PII: Soc. Sci. Med. Vol. 47, No. 9, pp. 1267±1276, 1998 # 1998 Elsevier Science Ltd. All rights reserved Printed in Great Britain S0277-9536(98)00192-0 0277-9536/98 $19.00 + 0.00 DOCTOR TALK AND DIABETES: TOWARDS AN ANALYSIS OF THE CLINICAL CONSTRUCTION OF CHRONIC ILLNESS RONALD LOEWE,1* JOHN SCHWARTZMAN,2 JOSHUA FREEMAN,3 LAURIE QUINN4 and STEVE ZUCKERMAN5 Department of Family Practice, Cook County Hospital and the Department of Medical Education, University of Illinois at Chicago, Chicago, IL, U.S.A., 2Christian Industrial League and the Department of Psychiatry, University of Chicago, Chicago, IL, U.S.A., 3Department of Family Practice, College of Medicine, University of Texas Health Science Center at San Antonio, San Antonio, TX, U.S.A., 4College of Nursing, University of Illinois at Chicago, Chicago, IL, U.S.A. and 5 Department of Family Practice, Ravenswood Hospital, Chicago, IL, U.S.A. 1 AbstractÐDuring the last two decades the illness narrative has emerged as a popular North American literary form. Through poignant stories, well-educated patients have recounted their struggle with disabling diseases as well as with the hospitals and health care bureaucracies from whom they seek service. However, much less has been written about the doctor's narrative construction of chronic diseases either in the process of learning medicine or through diagnosing, treating and counseling chronically ill patients. Indeed, following Kleinman's lead, the physician's narrative has been narrowly viewed as a discourse on the veri®able manifestations of pathophysiology. Drawing on contemporary theories of storytelling Ð including the conception of narrative as conversational interaction Ð the present paper argues that doctor narratives are equally complex if quite dierent than patient stories. Indeed, through an analysis of doctor talk centering on diabetes mellitus collected in several distinct venues Ð case presentations, narrative interviews and medical consultations Ð it is argued that physician stories not only employ very evocative tropes, but that these stories combine didactic, rhetorical and soterological elements in the telling. The research was conducted at two, urban family practice training sites in Chicago. # 1998 Elsevier Science Ltd. All rights reserved Key wordsÐnarrative, discourse analysis, diabetes, chronic illness, doctor±patient relations INTRODUCTION Over the last twenty years the illness narrative or illness biography has emerged as a popular North American literary form as well as a primary data source for medical ethnography. Through vivid accounts of their struggle with chronic or disabling diseases, patients have attempted to educate doctors and other health care professionals about the impact of major diseases on their family life, identity and self-image as well as detail their travails with the medical establishment. A short list of such works would no doubt include Norman Cousin's two popular volumes, The Anatomy of an Illness (1979) and The Healing Heart (1983), The Body Silent by Robert Murphy (1990), and At the Will of the Body (1991) by the sociologist Arthur Frank. Even physicians have contributed to this literature, although in their roles as patients. In A Leg to Stand On (1984), for example, the neurologist Oliver Sacks describes in Gomanesque terms the ritual quality of his communication with the sur*Author for correspondence. {(See also A. Broyard, 1993). geon who sewed his leg back on, and the isolation and reprobation he felt as a convalescing patient.{ Through narratives such as these the complexity of the illness experience is seen. Unlike the standard medical history which reduces the patient's experience to a chronology of symptoms and signs, the illness narrative is a multistranded discourse, an individual attempt to constitute a meaningful explanation of disease by drawing on cultural and personal resources at hand; religion, family history, and folklore as well as aspects of biomedicine that are familiar to the patient. Much less has been written about the doctor's narrative construction of chronic diseases either in the process of learning medicine or through diagnosing, treating and counseling chronically ill patients. Only recently have scholars (Sharf, 1990; Hunter, 1991; Mishler, 1995a) begun to think of the medical interview, the case presentation or the morning report as a story, a speech event which is amenable to the principles of narrative analysis. While Arthur Kleinman does include physician stories in The Illness Narratives: Suering, Healing and the Human Condition (1988), unlike the patient stories (or explanatory models generally), these vignettes are not built up around particular disease 1267 1268 R. Loewe et al. episodes, but are broadly biographical and, therefore, do not shed much light on problems which emerge in clinical encounters between physicians and chronically ill patients. Secondly, as Barbara Sharf notes, Kleinman continues to de®ne the physician's disease narrative, as stories which are ``narrowly focused on the veri®able manifestations of pathophysiology (Sharf, 1990, 222)''. While there is certainly an element of truth to this description of physician talk, it strikes us as overly simplistic. Indeed, notwithstanding Kleinman's avowed interest in reforming medical practice, we believe such a view ultimately reinforces the image of the physician as a neutral observer of physical processes, an observer unfettered by the cultural or moral lenses which direct the gaze of others. More interesting, in our view, is Kleinman's contention in an earlier work (cf. Patients and Healers in the Context of Culture, (1980)) that doctors and patients are able to communicate only because the actual medical consultation deviates greatly from the medical model. How does it dier? What speci®cally facilitates communication? Why does communication about chronic conditions remain so frustrating for both parties? And how is our understanding of a medical consultation transformed if the medical consultation is viewed as an exchange of narratives? Nor has much been written on the way doctors and patients' stories meld or fail to meld although this is certainly one way to account for problems which occur in the management of chronic illnesses like diabetes mellitus. Notable exceptions include Elliot Mishler's recent writing on narrative and the discussion by Cheryl Mattingly (1994) of ``therapeutic emplotment''. Building on these works this paper argues that doctor narratives are equally complex, if