (Ostrom & Ostrom, 1977) from the 1970s onwards, coproduction is a policy discourse with significant traction across policy areas and regions (Osborne & Strokosch, 2013). Nabatchi et al. (2017) argue that it is marked by definitional ambiguity and a 'bandwagon effect.' This ambiguity can be interpreted as conflict (Flinders & Wood, 2019), and some have argued the concept has become 'stretched' beyond its feasible limits, or critiqued its superficial 'cobiquity' (Dudau, Glennon, & Verschuere, 2019;Williams et al., 2020).
This diversity of use across multiple disciplines and over time (Filipe, Renedo, & Marston, 2017;Osborne & Strokosch, 2013) has prompted calls for greater clarity of conceptual definition in order to move beyond the exploratory to the explanatory and the comparative (Nabatchi et al., 2017;Osborne & Strokosch, 2013).
Definitions of coproduction within public administration literature reflect this ambiguity. One of the older definitions from Brudney & England (1983, p. 62) simply specifies 'voluntary, cooperative action in service delivery.' A quarter of a century later, Bovaird suggests.
"the provision of services through regular, long-term relationships between professionalized service providers (in any sector) and service users or other members of the community, where all parties make substantial resource contributions." (Bovaird, 2007, p. 847) Most recently, in an article bemoaning the evidence base around coproduction, Nabatchi et al. (2017, p. 769) define it broadly as 'a wide variety of activities…in which state actors and lay actors work together to produce benefits'. In an effort towards greater conceptual clarity they propose a matrix typology, identifying the levels (individual, group, collective), roles (client, customer, citizen) and benefits (personal, social), with useful practical illustrations within each category. They also distinguish activities temporally located in different phases of the 'service cycle': cocommissioning, co-designing, co-delivery and co-assessment (Nabatchi et al., 2017).
This typology potentially allows greater comparability across the coproduction evidence base. However, as an analytic device it also obscures other possibilities for studying and practicing coproduction. The urge to pin down and specify snapshots of what are widely acknowledged to be complex and often emergent processes (Bovaird, Flemig, Loeffler, & Osborne, 2019;Filipe et al., 2017;Osborne, Radnor, & Strokosch, 2016) focuses attention onto practices, which are relatively formalised and which are 'accounted for' in terms of clearly defined and broadly stable mechanisms and actors. There are echos here of critical accounts of New Public Management's 'projectification' of public services, in which the techniques and language of project management 'reduce social activity of all kinds to instrumental and rationalised action, and in doing so efface the political, social and ethical dimensions of activity structured as projects' (Hodgson, Fred, Bailey, & Hall, 2019). Segmenting coproduction into discrete and defined projects might aid the operationalisation of coproduction as a managerial strategy for public services, but may reduce our understanding of more informal, community-initiated actions and of coproduction as 'exploratory social space' (Filipe et al., 2017).
The inevitability and extant nature of citizen's contributions to public service delivery-acknowledgement that no service is delivered unilaterally without engagement from recipients and communities-has always been present in coproduction literature (Alford, 2016;Osborne & Strokosch, 2013). Indeed, Loeffler (2021, p. 24) highlights citizen militias, workers' education associations and voluntary firefighters as historical examples of citizen-initiated contributions which featured in early accounts of coproduction. However, as the literature has become preoccupied with strategies for institutionalising coproduction as managerial strategy, analyses of extant community practices became more marginal. Nabatchi et al. (2017, p. 767) argue that 'rather than being seen as an already occurring practice, coproduction was viewed as something to be introduced to organizations and integrated into operations.' One attempt to bridge community understanding of services, and their organisational design, is the concept of 'enhanced coproduction' proposed by Osborne and Strokosch (2013). They suggest a tripartite conceptualisation of coproduction as either individual and consumerist (at the operational level), collective and participative (at the strategic planning level) or 'enhanced,' a combination greater than the sum of its parts. Enhanced coproduction brings together the tacit knowledge generated within the 'involuntary and unavoidable' coproduction of service use, with collective mechanisms of participation in service planning to enable 'user-led innovation' (Osborne & Strokosch, 2013, p. 38). The article ends with a call for the design by public service organisations of fora for service user experience to be heard (Osborne & Strokosch, 2013, p. 44). In healthcare, this connection of service user experience with formalised decision-making is encapsulated in Experience-Based Co-Design methods (Robert et al., 2015).
