Taking sides with patients using institutional ethnography

Purpose – The main purpose of this paper is to document the first author’s experience of using institutional ethnography (IE) to “take sides” in healthcare research. The authors illustrate the points with data and key findings from a study of cardiovascular disease prevention. Design/methodology/approach – The authors use Dorothy E Smith’s IE approach, and particularly the theoretical tool of “standpoint”. Findings – Starting with the development of the study, the authors trouble the researcher’s positionality, highlighting tensions between institutional knowledge of “prevention” and other locations where knowledge about patients’ health needs materialises. The authors outline how IE’s theoretically and methodologically integrated toolkit became a framework for “taking sides”with patients. They describe how the researcher used IE to take a standpoint and map institutional relations from that standpoint. They argue that IE enabled an innovative analysis but also reflect on the challenges of conducting an IE – the conceptual unpicking and (re)thinking, and demarcating boundaries of investigation within an expansive dataset. Originality/value – This paper illustrates IE’s relevance for organisational ethnographers wishing to find a theoretically robust approach to taking sides, and suggests ways in which the IE approachmight contribute to improving services, particularly healthcare. It provides an illustration of how taking a patient standpoint was accomplished in practice, and reflects on the challenges involved.


Introduction
This paper draws on the first author's experience of "taking sides" with patients in a study of cardiovascular disease (CVD) prevention in English family practice (Cupit, 2018). The first author, Caroline, employed Smith's (1987Smith's ( , 1990Smith's ( , 1999Smith's ( , 2005Smith's ( , 2006) "institutional ethnography" (IE) as the conceptual frame for taking sides (a standpoint) and conducting an investigation from that standpoint. We answer the question: how can the IE approach practically support ethnographers with navigating the problem of different "sides"?
The paper is divided in four parts. First, we introduce this article by problematising the researcher's own position in relation to her study of CVD prevention. Along with Smith and other sociologists who have been critical of the practices of sociology, we make the familiar strange by recognising the danger of immersion in the institutional domainthe risk that she might become a "non-reflective dealer", who unwittingly promotes institutional interests (Widerberg, 2017). Second, we outline IE's theoretical approach to "taking sides". This is important because the IE approach provided a structured ontological toolkit which shaped every aspect of the ethnography and was integral to her reflections on taking sides that we discuss in this paper. Third, we describe the researcher's experience of employing IE during fieldwork and analysisthe process of identifying difficulties experienced by patients, and pursuing an investigation for patients that would go beyond those experiences to address their institutional co-ordination. Finally, we reflect on the challenges of conducting an IE. Despite these, we nonetheless argue that "taking sides" using IE enabled a different "way of seeing" (Wolcott, 1999) and an innovative analysis. We discuss ways in which the IE approach might contribute to improving services, and particularly healthcare.
Observing from the desk: troubling the researcher's own positionality The study we discuss in this article was not originally framed as taking the standpoint of patients. Instead, the researcher (Caroline's) formal introduction to the study, in an academic Health Sciences research centre, was through what we call "desk ethnography"with immersion in the online institutional domain. Although the study subsequently included traditional fieldwork (interviews with patients who had been the targets of preventative intervention; observation of clinical consultations and interviews with staff members in two family practice sites; and further interviews with other clinician-and policymakerinformants), the early part of her ethnography was dominated by reading and absorbing a huge body of academic and policy literature, and following key influencers in policymaking and clinical practice on social media.
Using IE's lens, we argue that these were sources of "institutional" (or "ideological") knowledge about CVD prevention, defined as "ideas and social forms of consciousness [which] originate outside experience [. . .] a forced set of categories into which we must stuff the awkward and resistant actualities of our worlds" (Smith, 1987, p. 55). As Griffith and Smith (2014) and many others have documented, public-sector management regimes (including those delivering healthcare) increasingly rely on such institutional forms of knowledge (standardised, measurable and textually based) to manage services across different sites. In the field of healthcare, ideological knowledge enables the "machinery of management" to reify problems and produce simple ("reductionist") solutions (Pigg, 2013;Davis and Gonzalez, 2016). Institutional knowledge is often carried from one setting to another using "shell terms" (Smith, 2005, p. 112) that stand in for people's real, everyday work and activities. Although the activities of service-users and frontline workers are co-ordinated by such institutional knowledge, they are not usually involved in creating the categories and concepts on which it is based. Indeed much management activity is invisible to those at the frontline.