quite dierent than, patient stories. In particular, through an analysis of doctor talk centering on diabetes mellitus collected in several distinct venues Ð case presentations, narrative interviews and medical consultations Ð it is argued that physician stories not only employ very evocative tropes, but that these stories combine didactic, rhetorical and soterological elements in the telling. Finally, the paper calls attention to some of the ways in which doctors' stories about diabetes are shaped by the cultural and class dierences between patient and provider. NARRATIVE IN A NUTSHELL As several writers have noted (Langellier, 1989; Riessman, 1993; Mishler, 1995b) the paradigm for analyzing orally produced texts is found in a study by Labov and Waletsky (1967), which examined stories that Black teenagers tell to members of their peer group (as well as Labov). According to this paradigm, narrative is a very restricted category of speech. To be considered narrative, in even a mini- mal sense, a speech event must contain a story about a speci®c past event which is remarkable, unusual, dangerous or strange in some way. A fully formed narrative, in their view, contains six additional formal and functional attributes: (1) an abstract, a summary or overview of the content which follows, (2) an orientation, a description of the setting including time, place, characters, etc., (3) a complicating action or sequence of events (e.g., a plot), (4) an evaluation, a statement by the narrator which interprets the meaning or signi®cance of the events described, (5) a resolution, an account of how things turned out, and (6) a coda, a statement which returns the listener or audience to present time. From Labov and Waletsky's standpoint narrative is representational rather than constitutive of reality. Elements of a narrative follow a chronology which mirrors the order of events in the narrated world. Events move in a linear way through time and the ``order can not be changed without changing the inferred sequence of events in the original semantic interpretation (Labov and Waletsky, 1967, 21)''. While the elements of Labov and Waletsky's model remain central to narrative analysis, they have also been criticized for failure to consider the context in which narrative texts are produced, speci®cally the relationships between the narrator and the researcher and the narrator and his peers. As Langellier asks: Which features of the Labovian model, for example, are by-products of the interview context? Is the presence of the abstract basically a summary response to an interview question for which narrative is an elaboration? Does the frequently absent coda signal the way in which the interview and the naturally-occurring conversation dier? What eect does the interviewer's outsider status have on the evaluative function (Langellier, 1989, 248). The idea that narrative events mirror or recapitulate ``actual events'', is even more problematic. How would we know that the events described in a narrative ever occurred, except through some other narrative? And how, then, are we to adjudicate between dierent narratives? Are some narratives privileged over others? The fact that narratives are by de®nition evaluative, e.g., that they embody the narrator's viewpoint or interest, suggests that there is no Archimedean point on which to test correspondence theories of reality. Nor is it clear why narratives are limited to stories about past events. Why can't narratives also be stories about events which are deemed likely to occur in the future, such as a doctor's prognosis? As Beach and Japp argue in another context, this view of temporality is overly restrictive, and ignores the ``way interactants switch from one time frame to another, from what has happened to what might happen (In Langellier, 1989, 258)''. Doctor talk and diabetes From criticisms such as these other modes of narrative analysis have emerged. Of particular interest to us are studies which fall broadly under the rubric of narrative as conversational interaction or personal interaction (Langellier, 1989). Associated with writers like Clark and Mishler (1992), Polanyi (1985), Sacks (1986) and Jeerson (1978) this perspective situates narrative in naturally occurring interactions which would, of course, include medical consultations, case presentations or a faculty precepting sessions. Rather than viewing narrative as a discrete unit of discourse, or a performance by a virtuoso speaker, personal narrative is mutually constructed by participants according to shared interaction rules (In Langellier, 1989). In fact, scholars working in this vein have been particularly interested in examining how the storytelling function is distributed among multiple speakers in larger discursive frames. Do they compete for the speaker's role, co-narrate the story, or just listen? While this approach implies that there are certain mutually understood rules governing the interaction, and the allocation of turns, it does not require that turns or time be distributed equally. As Polanyi points out, the recipient (e.g., the patient) of conversational storytelling is obliged to: (1) hear the story if it is proposed, or present a reason why it should not be told, (2) refrain from taking a turn except to ask a question or acknowledge that the story is being understood, and (3) demonstrate that the main point is grasped by making a comment, or by telling the next story (In Langellier, 1989, 258). In this frame, storytelling is also seen as a seamless part of the social relationships which exist or are being formed between narrator(s) and the audience. Rather than viewing the interlocutors simply as individuals occupying distinct speaker±listener roles, partners enact roles which are marked by dierences in gender, race, status and class. As with Labov, stories are evaluative, that is, they are told to make a speci®c point. For Polanyi, however, the story is only really eective if it is occasioned by some prior comment which makes the story pertinent. Or, as Langellier writes, ``a competent narrator builds a bridge from what is happening here and now in the ongoing talk to some state of aairs in the storyworld and back to the conversation again by relating the storyworld *One clinic is located on Chicago's near west side and serves an indigent, predominantly African±American population. The other clinic is located in Little Village on Chicago's southwest side, and serves a low-income, predominantly Latino population. While normally only interns