The Diagnostic Moment.pdf

The Diagnostic Moment: A Study in US Primary Care John Heritage, Sociology, UCLA Amanda McArthur, Sociology, UCLA Corresponding Author: John Heritage, 264 Haines Hall, UCLA, Los Angeles, CA 90095-1551. Tel. +1 310-206-5216 Fax +1 310-206-9838 Email: [email protected] Heritage ORCID ID: https://orcid.org/0000-0002-8603-6447 McArthur ORCID ID: https://orcid.org/0000-0002-0912-8006 2 ABSTRACT This paper conceptualizes the act of diagnosis in primary care as a 'diagnostic moment,’ comprising a diagnostic utterance in a 'diagnostic slot,' together with a patient response. Using a dataset of 201 treated conditions drawn from 255 video recorded medical visits with 71 physicians across 33 clinical practices in the Western United States, we investigate the incidence of diagnostic moments, aspects of their verbal design, and patient responsiveness. We find that only 53% of treated conditions in the dataset are associated with a diagnostic moment. Physicians present 66% of these diagnoses as hedged or otherwise doubtful, and deliver 30% of them without gazing at the patient. In the context of these diagnostic moments, patients are nonor minimally responsive 59% of the time. These findings underscore the different significance that may be accorded diagnosis in primary care in contrast to care in other medical contexts. The paper concludes that the analysis of sequences of action which empirically realize diagnosis are underrepresented in the sociology of diagnosis, and that better understanding of the diagnostic moment would enhance our understanding of diagnostic processes in primary care. KEYWORDS United States, Diagnosis, Primary Care, Conversation Analysis, Uncertainty, Verbal Design, Patient Response 3 INTRODUCTION This paper investigates the communication of diagnosis in the context of doctor-patient visits in US primary care. The significance of this process can hardly be gainsaid. It is widely understood within the medical profession that “a clinician's ability to diagnose accurately is central in assessing prognosis and prescribing effective treatments” (Heneghan et. al 2009:1003). According to the Institute of Medicine (2015:6), “the failure to (a) establish an accurate and timely explanation of the patient’s health problem(s), or (b) communicate that explanation to the patient” is to be considered a diagnostic error, because “an accurate and timely explanation of the health problem is meaningless unless this information reaches the patient.” Medical sociologists working in the field of diagnosis have also commented on the pivotal significance of what Jutel (2014:78) refers to as the ‘diagnostic moment,’ which, in the case of serious illness, “may turn the world on its head, instantly transforming the individual’s approach to life itself.” In this study, we begin from the perspective that, regardless of whether a diagnosis is serious and life-changing (Maynard 2003), or run-of-the-mill and self-limiting, the communication of diagnosis is a socially significant moment that is central to the expression of medical authority in society, and for the organization of patients’ experience of health and illness. For, as Jutel (2014:78) also notes, “even if your diagnosis was only bronchitis, you were not the same as when you went in. Today is somehow different from yesterday. You have a newly organized sense of what your experiences mean.” Despite these observations about the centrality of diagnosis in medical care, others suggest that diagnosis may not always be possible in the primary care context, and indeed may not always be necessary. Patients who visit the doctor early in the course of their illness often present symptoms that are not sufficiently developed to enable a clear diagnosis (Green & Holden 2003; Silverston 2016; Cox in prep). Even a fully developed cluster of symptoms may be 4 ambiguous as to their underlying disease process, and further testing may be required (Heneghan et. al 2009). Again, symptomatic treatment may not require full evaluation of an underlying diagnosis, especially in the majority of contexts where illnesses are self-limiting, or treatment based on an exact diagnosis is not necessary (Green & Holden 2003). Moreover, because a central role of primary care is to identify patients with serious illnesses and refer them to secondary care, diagnosis in this context may not require the same degree of specificity as in other contexts. As Heneghan et. al (2009:1004–5) observe, a GP’s job is to suspect an acute coronary event, and start appropriate treatment and referral, whereas in hospital the diagnosis relies on the precise classification of [an] acute coronary syndrome for appropriate management. Sometimes it is not practical to diagnose a specific microbiological or pathological cause. For