QSHC and BMJ distinguish research reports (presented as IMRAD – Introduction, Methods, Results and Discussion)¹ from quality improvement reports (presented as COMPASEN – Context, Outline of problem, Measures, Process, Analysis, Strategy for change, Effects of change, and Next steps).² In their taxonomy, research is seen as systematic and focused enquiry seeking truths that are transferable beyond the setting in which they were generated, while quality improvement is seen as real time, real world work undertaken by teams who deliver services.

Quality improvement research—that is, applied research aimed at informing change that improves policy and practice—has emerged as a tradition in its own right, embracing a broad range of methods including randomised controlled trials, cross-case comparisons, in-depth case study, and action research. The question of when quality improvement initiatives should be published as new knowledge in academic journals (and under what circumstances they should be subjected to the stringent ethical and data protection standards that now govern research) remains a contentious issue,^3,4 especially when the study involves the use of narrative (a term which, in this paper, we use interchangeably with “story”).

WHAT IS NARRATIVE?

Bruner distinguishes two forms of human cognition: logico-scientific (“science of the concrete”) and narrative (“science of the imagination”).⁵ Logico-scientific reasoning seeks to understand specific phenomena as examples of general laws, while narrative reasoning seeks to understand them in terms of human experience and purpose.^6,7 Conventional research relies mostly on the former.

Stories do not convince by their objective truth but by their emotional impact on the reader—achieved through such literary features as aesthetic appeal (the story is seen as touching, humorous, ironic, for example), metaphor (one thing is made more meaningful or vivid through subjective comparison with something else), and moral order (actors are constructed as heroes who get their just rewards and villains who deserve come-uppance).^5,8

In one of the great works of literary analysis, Poetics, Aristotle suggested that a story (narrative) has three key characteristics: an unfolding of events and actions over time; emplotment (the rhetorical juxtaposition of these events and actions to evoke meaning, motive, and causality); and trouble (peripeteia – the unexpected in the form of surprise, “twist in the plot”, and so on).⁸ Burke later proposed that story is about purposeful action in the face of adversity and risk, and comprises five key elements: the act (what is done); the scene (the context in which it is done); the agent or actor (who does it); the agency (how it is done); and the purpose (why it is done).⁹ Box 1 lists some unique advantages of narrative in relation to quality improvement research.

In this article (building on earlier taxonomies^7,10), we discuss four approaches to using narrative in quality improvement research (table 1):

• narrative interview;

• naturalistic story gathering;

• organisational case study; and

• narrative dimensions of collective sense-making.

For each we discuss the rationale, method, approach to data analysis, strengths and limitations. Readers unfamiliar with qualitative research might also find a general introductory text helpful.¹¹

APPROACH 1: NARRATIVE INTERVIEW

Rationale

Personal stories are readily collected and can provide a vivid window to the healthcare system within which people’s illness experiences are embedded. The study of end of life care based on the collection of the narratives of people dying from lung cancer or heart failure by Murray et al¹² is a good example.

Method

In narrative interview the researcher invites participants to “tell me what happened” and allows them to speak uninterrupted until the story ends. The key structuring devices are chronology (linking events in time) and emplotment (use of metaphors, imagery and rhetorical devices to imply causality and agency). The interview may be semi-structured (driven by a series of questions set out in advance) or unstructured (conducted in a more emergent, conversational style). In either case, prompts should only be used to preserve the flow of the story (hence “how did you feel at that point?” or “what happened next?”). The researcher might invite ideas for change in the form of a different ending to the story—as in “if you went through that experience again, what would make it easier for you?” Indeed, the prompt itself might be presented in the form of a short narrative fragment (vignette) and the respondent invited to continue the story. For a detailed methodology of narrative interviewing see Wengraf¹³ or Riessman.¹⁴

Analysis of data

Narrative interviews are, of course, qualitative data, and on one level can be approached using any mainstream method for analysing text. But narrative analysis per se takes the story as a whole, rather than segments of text, as its focus. Muller⁷ describes five overlapping stages of narrative analysis: entering the text (reading and preliminary coding to gain familiarity), interpreting (finding connections in the data through successive readings and reflection), verifying (searching the text and other sources for alternative explanations and confirmatory and disconfirming data), representing (writing up an account of what has been learned), and illustrating (selecting representative quotes).