are required to present their patients to the attending physicians, all residents were asked to present their diabetic patients when a member of the research team was present. {In one case, we blurred the conventional categories by interviewing a medical student who has had diabetes since age ®ve. 1269 and the interactional context (Langellier, 1989, 257)''. Therefore, good storytellers are not necessarily people with the best presentational skills, but individuals who can ad lib eectively, narrators who can formulate a story or adjust it to particular conversational context. Finally, it should be reiterated that narrative analysis need not be an end in itself. As Gee et al. (1992) points out there are two rather distinct uses of narrative or discourse analysis. In the ®rst category are studies which ``focus on the form, meaning and regularity of supersentence texts as representative instances of particular discourse genres (Gee et al., 1992, 229)''. Here, analyzing the structure of a text and comparing it to other genres of text is the goal. The other approach, as Gee notes, is to use discourse analysis as evidence in the investigation of some larger social or cognitive process. Our goal, in line with the second approach, is eminently practical: to understand the world view and practice of physicians treating diabetes as revealed in their narratives and interactions, and to learn why this is often such an unsatisfying process. METHODS As noted above, physician narratives were sampled in a variety of clinical and non-clinical contexts. In order to tap into the natural discourse between physicians regarding Type 2 diabetes, clinical precepting sessions between attending physicians and family practice residents were recorded at two family practice training sites in Chicago.* In addition, several medical consultations involving residents or attending physicians and their diabetic patients were videotaped and analyzed. This data collection technique was, of course, limited by the diculty of getting informed consent (and permission to videotape) from two individuals simultaneously (e.g., doctor and patient), but provided the research team with an opportunity to compare the intersubjective context of clinical care with statements made about clinical care by physicians and patients in narrative interviews. Finally, residents and attending physicians were interviewed outside the clinic using a ``minimally structured'' interview guide (see Appendix A).{ These interviews typically began with an invitation to discuss particular cases involving diabetic patients which the interviewee found interesting or illuminating in some way, and, thus, generated many rich stories about clinical care. By referring to precepting sessions between family practice attendings and residents as natural discourse, the authors do not intend to suggest that such sessions were not in any way aected by the presence of an observer. Attendings and residents were both aware of the observers, and attendings, in some cases, were asked to clarify medical terminology between precepting sessions. The dialogues 1270 R. Loewe et al. were, however, constrained by the fact that the residents were seeing actual patients at the time and, therefore, focused on issues the two parties deemed essential to patient care. In contrast, the interviews we conducted with physicians seemed less constrained by the exigencies of the situation, and enabled them to talk about patient care in a more ideal or abstract manner. The fact that the interviewer/interlocutor was in most cases a Ph.D. rather than a MD or a clinician also most likely aected the nature of the narratives produced. All of the speech events described above were audiotaped or videotaped, transcribed in their entirety and distributed to members of our research team which included three anthropologists, two family therapists, a family practice physician, a clinical psychologist and an endocrinologist.* The transcripts were, thus, coded and analyzed independently (in batches of ®ve or six) by each of the members and then discussed in monthly data analysis sessions which were also tape-recorded. In these meetings the interviewer/observer provided additional information about the context in which the stories were recorded, noting circumstances which may have aected the production of the text under discussion. Through this process, and the development of several concept papers, the team identi®ed a series of themes and strategies relating to problems in the management of diabetic patients, and further re®ned the interview protocols.{ Our decision to elicit stories in multiple settings is rooted in the belief that stories, like other forms of social action, are sensitive to the context in which they are told. It is also, however, a measure of our *Both of the family therapists are also Ph.D. anthropologists. A few months after the project began, the endocrinologist left and was replaced by a Ph.D. in Nursing who is also a certi®ed diabetes educator. {As the data set grew to unmanageable proportions the transcripts were entered in Nud.ist, a qualitative data management system, and indexed according to format (e.g., precepting session, interview, etc.), speaker attributes (e.g., ethnicity, gender, professional status, etc.) clinical site, and other features. This procedure not only enables us to quickly identify units of text relating to a particular theme, but to compare the use of key concepts like ``control'' across a variety of speakers. {In this regard, we note that an important aspect of residency training involves teaching doctors to restrict ``doctor talk'' to doctors. For example, in the course of preparing a workshop on doctor/patient communication, the Director of the pediatrics residency complained to one of the authors about a resident who reveled over the clarity of an EKG during a pediatric cardiology consultation. ``I'm sure the family didn't hear a word after the resident mentioned that the EKG came out crystal clear, but they have a very sick kid.'' A second intern, according to the Director, had to be admonished for telling a man that one of his relatives had died, ``but in perfect Harvard balance''. (Personal communication, Director of Pediatrics, September, 1996, Department of Pediatrics, Chicago, IL). dissatisfaction with the direction in which qualitative health research has moved in recent years. While we are grati®ed to see that other researchers have been making the case for the use of qualitative methods in health care research, we are disappointed by the tendency to reduce qualitative research to semi-structured interviews or to focus groups. While these are clearly more ecient ways of collecting data and facilitate comparisons across sites and across researchers, they ultimately produce a set of ``representations'' (Durkheim, 1915), ``ideal types'' (Weber, 1968), or ``doxa'' (Bourdieu, 1977), and tell us little or nothing about the more ineable realm of practice.