example, conjunctivitis requires culture and polymerase chain reaction to identify the infective agent in 80% of children; however, this does not affect clinical management, and GPs are adept at identifying acute infective conjunctivitis, differentiating it from other causes of red eye, and initiating appropriate management. Finally, the largest subset of primary care patients present with significant symptomatic discomfort associated with relatively minor, self-limiting illnesses. For these patients, symptomatic treatment accompanied by reassurance that the illness is not serious may assume a greater relevance for the clinician than diagnosis per se. These contingencies may influence whether and how a diagnosis is determined and communicated to patients in primary care. This paper investigates the extent to which this may be true, by examining the prevalence of diagnosis in US primary care, its positioning within the clinical encounter, the content and manner of its delivery, and the nature of patient responsiveness to the information conveyed. 5 BACKGROUND Approaches to diagnosis Although medical sociologists have long examined social processes in which diagnosis is implicated, such as medicalization (Conrad 1975), it is only within the past decade or so that, taking up earlier initiatives from Blaxter (1978) and Brown (1990), a distinct sociology of diagnosis has begun to emerge (Jutel & Nettleton 2011; McGann & Hutson 2011). In the growing corpus of studies in this field, diagnosis is conceptualized in three main ways (Blaxter 1978; Jutel & Nettleton 2011): (i) as a socially constructed set of categories that are used to divide up the world of medical conditions (e.g. Armstrong 2011; Nettleton, Kitzinger & Kitzinger 2014); (ii) as a process (or processes) by which those categories are applied to individuals, through social and/or cognitive processes leading up to the ‘moment’ of diagnosis (e.g. Gardner et. al 2011; Armstrong & Hilton 2014); and (iii) as a set of consequences resulting from (i) and (ii), such as patients’ access to resources and the legitimization of sickness or deviance (e.g. Trundle 2011; Schaepe 2011). Largely absent from these studies, and the conceptualizations of diagnosis around which they are organized, is an examination of the actual moment of diagnosis delivery, and its reception, as a form of social action. Herein lies the significance of Jutel’s (2014) notion of the diagnostic moment. It allows us to distinguish between, on one hand, the patient’s gleaning of diagnostic information from the doctor’s conduct during the course of a visit, and on the other, the doctor’s action of diagnosis and the patient’s reciprocal experience of ‘being diagnosed,’ such that the patient can subsequently report that “The doctor said I have X.” In effect, although sociologists of diagnosis have highlighted the social importance of the diagnostic moment, they have empirically and theoretically focused their attention only on either side of it – i.e. the processes leading up to it and/or its subsequent social ramifications. Yet the Institute of Medicine 6 account quoted above clearly treats the diagnostic moment as an important element of successful medical practice. Moreover, as a long line of medical sociologists have observed, the medical profession’s exercise of authority is inextricably tied to patients’ reliance on physicians’ epistemic rights to define diseases and apply disease categories to patients (Parsons 1951; Freidson 1970; Starr 1982; Abbott 1988). To more fully understand the role diagnosis plays in medical practice and society, it is important to empirically explore the role diagnosis-as-socialaction plays in the lived worlds of doctors and patients as they interact with each other in clinical encounters. In this paper, we build on the conversation analytic tradition of examining clinical social actions like diagnosis in situ. Such research has explored uncertainty and diagnostic valence in the context of serious diagnoses in US secondary care (Maynard 1991, 1992, 1996, 2003, 2004; Maynard & Frankel 2006; Stivers & Timmermans 2016, 2017), and authority and accountability in the context of more routine diagnoses in British and Finnish primary care (Heath 1992; Peräkylä 1998, 2002, 2006). By examining diagnosis from the perspective of social action using a large collection of clinical encounters, we aim to contribute an empirical specification of the diagnostic moment as a routine part of primary care, and to expand its conceptualization as a component of the sociology of diagnosis. The 'Diagnostic Moment' in Medical Interaction In this section, drawing from existing scholarship on doctor-patient communication, we conceptualize diagnosis and the diagnostic moment as an interactional activity lodged