These analytical stages can be approached through one of several disciplinary lenses. All share what Muller calls “the focus on the broad contours of the story”—that is, the context in which it is told, its structure, the dynamics of how the plot unfolds, and any patterns that emerge from multiple stories about the same event.⁷ Riessman, for example, suggests that narratives can by analysed conversationally (as teller-listener dialogue), performatively (as drama), or politically (the unfolding of events is seen as constrained by prevailing social and institutional norms).¹⁵ Frank uses a literary framework to analyse the stories of people with serious illness.¹⁶ Their narratives, he suggests, fall into three basic categories: “restitution”, “quest”, and “chaos”, corresponding to the literary genres of adventure, [coping with] tragedy, and nonsense. The narrator constructs their experiences respectively as restorative, heroic, or absurd using devices such as metaphor, exaggeration, understatement, and humour.

Strengths and limitations

The main strength of the narrative interview is its inherent subjectivity. The story is irreducibly perspectival. A frequent theme in classical literature (consider Great Expectations or The Grapes of Wrath) is the struggle of society’s underdogs against social injustice or institutional incompetence. Perhaps for this reason, the narrative interview comes into its own when considering a quality improvement initiative from the perspective of disadvantaged groups such as the socially excluded, the seriously ill, and the very old.¹⁷

But the perspectival nature of stories is potentially a major limitation when they are used as research data. Furthermore, a story is an interaction—an artistic and rhetorical performance for an audience who (actively or passively) shapes the telling. The narrative interview has been described as “practical production, the meaning of which is accomplished at the intersection of the interaction of interviewer and respondent”.¹⁸ A different interviewer on a different day will never be able to collect the “same” story from a respondent.

The challenge of narrative research is not to “control for” the inherent subjectivity, inconsistency, and emotionality of stories but to capture these phenomena as data and interpret them appropriately. Gabriel¹⁹ offers some sound methodological advice: “It is the researcher’s task not merely to celebrate the story or the narrative but to seek to use it as a vehicle for accessing deeper truths than the truths, half-truths and fictions of undigested personal experience.”

The process of “accessing deeper truths” is not straightforward, and narrative research should not be equated with privileging the judgement of the researcher over that of the informant. The validity of the research process rests heavily on evidence of the researcher’s reflexive awareness. Aristotle’s definition of good literature is that it has a powerful emotional impact on the reader.⁸ The researcher must acknowledge and engage with this emotional dimension—thus turning sympathy, joy, revulsion, and even “mixed emotions” into research data. As in all qualitative research, there should also be a transparent account of how the researcher decided what aspects of the story to include and exclude as data, and how inferences were made.

Incidentally, the need to systematically and repeatedly interrogate the “truth” of narratives collected in a research study is one reason why it is generally poor practice to use stories supplied for one purpose (for example, in a clinical encounter) for research. It would also be unethical. A more in-depth discussion of these and other issues of methodological rigour in narrative research can be found elsewhere in the literature.^15,20,21

APPROACH 2: NATURALISTIC STORY GATHERING

Rationale

As table 1 shows, stories told informally may be especially valuable for accessing that elusive composite of shared values and meaning systems that can make or break a quality improvement initiative—namely, organisational culture. Through the interplay and exchange of stories, members of an organisation interpret, contextualise and collaboratively reframe the events they hold in common. Clifford Geertz²² states: “Believing, with Max Weber, that man is an animal suspended in webs of significance he himself has spun, I take culture to be those webs, and the analysis of it to be therefore not an experimental science in search of law but an interpretive one in search of meaning” (page 5).

This semiotic (meaning based) view of culture is strongly echoed by three leading organisational anthropologists. Czarniawska¹⁰ states that “Modern institutions … run on fictions, and the task of the scholar is to study how these fictions are created and sustained” (page 10). Gabriel²⁰ highlights that stories exchanged by people in organisations have multiple functions in the creation of meaning: they variously inform, entertain, warn, advise, justify, explain, reassure, console, educate, sustain and transmit ideas or values, and draw moral lessons. Boje²³ has observed that informal stories in organisations are generally multi-authored (with different members alternating the role of teller and listener); highly reflexive (that is, the past is continually re-created and re-interpreted in the light of the present); dialogical (that is, the narrative is co-constructed through a dynamic interaction between each teller and listener); and often allusory and fragmented (emerging “in bits” rather than as fully formed narratives with a fixed cast of characters).