{ Listening to medical stories in diverse settings also aids in the interpretive process. For example, after reviewing numerous precepting sessions where residents justi®ed their interventions to attending physicians on the basis of medical tests and epidemiological information, it is easy to understand why Dr. M (below) thinks that presenting statistics to her diabetic patients is an eective rhetorical stratagem for motivating poorly controlled diabetics, even though her supervising attending would disagree. Similarly, Dr. M's complaint, incomprehensible to the Ph.D. interviewer, that ``[patients] really aren't aware that dieting doesn't do any good without exercising three times a week and getting your heart rate up to 220 minus your age times 0.7'', serves as an analog of the problematic consultation. It exempli®es the diculty that this doctor has adjusting the message to audiences with less medical knowledge. In short, operating in multiple research venues enabled us to see how physicians sometimes innocently transfer one rhetorical strategm from a setting where it is appropriate to one in which it is inappropriate or incomprehensible. Similarly, the physician interviews not only produced stories which were qualitatively dierent from those that emerged in precepting sessions or medical consultations; they served as meta-narratives as well, or as we discuss in Section 5, commentaries on what constitutes a good story in the context of a diabetes consultation. ELEMENTS OF NARRATIVE Time and agency Patient stories generally begin with the discovery that they have diabetes, or one of the events leading up to their diagnosis, and move forward in time to the present or an inde®nite future. Physician narratives, on the other hand, begin at the moment of onset, at media res as Hippocrates suggested (Hunter, 1991), and move forward toward a more or less predictable future, either because the patient is thought to have little eective control over the Doctor talk and diabetes disease (i.e., biology and genetics supersede selfregulation) or the patient is unable or unwilling to regulate him or herself (i.e., environment and culture supersede self-regulation). In¯uenced by statistical reasoning, physician time becomes ``organic'' time in the sense that it is marked by the progressive and relatively predictable breakdown of speci®c organ systems. ...so nephropathy we believe is related to the length of time the person has had diabetes and the control they've had over the years.* There's some study that suggests that other things come into rule like genetics and race... but that's all theory... so its important to identify early, uhhh whether a patient's having [proteinuria] or not, because once a patient has gross proteinuria end stage renal failure or kidney transplant is within ®ve years... what I've heard from the nephrologist is once you have microalbuminuria it's like a ®fteen year type of progression from microalbuminuria to end stage renal disease, but its dierent for all patients... 50% of the patients that have diabetes for longer than 25 years had some signs of peripheral neuropathy... the ophthalmologists at Rush believe that anyone who's had diabetes over ten years, you can detect some degree of retinopathy [Taken from a presentation by a 3rd year medical student to a group of attendings and family practice residents.] But if the doctor's narrative tends to be progressive and episodic, a timeline which can be presented back to the patient who is ``caught up in the act of daily life'', there is also a de®nite tension between this aspect of narrative construction and other aspects of medical ideology and practice, namely the emphasis on intervention and individual responsibility. Touching on the issue of individual responsibility, Dr. G., a third year resident remarked: I tell them that the main, key player is themself, the patient. And all these people who are around, they should help him or her to improve themself, but like in our exercise program, I make it very clear that I cannot have a diet for you, you know. And I remember that one patient told me, ``Well'', he said, ``you are telling me to exercise. How come you don't do it yourself?'' I said, ``I'm not diabetic.'' You know, [laughing] the whole point is that I always make sure that they are the people who can control the disease. In short, while ambivalence about the bene®ts of glycemic control and a pervasive doubt that patients will exercise or eat well, leads to the formulation of a weak, unheroic subject, other ideals of medical practice and culture place the diabetic subject back at the helm. Or do they? Here it should be noted that following the soliloquy on ``responsibility'' cited above, the resident went on to discuss the great lengths he goes to in order to control his out of control patients. *The diabetes control and complications trial (Diabetes Control and Complications Trial Research Group, 1993) has shown that nephropathy, neuropathy and retinopathy can be slowed or prevented by good blood glucose control, but the bene®ts of tight control for Type 2 diabetes remains questionable. 