This urge to offer a practical strategy for public service organisations to implement, though, returns us firmly to the realms of the formal and invited; what Bovaird (2007, p. 847) depicts as 'institutionalised coproduction.' This article advances an alternative approach, in which community roles are recognised as an existing dimension of local services, rather than an input to be assembled by decisionmakers.

| STUDY BACKGROUND, DESIGN AND METHODS
This article reports an interpretive qualitative study of three changing hospitals in Scotland, a country where coproduction is rhetorically entrenched in policy discourse . Public hospitals in Scotland (the overwhelming majority of all hospitals) are owned and operated by the NHS, which has since 1998 been distinct from the other UK NHSs in England, Wales and Northern Ireland. Scotland's NHS has a fairly flat organisational structure (made possible by a population of only 5.4 million), in which a central ministerial department directly manages territorial 'Health Boards' who are responsible for the planning and provision of the overwhelming bulk of healthcare provision in their given territorial area (Greer, 2016). The appointed Board members who run Health Boards are directly accountable to central government (Greer, Wilson, Stewart, & Donnelly, 2014). Since 2016, efforts to integrate healthcare and social care have involved greater joint working with directly elected local authorities (Pearson & Watson, 2018), but at the time of this study (2016)(2017)(2018) these changes were only beginning to have significant impact on the local administration of healthcare.
Scotland is an intriguing focus for this study due to claims of a 'Scottish approach' to public policy and administration (Elliott, 2020). While related to distinctive aspects of the parliamentary context in Scotland (St Denny, 2020), this also refers to a style of governance associated with Scottish National Party government since 2007. This 'consultative and cooperative style' (Cairney, Russell, & St Denny, 2016) has, Rolfe (2016) argues, been distinctive for its use of legislative powers to strengthen its focus on community participation. However, empirical evidence for the impact of the Scottish approach to public policy at the frontline is thus far lacking. While the rhetoric of coproduction is well-established in Scottish policy (Scottish Government, 2017), at the time of this study the Scottish health system remained hierarchically-organised, and far more centralised than its counterpart in the contemporary English NHS.
The study aimed to contribute to existing debates about how best to involve the public in intrinsically contentious decisions to close or significantly change hospitals in the UK NHS (Djellouli et al., 2019). Taking an interpretive approach to case study research (Schwartz-Shea & Yanow, 2006), I set out to explore everyday public practices and meanings around changing hospitals, and had not sought out coproduction. As the research progressed, it became clear that community groups who were active in debates about hospitals' future were deeply enmeshed in not just supporting but actually constituting these facilities. The interpretive approach to generating knowledge of the cases, and case studies as a research design, permitted me to explore the perspectives of a range of people around a single hospital through multiple methods of data collection.
Case selection began by assembling a list of statutory 'major service changes' in the decade 2004-2014. I then supplemented this by searching Scottish newspapers for cases 2004-2014 which had either not been declared major service changes, or had been abandoned before formal proposal. Combining these two lists generated a 'population' of potential cases, from which I selected a range of outcomes, level of apparent contentiousness and settings (urban, semi-rural, rural). The three hospitals studied were: • Case 1: an NHS service redesign which involved closing two community hospitals 1 and building a new community hospital and health centre in a town situated between the two closing hospitals.
• Case 2: a small community hospital which had prompted repeated community mobilisations in response to proposals to remove or reduce services. I focused on the local 'Friends of the Hospital' 2 group who were leading fundraising and planning for the future of the hospital. During the period of the study, the NHS began public engagement around centralising the 24 hr 'minor injuries service' onto fewer sites.
• Case 3: an NHS service redesign which included closing an acute hospital in a Scottish city.
Each case study involved repeated visits to and stays within the community over a period of months, with empirical data collection taking place between March 2016 and May 2018. Fieldwork included non-participant observation of 11 events (consultation meetings, campaign meetings and other relevant community meetings), analysis of local official documents, along with local media and, where available, social media (e.g., Facebook and Twitter coverage of the change) and semi-structured interviews with NHS staff (clinicians and managers), local politicians and journalists (N = 26) and citizens (N = 44). Interviewees were recruited through NHS routes (e.g., lay representatives on committees), through non-NHS routes (e.g., members of campaign groups and other community associations) and informally (through contacts met while spending time in the communities). A fuller description of data collection, including my positionality and reflexivity in the project, is included in supplementary materials. In the interests of anonymity, all descriptions of or quotes from interviewees are reported with a pseudonym, and generic descriptors (citizen, manager, clinician, local politician). Ethical approval for the research was given by University of Edinburgh Usher Research Ethics Group, and NHS Research & Development approval was gained from each Board area.