In the field of CVD prevention, the vast body of academic and policy literature Caroline was absorbing during the early part of her study was full of institutional knowledge about CVD prevention and related concepts such as "risk conditions", "healthy lifestyles", "motivation" and "adherence". Caroline found these ideas troublesome to navigate as she attempted to design a patient-focussed study; it was often impossible to see clearly how these concepts related to the everyday lives and activities of patientsespecially those from socially disadvantaged contexts, and in poor metabolic health, who were the object of preventative interventions. Despite her engagement with a rich body of critical social science literature, and serious intention to foreground patient experiences and concerns, we argue (retrospectively) that it was inevitable that Caroline would be influenced by (drawn into) a standpoint that originates in the "factual and value assumptions" of healthcare management (Ong, 2013).
Examples of institutional knowledge could be seen throughout policy and research literature. The following excerpt, from a key policy document, spoke in urgent, but abstracted, terms: The future health of millions of children, the sustainability of the NHS, and the economic prosperity of Britain all now depend on a radical upgrade in prevention and public health (NHS England, 2014) The message was clear if unspecific: the whole health community (including patients) should start to "take prevention seriously" by identifying and conscientiously managing "avoidable" risk factors (e.g. high blood pressure, high cholesterol, (pre)diabetes, atrial fibrillation) (NHS England, 2014;Department of Health, 2013). In other texts, policy aspirations were translated into practical instructions for healthcare workers to proactively provide information to patients about their risk. Patients were expected to then "avoid" future disease by taking particular forms of actionnotably pharmaceutical interventions.
Authoritative, institutional knowledge about CVD prevention was however in tension with anecdotal accounts from Caroline's personal networkspeople's everyday struggles to make "lifestyle changes", their apparent indifference to risk statistics and their concerns about taking preventative medications, for example. Although some of these concerns were found in the critical social science literature, they were rarely discussed within dominant CVD discoursein which these types of responses were typically framed as barriers to be overcome. As well, Caroline increasingly noticed such tensions between authoritative knowledge and patient experience as she listened to debates about "overdiagnosis" and "medicalisation" on an online professional forum for GPs (McCartney and Treadwell, 2014). These professionals highlighted examples from their everyday clinical practice, illustrating the harms (as they understood them) which arose from their practical enactment of authoritative forms of CVD prevention. Such tensions surfaced repeatedly during the study.
Of course, researchers using ethnographic methods aim to circumvent institutional knowledge (ideological frames) by exploring people's mundane practiceswhat actually happensin contrast to what is documented. However, this is easier said than done. Dominant ideological concepts and categories are affective in shaping researchers' own knowledge and values (Bowker and Star, 2000) which, in turn, determine topics and questions for investigation, and what may be known from the researcher's position (Stones, 2017). Even researchers employing ethnographic and other inductive approaches are in danger of (re)producing abstract, ideological concepts that "avoid describing people and their actions" (Billig, 2013). This may particularly arise as researchers studying the complex problems of healthcare and healthcare improvement grapple with time constraints driven by demands of funding or quality improvement agendas . They may be inadvertently drawn into particular value-laden missions, which do not necessarily reflect patients' concerns (Leslie et al., 2014;Galasi nski, 2011) despite commitments to "patient-centred care" or "patient involvement" in research.
It was only after a considerable period of reading and networking, whilst formulating her research proposal, that Caroline discovered IE. In Smith's methodology, she found a well-articulated framework to orientate fieldwork and navigate the different "forms of knowledge" about CVD prevention that she had already begun to identify. The approach, originally rooted in feminist concerns, provided an ontological starting point for positioning herself and her research. It allowed her to explicitly "take sides" with patients, using observations and accounts of everyday health improvement activities and interactions with healthcare services to elevate patients' concernsto consider how preventative care might be improved from the standpoint of patients. It was an approach which would turn out not just to provide a starting point, but a framework for the whole research studythrough literature and policy review, observations of clinical consultations, interviews (with patients, healthcare professionals and policymakers), analysis, write-up of the ethnographic record and reflexive practice.