within the unfolding clinical encounter in primary care. In the literature on this subject, diagnosis occupies a particular position in the structural organization of the medical visit, which is traditionally divided into six phases, comprising (i) opening, (ii) problem presentation, (iii) history-taking and 7 physical examination, (iv) diagnosis, (v) treatment, and (vi) closing (Byrne & Long 1976; Robinson 2003; Heritage & Maynard 2006a, 2006b; Koenig 2011). Figure 1 portrays the standard model widely employed to conceptualize the overarching structural organization of acute care visits. Figure 1: The ideal-typical phases of a primary care visit, based on Koenig (2011) This model depicts an orderly succession of phases, each of which builds upon the one before (Robinson 2003). An idealization based on empirical observations of actually-occurring primary care visits, it incorporates a number of assumptions about diagnosis in the structure of the visit. First, the diagnosis phase occurs in a particular temporal location, after history-taking and physical examination, and before the onset of treatment recommendations or other forms of medical counseling (Peräkylä 1998, 2002, 2006; Tarn et. al 2006; Koenig 2011). Second, while each of the six phases may incorporate a number of sub-components, the model suggests that both clinicians and patients orient to the phases as an ordered succession of distinctive activities, and tend not to experience difficulty in recognizing transitions between one phase and the next (Robinson & Stivers 2001; Robinson & Heritage 2005; Heritage & Clayman 2010). Thus, for example, a diagnosis is recognizable as the project, culmination, and outcome of a previous sequence of actions that have been investigative, and as the point at which the physician shifts from gathering information to delivering it (Roter & Hall 2006). While diagnostic information may emerge at other times and places in the medical visit, the diagnostic phase is the moment 8 that is officially devoted to the transmission of this information. Thus it is within the diagnosis phase that the ‘diagnostic moment’ normatively and ordinarily occurs. In this paper, we conceptualize the diagnostic moment as minimally comprising a twopart sequence. First, there is the production of an utterance that names (e.g. “this is bronchitis”), describes (e.g. “you have a nerve that’s pinching”), or provides an explanation for the patient’s problem (e.g. “my suspicion is it’s related to the smog”), which we refer to as the diagnostic utterance (DU). The moment this utterance is recognizably complete is the first possible moment a patient may transition from the status of unknowing to knowing, and the first moment that a patient can possibly have a ‘newly organized sense’ of what his or her symptoms mean. The DU occurs within a particular temporal ‘slot’ – the diagnostic slot (DS) – in the medical visit that is constituted by the ending of the history and physical examination. Just as Sacks observed that “something is a ‘greeting’ only if it’s a ‘greeting item’ in a ‘greeting place’” (Sacks 1992, Vol.1:97), we argue that something is a ‘diagnosis’ only if it is a ‘diagnosis item’ in a ‘diagnosis place.’ This conjoining of the diagnostic utterance with the diagnostic slot renders the delivered information hearable as a diagnosis, i.e. as a discrete social action concerned with naming, describing, or explaining the patient’s problem as a culmination of the prior investigation. The second component of the diagnostic moment is patient uptake, in which patients overtly recognize, acknowledge, accept, or ask for further information about the diagnosis. Patient response is an important part of a conceptualization of the diagnostic moment in medicine because, as discussed earlier, the Institute of Medicine stresses the importance of being sure that diagnostic information ‘reaches’ the patient. Physician recognition that this has indeed happened can only emerge as a product of patient response, which can variously embody claims or demonstrations of (non-)understanding, as well as inquiries exploring the nature of the 9 information conveyed. Absent patient response to a diagnosis, physicians cannot arrive at even a minimal determination of whether the information has indeed 'reached the patient' or not. In sum, we propose an operationalization of the 'diagnostic moment' that comprises a (minimally) two-part sequence including (i) the delivery of diagnostic information in the diagnostic slot and (ii) the patient’s acknowledgment of that information. In the remainder of this paper, we describe the incidence and characteristics of this moment as a matter of concrete empirical practice in US primary care. DATA & METHOD The following analysis is based on 255 video-recorded