1271 RL: Uhh huh [pause] How often do you schedule appointments for diabetic patients, or does it depend on the severity of their illness? Dr. G: Well it depends on how controlled they are... If it's somebody who is not controlled and they have a problem to monitor their sugar then I see them more frequently. Another trick, trick that I've been using lately, especially at [Little Village Clinic]... I ask them to come at least two or three times a week to the clinic to have their glucose checked by the nurse. Then they, most people come a couple of times a week fasting, then we check... if they have it [a glucose reading] in their chart or they write on a piece of paper, then they bring it with them. In light of this, it is easy to understand why Matthew Timmerly, Arthur Kleinman's diabetic patient, felt that the doctors were constantly ``switching signals'' on him. They tell you to run with the ball and help plan the play, then when the situation changes [the patient is out of control] they take the ball away from you and tell you to stay out of the decision making, leave it to the experts. It can be terribly frustrating (Kleinman, 1988, 171). The larger question, however, is whether the idea of ``taking responsibility'' makes sense to diabetics or other chronically ill patients. Even if Dr. G. did not attempt to ``tether'' his patients to the clinic by having them come ``at least two or three times a week'', would they feel capable of controlling their bodies and their diabetes, or do chronically ill patients begin to think that the body is ``alien to the acting self (Good, 1994, 124)?'' As Byron Good points out in the case of Brian, a patient diagnosed with temporomandibular joint syndrome (or TMJ), pain is what has agency. ``His body dominates consciousness, undermining his sense of being an `undivided total self' who is the `author' of activities, threatening to unmake the everyday world (Good, 1994, 125)''. In other words, unlike the doctor, the patient may not be able to separate herself into subjective and objective states Ð will and body Ð and use the former to control the latter. Rhetoric which cuts to the bone A recurrent theme among the health care providers (physicians and physician assistants) interviewed, is the diculty of getting asymptomatic patients to see the long term consequences of uncontrolled diabetes. This not only requires that patients accept the physicians' ``timeline'' noted above, but that they begin to visualize the invisible processes occurring in their bodies or their tangible eects. As one physician assistant noted: We can present all of the information, but are we gonna present it in a way in which they can understand it, and that can get through, you know, the 10 or 15 things that are ahead of diabetes as being important? So I think that's our problem as health care providers... they don't see that the changes they make really aects their life and keeps their kidneys and eyesight and everything. It's too far along, in the future to see any immediate... One solution to this dilemma, though clearly not the only solution, is to highlight the connection 1272 R. Loewe et al. between internal processes and external (i.e., visible) states through vivid imagery and synecdoche. Thus, rather than telling a common tale about the pancreas and insulin production, one experienced physician oered a very visceral tale about diabetes and foot problems in the course of a medical consultation, thereby reducing a complex phenomena to a highly cathected symbol. The patient, as Langellier suggests, simply demonstrates his understanding of the issues presented. on an ``inner narrative'', a subject reserved for case presentations and precepting sessions, at least one third year resident felt they might be used to persuade non-compliant patients: DR: O.K. when you're a diabetic you lose sensation in the bottom of your feet sometimes so you may have a cut that you don't even feel. RL: Do you think that Ð you emphasize the science and the statistics. Do you think that's really what motivates them to comply? PATIENT: Uh huh. DR: ...but I think if you do use numbers also I think that the people feel you have respect for them, that they have some intelligence and that you're talking to them uh, not in glittering generalities, but in numbers, I mean they are numbers, but they're not that scienti®c and I think that they have a little more respect for themselves and their ability to control if they can measure sugars, do this and that, and when you consider measure and write down things, then people can see for themselves. DR: And the only time you might notice it is when you take your socks o and you notice pus in your socks and by that time it's time for the knife. PATIENT: Serious. DR: You heard about diabetics losing their feet and stu like that? PATIENT: Right. DR: Well see that's why it happens because they don't check their feet every single day because decreased sensation in your feet allows infections to go a long way before you realize it is there. DR: But it seems that if you back things up with numbers sometimes, and say okay you have to exercise this many times a week, this many minutes, with your heart rate at such and such, then they seem to think it has a scienti®c reason. And I always let them know it's not a beauty issue, that this, this weight thing isn't meant to be a beauty issue, that it's a health risk. Such a view, of course, assumes that the scientism of the physician is shared by the patient and that quanti®cation provides the best window onto reality. PATIENT: Uh huh. DR: And by the time you realize it's there a strain of pus is already to the bone and it's time for the man with the knife in his hand to have a stab at your foot. That right. PATIENT: O.K. DR: And once they start cutting generally they cut a little bit here, cut a little bit there and next thing you know they're up to your knee and once you get one leg cut o generally the other leg is cut o within ®ve years because if you don't move around the circulation goes bad in the other leg real quickly. PATIENT: O.K. in real life you see what the reading is today. Therefore, while ``educating the patient'' is the phrase most commonly used by physicians to describe what they do in the context of a diabetes consultation, education does not simply involve presenting information to poorly educated diabetics Ð as several physicians imply Ð or even reminding patients of the things they repeatedly forget (e.g., proper nutrition), but returning to the things patients know best. Indeed, a second physician's comments about ``fear'' could serve as a commentary on the ®rst: Well fear's always there. I hate to say always a good tool to me, sometimes, but umm I think that that's one thing especially `cause most diabetics will know someone who had to get an amputation, you know. They de®nitely associate that with diabetes. Statistics as rhetoric In other cases, a more impersonal rhetoric is employed. While the discussion of numbers, hemoglobin A1C and creatinine levels, is generally part The physician as dramatis personae: the educator, the nagging wife, the compulsive doer and god In the course of interviewing physicians and eavesdropping on clinical precepting sessions, the authors were privy to a variety of dierent comments and re¯ections on the doctor's role in the diabetes drama, and their de®nition of the doctor± patient relationship. While several physicians discussed the importance of negotiating