IE's method of enquiry for taking sides: exploring socially organised practices Since Becker (1967) first opened the debate about how to manage different "sides", scholars have continued to develop strategies to address values, bias, sympathies, partisanship, politics, ethics and so forth (see, amongst others, Doran, 1989;Scott et al., 1990;Collins, 1991;Lawson, 1991;Martin et al., 1991;Hammersley, 2001;Allmark et al., 2009) that may result in a "one-sided view" (Becker, 1967). By contrast, in the 1980s, sociologist Smith (1987) originated a sociology that requires the researcher to adopt an explicit standpoint when commencing ethnographic fieldwork. Smith (1987, p. 177) expressly directs "taking sides, beginning from some position, with some concern" (emphasis added). Dorothy Smith describes the IE approach as an "alternative sociology"a "sociology for people" rather than a sociology which follows conventional approaches "in which people [are] the objects [. . .], whose behaviour [is] to be explained" (Smith, 2005, p. 1). IE's mode of enquiry is ontologically very different to other forms of ethnography (see Smith, 2005 (including glossary on p. 223), 2006; Campbell and Gregor, 2002), providing an integrated theoretical toolkit that enables the researcher to explicitly track social ("ruling") relations "that stretch across time and place to coordinate people's activities" (Bisaillon, 2012). (See Smith's (2006, p. 3) "small hero" illustration.) For this reason, IE research tends to focus less on producing thick descriptions of local culture (from extensive periods of fieldwork in a particular geographic setting) and instead utilises case examples as the springboard for an analysis of ruling relations. The ethnographer may undertake extensive traditional fieldwork (and/or have pre-existing experience of the research topic), but it is the analysis of social relations, coordinated by texts (policies, protocols, electronic systems, etc.), and the dominant forms of knowledge embedded in them, that are the primary object of study.
Mapping these often-invisible relations, and how they co-ordinate "the actualities of people's everyday lives and experiences" (Smith, 2005, p. 10), enables the ethnographer to work in the interests of the standpoint group. The research goal is to empirically map and describe how people participate in ruling relations, how activities across the institution are co-ordinated in ways that create difficulties for those in the "standpoint location", and to discover "how the people 'we take the side of' are implicated in social organisation that extends beyond them" (Campbell and Gregor, 2002, p. 44). The task of explicating ruling relations necessarily involves tracking what people do (and know) who are variously positioned in relation to the institutional complex (not just from the standpoint position). However, it is the experiences of people in the standpoint location that contribute to the development of a "problematic" -"a guiding perspective" to orientate the research. As Smith (1987, p. 89) outlines, "[the problematic] directs attention to a possible set of questions that have yet to be posed or of puzzles that are not yet formulated as such, but are 'latent' in the actualities of our experienced worlds". Similar (but not equivalent) to a "research question", the problematic is often a summary statement or series of questions that highlight "disjunctures" (Rankin, 2017) tensions between what standpoint informants know from an experiential perspective and what is known about them from a ruling perspective.
Recognising different "forms of knowledge" is an important element of IE, and we discuss some of the challenges we found in navigating these later in this paper. To simplify, the ethnographer conceptualises alternative forms of knowledge: "experiential" knowledge (local, situated) and "ruling", institutional knowledge (abstract, ideological). Both of these different knowledge forms may be activated by an individual. As highlighted in our introduction, this knowledge is often translated into categories and concepts that "decentre" the experiences of people who are the objects of such institutional knowledge (Smith, 1987, p.8); knowledge is generated about but not from the located, embodied particularities of their everyday worlds (their standpoint). This knowledge is often carried via texts that are "reproduced many times so that people can read the same text in different places or at different times" (Smith and Turner, 2014, p. 5). They enter into people's activities as sequences of action (work → text → work) (Smith, 2006, p. 67). Everyone, including patients, "works" (even if their activities are not recognised as "work") (Smith, 2005, p. 229), and people activate texts as they co-ordinate their activities in relation to them.