Data analysis began during the data collection phase, allowing me to explore emerging themes in interviews and, in some cases, seek additional interviewees who could speak to them. Interviews were audio-recorded, and transcripts, documents and my fieldnotes were analysed in NVivo using an approach based on Charmaz's (2014) grounded theory: beginning with close line-by-line coding of transcripts, focusing on gerunds and 'staying close' to interviewees' descriptions. I then reviewed these to develop thematic codes, which were checked back against data and refined. Much of the data discussed in this article came from a single, overarching and unexpected thematic code: 'making the hospital.' I first created and then heavily populated this code as it became clear that local communities and local staff were describing not merely 'fighting to save' or 'working/being treated within' hospitals, but were much more active participants in the creation of care within hospitals. I assigned sub-codes including' volunteering,' 'making decisions' and 'shift from protesting closure to supporting hospital' (see supplementary information S1). The data within these codes seemed intimately connected to coproduction, and yet jarred with currently dominant accounts of coproduction as governance strategy.

| FINDINGS: MAKING HOSPITALS
The three case studies were significantly different from each other. Case studies 1 and 2 both concerned relatively affluent semi-rural areas with small and historic community hospitals (with some acute provision). By contrast, case study 3 was a larger and more modern acute hospital in a socio-economically deprived urban area. In all three case studies, it became evident that the hospitals were already significantly co-produced with their communities. This was not presented to me as a 'fact' in any case. The only occurrence of the word 'coproduction' in my dataset is an interview with a senior manager in case study 1, when it does not refer to existing local practices but to a future aspiration. Indeed, campaigner interviewees often downplayed or denied the significance of their own contributions. In this section, I describe practices through which I propose communities were coproducing hospitals. Then, in the discussion, I argue for these as a particular 'fugitive' form of coproduction, manifested differently in communities with different socio-economic profiles.

| Case study 1
Case study 1 was an example of a service change where public engagement and involvement had played a central role in the process, which aimed to close two hospitals and provide services in a new hospital in a third location.
Opposition to the closure of the two local hospitals was forthright but fairly localised, and key opinion formers had been closely involved in and were 'signed up to' the proposals. Indeed, I heard from multiple interviewees that the suggestion to close the hospitals had been made by a member of the public at a meeting between NHS managers and one of the Friends of the Hospital groups: "It was a locality meeting and…actually it was [named member of the public] said 'have you ever thought actually about closing this one and building a new hospital somewhere centrally?' Because I think they were starting to see the problems that were coming with a big old building." (Margaret, local politician) While it was, then, fair to say that the proposal to close had come from the community, this had been brought about because of a purposeful plan from the local Health Board. Members of staff were frank about having enthusiastically 'leapt on [the suggestion], to some extent' (John, NHS manager) but the process had begun with a programme of open-ended conversations with local communities. This was characteristic of service planning in this locality, where several managers leading change projects also lived in the communities: "We knew an awful lot of people, we continue to know an awful lot of people…That kept us, what I would describe, is kept us safe and it kept us steady" (Elizabeth, NHS manager) This steadiness made for a painstakingly slow process. In meetings I observed, NHS managers would take time to establish their connections with attendees, and by taking part in the mundane rituals of the meeting (such as James, a senior clinician, getting up from the table to refill the teapot in one meeting) there was a sense of a shared endeavour between community members and managers.
In this joint endeavour, communities and local NHS staff were to some extent allied against both the wider NHS Board and the Scottish Government as the financers of capital funding. Staff repeatedly argued that this change was to 'futureproof' provision, on the basis that the older hospitals may not be replaced at all if a few years passed: "The choice which we could have put on the table but you arguably could've put on the table is not to have anything at all. You know…given the population size, given it's not far from [major regional hospital]. Imagine that we move to regional boards would somebody in [a city] think actually you really need a hospital in that area for the population you've got? My argument would be probably not." (Elizabeth, manager) The change project was portrayed, in my interviews, in media coverage and presentations at public meetings, as a 'last ditch' attempt to preserve local services from the perceived inevitability of the buildings being condemned and not replaced. James, clinical lead, described: "The bad thing would be 'the building has now failed all fire regs' or 'the building has contributed to the death of that patient due to C-diff or name your superbug' or you know, whatever other thing.
'The building no longer fits European or Brexit sort of criteria for dignity in care.' Whatever it might be, something will fail the building." (James, clinician) In this the external risks of both regulations and money were looming spectres, prompting the sense of a collective endeavour between local staff and communities to defend local services.