consultations that were gathered from 33 clinics involving 71 physicians across Southern California (2003-2005). Original collection of the data and their re-use for this study were approved by UCLA’s Institutional Review Board. For a detailed description of the procedure originally used to gather the data, see Robinson and Heritage (2006). Of the 255 patients recorded, 212 presented at least one new and/or acute problem that was potentially subject to a diagnosis. Observations were made at the problem level; some visits contained more than one new and/or acute medical problem, resulting in 244 observations in the 212 recordings under investigation. Of the original 244 new and diagnosable medical concerns presented by patients in the dataset, 201 (82%) received a treatment recommendation. These were selected as the basis of our analysis for the following reason: if a doctor does not diagnose a patient’s problem, but rather recommends further testing or provides a referral to a specialist, then a plausible explanation for the absence of a diagnosis is that no diagnosis is possible, given the available evidence. However, because treatment recommendations are ideally based on a physician’s diagnostic inferences, a dataset of treated problems provides a consistent analytic landscape within which to 10 consider the nature of the diagnostic moment in primary care, because it encompasses both occurrences and relevant non-occurrences of diagnoses within the database. The principal method of analysis used was conversation analysis (CA), from which a variety of coding categories were derived. Initial coding was performed by student researchers working in pairs. This coding was subsequently reviewed by the authors, who revised some coding categories and added a number of others, in a joint and iterative process (Glaser & Strauss 1967) that moved between conversation analytic observations and coding categories (Stivers 2015). Our revisions of the coding scheme concentrated, in particular, on relatively granular issues concerning the type, positioning and timing of patient uptake (see below). Coding the Diagnostic Moment First, in coding the diagnostic slot, we examined the clinician’s turn(s) occurring immediately after the completion of the physical exam or, where there were intervening activities (e.g. disposing of, or replacing, medical tools used in the exam, hand-washing, or chart entries), we looked to the next substantive turn at talk. In circumstances where either physician or patient initiated talk about another topic, we looked to the clinician’s subsequent turns at talk recognizably reengaging the issue investigated in the previous physical exam. Diagnostic utterances themselves presented several coding contingencies. In the simplest cases, as in (1), a DU comprises a single sentence or other turn-constructional unit (hereafter TCU; Sacks, Schegloff & Jefferson 1974) (e.g. “You have an Achilles’ tendon tear” in line 6) following the physical exam (PE in the transcripts). This provides an opportunity for patient response (PR in the transcripts; e.g. “Uh huh” in line 8), after which clinicians ordinarily move to the next relevant activity, such as recommending treatment (Rx in the transcripts; e.g. “it may require surgery” in lines 10-11). 11 (1) [19-05] Achilles’ tendon tear 1 2 3 4 5 6 7 8 9 10 11 12 13 DOC: PE-> Right in this area? [right here? PAT: [Yep, DOC: Yeah, .h [You have an aPAT: [The whole li:ne that you [wereDOC: [You- You haveDU-> You have an Achilles’ tendon tear. (1.2) PAT: PR=> Uh huh. (1.5) DOC: Rx-> ((clears throat)) A::nd u::h (2.0) it may require surgery. (0.5) PAT: #Oh Lord.# As (2) illustrates, DUs can also be composed of more than one TCU, followed by transition to the next relevant activity (here a treatment recommendation in lines 15-16). In this case, the patient responds to the first recognizably complete diagnostic TCU (“Sounds like (0.2) upper respiratory infection” in line 6) with “Okay” in line 8, but not to the expansion in line 9 (“an: most of the ti:me they’re virus”), nor the secondary diagnosis in line 11 (“An’ also you have some component of allergy”), or the evidentiary expansion of that diagnosis in line 13 (“I saw some swelling of your nasal mucosa”). (2) [25-01] Upper respiratory infection 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 DOC: DOC: DOC: PAT: DOC: DOC: DOC: DOC: PE-> (2.5) ((DOC’s listening to PAT’s lungs)) Okay. ˚khm˚ (1.0) That’s good. (2.0) ((DOC moves away from PAT)) DU-> Sounds like (0.2) upper respiratory infection, (0.5) ((DOC nodding)) PR=> Okay, Mka:y, an: most of the ti:me they’re virus. (0.2) .pt .hh An’ also you have some component of allergy. (.) I saw some swelling of your nasal mucosa. (.) Rx-> .h So I’m gonna treat you: (0.5) r-r-right now for symptomatic treatment. In a final set of cases, initial DUs undergo subsequent expansion that involves some etiologic, prognostic, or explanatory extension of the DU. In (3), following a diagnosis of dermatitis in lines 1-2, the physician expands the DU to discuss a possible etiology related to an 12 allergic reaction to jewelry (lines 4-10), and it is this to which the patient responds with “Right” in line 11 and “Okay” in line 12. (3) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 [14-05] Dermatitis DOC: DU-> No:w. .h What I think you ha:ve is u:h- is a type of=u:h dermatitis. (0.8) DOC: U:m (0.2) that (0.8) um (0.5) ↑we don't know the cause.