Identifying a "standpoint position" is a heuristic to guide the research process (Smith, 2005, pp. 10, 32). While no two people in a "standpoint group" will share exactly the same experiences, the researcher is likely to find "patterns" of difficulties (Campbell and Gregor, 2002, p. 69)as experiences are co-ordinated within the same set of institutional ruling relations, that are active across multiple settings. This also means that empirical mapping from different standpoints will intersect; many of the same work processes will be troublesome (albeit with somewhat different consequences) from different institutional locations. A researcher may take up a standpoint which is not their own. However, they are "committed to knowing on behalf of those whose lives [they study]" (Campbell and Gregor, 2002, p. 48). Smith's (1987, p. 177) conceptualisation of "taking sides" therefore contrasts with those embedded in the discussions precipitated by Becker. Smith does not employ standpoint to take sides in conceptual debate between different groups of people (although debates will undoubtedly be uncovered during fieldwork), or to take a moral stance (Becker, 1967) (the creation of "goodies and baddies" (Hammersley, 1999, p. 11)). Rather, IE empirically examines the textually mediated institutional complex to show how people in both the standpoint and other positions are involved. Although ruling relations may be experienced differently by people in different standpoints, they are nonetheless "mappable" in the sense that actualities may be "known as known in common", as people from different perspectives engage dialogically with the findings of the enquiry (Smith, 1996). Perhaps ironically, the explicit adoption of a research standpoint, and the empirical mapping of "things happening" have been critical to countering accusations that the research is biased and/or distorted. The aim is for a warrantable analytic account that empirically explicates practices across institutional sites and that may help people to reformulate their ideas about problems and solutions. This is an account that holds up to scrutiny from divergent perspectives. (For elaboration on the IE stance regarding truth claims, see Smith (1996)).
An ethnography from the standpoint of patients Reframing "prevention" and "good care" There are frequent calls for research that is orientated to the priorities of patients, and useful in practice (James Lind Alliance, 2020). As highlighted in our introduction, there may also be tensions between different ways of knowing about healthcare. In her study, Caroline therefore chose to take the standpoint of patients who had experienced difficulties in their interactions with the systems and processes of prevention. This meant putting aside the policy and research literature in order to discover those difficulties from the "the actualities of people's everyday lives and experiences" (Smith, 2005, p. 10). Caroline started to interview people who had experienced preventative interventions through their local family practice. Many were keen to talk about their compliance with preventative protocolsexpressing an obligation to enact healthy diet and exercise patterns, and take prescribed medication to minimise CVD risk. Others however, found CVD prevention more troublesome. Here, we illustrate by drawing particularly on one patient's account -"Dan's story" (see Cupit, 2018, chap.1).
Dan (pseudonym) was in his 60s and had retired early from lorry-driving work due to an accident. He described having lived an "unhealthy lifestyle" shaped by the conditions of his job; he had become very overweight and had also been diagnosed with risk conditions for CVD (e.g. atrial fibrillation, high blood-pressure). However, following retirement, he had overhauled his diet and exercise patterns and lost a large amount of weight. In the following except from Dan's story, he recounts a consultation with his general practitioner (GP) in which he had suggested stopping some prescribed medications due to unpleasant side-effects: Dan's GP tried to persuade Dan against stopping his medications by citing research studies that showed they prevented heart attacks and strokes. However, Dan says they did not apply to him because they related to "population lifestyles [which he did not have]". Dan reports that his GP "was always, you know, "you need this, you need that, you need the other", and I am thinking 'well hang on a minute, I do not really need them'." The conversation came to a head when "[the GP] got very, very vociferous and "you know you will have to", [. . .] the word that comes to mind is belligerent, he was very adamant to the point of raising his voice. [...] There was no meaningful conversation between the two parties. It was "you will", "you must'". Dan tells me that he responded with "I will not" and ended the consultation with "thank you doctor, I am gone, and I walked out". [From fieldnotes summarising Dan's story] Using IE's social-organisation-of-knowledge approach, Caroline was able to see a disjuncture between Dan's experiential knowledge about how to improve his health (his way of thinking about "prevention"), and the GP's institutional knowledge about how to prevent CVD. These knowledges contained many overlapping elements; Dan had, for instance, overhauled his diet and exercise patterns in line with previous advice through his family practice. However, Dan's knowledge, as expressed in his conversation with Caroline, was rooted in his current well-beingfeeling good. In contrast, his GP appeared to have prioritised statistical knowledge about future CVD risk (and the pharmaceutical interventions understood to treat this risk) above Dan's immediate concerns.