Method

The subtle complexities of organisational culture will be inaccessible to the researcher who arrives with a tape recorder and only collects narrative interviews as in Approach 1. Naturalistic enquiry, in which researchers undergo “immersion in the field” to study actors in their own environment, relatively free of intervention or control, offers scope to produce what Geertz has called “thick description”—that is, multilayered interpretation of social actions in context.²² The main data source for such enquiry is the stories and story fragments exchanged in informal interaction with staff. The researcher must, of course, be selective in deciding which of the hundreds of stories heard during the course of the field work to capture as data, and also in assigning the status of “story” to what might be no more than a sentence, and (as with ethnography in general) the naïve or incompetent researcher will lack the skill and sensitivity to select appropriately. For more detail on these methodological challenges in organisational research, see Boje’s recent book.²⁴

In one variant of the naturalistic approach, the researcher is appointed from within the quality improvement team rather than from an external research team (an example of autoethnography). In another variant, known as applied ethnography, the researcher explicitly feeds their ethnographic findings back into the organisation in order to effect change.²⁵

Analysis

The principles of narrative analysis set out in Approach 1 above are also pertinent to the analysis of stories gathered in naturalistic settings. Close analysis of the stories as texts will not, of course, be possible if they have been recorded second-order as retrospective field notes (as is usual practice in ethnography). However, collecting and interpreting particular stories within a wider ethnographic study of the organisation gives the researcher a rich context within which to interpret their significance. Atkinson (cited in Hosburgh²⁶) has strongly criticised the conventional narrative interview as characterised by “… an extraordinary absence of social context, social action, and social interaction [and]… remarkably little sense of how narratives are forged in face-to-face interaction or how they are elicited in given social contexts”.²⁷ In other words, Atkinson sees a very positive trade-off between the accuracy with which the text of a narrative is recorded and the richness (and hence validity) of the context in which it is captured. For a more detailed methodology of ethnography in general, see Geertz,²² and for organisational ethnography, see Gabriel²⁰ or Boje.²⁴

Strengths and limitations

A naturalistic approach enables the collection and comparison of multiple stories about a single issue or event. “Trouble” within the organisation will generate stories from different actors, and each person’s story will change with repeated tellings. Not only is the “real” version of events an unhelpful concept, but the very plasticity of stories is the key to what Gabriel calls the “organisational dreamworld”—with multiple narratives interacting and challenging one another.²⁰ Gabriel’s own field work has highlighted the contrast between organisations’ official version of their own story (“well oiled machine, cutting edge technology”) and the subversive metaphors used by the members (“the [pompous, incompetent] management, nothing works round here”).²⁰

The limitations of naturalistic story gathering are both practical and theoretical. “Prolonged immersion in the field” holds little currency with today’s cost conscious research funders and, in any case, quality improvement initiatives may move too quickly. Being an “insider” to the quality improvement initiative has both advantages (in-depth knowledge and understanding of issues, rich social networks, mutual trust, timeliness, a longitudinal perspective, ability to effect change and integrate research with development) and disadvantages (lack of critical distance, a specific role in the organisation, prejudices arising from past personal experience, lack of knowledge of wider context). These issues are discussed further by Winter et al.²⁸

APPROACH 3: ORGANISATIONAL CASE STUDY

Rationale

Case study research considers a social system (“case”) in context and explores it in sufficient detail to illuminate relationships and processes and provide insights into why particular events unfold as they do.^29,30 Like ethnography (with which it overlaps considerably), case study involves detailed reflexive field work leading to rich authentic description.

Method

Case study requires the prospective in-depth investigation of an organisation, team, or inter-agency initiative using multiple methods—typically, a combination of formal interviews, focus groups, participant observation, and collection of contemporaneous materials (minutes of meetings, emails, memos, etc). We must distinguish case study research, which can take years to produce, from the much more common quality improvement case report in which one or more members of a team tell a personal story about their initiative and its impact.

Constructing a case study requires considerable judgement and skill. The elements of the case must be iteratively defined through a sequence of sampling (to identify somewhere to start), progressive focusing (to refine and systematically explore what has been sampled), theorising (about interactions within the arbitrarily defined case and across the boundary with the world beyond it), analysing (testing how well the data fit the theory), and interpreting (deriving meaning from the data).^29,30

Analysis

Organisations are complex, containing much social action (and a good bit of trouble) so, in practice, most organisational case studies get analysed and written up in a story-like way. An organisational case study is generally based on a large, heterogeneous, dynamic and complex collection of empirical data, each component of which will first need to be analysed separately—for example, quantitative data statistically; qualitative data thematically—before being woven into a higher order interpretation of the whole picture and how and why it has changed through time.