↑ (0.2) DOC: But there are so:me (.) things that you might be allergic to. (0.2) DOC: Like the nickel that comes in the jewelry that is not pur:e wi[th gold. Y'kno:w, PAT: PR=> [Right. PAT: PR=> [Okay, DOC: [.hh E-The same thing with earrings.<Y'know if you don't have all gold earrings you'll end up with a ra:sh, ... In sum, DUs can take a single sentential form, or undergo a variety of forms of expansion. We treated all DUs, regardless of expansions, as single, codable diagnostic events. Finally, in coding patient responses to diagnoses, we began by focusing on the transition relevance place (TRP; Sacks, Schegloff & Jefferson 1974) following physicians’ first recognizably complete DU in the diagnostic slot, using the standard criteria of intonational, syntactic and pragmatic completion (Ford & Thompson 1996). Because, as illustrated above, diagnoses are sometimes composed of multiple parts, patients have multiple opportunities for response. However, the initial DU is the first possible point at which patients can acquire a "newly organized" sense of their condition, therefore we regarded this location as a primary observation point for patient response. We did not code further if, as in (2), the patient responded in this slot, regardless of whether or not the patient produced subsequent responses. If a physician rushed into an additional unit of talk or otherwise obscured the TRP and hence the patient’s opportunity for response, we coded for patient response at the first subsequent open TRP. 13 In the cases of expanded diagnoses with no initial patient response, we looked for response in subsequent TCUs. In two-thirds of these cases, patients remained unresponsive. In the remaining one-third of cases, we looked at whether the clinician persisted in diagnostic actions (reformulating, explaining, or otherwise expanding on the diagnosis). Patient responses to these expansions of diagnostic actions, as in (3), were coded as responses to the diagnosis, and thus as bona fide elements of the diagnostic moment. ANALYSIS & RESULTS Position of Diagnostic Utterances We distinguished three positions in which DUs occurred in the data: (i) in the diagnostic slot as previously defined, (ii) before completion of history-taking and/or physical examination, and (iii) during or after treatment recommendations. We did not treat DUs in positions (ii) and (iii) as initiations of fully-fledged diagnostic moments, because they are observably occupied with projects other than delivering a diagnosis per se, for the following reasons. Early diagnoses are frequently presented as overtly preliminary, speculative, or hypothetical (e.g., “I hope you don’t have a small fracture there”), and are often delivered as part of a doctor's project, for example, to manage perceived patient pressure for antibiotics (Stivers 2007; Stivers et al. 2003) (e.g., following a patient’s candidate diagnosis (Stivers 2002) of bronchitis: “you likely do have some inflammation in your bronchial tubes… bronchitis in a nonsmoker is something that generally does not benefit from antibiotics”), or to reassure a patient that their problem is not likely to be serious or life-threating (e.g., following a patient’s stated concern about cancer: “well let’s take a look… the most common thing to be on the back like that is something called seborrheic keratosis, which is completely benign”). Because they are recognizably delivered before some or all evidence has been collected, early DUs are not 14 hearable as the culmination of the doctor’s investigation. Similarly responses by the patient cannot be treated as responses to "the diagnosis." Indeed, doctors routinely confirm or disconfirm early DUs at the end of the physical exam, displaying an orientation to the insufficiency of the preliminary diagnosis as a diagnosis. In cases where there is no overt (dis)confirmation of early DUs, patients may grasp the general drift of the clinical reasoning behind their treatment by the end of the visit, but entertain residual uncertainties as to the condition for which they are being treated. At the other end of the visit, ‘late’ diagnostic utterances occur subsequent to, or