The hospitals in case study 1 both had proactive 'Friends of' committees. Most of the work here was fundraising for additional equipment to supplement NHS provision, and interviewees in the community repeatedly talked about 'our' hospital. Mary, a retired nurse and member of the 'Friends of' group, described pathways for local people to use the hospital: "We've managed to get it now that you can go into [hospital 2] for terminal care, you can go into [hospital 2] post operatively if you've had hips and knees done, if you're of that age and you've nobody to look after yourself, so you're getting rehab. You can go in post stroke, you can go in post coronary and just general surgery, or just if you are ill, pneumonia, anything like that and there is a space, the local doctors can admit you… If there's a bed we'll get you in." (Mary, citizen) The easy use of the word 'we' here reflected both Mary's previous career in the NHS, but also a sense of ownership which was widespread among interviewees involved with the 'Friends of' groups. The hospital was presented as a community asset for the collective care of vulnerable members of the community: despite the NHS case for closure relying partly on evidence that most inpatients were from outside the local area.
Both the closing hospitals were historic, dating from decades before the creation of the NHS, and yet interviewees described ways in which they were continued to evolve. Hospital 2 had a community-run gardening project which had come about when Catherine, a therapeutic horticulturalist, moved into the area.
"one of the local councillors approached me and said 'do you know about the little patio area up at [hospital 1], it's fallen into disrepair'… So we then set up a steering group, had a couple of interested people, we wrote to the NHS and asked them would it be possible to use the patio area…We got permission, then we got some funding and raised money to get a summer house…adjacent to where the patio area was, that we could use for indoor work. And then we applied for planning permission to get a ramped area built down to the summer house and we got permission from the hospital to put a disabled toilet inside" (Catherine, citizen) The gardening project enhanced the view outside for patients inside the hospitals. However, it also altered the physical grounds in more permanent ways, and integrated into the clinical services provided on site. Catherine explained: "Patients from the psycho-geriatric ward, if they were able, would come down and do a one-to-one session with me. Sometimes just the sensory input of being outside and they talked about their previous experiences of having a garden and what they grew in it. Also maybe one or two might come out and watch while the other might do some seed sowing, some transplanting of bud plants etcetera." (Catherine, citizen) Over time, and as patients in the hospital became less physically able to access the garden, some of this work moved inside to the wards. Anne, a local retiree who volunteered in the garden, described these activities: "as a volunteer, [friend's name] and I go in on a Monday morning…and we take garden-related things in, so we do collage work and flower arranging…all related to nature and gardens and get people to talk about what they grow in their garden…And then after a number of weeks we have a finished product which then the hospital display for us…The staff are very good at sharing information about the patients with us and [we're] certainly good at giving feedback after our sessions if somebody's been very unsettled." (Anne, citizen) This community-led and charitably-funded project became integrated as a physical and clinical enhancement to the hospital. Significantly, though, the gardening project had emerged during, and continued developing after, the plan to close the hospital. The project was designed as far as possible to be portable (Anne explained that 'the raised beds can be moved and the summer house can be moved and we can dig out some of the plants and take them elsewhere'). Thus, this emergent enhancement to the hospital was always understood as temporary.

| Case study 2
In case study 2, a community group had, several decades earlier, worked closely with the local GPs to develop an integrated day service within the hospital for local elderly people. The service was funded by a combination of GP fundholding money, social care contributions and charitable funds. Linda, involved from the beginning of the project and still living locally, described the project's genesis: "[A local GP] put the idea in our heads that if we made this big extension we could have day care in the hospital, and that's what we did. And I have to say we were told by the NHS that there was no way we would ever be allowed to do day care because it didn't fit with the NHS sort of normal thing, cause you would normally have day care in a social setting rather than a medical setting. But…[the local GP] wanted this holistic approach to looking after, it tended to be frail elderly people… And there was a person employed to look after them but it was just held in the main hospital day room. It was very innovative and it was very successful." (Linda, citizen) The community group fundraised to build an extra room onto the hospital, extending both the physical space and the type of services provided within it. Linda described how the service operated as a partnership between local staff and the community: "[Community group] were very much seen as part of the team… so we went to the meetings, you know, I usually did the minutes and it was just, we were seen as an equal partner in this holistic approach to dealing with elderly and frail people." (Linda, citizen) However Linda, who remained involved for years, described how 'not everyone was comfortable with [volunteers] having such a stake in things I think. Depends who the management was at that time over the years' (Linda).