Caroline had previously caught a glimpse of such tensions through reading the literature and engaging with clinician/patient anecdotesand she suspected that systems and processes such as clinical guidelines might be involved. However, dominant conceptualisations directed her to read accounts like Dan's either as relating to "adherence" (producing questions about how barriers to adherence might be overcome) or as relating to doctor "communication" (producing questions about professional training and development). In IE, Caroline found an ontological frame which provided structured "thinking tools" (such as different "forms of knowledge" and different "standpoints" inside knowledge regimes) to enable her to connect real, everyday problems experienced by patients with the bigger institutional context in which these problems occurred. Using IE promised a way of "seeing differently" during fieldwork, that would make institutional systems the object of investigation (things happening out of sight of patients and frontline clinicians) rather than patients or clinicians groups as the focus.
Using IE's conceptualisation of different standpoint locations inside knowledge regimes enabled Caroline to connect experiences that were at first glance quite different. Another informant, "Naomi", for example, had been unsuccessful at diet/exercise changes and was taking several preventative medications that were failing to control her metabolic biomarkers. In interview, she repeatedly cast herself as "unmotivated" whilst recounting desperately challenging socio-economic circumstances in which she was struggling to afford food, look after her children and manage everyday harassment and depression. In Naomi and Dan's very different experiences, a similar knowledge disjuncture could be identified in relation to the authorised knowledge of prevention; in Naomi's case, she internalised the tensionapplying institutional ideas about "motivation" to herself (how she should go to the gym, stop eating biscuits) that jarred with the realities of her emotional and physical resources. Both Naomi and Dan were objects of the institutional complex of "prevention".
Interviews with patients involved questioning them in such a way that Caroline could start to trace how their knowledge and activities intersected with those of frontline clinicians. Disjunctures such as those highlighted above provided a "way in" to direct her investigation. As DeVault and McCoy (2006, p. 24) have described, "the process of inquiry is rather like grabbing a ball of string, finding a thread, and then pulling it out". Finding specific "threads" was aided by selecting and working in detail with a limited number of particularly troubling patient accounts (such as Dan's and Naomi's) that illuminated tensions that were less obvious in more mundane accounts of the everyday "prevention work" of patients and providers. These precipitated questions to be explored. For example: Why did Dan find it difficult to achieve a constructive conversation with his GP? Why was his GP unwilling to support him in pursuing lifestyle change as his preferred course of preventative actiondespite policy and professional guidelines instructing him to "involve" the patient, respect their "needs and preferences" and "share decisions" with them? (see, for example, NICE, 2014). Why did Naomi's GP prescribe numerous tablets and hand her diet sheets but provide her with little individualised support?
In order to investigate these puzzles (to follow threads), Caroline then drew on observations of clinical consultations in family practice and interviews with healthcare workers and policymakers. This took her into the realm of the institutional systems and processes designed to co-ordinate good preventative care. Glimpses of these systems and processes could be seen in patient accounts. For example, Dan's GP told him that he "needed" to take particular medications and cited research studies. This was not based in knowledge about Dan and his individual needs but a textual knowledge arising from elsewhere. It was so strong that the GP became "adamant to the point of raising his voice". When Dan told Caroline about this interaction, she questioned him about it: perhaps the GP's insistence was a sign that he was genuinely concerned about Dan's health if he came off the medications? Dan however took a different view, arguing that the GP had a high sense of "self-importance" and was probably influenced by payments from pharmaceutical companies. To Dan, this was poor care: "a doctor's first priority is to look after the patient, not the medical companies".
Uncovering what goes onfrom different "sides" In this study, standpoint orientation was combined with an approach to analysis that depended on the rigorous collection of empirical evidence; the researcher aimed to "see [and show] every step [of particular processes] without having to imagine pieces" (Campbell and Gregor, 2002, p. 77). Although some processes could be mapped through observation, there were many barriers to undertaking traditional fieldwork, as is typical in healthcare organisations. As a result, Caroline did not have research ethics permission to observe back-office work, or the sites in which policy was made and enacted. Even to observe clinical consultations required that she negotiate access with individual clinicians and patientsand CVD prevention activities were often interspersed with other clinical work. In addition, observations provided limited answers, as care revolved around electronic platforms that could not be easily seen or understood by the ethnographer. For reasons such as these, and to build her empirical investigation, Caroline made extensive use of interviews with people working at different locations within the institutional system, beyond her interviews with patients. These interviews "extended ethnographic observation", enabling Caroline to access parts of the organisations involved in CVD prevention to which she did not have "time or license to go" (Mol, 2008, p. 11). Caroline used IE's ethnographically focussed interviewing style to ask informants "How do you know what to do?" "What do you do next?" in relation to various scenarios.