embedded within, activities that typically follow the diagnosis phase, e.g. recommendations for treatment, and often serve as embedded or post hoc justifications for those recommendations (e.g., “I’m going to prescribe you something stronger for your neck called diclofenac, it’s – you probably just pulled a muscle – it’s like ibuprofen”). Diagnostic information delivered in this way is not hearable as a discrete social action primarily concerned with naming, describing or explaining the patient’s problem. Rather, it is entangled with, and often subordinated to, other advice-giving actions (see also Jefferson & Lee 1981; Whalen, Zimmerman & Whalen 1988; and Heritage 2011 on ‘activity contamination’). In these cases, patients may likewise gather information about their condition. However, the information is presented tangentially and, in most cases, patient acknowledgment is addressed to the treatment recommendation or other action in which diagnostic information is embedded, rather than to the embedded diagnostic information. Table 1 shows the distribution of diagnostic utterances relative to the diagnostic slot across the dataset. 15 Table 1 Distribution of diagnostic utterances in visits with treatment recommendations Position in Visit Frequency Percent DU in DS 106 53% DU outside DS 49 24% No DU anywhere 46 23% Total 201 100% As this table shows, a total of 77% of treatment recommendations were supported by some kind of DU. However, only 53% of these occurred in the DS, with the remaining 24% (n=49) occurring in locations outside of the DS. Of these 49 cases, slightly more than half occurred ‘early,’ prior to the completion of history-taking and/ or physical examination, while the remainder occurred ‘late,’ embedded within or after the treatment recommendation. Finally, 23% of treated problems were not supported by any DU. In cases in which a DU did not occur in the DS, a range of other actions occurred in place of diagnosis. Non-Diagnostic Utterances in the Diagnostic Slot Prominent among non-diagnostic utterances in the diagnostic slot were treatment recommendations (e.g., immediately following physical examination, “if this has been going on for five days, let me go ahead and give you an antibiotic”). These were common, representing 32% the cases. Less frequent were recommendations for further tests (e.g., “because of the swelling I’m going to go ahead and x-ray to make sure there’s not a bony deformity”; 8%) and uncertain diagnostic talk (e.g., “you don’t have anything when I examine you that makes me 16 pinpoint it to one spot like you have a cartilage tear or something like that”; 7%). Table 2 summarizes this distribution. Table 2 Distribution of actions in the diagnostic slot Action Type Frequency Percent DU (‘Diagnostic Moment’) 106 53% Treatment Recommendation 64 32% Test Recommendation 16 8% Uncertain Diagnostic Talk 15 7% Total 201 100% To investigate the incidence of these non-diagnostic utterances further, we examined their distribution in relation to particular illness contexts. Table 3 compares the incidence of three categories of actions in the diagnostic slot: diagnostic utterances, treatment recommendations, and assertions of underdetermined diagnostic conclusions (including recommendations for further tests and uncertain diagnostic talk). Table 3 Distribution of non-diagnostic actions in the diagnostic slot, by condition Action Type Diagnostic Utterance Assertion of Underdetermined Treatment Diagnostic Recommendation Conclusions Condition % (N) % (N) % (N) Total % (N) Upper Respiratory 43 (27) 48 (30) 10 (6) 101 (63) Musculoskeletal 56 (30) 20 (11) 24 (13) 100 (54) Remaining Cases 58 (49) 27 (23) 14 (12) 99 (84) All Cases 53 (106) 32 (64) 15 (31) 100 (201) Chi2 (4) = 13.54, p<.01 (percentage totals subject to rounding effects) 17 As Table 3 shows, in cases where patients presented with upper respiratory symptoms (e.g. congestion, cough, runny nose, ear and/or throat pain), clinicians were marginally more likely (48% vs. 43%) to recommend treatment in the diagnostic slot than to diagnose the condition. In such cases, the orientation of the physician is clearly directed toward treatment, predominantly antibiotic medication, as the primary outcome of the visit, with diagnosis playing a less central role. In visits for musculoskeletal conditions, by contrast, discussions of diagnoses, including diagnostic utterances (56%) and assertions of underdetermined diagnostic conclusions (24%) were together significantly more prominent than treatment recommendations (20%) in the diagnostic slot. In these cases, clinicians are generally oriented to the relevance of diagnosis and, often, to the value of diagnostic imaging in pursuing a determinate diagnostic outcome (McArthur 2018). In sum, upper respiratory and musculoskeletal