Helen, another committee member and carer for an elderly parent who had used the service, described the rationale to stop its funding: "The NHS kind of noticed what was happening at [the hospital], really, I think, and decided that…It's ridiculous, I remember being at meetings and them saying 'you know that's social care, it's nothing to do with us, we're not doing it anymore.' But then I can remember sitting at the meeting and saying… 'but you can't separate health and social care!' But ten years ago, you could. Now, you can't, but you could then." (Helen, citizen) The 'day room' was now simply a part of the hospital for inpatients to use, but community groups continued to buy equipment for the inside of the building and especially enhancing the outside of the building through voluntary labour, and purchases of plants and flowers.
Key members of the community in case study 2 were conscious that the significant amount of fundraising within the community sent a political message to regional decision makers. Alongside improving services, these activities built visible support. George, local resident and long-time committee member, described how this had played out during past NHS attempts to reduce services: "We were showing ourselves to be politically very effective…We'd also shown ourselves generally as a community campaigning group to have a lot of support which is well evidenced, you know, with the number of collections there are at funerals, the number of people that turn up for our events, the number of legacies that are left to us." (George, citizen) While this quote points to the political efficacy of discursively framing a hospital as widely supported within the community, evading closure was not always an explicit goal of the group. The group's existence did however, create opportunities for easy mobilisation and facilitated the monitoring and pre-empting of local NHS plans for the hospital, including convening public meetings in rapid response to rumours of closure. Health Board documents proposing a collaborative relationship with the group to discuss the future of the hospital echoed this: 'Whilst the "Friends" have always been able to mobilise the considerable efforts of the local community in a number of innovative ways, these efforts have tended to be in response to, rather than in partnership with the NHS.'

| Case study 3
In case study 3, the mechanisms of community coproduction were less evident than in case studies 1 and 2. There was no 'Friends of' group, and thus no discussion in interviews of fundraising for or donating to the hospital. The hospital was an acute facility, therefore not run by local GPs, and located in a socio-economically deprived urban setting. The absence of data in the columns for 'fundraising as mobilisation' and 'material provision' in the supplementary data table reflects the less affluent population living around, and using, the hospital. However, where I did identify practices of coproduction around this facility they were, if anything, more analytically interesting. A recent history of attempted closures of the hospital had created an unresolved conflict which meant that in this case study, more than the other three, claims about the function and value of the hospital were fundamentally contested.
Community groups in the area were clear that this small hospital, located in a relatively socio-economically deprived area of the city, was a 'centre of excellence' for the care of a particular chronic condition. They referred to 'the [condition] clinic,' and several were effusive in their praise, even love, for the facility. Alan, a service user and campaigner described his perception of the uniqueness of care in the hospital for people with his condition.
"It was really fun meeting other people with [condition] and doing things and you wouldn't have got that anywhere else. It's still happening just now, people can still get referred … people say that, they just feel so much part of it." (Alan, citizen) The condition-specific group had started holding weekly support meetings within the hospital (in a 'day room') for patients and their carers. Observing these meetings suggested the group's comfortable, albeit temporary, ownership of the room. These were sociable occasions, led by one of the patients with a nurse present but (on the occasion I attended) popping in and out. Refreshments were brought into the hospital from outside, and organised among the group. Members of the group confidently moved furniture around the room, bringing more in from other rooms, to accommodate the format planned and the sometimes complex access needs of attendees.
These meetings were long-standing, and seemed in interviews closely intertwined with patients' experience of services within the hospital. Together, they meant that the hospital as a site was experienced as offering an unusual level of service for people suffering from the condition, even if the clinical services delivered by the NHS on site were as standard. In responses to online petitions to 'save' the hospital, parliamentary debates and some public interviews, the hospital was described as a specialist facility for the condition, or containing a specialist 'unit.' This narrative seemed frustrating to managers trying to redesign services in the area, and in interviews I was repeatedly corrected on this point, with one member of staff exclaiming 'why aren't you asking me about falls? [Condition-specific group] are just like any patient group treated at the hospital!' (fieldnotes from conversation after staff interview). Notably, the Support Group declined to cooperate with the formal NHS engagement process, instead going directly to elected politicians, including Government Ministers. Manager Derek explained.
"One of the things that surprised us a wee bit was the fact that the local support group who had been very vocal in the first attempt at closing [the hospital] effectively refused to engage as part of the public involvement… An interesting tactic" (Derek, Manager) The campaign group had refused offers of a meeting to discuss the latest change proposals: 'no matter how hard we tried, and we did try, they just never engaged with us at all' (David, manager). In an observed event, members of the support group refused to offer individual or small group feedback to regional NHS staff in attendance, preferring to refer to a collective written response.