GPs' accounts of their work processes revealed their "position" inside the systems and processes of CVD prevention. They described, for example, what they would do if a patient wanted to avoid preventative medication and the system of diagnostic coding that directed their work. They were pulled into specific work processes when patients whose biometrics indicated that they were "at risk" did not receive standard treatments. For instance, they had to respond to drop down options on their computerised records indicating that the patient was "unsuitable" or had "declined" medications. These categories had consequences, and probing further (learning about and following work processes) led to systems for monitoring the performance of the family practice, and incentives embedded within those systems.
Following work processes was laborious but a crucial part of the study. During fieldwork (after each interview), Caroline sketched them out, often using simple diagrams such as in Figure 1, employing Smith's (2006, chap.4;2014) approach. These diagrams highlighted how texts (e.g. codes in an electronic patient record, quality metrics) were important in shaping everyday institutional processes (in Figure 1, the actions of healthcare professionals). Sketched diagrams were compared, corroborated and extended in interviews with other informants in order to empirically show threads of "things happening".  One valuable data source was the professional group concerned with "overdiagnosis" that we mentioned in our introduction. Caroline interviewed several of these GPs who expressed concerns about the preventative care they were able to provide. Similar to in Dan's story, their accounts highlighted disjunctures embedded in their workdisjunctures that often led to considerable frustration (see Cupit and Tobin, 2020). Although these activists were more attentive to (concerned about) the problems that arose for patients as part of preventative care, they were still bound by the institutional processes of the risk/prevention regime. We argue that similar empirical analyses would have been produced if Caroline had taken a standpoint with these healthcare workers; whilst the research would have been orientated differently (making visible the problems arising for professionals rather than for patients), the same systems and processes generated problems from either standpoint. Notably, Caroline found that mechanics of the institutional processes healthcare workers described closely matched those of both their non-activist colleagues and patients; across different groups, the co-ordination of systematic work processes were never disputed.

Mapping ruling relations of CVD prevention for patients
The heuristic of patient standpoint provided the ground from which Caroline began the process of identifying and mapping dominant textual processes. Uncovering the institutional relations that extended well beyond each provider/patient interface provided insight into Dan's storywhy his GP may have been so "vociferous" in persuading him to stay on preventative medications. Caroline discovered healthcare workers' onerous record keeping work, which was intensified when patterns of medication prescribing were out-of-step with clinical guidelines for the management of cardiovascular risk. Her investigation led to extensive descriptions of the way in which ruling relations organised healthcare professionals' (and patients') work in ways which made it difficult for them to have open and productive conversations about improving patients' healthto constructively "share decisions". As one GP expressed it "the temptation is there to push people to do stuff that you would not feel you need to do". There is insufficient space in this article to detail the complete analysis undertaken. However, Figure 2 summarises ruling textual processes that she found were orientated towards metrics of lives and costs "saved" (modelled as an accountability circuit (Griffith and Smith, 2014, p. 10)). Reading the circuit from bottom to top, her analysis showed how Dan's concerns to improve his immediate well-being, and his queries about the benefits and harms of medications (Box A), were not of primary importance within the institutional organisation of preventative care. GPs were institutionally organised to overrule patients' individual priorities and focus instead on ensuring that risk conditions (and statistical calculations of CVD risk) were identified and treated in accordance with clinical guidelines (Box B). Quality and pay-for-performance metrics (Box C) incentivised particular activities within the mass of guidelines. These metrics were used to demonstrate lives and costs saved in local and national policymaking (Box D). Clinicians' work was increasingly pulled into achieving these metrics by local and national policymakers who used them to undertake governance activities (Box E).
