Background Translating and scaling healthcare quality improvement (QI) and patient safety interventions remains a significant challenge. Context has been identified as a major factor in this. QI and patient safety research have begun to focus on context, with ethnography seen as a promising methodology for understanding the professional, organisational and cultural aspects of context. While ethnography is used to investigate the context of a variety of QI and safety interventions, the challenges inherent in effectively importing a qualitative methodology and its social science practitioners into this work have been largely unexamined.
Method and results We explain ethnography as a research practice grounded in theory and dependent on observations gathered and interpreted in particular ways. We then review the approach of health services literature to evaluating this sort of qualitative research. Although the study of context is an interest shared by both social scientists and healthcare QI and safety researchers, we identify three key points at which those ‘exporting’ ethnography as a methodology and those ‘importing’ it to deal with QI and safety challenges may diverge. We describe perspectival divergences on the methodology's mission, form and scale. At the level of mission we demonstrate how ethnography has been adapted to a ‘describe and feed back’ role in the service of QI. At the level of form, we show how the long-term embedded observation at the heart of ethnography can be adapted only so far to accommodate QI interests if both data quality and ethical standards are to be upheld. Finally, at the level of scale, we demonstrate one ethnographic study design that balances breadth of exposure with depth of experience in its observations and so generates a particular type of scalable findings.
Summary The effective export of ethnography into QI and safety research requires discussion and negotiation between social scientific and health services research perspectives, as well as creative approaches to producing self-reflexive data that will allow clinicians to understand their own context and so improve their own processes.
- Qualitative Research
- Quality Improvement
- Patient-Centred Care
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Despite a supply of good ideas and local successes, ‘many quality improvement (QI) projects often fail to reach their goals.’1 Faced with translating and scaling QI and safety interventions, the field has turned towards a study of context as a mitigating factor in the uptake of evidence-based findings at clinical and organisational levels.2 This is part of a broader acknowledgement that exclusively protocol-based interventions,3 ,4 and interventions without robustly theorised mechanisms for success which include attention to context, are likely to fail.5 ,6 As understandings of context have improved, the literature has moved from defining it as everything but the QI intervention itself,7 ,8 to more nuanced theories and empirical work in which professional, organisational, cultural and structural factors are seen as contextual influences on how quality and safety interventions are put into practice.9–11
Ethnography as practised in the social sciences shares this theory-driven approach to understanding how these contextual factors shape action.12 Longstanding social scientific interests in how professionalism13–16 and organisational factors17–20 influence human behaviour suggest a synergy between ethnographers and healthcare QI and safety researchers. After repeated calls for this synergy to be built into practice,21 ,22 ethnography has been used to understand the contexts surrounding a range of healthcare QI and safety topics. These include adverse event reporting23; intravenous medication errors24; healthcare information technology platforms25; continuous QI programmes26; ethical behaviour amongst clinicians27; preoperative checklists and operating room errors28–30; the practical application of abstract teamwork principles31; distributed team dynamics32; the generation of nosocomial infection data10 and the translation of large-scale QI interventions into new locations.33
Although ethnography has established itself as a powerful method for understanding healthcare context, as it has travelled from the social sciences into QI a number of key challenges to producing high-quality analyses have emerged. This paper takes up these challenges. It shows that assumptions made by ethnography's social scientist exporters and healthcare QI importers about the methodology's mission, form and scale are not necessarily synchronised. It offers practical solutions for improving the degree of match between the two perspectives and more effectively integrating ethnography into QI and safety efforts.
Ethnography: an overview
Ethnography involves direct observation of people's behaviour and their social environments. It is carried out over a sustained period of time, with a technically and theoretically trained ethnographer:
observing context through written, audio or video recordings;
performing more or less structured interviews;
analysing the use of artefacts such as documents and equipment.34
The ethnographer's goal in collecting these data is to develop a nuanced understanding of the social relationships and technical activities of interest in a study. It proceeds from the assumption that people create their own realities in the course of interacting with one another and their environment. This social constructivist approach to the facts ethnographers observe emphasises the role of theory and interpretation in the scientific process to an extent that is absent from most quantitative research. Indeed, the development of a theoretically informed reading of the social, artefactual and technical phenomena under observation is the central task of contemporary ‘interpretive’ ethnography.35
To accomplish this, ethnographers generate a ‘thick description’ of the people and environments they observe.12 Such a thick description does more than record surface appearances36: it interprets ‘how [people] shape and trim their actions to fit their principles,’ and vice versa.37 As Schwandt notes, ‘it is this interpretive characteristic of description rather than detail per se that makes it thick.’38 Indeed, focusing on the mere accrual of detail risks turning ethnography into ‘a laundry list of features and considerations.’39 As much as amassing a great volume of detail may be part of the sometimes ‘boring’ work of ethnography,40 this is not its central focus as it might be in a more Taylorist research programme engaged in watching and counting instances or time segments. Rather, ethnography focuses on using social scientific theory to sift through and interpret the most relevant and telling details.