These campaign tactics reflected a sense from campaigners that regional NHS decision-makers were not to be trusted: "this new model of care is complete nonsense and they say it's not about money but it is about money" (Alan, citizen) In case 3 what began as a small, informal use of a day room had over time (and two sustained conflicts with the regional Health Board) come to discursively redefine the hospital's core mission in the public eye, and had materially contributed to it being 'saved.' From a position of defending the hospital, the group had not only developed their own 'reading' of the site, but had made significant inroads into establishing their story of the hospital as the dominant narrative.
What had not developed were the community structures of volunteering and fundraising which characterised the other cases. In the aftermath of the hospital being 'saved' again, one interviewee, who was a retired health professional reflected on the state of the relationship between local concerned members of the community and Board: "I, kind of, wish that the community would get more involved in shaping that service, which would be fantastic. But then a wee sigh, because I thought, I wonder if the Board have changed and they'd be up to actually doing that. What an opportunity to have real, true community participation." (Moira,

citizen)
This most politically contentious case of the three I studied, was marked by intense collaboration at the service level within the hospital, but an entrenched opposition between patient groups and regional decision makers.

| DISCUSSION
Coproduction of public services is no longer a radical, outsider demand, but something sanctioned, mandated and, in Scotland, central to an official approach to public service delivery (Elliott, 2020). This growth in popularity has prompted expressions of frustration about the 'wooliness' (Bovaird et al., 2019) of conceptual definitions (Dudau et al., 2019). This article explores community practices around healthcare and identifies instances of 'fugitive coproduction': informal, unsanctioned, cooperative practices between communities, patient groups and local staff which are productive of particular forms of valued care. Each case may be mapped onto Nabatchi et al.'s (2017) 'who, what, why' typology: as instances of group coproduction, generating social benefits for their community, and primarily focused on the delivery of services. However, this also obscures aspects of these cases, specifically their informality, and an orientation towards 'doing,' not designing. A more holistic perspective, in which we explore 'how' coproduction is achieved (Crompton, 2019), aligns more comfortably with Filipe et al.'s (2017, p. 5) vision of coproduction as a complex and dynamic 'exploratory social space.' First, the practices identified above were informal collaborations between community members and the local staff of the hospitals, not simply uninvited, outsider action. Like Petchey, Williams, and Carter's (2008) 'street-level policy entrepreneurs,' in each case a handful of energetic individuals brought local knowledge to bear on the issues, but in this study these actors worked more informally and more collectively than Petchey et al's nationally-funded projects. The cases varied in how far groups were integrated into and recognised by the management of local hospitals. In cases 1 and 2, 'Friends of the Hospital' groups operated as semi-formal associations, and had regularised patterns of communication with local staff. In case 3, the support group was more informal, and had been quietly supported by local staff. However, in all cases, the personnel and goals of the groups shifted over time, and, most pertinently, all these local collaborations were conducted within the context of, and arguably drew their mobilisational energy from, a currently or historically agonistic relationship with regional administrators. Coproduction typologies have tended to focus on the 'who' of the citizen side of collaborations (Alford, 2014), while being broadly silent on the multi-level complexity of public service organisations, let alone on the potential for oppositional relations (Dean, 2018) within organisations. However, these cases demonstrate the possibility for citizen collaboration with staff at some levels of an organisation, in the context of conflict with other levels.
Second, while often unstrategic, these practices did not just produce 'nice extras. ' Brandsen, Steen, Verschuere, Steen, and Verschuere (2018) distinguish coproduction of 'core services' from 'complementary tasks,' but while most of the practices in these case studies were complementary (in that they did not supplant medical care), they were in their cumulative effects, significantly productive of practices of care within the hospital. When campaigns to 'save' the hospitals were mobilised, these practices of care became bound up with other valued aspects of the hospitals into a defence of its material presence. The initiation of co-delivery was usually unstrategic: practices emerged pragmatically in response to observed local need (e.g., the elderly day care project) or recognition of a valuable skillset in the locale (such as the therapeutic gardening). This resonates with Guerlain and Campbell's (2016) argument, in the context of community gardens, for better recognition of a range of activities despite the absence of an over-arching strategic intention on behalf of the group. It sits far less comfortably with Scottish Government policy, which locates coproduction as one route to a more strategic state (Elliott, 2020).