This IE investigation provided a platform from which standpoint informants (in this case, patients) or their advocates might be better able to "talk back"to highlight problems, and potentially to intervene in the institutional mechanisms that are implicated. It provided alternative explanations to those that were visible by those in the standpoint and other locations. Dan, for example, assumed that his GP was being canvassed by pharmaceutical companies (see above and "Dan's story" (Cupit, 2018, chap.1)). However, despite the undoubted influence of the pharmaceutical industry within the social organisation of preventative care (a topic for another study), the analysis showed that problems for patients were co-ordinated at a national level through performance management metrics and processesnot by individual pharmaceutical representatives influencing local GPs. The analysis also highlighted that, however forcefully clinicians or policymakers may advocate for better communication practices (a frequently repeated imperative), such performance measures (that incentivise pharmaceutical treatments and are orientated towards demonstrating lives and costs saved) are likely to continue to dominate preventative care practices (see Cupit et al., 2019).
In our example, Caroline explicitly maintained her interest in the standpoint of patients; we have described how the whole research process was orientated around disjunctures between patients' work and the institutional work into which they became enrolled. At the same time, the tensions embedded in GPs' work were also exposed, and we gained insight into the challenges of people who were differently located within the institutional complex. Anyone who has access to the map can see how people's activities in various standpoints (institutional locations) are co-ordinated (ruled) within from the social organisation of prevention as it is currently modelled and formulated.

Challenges of taking sides
Caroline encountered two main (and connected) difficulties in this study. The first was related to the different perspectives that were produced from various locations within the institutional complex. The second was more specifically related to the IE method of addressing the problem of "sides". First, Caroline found that informants whose paid work was to undertake prevention activities frequently activated institutional (ideological) knowledgeproviding what Campbell and Gregor (2002, p. 71) have described as "professional accounts". These accounts (from a different "side" to patients) typically utilised concepts and shell terms (see introduction) that threatened to obscure the problems that patients faced. For example, several GPs contended that they "shared decisions" about preventative medications with patients; for them there was no problem with preventative care practices, as patients were free to choose which interventions they would take up. Rather, these GPs located problems with patients' individual characteristics such as their "motivation" or "social problems". It was all-too-easy for Caroline to assume that so-called "shared decisions" were consisting of balanced conversations about the options available and the patient's preferences. She could create an illusion of understanding based on this abstracted, ideological frame, which in fact prevented her from "[seeing] every step without having to imagine pieces" (Campbell and Gregor, 2002, p. 77), as her own preunderstandings filled in the shell concept. However, when Caroline asked GPs and other healthcare workers to describe the process of "shared decision-making" in detail, they etched a very different picture to that which Caroline had initially envisioned. These detailed accounts often undermined the notion that so-called "shared decisions" were actually involving attentive discussion about patients' individual needs and circumstances. Instead, they resonated with troubling patient experiences, and observations of practice (see Cupit et al., 2019 for an extended analysis). Institutional (ideological) accounts were consistently produced from multiple institutional locations (even the standpoint location), obscuring problems for both patients and healthcare workers. They were challenging to navigate, involving detailed empirical mapping of people's textual activities in order to unpick what was really happening in practice.
Second, Caroline found it challenging to negotiate the boundaries of her investigation; the process of mapping institutional processes from a standpoint position extended fieldwork into sites of policy and management that are ordinarily not included within the local ethnographic setting (Rankin, 2017). As an indication of the volume of textual processes that could be tracked, each individual field on an electronic template may be understood as a "text" representing one end of a "thread" that extends deep into the institutionthat is made up of a long string of work → text → work processes. There were many "texts-within-texts" such as those in policy documents; contractual agreements; data management arrangements; performance measures and so on. Mapping people's sequences of action (and the texts involved) was laborious, and required alertness in interviews to judge how and when to drill down into the micro-detail of a work processand when to only to sketch the most major systems. Despite being guided by the research problematic, the research process often necessitated that Caroline follow a thread for a while and then back-up, rethink and adjust in order to focus on the most dominant organising processesthose with troublesome consequences for people in the standpoint group such as Dan and Naomi.
In many ways, the constant "digging" for detailed answers ("what happened then?"), and navigating an expansive field of enquiry, felt more like a forensic analysis or investigative journalism than the more informal "deep hanging out" of traditional ethnography (Gusterson, 2008).