These interpretations present a mirror image of the cultural context, relationships and technical activities under observation. It has been argued, however, that ‘invariably, the mirror distorts’ and the reflection may not be the observed subjects’ own, but rather one shaped by the ethnographer's interpretation.41 In this sense, ethnography is ‘highly dependent on the individual researcher's subjectivity, sensitivity and interpersonal skills,’42 with his or her theoretical grounding, professional and biographical experience shaping how the thick descriptive material is gathered and the ‘mirror image’ interpretation of that material developed.43–45
Evaluating ethnography in QI and safety
To address its inherent subjectivity, ethnography ‘has become more self-reflexive in recent years,’ substituting transparency of method for objectivity in observation.46 Transparency here is an adherence to, and presentation of, methodological detail and research best practices. Following the more positivist approaches championed in an ongoing debate within traditional ethnography,47 ,48 transparent, high-quality ethnographic work in the service of healthcare:
thinks about and presents its authors’ social biases to readers;
clearly describes its methods of sampling, data collection, analysis and the theories underpinning the authors’ interpretive thick description;
adequately triangulates data sources (such as observations, interviews and documents);
seeks out a wide range of perspectives when collecting data;
is attentive to cases that are exceptions to the authors’ interpretation;
demonstrates its relevance to problems identified in the literature;
A recent review of intensive care unit (ICU) ethnographies suggests that much of the research does not meet these criteria.53 To produce higher-quality data, ethnographers seeking to contribute to QI must be transparent in how their ‘mirror image’ interpretation has been generated and must point out the most likely ways this image of context might be distorted. In addition to adhering to these principles, we suggest, higher-quality ethnographic accounts of context also require attention to key differences in how the methodology's exporters (social scientists) and its importers (QI researchers) approach three of its core elements (see table 1).
Social science ethnographers generally, and medical ethnographers specifically, have often viewed their work as an opportunity to critique the present arrangement of social relations, revealing inequalities.22 Inherent in much of the theorisation that underpins their work is a search for how power is distributed and exercised in society.54 ,55 The end goal of their work is to offer plausible explanations for the way contextual elements like culture, professions and organisations are arranged. This ‘describe and critique’ model stands in contrast to the more utilitarian approach of evaluative and applied health services research. Often this has led to contextual descriptions of clinical practices, which are not necessarily intended for clinicians’ use or QI purposes. In this way ethnographic data from clinical settings have been used to illuminate present social relationships and power, but not necessarily to re-engineer social and cultural context to optimise quality and safety.
Recently, there have been suggestions that the traditional mission to ‘describe and critique’ be modified to one of ‘describe and feed back.’ Concurrent evaluative fieldwork—ethnographers feeding their descriptions back into the quality and safety process—has been proposed as a way to improve intervention success.5 Re-tasking ethnographers to produce output that serves quality and safety interests—shifting them from contextual analyses of health services to analyses of context for health services professionals—is, importantly, a shift in mission from one of distant critique to one of engaged service.56 ,57 It is a shift from describing action to ‘action research’ where thick description is focused on change.58 ,59
A key element of this shift in ethnography's mission is that it shapes the practices and perspectives of both exporting social scientists and importing QI researchers. As ethnography is imported into the QI field, its adapted ‘describe and feed back’ mission provides that same field with a self-reflexive resource for better understanding its own context. Kitto and colleagues argue that this increased self-awareness will offer QI greater freedom to grapple with the complexity inherent in translating and scaling interventions.60 Early efforts at acknowledging the specific norms and knowledge forms that shape the QI field itself have proved promising.61 Given that ethnography excels at revealing complexity and nuance, QI importers will need to brace themselves for a stream of complicating and confounding—rather than easy and transparent—findings on how clinicians and healthcare organisations interpret and operationalise policies, programmes, protocols and checklists aimed at improvement. There is rarely a single culprit or barrier in an ethnographic account but rather a complex interaction of contextual factors. Transparent discussions of both ethnography's strengths and adaptations as well as implementation science's own contextual limitations will need to continue if the exported method is to retain a critical vigour in its new environment.