with patients to get them to eat fewer sugary treats, and one practitioner even spoke of forming a ``therapeutic alliance'' with her patients, several physicians told stories which portrayed the doctor as an adversary or authority ®gure. This self-characterization was particularly common when physicians chose to talk about the dierences between themselves and other providers (e.g., nurses, dieticians) who work with diabetic patients. While most of the physicians interviewed would readily admit that they learned little about diet and nutrition in medical school, a good part of the diabetes consultation was devoted to discussing this issue, notwithstanding the presence of dieticians and the severe constraints that a crowded public hospital or clinic places on a physician's time. The rather unremarkable revelation, as one resident stated, is that the physician speaks with the authoritative voice of medicine. It'll be great if some of those things the dietician is going to tell them, that we tell them, because some of them takes what we tell them as the gospel truth, [but] they take the dietician as a dietician. Doctor talk and diabetes A second physician, however, was less sure of whether she spoke with the ``authoritative voice of medicine'', or the voice of a nagging mother*: I guess what strikes me is it's just sort of a real bummer to be asking people to change their lifestyle and watch what they eat `cause it's just not very fun for them, and you just kinda feel sorry for them for being in that position, ummm it's just sorta like you're their mom, nagging them over and over again... What are the implications of this attitude for the management of diabetic patients? Likewise the need to intervene, to ``do something'', is noted by one physician and team member during a discussion about the absence of scienti®c evidence showing that tight glycemic control prevents many of the more common complications from developing: Ð the blood sugar's a marker of one of the manifestations of diabetes, but they're only manifestations of diabetes, and it isn't clear that by stepping on one [symptom], that you prevent the others, that by bringing down the fever in a cold you prevent them from having running noses... and I think you'll never convince doctors that controlling blood sugar doesn't help, because nobody's gonna do a study to try and prove it, and until we ®nd something else that we can watch to control, it helps... I certainly want to believe that John, because I can do something, that's something I can impact on... Indeed this comment echos a poignant remark made by the British physician Oliver Sacks after he sustained a leg injury which required a long period of convalescence and inactivity. Re¯ecting on his pre-injury status as a physician, Sacks states that he believed: ``in the noonday light of reason that whatever was worth accomplishing in life could be accomplished, by that strong masculine sense... enterprise... vigilance and activity which had characterized my previous endeavors (Sacks, 1984, 88)''. After the injury he was much less sure. What are the implications of this attitude for the care of diabetic patients? THE CONSULTATION AND ITS CONUNDRUMS: WHAT SHOULD THE PATIENT HEAR? Diabetes narratives not only convey a certain mood, but are told with certain purposes in mind. They can be used to frighten patients, to humor them, or to give them hope. Thus, one thing we learn through stories is the physician's view of human nature, what he or she believes motivates patients to action. By having physicians comment on these stories (in the course of an interview), we *While this somewhat negative, self-characterization might be viewed as a dilemma speci®c to female physicians, the concept of the physician as a nagging parent is echoed by male practitioners as well. {How physicians' stories resonate with patients, or what patients consider a good story, is unclear at this point. These are questions we hope to answer in the next phase of our research. 1273 create a meta-narrative, and learn more about what a good story entails. What are the canons of good storytelling in the context of a medical consultation, of a diabetes consultation? What mood should the storyteller create? While the research team encountered a variety of dierent narrative styles, some overarching similarities were also identi®ed. While fear is often necessary or inevitable, and providing hope is highly valued, providing ``false hope'' is anathema. Finding the balance between fear and hope is clearly the unstated key to being an eective storyteller from the physician's standpoint.{ As one third year resident clearly stated: The only problem is with fear, it can initiate actions, but I don't know if it maintains actions, and so I'm not convinced that's the way to go,... it's helping her now, but I'm not sure what it takes to maintain it. I think she's gonna need some other support and back up to keep her going because I think, you know, once the existence of what it was that scared her wears o I'm not sure she'll be able to continue... it has some bene®ts initiating, but I'm not sure how much more. What's most interesting about this aspect of the diabetes narrative, is that it represents one area where physicians do not want their diabetic patients to share their perspective on the disease. Whether because of their experience as medical students, or because of the questions that remain about the value of tight glycemic control in Type 2 diabetics, many physicians were horri®ed by the disease, and ambivalent about the standard medical treatment. As one attending put it: I think medical students get the experience of diabetes in the hospital frequently, so they see diabetics that are having their second cardiac bypass, or having their legs cut o, or on dialysis. They see the worst end of diabetes instead of seeing the beginning. And so we come into working with diabetics without a lot of hope, because what we see when we're training is the end product of people with uncontrolled diabetes... and so we sort of go, ``this is a horrible disease''. And someone who is diabetic doesn't want to hear that. They want to say, ``Oh there's lots of things you can do to live a long happy life.'' But what medical students get is this horrible picture of what a diabetic's life is. So, a lot of times we don't have a lot of hope... Similarly, other physicians and physician assistants we spoke with made invidious comparisons between diabetes and cancer, or diabetes and AIDS, and one third year resident we interviewed refused to recount her pessimism while the tape recorder was running. Therefore, while the physicians in this study clearly wanted their patients to recognize the progressive nature of the disease (without becoming fatalistic) and to realize that they can be sick without feeling ill, the unenviable task Ð the key to good storytelling Ð was to instill these perspectives without, at the same time, conveying one's sense of melancholy or pessimism. 