conditions are distinctive in that physicians are generally oriented either to the reduced salience of diagnosis relative to treatment or, alternatively, to its continuing relevance to the clinical project. Design of Diagnostic Utterances in the Diagnostic Slot An important feature of the diagnostic moment is the design of the DU itself. In this section, we examine two main elements of design: (a) the extent to which diagnoses are mitigated or flatly asserted as matters of fact, and (b) the role of gaze in diagnosis delivery. (a) Mitigation In her seminal paper 'Training for uncertainty,' Fox (1957:227-8) noted that student physicians at the time were trained to express their diagnoses and other clinical judgments as definite and certain, and to suppress expressions of doubt or indecision. Subsequent research on diagnosis design, from Byrne and Long (1976) onwards, indicated that a plurality – if not a majority – of diagnoses were presented using a format that we term, following Peräkylä (1998), a 18 ‘plain assertion.’ These assertions simply and authoritatively state the physician’s diagnostic conclusion (e.g. “you have a throat infection”). In the present study, we contrast plain assertions with diagnoses that are mitigated in some way. The mitigation formats we found, which are exemplified in Table 4, include diagnoses delivered using epistemic modality, evidentialization, and epidemiologic generalization. As the examples illustrate, mitigated diagnoses are just that: named medical conditions presented with some element of epistemic distancing. These diagnoses are not part of what we have coded as ‘uncertain diagnostic talk,’ because the latter does not assert a single condition, uncertain or otherwise. Table 4 Examples of diagnostic utterance formats Format type Example from data Plain assertion You have a throat infection. Epistemic Modality You might have cracked one of your ribs over there. Evidentialized It feels like an early plantar fasciitis, you know like a heel spur. Epidemiologic Generalization It’s like the lining is irritated. The most common reason for the lining to be irritated is because of the acid from your stomach. In diagnoses deploying epistemic modality (Palmer 2001), expressions like “could be,” “maybe,” “probably” etc. are used to present a diagnosis as uncertain to a greater or lesser extent. Similarly, mental state verbs like “I think” express a diagnosis as less than certain and as a matter of personal judgment (Kärkkäinen 2003). In evidentialized diagnoses, verbs referencing a sensation, e.g. “feels like” and “looks like,” index the source of evidence upon which the diagnosis is grounded (Cornillie 2009). 19 According to Peräkylä (2006:219), these verbs make “reference to an observational and inferential process, marking the diagnosis as a conclusion that arises from the information that has been made available to the doctor.” However, these diagnoses do not specify the details of this evidence (ibid.) and, unlike diagnoses delivered using epistemic modality, do not make an explicit judgment or evaluation of the likelihood that what the evidence points to is true (de Haan 1999; Peräkylä 2006). This can communicate a lack of commitment to the evidentialized proposition, and hence uncertainty (McArthur under review). Finally, diagnoses that are described epidemiologically do not explicitly tie a diagnosis to a patient’s condition, but rather imply or presuppose a diagnosis by asserting the widespread linkage of the condition to some diagnosis or causal process. Thus, like diagnoses using epistemic modality and evidentialized diagnoses, these utterances convey an element of uncertainty about the diagnosis. Table 5 shows the distribution of diagnostic utterance types in the consultations we analyzed. Table 5 Distribution of diagnostic utterance designs Design Type Frequency Percent Plain Assertions 36 34% Mitigated Assertions* 70 66% Epistemic Modality 41 39% Evidentialized 29 27% Epidemiologic Generalization 11 10% Total 106 *Sub-categories of mitigation are not mutually exclusive 100% 20 As Table 5 shows, in a full two-thirds of cases, diagnoses were delivered ways that communicated some element of uncertainty. (b) Gaze The second element of diagnosis design considered here is physician gaze. When a doctor steps away from a patient’s body and establishes a face-to-face relationship with the patient (Figure 2), this not only marks the transition from the physical examination to the diagnosis, but also establishes an interactional framework in which the patient is treated as the direct recipient of the diagnostic utterance, and response is treated as relevant (Lerner 2003). A large body of Figure 2: Physicians gazing at patients while delivering diagnoses social psychological and conversation analytic