Conclusions: reflection on IE's contribution to improving healthcare IE research conducted on behalf of standpoint informants is different to other forms of ethnography, such as that of business school ethnographers who explicitly aim to conduct their studies on behalf of an organisation (Neyland, 2007). It also radically diverges from more apparently "neutral" theoretical stances. We have shown how IE provided an integrated theoretical toolkit for a healthcare researcher to take sides with a group of people (in this case, patients) who were similarly located in an institutional matrix (the standpoint location), and to develop a robust empirical analysis into organisational work processes.
The IE approach offered more than sociological critique; we illustrated the process of describing, tracking and linking institutional processes. We have argued that, although the difficulties faced by the standpoint group grounded the project, the researcher used these experiences to go beyond these accounts and show socially organised work processes -"the institution". Mapping institutional relations provided a different view into the problems experienced by the standpoint groupand we highlighted the accountability practices that shaped frontline preventative care work (including these patient experiences).
IE provided an empirical ground from which to circumvent partisan opinions about problems and solutions, allowing the researcher to break into institutional (ideological) knowledge of CVD prevention, which was institutionalised within systems and processes. The research provided a way of producing research accounts that were aligned with what was important to the standpoint group, and examining whether and how the espoused aims of the organisation were actually being delivered in frontline practice. It troubled how well-intentioned aspirations were being operationalised, often in contradictory ways. The study therefore has the potential to support patients' (or their advocates') work in talking back to precipitate institutional change.
Within the domain of healthcare, the value of ethnography is increasingly being recognised (Savage, 2000;Dixon-Woods, 2003;Webster and Rice, 2019). The study by Cupit (2018) however is one of only a small number of previous studies in which the researcher has taken a patient (service-user) standpoint (researchers with clinical roles have tended to conduct research from their own standpoint). However, we argue that there is an important role for IE researchers in problem definition (identifying targets for improvement) from a patient standpoint, as well as in the evaluation of improvement interventions (for which ethnographic approaches are more commonly employed). Taking this approach, routinely collected "patient experience" data (e.g. https://www.careopinion.org.uk), or case reports of patient safety incidents, could be used as the springboard for an investigation of institutional systems and processes, for example. A patient (or other frontline) standpoint, combined with a research interest in in institutional systems (not individual/group responsibilisation) has the potential to generate constructive investigation and improvement . Instead of "making soft intelligence hard" by encoding the concerns of patients (and others) so that they are amenable to management interests (Martin et al., 2018), IE research, if undertaken with integrity, can make soft intelligence meaningfulthe foundation for changing healthcare systems and processes.
In addition, IE's work-focussed interviewing approach may be particularly useful for investigating what happens in clinical settings, where access is often difficult. It is wellknown that healthcare ethnographers often face research governance and management barriers to undertaking investigations (particularly those which involve investigation of administrative work/electronic systems). These difficulties are accentuated where the research is perceived to be burdensome, sensitive or otherwise inappropriate (e.g. with vulnerable people; coronavirus disease 2019 (COVID-19) infection control measures), or to have the potential to raise uncomfortable questions about institutional projects. Although interviews cannot entirely substitute for more traditional forms of ethnographic observation, analytical mapping ensures that they not only capture opinions, but also generate empirical accounts of what goes on and how work processes are organised.
Nonetheless, the process of undertaking an IE is challenging; researchers need to constantly make decisions about which threads to follow and where to set the boundaries of their investigation. Researchers should not underestimate this work. Whilst it is important that the research remains inductiveflexible to pursue unexpected lines of investigationstudies should have a clear focus around investigating a particular problematic; we recommend that decisions that deviate from this focus should be recorded in ethnographic fieldnotes and new boundaries redrawn. IE researchers should recognise that they cannot map all the social relations involved, but should concentrate on ruling institutional relations that create real and substantial difficulties for those in the standpoint position.
Research commissioners need to appreciate that service-users and/or other stakeholders may resist institutionally defined notions of quality and improvement (Swinglehurst et al., 2014). It should also be noted that, by definition, there are powerful interests embedded in ruling relations which may lead to institutional research from a service-user standpoint being overlooked or challenged. IE research is therefore always at heart an "underdog sociology" (Hammersley, 2001) requiring ethical commitment to making the difficulties of less powerful groups (e.g. patients) visible. Within healthcare, our experience suggests that such difficulties are often produced when institutional knowledge, systems and technologies are delivered into patients' lives without the individualised practices of "care" (Mol, 2008) that they need.