1274 R. Loewe et al. CONCLUSION ``When we hear stories'', Catherine Riessman notes, ``we expect protagonists, inciting conditions and culminating events (Riessman, 1993, 18)''. In this way doctor narratives are like patient narratives or the narratives of any other group of storytellers. What's dierent is the nature of the protagonists, the conditions that do or don't incite them and how events actually culminate. As always, the devil is in the details. For this reason, we believe that it is time for researchers to begin delving into the details of doctor talk to learn why doctors and patients talk past one another when it comes to the management of chronic illnesses like diabetes. Despite invocations to study up the vast majority of social scienti®c research relating to health belief, action and practice still focuses exclusively on the patient. This study seeks to begin rectifying this problem. While physicians generally describe their task as ``educating the patient'' about the risk of diabetes, education is really a euphemism for a style of rhetoric or argumentation which involves graphic descriptions of patients in advanced stages of the disease, references to scienti®c reports, or entreaties to observe shared cultural ideals such as personal responsibility. In this regard it should be noted that ``control'', the centerpiece of many narratives, is a cultural ideal as well as a measure of metabolic functioning, and may be used in both ways in the course of a single consultation. While physician narratives are produced interactively and embody features of the setting in which they are produced, there are a variety of external factors which aect what gets told. One of the more salient ones seems to be the experience young physicians had with diabetic patients during a hospital rotation, or the stories he or she heard as a medical student. As one senior physician noted, medical students hear a lot of ``locker room talk'' about the diculty of changing patient behavior and ``come out of school somewhat jaded''. Thus, physician stories have a life of their own. Given the numerous discussions of fatalism and its implications for compliance in the anthropological literature it is also interesting to note the pessimistic tone which colors many physician narratives. Either because the patient is thought to have little eective control over the disease (i.e., biology and genetics supersede self-regulation) or the patient is *This point, in fact, is noted by others who have studied interactions between physicians and diabetic patients (Personal communication, Linda Hunt, at a panel discussion on diabetes at the Society of Medical Anthropology in Seattle, Washington, March 5, 1997). {This comment was made in a lecture titled, ``When the Patient is a Linguist: Re¯ections on the Language of Medicine'', at Rush Medical College in Chicago on Thursday, January 8, 1997. The lecture was part of the James Campbell Distinguished Lecture Series. unable or unwilling to regulate him or herself (i.e., environment and culture supersedes self-regulation), diabetes is thought to have a predictable and dire course. Its course can be plotted and timed. Physician narratives, in short, are teleological: they reveal a bleak prognosis tempered only by the belief that medical science will eventually rescue both the doctor and the patient from this dilemma. Such pessimism not only belies the more optimistic tone of the medical consultation, but leads to a series of conundrums. For example, while physicians privately acknowledge their doubts about the bene®ts of glycemic control, or even the possibility of losing weight, they often end up blaming their patients for poor compliance or negative outcomes.* As Suzanne Fleischman noted in a recent lecture on the language of medicine at Rush Medical College{ it is not unusual to uncover a ``metonymic slippage'' between the patient's aiction and the patient's personality in medical discourse. We would make the same point even more forcefully by arguing that the diabetes narrative Ð especially the emphasis on control Ð is part of a larger social discourse in which disease (as poverty or homelessness) is considered the result of an individual's failure to achieve autonomy. In short, ambiguity over locus of control is the core feature of the diabetes narrative. Secondly, while physicians may encourage patients to take control of their diabetes, they often subtly, or not so subtly Ð as Dr. G. illustrated Ð usurp the role of controlling agent as the drama unfolds. And ®nally, while the physicians in this study clearly wanted their patients to recognize the ``progressive'' nature of the disease and to realize that they can be sick without feeling ill, they also felt a need to instill these perspectives without, at the same time, imparting a sense of melancholy or fatalism. Herein, perhaps, lie some of the many reasons why management of diabetes is frequently a dicult and frustrating process for both parties. AcknowledgementsÐWe would like to express our sincere thanks to the American Academy of Family Physicians' Foundation (AAFP/F) for the generous ®nancial support they provided. Special thanks also go to DeAnn Pendry, Barbara Sharf, Anne Larme and Linda Hunt who read and commented on the manuscript, to Anamari Golf, who helped design the interview guide and assisted in the data collection process, to Mike Lieber, Eve Pinsker, Sean O'Sullivan and Pam Stewart who participated in the research design phase of this project and to Dionne Hart who interviewed patients at one of the clinical sites. Finally, we would like to acknowledge the tremendous support of Stephanie Tillman who transcribed most of the audio-tapes and helped coordinate group meetings and other research activities. REFERENCES Bourdieu, P. (1977) Outline of a Theory of Practice. Cambridge University Press, Cambridge. Doctor talk and diabetes Broyard, A. (1993) Intoxicated by my Illness and Other Writings on Life and Death. Fawcett Columbine, New York. Cousins, N. (1979) Anatomy of an Illness as Perceived by