Better recognition of informal, 'doing' projects requires us to expand our understandings of service design beyond planned events. While Osborne & Strokosch's 'enhanced coproduction' requires the creation of fora for service users to share their experiential views to shape service design, these may either already exist or be accessible through recognition of a wider extant range of modes of citizen participation (Stewart, 2016). Cooper captures something of this in her notion of 'prefigurative publics'; directing our attention 'away from the conventional focus on opinion formation, debate and decision-making to practically enacting alternatives' (Cooper, 2016, p. 327). In this study, the alternatives were small, often quiet practices of care to enable and sometimes enact an alternative vision of contemporary healthcare. Community groups were 'fixers' rather than 'designers' (Mattern, 2018).
Lastly, the coproductive practices explored in this study were unsanctioned in that formal decision-making bodies (in Scotland, Health Boards) had often been unaware of or uninterested in such small local projects. Indeed, some of them were rooted in an assertion of community ownership against Board authority. As Bovaird and Loffler (2012, p. 1119) state, 'user and community coproduction has always been important, but rarely noticed'. Projects took place in the shadow of, as Helen put it in case study 2, 'the NHS notic [ing] what is going on' (Helen). Practices were often explicitly acknowledged to be temporary, as in case study 1's gardening project where the summerhouse was 'bolted together and not cemented in' (Catherine). Practices of fugitive coproduction continue in the interstices of mainstream NHS provision, and then fade or are shifted to other spaces, having contributed to people's experience of a service for a period. Significantly, these case studies suggested that temporally bounded projects could nonetheless have more lasting consequences for a community's sense of what a service is, and is for.
A key finding of this study, with wider relevance for scholars promoting the institutionalisation of coproduction, is that interaction with official, "invited" public involvement was often experienced as a risk to these practices, rather than an asset. The key mechanism here was one of formalisation: as the NHS responded to the policy-driven demands of community empowerment, organisations required collaborative practices to be audited and justified through talking (in meetings), writing (of plans and evaluations) and formal evidence-gathering (including needs and impact assessments). This was a change in activities and in tone from informally meeting the needs that community members saw in "their" local hospitals, and at times community groups seemed required to represent the local population in uncomfortable ways. My interviewees were in most cases keen to distance themselves from any suggestion of control over decision-making in the management of these hospitals. Organisational logics of how and when to involve public sat uneasily with the informality of groups within communities who persisted in 'acting not talking' (Cooper, 2016): turning a peripatetic clinic into something discursively recognised as a 'unit,' bringing gardening into wards, setting up services at the boundaries of organisations to meet identified unmet needs. Community actors can be characterised as having a shape-shifting potential in which 'groups or movements can be expected to have periods of latency, or public quiescence, without ceasing either to exist or to operate effectively as submerged networks' (Scambler & Kelleher, 2006, p. 228).
Acknowledging the significance of fugitive coproductive practices in communities requires consideration of how they might be unevenly distributed across society, and their implications for equity (Matthews & Hastings, 2013;McMullin & Needham, 2018). In her discussion of "fugitive libraries" in the USA, Mattern (2019) identifies the creation of "itinerant, independent" spaces as a response to the injustices and exclusions of historically white public libraries. By contrast, coproducers I identified in my cases were older than the general population, in cases 1 and 2 were from more middle class occupations, and in all cases were almost exclusively white (albeit in cases 1 and 2 in overwhelmingly white communities). Questions of the equitable distribution of capacity to engage in coproduction come to the fore when comparing cases 1 and 2, affluent semi-rural locations, with case 3's socio-economically deprived inner-city neighbourhood. This is most pronounced when considering fundraising and donation. In cases 1 and 2, a history of fundraising for the hospitals mobilised the communities into associational forms, created regular channels of communication with local management (for the purposes of donating material goods and services) and functioned to galvanise a sense of community ownership of the hospitals. By contrast, in case 3, the association which sprang up was rooted in service use, not in fundraising and donation; this hospital had no 'Friends of the Hospital' group. Case 3's coproductive practices were more 'outsider' tactics, especially their use of media coverage and direct appeals to elected politicians. As Fox (2003) notes, coproduction has tended to be a 'politician averse' literature, which focuses on collaborations between citizens and administrators, not lobbying via elected politicians. Nonetheless, the overtly political work of the support group in case 3 both 'saved' the hospital (twice) and ensured the provision of a form of care (supplementing specialist support with peer support) that they valued therein. Thus, while this study provides evidence for the abilities of advantaged social groups to consolidate their advantages through coproduction (Jakobsen & Andersen, 2013;Matthews & Hastings, 2013) it also demonstrates that a 'fugitive' lens can keep in sight alternative tactics utilised by less advantaged communities.