Finally, although the traditional commitment to ‘describe and critique’ has been challenged as stultifying62 it is important to note that it is based in pragmatism. Ethnographers have learnt that, to have their work accepted by the widest possible audience, analysis ought to flow from a critically distant and disinterested position, rather than that of an interested agent of change.60 Sociologists are well aware of the challenges to their authority that are associated with moving from being ‘interpreters’ to ‘legislators’ of change.63 As Bosk points out, ‘all ethnographers know that the accounts of the partisans in action are accepted with great scepticism.’46 Taking up a ‘describe and feed back’ approach requires an acknowledgement of the possible loss of authority inherent in such a mission choice, and concrete design strategies to counter that possibility. Rather than abandoning critique, exported ethnography needs to work with its QI importers to negotiate a place for the self-reflexive challenge of power relations in both clinical and QI research environments.
A second point of mismatch between ethnography and quality and safety can be found at the level of form. Ethnographers tend to prefer long-term observation over short-term action. This preference has normative, technical and financial implications, each of which interacts with QI and safety. An ethnographer's training informs her that good work is, ‘predicated on sustained time in the field and taking time to build trust relationships with participants.’44 This is a commitment that is at once normative and technical, with ethnographers relying on long-term, trusting relationships to provide them with nuanced rather than surface and self-conscious contextual data. Sustained observation is, in this sense, a check against reactivity, or the so-called Hawthorne effect, in which subjects are assumed to alter their behaviour in response to being observed.64 ,65
As part of re-tasking ethnography to address the problem of context in QI and safety, some health service ethnographers are acknowledging that a flexible and pluralistic approach to research forms is required.44 ,66 Although there will always be a place for the long-term embedded ethnographer, ethnography conducted over shorter periods has been shown to produce data that are highly relevant to QI and safety.10 ,44 Similarly ‘video-reflexive ethnography’ in which time-delimited interactions are filmed and co-analysed by ethnographers and subjects has informed the development and implementation of interventions to improve handovers.67 ,68 Even with these advances considered, the way in which trust-building embeddedness and production-oriented efficiency ought to be balanced when importing ethnography into quality and safety research, remains an open question. At stake are not just ethnographers’ professional ethics and traditional mission of challenging power relations, but the quality, and so authority, of the data that are gathered.
Traditional ethnography aims to produce an in-depth account of the social context at a single site. It has been decried as ‘useful for describing social process in very specific circumstances but not for generalising across settings.’42 Health services ethnographers have used theoretical and technical tools to overcome this handicap and so ‘scale-up’ their findings. As an example, the ethnographic observations of a specific group of surgical residents responding to a particular regulatory intervention to improve quality and safety can be linked to a broader research literature on change in organisations.41 Another, more technical approach to scaling up health services ethnography adopts a multisite approach. Rather than one or two sites covered by a single ethnographer, this approach to studying context multiplies both field sites and the number of fieldworkers. In this way what might have been an ethnography of a specific ICU responding to a particular quality and safety intervention becomes a comparative study of 19 ICUs interpreting and implementing a national directive.10 The move by Dixon-Woods and colleagues to scale-up ethnography in this manner was, to our knowledge, the first instance of such an effort directed at examining QI and safety in the health services research literature. Although pioneering work, their alterations to ethnography's traditional scale raise important questions about how multisite fieldwork ought to balance breadth of exposure with depth of experience. As these questions are debated and research teams adopt their own balances it is worth recalling that conducting ethnography on a grand scale is not necessarily the only or best solution. Just as thick description is not the pursuit of detail for detail's sake, so ethnography in the service of QI is not simply about ‘scaling up.’ A small scale, but high-quality ethnography asking the sort of complex, nuanced questions the methodology excels at is a powerful way to gather actionable data that survey instruments can struggle to achieve.
Based on sustained, systematic and theoretically informed observations, ethnography produces an account of professional, organisational, cultural and structural factors. Although it is an inherently subjective methodology, criteria for evaluating the quality of ethnographic work—and so the fidelity of the descriptions it generates of context influencing action—are widely agreed on. In addition to following these best practices, the effective integration of ethnography into QI, involves negotiating operational definitions of the methodology's mission, form and scale that are consistent with high-quality work and the goals and values of those undertaking the research. Our own experience of multisite ethnography in support of QI in ICUs has included discussion of adaptations across these three dimensions. We share these here in an effort to encourage further discussion of these key points between social scientific exporters and QI importers.
Our core research team presently includes two social scientist leaders and two ethnographic field workers trained in social science. This team is presently conducting a comparative ethnography of ICUs in four hospitals in the USA. Funded by the Gordon and Betty Moore Foundation, the project seeks to describe the nature of interprofessional collaboration and family involvement in care, and to develop a diagnostic tool that will feed these ethnographic data back into ICUs, enabling them to improve their practice. We are committed to the ‘describe and feedback’ rather than the ‘describe and critique’ approach to the methodology's mission.