the Patient: Re¯ections on Healing and Regeneration. Bantom Books, New York. Cousins, N. (1983) The Healing Heart: Antidotes to Panic and Helplessness. Avon, New York. Clark, J. and Mishler, E. (1992) Attending to Patients' Stories. Sociology of Health and Illness 14(3), 344±371. Diabetes Control Complications Trial Research roup (1993) The New England Journal of Medicine 329, 977± 986. Durkheim, E. (1915) The Elementary Forms of Religious Life. Allen and Unwin, London. Frank, A. W. (1991) At the Will of the Body: Re¯ections on Illness. Houghton Miin Company. Frank, A. W. (1995) The Wounded Storyteller: Body, Illness and Ethics. University of Chicago Press, Chicago. Gee, J., Michaels, S. and O'Conner, M. C. (1992) Discourse Analysis. In The Handbook of Qualitative Research in Education, ed. M. LeCompte, W., Millroy, J. Preissle, pp. 227±291. Academic Press Inc., Harcourt Brace Jovanovich Publishers, London. Geertz, C. (1983) Local Knowledge: Further Essays in Interpretive Anthropology. Basic Books, New York. Good, B. (1994) Medicine, Rationality and Experience. Cambridge University Press. Hunter, K. M. (1991) Doctor's Stories: The Narrative Structure of Medical Knowledge. Princeton University Press, Princeton, NJ. Jeerson, G. (1978) Sequential Aspects of Storytelling in Conversation. In Studies in the Organization of Conversational Interaction, ed. J. Schenkein, pp. 219± 248. Academic Press, New York. Kleinman, A. (1980) Patients and Healers in the Context of Culture. University of California Press, Berkeley. Kleinman, A. (1988) The Illness Narratives: Suering, Healing and the Human Condition. Basic Books, New York. Labov, W. and Waletsky, J. (1967) Narrative analysis: oral versions of personal experience. In: Essays in the Verbal and Visual Arts, ed. J. Helms, pp. 12±44. University of Washington, Seattle. Labov, W. (1972) Language in the Inner City. University of Pennsylvania, Philadelphia. 1275 Langellier, K. (1989) Personal narratives: perspectives on theory and research. Text and Performance Quarterly 9, 243±276. Mattingly, C. (1994) The concept of therapeutic ``emplotment''. Social Science and Medicine 38(6), 811±822. Mishler, E. G. (1995a) Narrative accounts in clinical and research interviews. In The Construction of Professional Discourse, eds. B.-L. Gunnarson, P. Linell and B. Nordberg. Longman, London. Mishler, E. G. (1995b) Models of narrative analysis: a typology. Journal of Narrative and Life History 5(2), 87±123. Murphy, R. F. (1990) The Body Silent. Norton, New York. Polanyi, L. (1985) Conversational Storytelling. Discourse and Dialogue. In: Handbook of Discourse Analysis, ed. T. van Dijk, Vol. 3, pp. 98±115. 4 Vols. Academic Press, London. Riessman, C. K. (1993) Narrative Analysis. Qualitative Research Methods Series, No. 30. Sage Publications, Newbury Park, CA. Sacks, O. (1984) A Leg to Stand On. Summit Books, New York. Sacks, H. (1986) Some considerations of a story told in ordinary conversation. Poetics 15, 127±138. Sharf, B. (1990) Physician±patient communication as interpersonal rhetoric: a narrative approach. Health Communication 2(4), 217±231. Weber, M. (1968) Basic Concepts in Sociology. Citadel Press, New York. APPENDIX A Health Care Provider Interview Guide Introduction This research project is designed to help us gain a better understanding of the diculties of treating Type 2 diabetes. There are several components to our study. We will be looking at both patient and physician perspectives on diabetes. For now, however, we are spending time talking to doctors and other health care practitioners in order to better understand how they work with their diabetic patients. 1276 R. Loewe et al. Background Information 1. Tell me what it's like working with diabetic patients? Perhaps, you've had some recent experiences with Type 2 diabetes patients which you consider signi®cant or typical. Probes: a. Does working with diabetic patients dier from working with other types of patients? b. Is there any such thing as a ``typical'' diabetes patient [If yes, have the practitioner describe the typical diabetic. The point of this question is to elicit stories about diabetic patients which encapsulate some of the diculties in managing the disease and capture the practitioners conception of patient behavior, motivation, etc.] 2. Can you describe good treatment for diabetic patients? Probes: What are the essential ingredients of good treatment of diabetic patients? What advice would you give to medical students (nursing students)? [For interviewees who focus on medical outcomes, ask them to describe what constitutes a good (e.g., eective) relationship with their patients. Another possible probe would be: Do you think of your relationship with NIDDM patients in terms of some other type of relationship (e.g., student/teacher)? In this case, however, it would be better not to suggest a type of relationship! Let them come up with it.] 3. How often do you schedule routine visits for your diabetic patients? How much time do you spend with diabetic patients during a routine visit? What do you do during the course of a visit? 4. What's the most frustrating aspect of working with NIDDM patients? 5. What are some of the factors which lead to poor compliance? 6. Is there anything that health practitioners do which may contribute to poor compliance? 7. What, if any, background information do you need about your patients in order to manage their diabetes successfully? Probes: How, if at all, does the patient's culture, ethnic background or family aect management of this disease? 8. What, if anything, do you have to know about their families? Follow up questions: (a) Do you know the family members of your diabetic patients? (b) Do you treat the family members of your diabetic patients? 9. What should patients know about their diabetes? Probes: (a) Should the subtleties of diabetes be explained to the patient? [If yes, by whom?] (b) Should the health practitioner explain that adhering to a low glucose diet, taking medications, exercising, etc., may not result in improved outcomes? 10. What sorts of questions do patients come in with? Can you describe some of the more surprising questions you've been asked? 11. What were you taught in school (or during your residency) about Type 2 diabetes and the treatment of Type 2 patients? 12. How long have you been working with diabetic patients? 13. [If the interview is being conducted in a setting where there are a lot of diabetic patients, ask whether the practitioner chose to work with a diabetic population.]