Professionals are not only socialised into but also held accountable for consistency and equity in the provision of services (Fenwick, 2012) and regional management's ambivalence to public initiative was reflected in official documents and in national policy (Stewart, 2017). It is easy to critique this as an (anti-democratic) failure to cede power, but it also relates to managerial concerns about hospitals as assets to be fairly equalised across the health system, in recognition of the eccentric distribution of hospitals across the UK (Mohan, 2002). Consistent and equitable service provision can be in direct conflict with calls for the more distributed approach to decision-making advocated by Ostrom (Alford, 2014;Ostrom & Ostrom, 1977). In empirically identifying the bottom-up, unsanctioned coproduction of clinically designed and NHS-delivered services, we should not assume that these practices are a straightforwardly 'good thing' simply because they are led by community members. It is possible that community-led initiatives might be clinically ineffective (e.g., making inpatient stays more appealing than a return home for isolated elderly patients) or might advance a narrow group interest. It does not, though, follow that the appropriate response is to ignore or quash them. Recognising fugitive practices of coproduction enables a more realistic and sustainable set of policy tools to be employed.

| CONCLUSION
Advocates of coproduction assert that 'service users and their communities can-and often should-be part of service planning and delivery' (Bovaird, 2007). While supporting this normative goal, this paper asserts that mainstream public administration needs to expand its engagement with the ways that service users and communities are already part of service delivery, especially through informal and unsanctioned practices. Especially given current rhetoric and lofty aspirations around coproduction in Scotland, the disconnect between policy ambitions and day-to-day community practices within these case studies, is notable. Indeed, the general lack of interest in citizen-initiated practices unless they are immediately and instrumentally 'useful' for policymakers (see e.g., Brudney & England, 1983), resonates with Nisar's (2020) recent critique of public administration scholarship's imagined audiences. Perhaps researchers should seek to inform community practices, not just those of public service organisations (see e.g., Durose et al., 2016;Nickels & Clark, 2019).
Asserting the existence of fugitive forms of coproduction within healthcare is ripe for accusations of conceptual stretching given the ubiquity of the term. While Dudau et al. (2019) call for 'suspicion' and 'constructive disenchantment' with coproduction, they acknowledge that there is more, empirically, to understand about collaborations between citizens and state. I argue there is particular analytic purchase in exploring conflicts of interest between state institutions and different local publics. Ostrom and Ostrom's (1977) interconnections between consumers and producers of services-which Needham (2008) asserts as negating more adversarial consumeristic approaches to governance-do not entail shared interests. While inviting further inquiry to explore generalisability to other contexts, this study has identified a set of practices which, while meeting broad definitions of coproduction, fit poorly within Nabatchi et al.'s (2017) typology of coproduction. The consequences of this conceptualisation for practice are considerable. Organisations need to recognise the myriad actors and actions which shift the dynamics of public ser- vices. An ethos of coproduction entails shifts to more flexible and emergent strategies of governance (Bovaird et al., 2019). Acknowledging fugitive coproductive practices means accepting that local services may be tailored in unpredictable ways that shift over time as individuals and groups become more or less active. Embracing the rhetoric of coproduction at a system level thus requires a conscious loosening of expectations of control from the centre, resonating with Loeffler's (2021, p. 24) suggestion of an 'outside-in' pathway for coproduction, where 'public service organisations map what service users and communities are doing already… and build on this'. Contributing to debates about citizen motivation for engaging in coproduction (Alford, 2002), this mapping might additionally reveal whether and how practices of fugitive coproduction in healthcare develop differently in cases without a mobilising campaign to 'save' hospitals.
The considerable differences in practices identified across the three case studies also suggests that a more respectful attitude to fugitive coproduction requires active attention to questions of equity. Fugitive coproductive practices are often found in apparently mundane acts, and may look very different in communities where resources of social capital, professional skills and spare time are lacking (Matthews & Hastings, 2013). As case 3 suggests, where these resources are limited, fugitive practices might appear more oppositional than is comfortable. Given the dual risks of entrenching inequalities by creating routes for affluent communities to enhance their services, and of crowding out practices of care by seeking to formalise them, it is vital that decision makers act sensitively in supporting marginalised community practices within their social contexts.