“Storying” the case—that is, constructing a chronological emplotted account of the key actions and events—is a way of selecting which data to focus on and which to omit. It is also a way of drawing meaning from different data sources and making causal links between aspects of the case, either tentatively (as hypotheses to be tested in further research) or more firmly as lessons or conclusions (if the links are particularly strong and plausible). In practice, organisational case studies tend to be the product of several researchers working together over months or years, and the task of processing, interpreting and integrating the data into a coherent story is achieved through interaction between team members—especially the repeated exchange and negotiation of stories.

Stake³⁰ draws on Van Maanen³¹ to suggest four approaches to “storying” that researchers may use to present in-depth case studies:

• realist tales: a direct, matter-of-fact portrait, a chronological or biographical development of the case;

• confessional tales: the researcher’s personal account of coming to know the case and the challenges they faced;

• impressionist tales: a sequential description of several major components of the case, “personalised accounts of fleeting moments of fieldwork case in dramatic form”; and

• illustrative tales: the use of vignettes (storied episodes) to illustrate particular aspects of the case.

In all these approaches a good case study researcher, like a good storyteller, will use literary devices to place emphasis and convey surprises and ambiguities, and will “zoom in” judiciously to analyse the behaviour of individuals within (and as influenced by) the wider system. Once again, an important criterion for judging the rigour of a case study is evidence of the researchers’ reflexive awareness and the transparency of their inferences from the data.

Strengths and limitations

Case study has been described as “strong in reality”—that is, as having high potential for validity within the confines of the case itself.³⁰ But researchers who have been raised on the conventional hierarchy of evidence (with randomised controlled trials at the top and anecdote at the bottom) often find it hard to identify much value in case study research. The central epistemological question might be put: to what extent does case study trade external validity (i.e. direct transferability to other contexts) for internal coherence and richness, and (conversely) to what extent will a detailed and systematic analysis of one unique “case” give us robust, transferable lessons for elsewhere?

This question is much debated amongst case study theorists (see, for example, a recent compilation³²). Yin²⁹ takes a conventional scientific view that a case is only meaningful as a member of a sociological family of cases which provide the analytical framework to understand it (“previously developed theory is used as a template with which to compare the empirical results of case studies”). Stake,³⁰ in contrast, argues from an interpretive perspective that the case is meaningful in its own right (the “sample of one” or what he calls “the intrinsic study of the valued particular”).

May, quoted in Simons,³³ describes how his understanding of trees was changed forever when he saw a painting by Cézanne. The tree in the painting was not statistically representative of trees in general, nor did it contain features present in every tree. Nevertheless, the qualities that Cézanne had illuminated in his particular tree enabled the author to see every subsequent tree through new eyes. Simons argues that the hallmark of a good case study is this metaphorical (rather than scientific) generalisability.

Thor et al³⁴ describe a five year case study of an ambitious quality improvement initiative in an acute hospital, based on “learning facilitators” who helped a total of 93 project teams. The researchers attributed the success of the facilitators to allowing each clinical team to remain in charge of their ideas and adopting a supporting role that comprised (a) providing feedback on ideas and progress; (b) helping with demanding (and sometimes menial) tasks; (c) developing specialist skill and experience in quality improvement; and (d) taking responsibility for small practicalities such as refreshments for meetings. Arguably, the validity of this case does not rest on (nor would it be enhanced by) the presence of a “control group” or “comparative cases”: it rests on the authenticity of the observations and interpretations about what happened in this case. Of course, we cannot extrapolate these findings to every quality improvement project (for example, we cannot say that providing cookies at meetings will always improve the quality of decisions). But we can learn a general lesson from a facilitation approach characterised by “mucking in” and taking account of specific contextual features (in this case, that meetings were often held over mealtimes).

Another potential limitation of in-depth case study is that, because of the detailed contextual information necessary to understand the case, organisations (and the individuals within them) may be identifiable. Elwyn et al³⁵ have described a way of fictionalising organisational case studies by first abstracting the key themes from a sample of cases—for example, high user expectations, lack of cash, external policy mandates, and so on—and then writing a new story that includes all these key themes.