Despite this commitment, QI and patient safety research tends to take place in environments requiring rapid improvement, with funders, hospital executives, researchers and clinicians strongly focused on producing demonstrable results in relatively short periods of time. Disseminated as tools, checklists and protocols, this approach has yielded many helpful fixes, although their transferability and scalability remain stubborn problems. As the study of context, ethnography is time consuming, producing incremental insight that is suited to stimulating local introspection rather than the development of a generally applicable checklist. Negotiating at this point of mismatch, we have entered into conversation with the Moore Foundation and our advisory group of healthcare QI and ethnography experts. As part of negotiating the contextual challenges presented by the QI environment itself to ethnographic processes, we have begun developing a diagnostic tool that encourages ICUs to take a self-reflexive approach to identifying problems of interprofessional collaboration and family involvement in their practice. Each conversation between QI importers of ethnography and its social scientific exporters will, we suggest, require similar negotiations in which the ‘describe and feed back’ mission is tailored to the methodology's strengths.
At the level of form, we are negotiating a new approach to gathering ethnographic descriptions at the point where our unit of analysis—the ICU—meets our interest in patient and family involvement. In its design, our ethnography is of and for, the ICU clinicians we are observing, with our ethnographers embedded in ICUs and their context. At the same time we have taken up the challenge of improving understanding of families and their experiences as potential contributors to care. We are confident, after our first 600 h in the field, that we are gathering high-quality data of the ICUs and their processes. However, similar levels of exposure, familiarity and trust have proved technically and ethically challenging to attain with patients and families who traverse ICUs more or less rapidly and in various stages of illness and crisis. Again, in discussion with our QI partners we are developing study design adaptations at the level of form that will supplement the rich descriptions of patients and families we have gained through the eyes of the ICU clinicians. Our hope is to leverage our long-term relationships with the ICUs and their staff to attain as great an exposure to families as is possible and ethically appropriate. Similar conversations negotiating multiple QI priorities and the design of an ethnographic study to answer specific questions about context, will remain important.
At the level of scale, our project's comparative ethnography of four ICUs seeks to balance the rapid, broad exposure design of Dixon-Woods and colleagues10 with a more traditional ethnographic approach to depth and nuance of experience. Our intention in striking this new balance and in designing a project-specific data management framework that will bring the comparative observations of our two ethnographers into alignment for analysis and coding, is to create a space where local ICU differences, as well as differences in ethnographic style, are accommodated. This data management framework has been designed to facilitate analysis and, perhaps even more importantly, to give rise to scalable ‘describe and feed back’ output. To this end, our emergent efforts to draw general lessons from the specifics of our observations have focused on identifying ‘sensitising vignettes’ in developing our diagnostic tool. We are building these vignettes out of the full comparative dataset, selecting them in ways designed to facilitate local, rapid, QI oriented self-analysis. Our intention here is to have ICUs use the vignettes to start QI conversations locally, thus achieving the levels of buy in and ownership that have been shown to be important in achieving change.69 This effort to feed ethnographic findings back into the ICU is aligned with the QI mission, and also reflects trends in the literature towards self-reflexivity as a method for identifying local strengths and solutions.67 ,70 ,71 This innovative approach to bringing ethnographic data to bear on the contextual challenges presented to QI is the result of talks between the QI importers and exporting social scientists. These conversations about how to scale ethnography for optimal effect on healthcare QI are essential.
The effective integration of ethnography into QI and safety requires that it be designed into context research. These discussions of the methodology's mission, form and scale need to focus on the innovations and trade-offs involved in its importation. Ethnography will produce the highest-quality insights when the perspectives of both its QI importers and social science exporters are taken into consideration. The effective uptake of its findings also requires that both sides embrace its inherently self-reflexive approach not just to a particular research question, but even in the QI and safety environment in which it is being used. This openness to self-analysis will lead to better understanding of the complexity of the challenge that context poses to QI; improved research questions guiding ethnographic fieldwork; and creative approaches to using ethnographic data to allow healthcare clinicians to understand their own context and so improve their own processes.
This study was conducted with the support of the Gordon and Betty Moore Foundation.
Contributors ML developed the initial concept for the paper and led in its production. EP and SR contributed to the paper's theoretical and substantive development, commenting on and editing drafts. MAG read and commented on drafts. SK contributed to the paper's theoretical and substantive development, revising sections of the paper and commenting on and editing drafts.
Funding The Gordon and Betty Moore Foundation.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.