APPROACH 4: USING NARRATIVE IN COLLECTIVE SENSE-MAKING

Rationale

An alternative conception of narrative research is developing (and enacting) shared meanings rather than collecting or writing stories.³⁶ Contemporary theories of organisational development centre on the construction, acquisition, transfer and use of knowledge, and on the need for the members of an organisation to make sense of change efforts by assimilating them into their cognitive schemas.³⁷ Organisational members are active “framers”, cognitively making sense of the events, processes, objects and issues that comprise (say) a quality improvement initiative and fitting it into their current cognitive schemas. When the initiative (or the goals that justify it) are perceived as new, the individual may retain an outdated schema of the organisation rather than discarding or modifying it. The result is cognitive inertia—that is, the tendency to remain with the status quo and resist changes that lie “outside the frame”.³⁸

To be successfully assimilated by staff (and service users), a proposed change must make sense in a way that relates to previous understanding and experience—and this is where story comes in. People caught up in a change initiative (or who are trying to “stay afloat” in a changing environment) engage in a continuous stream of experience and action which generates a succession of equivocal situations.³⁹ They then retrospectively impose a structure or schema to make these situations sensible. In other words, the act of sense-making is itself the construction of a narrative, requiring elements to be selected out, highlighted as significant or surprising, juxtaposed with one another (and with the existing cognitive schema), given meaning, and so on.^40,41 For a taxonomy of such interpretive approaches, see Czarniawska.¹⁰

The notion of narrative as sense-making fits closely with a rapidly expanding “knowledge based” tradition within organisational research (including Senge’s work on the learning organisation⁴²). In this tradition, “management” is seen not as a technical process of keeping the cogs oiled and maintaining throughput, but as the effective creation and circulation of knowledge. Leadership is conceptualised not as “command and control” but as providing the opportunities and facilitation needed for people to build and exchange knowledge.⁴³ In a change effort the leader’s role is to pull a team together around a shared story of innovation, improvement, and action. In practice, this means initiating and leading discussions around the meaning of any proposed changes.

Method

A good example of this research approach is Paul Bate’s story of a quality improvement initiative in a “failing” UK NHS hospital trust.⁴⁴ Change was achieved through an action research design (defined as “a mutual learning process within which people work together to discover what the issues are, why they exist, and how they might be addressed”) in which members of the organisation met periodically for facilitated discussion, reflection, and action planning.⁴⁵ A striking aspect of the change effort in this study was the perception that all the members of the change team, from the chief executive to the cleaner, felt that they were all pulling together in a common ethical endeavour about which they had strong (and positive) personal feelings. The conventional narrative of organisational change, Bate argues, is couched in “the vocabulary of coercion, competition, tyranny, hegemony, control, subjection, engineering, manipulation, domination, subordination, resistance, opposition, diversity, negotiation, obedience and compliance”. The more contemporary “knowledge creation” approach uses different vocabulary: “cooperation, convergence, coherence, integration and consensus”, for which the development of a shared story can prove the critical mechanism.⁴⁵ In emphasising the critical role of enacting stories as a vehicle for collective action, Bate cites the commentary by Kling⁴⁶ on social movements:

“Social movements are constituted by the stories people tell to themselves and to one another. They reflect the deepest ways in which people understand who they are and to whom they are connected … They are constructed from the interweaving of personal and social biographies – from the narratives people rehearse to themselves about the nature of their lives … The construction of collective action, therefore, is inseparable from the construction of personal biography, from the ways, that is, we experience the imprecation of our individual and social selves”.

Strengths and limitations

One great strength of the sense-making approach is its action orientation. Collective sense-making of past and present events feeds into creative team-driven action—less narrative, more organisational drama. Buckler and Zein⁴⁷ emphasise the critical role of informal stories in organisational innovativeness and change. Stories, they claim, are inherently subversive. They create the backdrop for new visions and embody “permission to break the rules”. In an old fashioned machine bureaucracy, behaviours and events that go beyond the existing structures and systems are implicitly (and often explicitly) “wrong”. But, by presenting innovation and change in story form, its agents can be imbued with positive virtues such as courage, tenacity and creativity (“Mrs Patel from the records department went in and told them straight”). Rule breakers become hero[in]es and the change effort a moral epic, enabling members of the organisation to reject the rules and routines more confidently. By using the enacted story as the vehicle for change, a current of subversion can be run through a previously indifferent organisation.

The main limitation of collective sense-making is that it is a highly specialised and intensive approach, of which there are as yet few rigorously conducted examples in the literature.

CONCLUSIONS

The richness and flexibility of the story form make it an enticing addition to the researcher’s toolkit in the hard to research world of quality improvement. This paper has attempted to classify the different theoretical positions that might underpin a narrative approach to quality improvement research, and to provide some practical advice for those who seek to use these methods in empirical work.

The slippery nature of narrative truth means that we must be wary of so-called “narrative research” that is nothing of the sort. Box 2 lists a provisional checklist of questions to ask when considering whether stories elicited, interpreted, analysed, or constructed in relation to a quality improvement initiative should count as research at all, and to assess the rigour of such research.