studies reaching back over forty years identifies gaze as a significant factor in response (Rossano 2012a). Early on, Argyle et. al (1973) and Kendon (1967) observed that gaze enables a speaker to monitor a recipient’s reaction to a turn at talk. This argument was enlarged by a variety of findings indicating that gaze direction solicits response from a particular gazed-at interlocutor (Sacks, Schegloff & Jefferson 1974; Goodwin 1979; Heath 1986; Lerner 2003). Work by Stivers and Rossano (Stivers et al. 2009; Stivers & Rossano 2010; Rossano 2012b) has gone on to show that speaker gaze also mobilizes recipient 21 response and sequence expansion. These findings strongly suggest that physician gaze at a patient is likely to be an important feature of diagnosis delivery. Thus it is somewhat striking that, in the data, physicians gazed at patients while delivering diagnoses only 70% of the time. In the remaining 30% of cases, physicians were, for example, standing behind a patient, or walking away from the patient, or looking in the patient’s chart, etc. While these behaviors display a ‘reason’ for not looking at the patient, in many cases where physicians do gaze at patients during diagnosis delivery, they interrupt such other engagements to do so. This contrast suggests that these 'no gaze' actions are chosen. In the 'nogaze' diagnosis deliveries, the sense that the diagnosis is directed to the patient and should be responded to is undermined. Patient Response The final feature of the diagnostic moment we explore is patient response. In developing a framework to establish a distribution of response types, we distinguished between no response and (i) nods unaccompanied by verbal response, (ii) minimal verbal acknowledgements like mm hm and yeah, (iii) responses that accept diagnoses as informative, e.g. okay, (iv) items such as oh and newsmarks that treat the diagnosis as news, and (v) more expanded responses such as questions and commentaries. As Table 6 below shows, patients made no verbal response to diagnoses 25% of the time. As previous studies have shown, aspects of the design of a DU, e.g. its assertiveness, may influence the likelihood of patient response. In their study of British primary care, Byrne and Long (1976:106–7) suggested that plain assertions are not in general designed for expansion, via either physician elaboration or patient questioning. Pursuing Byrne and Long’s original findings, Heath (1992) and Peräkylä (1998, 2002, 2006) documented increased patient response when 22 diagnoses are not delivered as plain assertions. In the British primary care context, Heath (1992:247–8) observed that patients were more responsive to diagnoses presented as uncertain, while in the Finnish context, Peräkylä (2006:219) showed that patients’ responses were more expanded when physicians explicitly made reference to the evidence upon which the diagnosis was grounded. In a statistical analysis of the relationship between DU design and patient response, we found that, in a logit model that included physician gaze, diagnoses delivered as plain assertions decreased the likelihood of patient response by 66% when compared with diagnoses delivered using one of the mitigation formats described above (OR=0.34; p=0.048). This finding suggests that, in contrast to treatment recommendations (Stivers et. al 2018), where patient response is highly robust to the contingencies of recommendation delivery, patient responses to diagnoses are comparatively sensitive to features of their design. In addition, as described above, gaze plays a well-documented role in displaying that an utterance is intended for a particular recipient, and is instrumental in soliciting response. Hypothesizing that the absence of physician gaze would be associated with reduced rates of patient response, we found that this was indeed the case. In a logit model that included diagnostic utterance design, diagnoses delivered without physician gaze to the patient decreased the likelihood of patient response by 83% (OR=0.17; p=0.003). Turning to our analysis of the types of acknowledging responses to diagnoses, we distinguished them in the following way: i. Nodding can index patient orientation to diagnostic information and the boundaries of the turns in which it is housed (Stivers 2008; Zama & Robinson 2016). However, nodding performs these tasks outside the verbal communication channel and hence in a fashion that is ‘off-record’ (Brown & Levinson 1987). Ordinarily, nodding is used to 23 enable speaker continuation and, when occurring at the completion of social actions like diagnosis and unaccompanied by verbal acknowledgment of any kind, may thus be regarded as a weak and unsatisfactory form of response (Stivers 2008). ii. Minimal acknowledgments, e.g. mm hm, acknowledge a previous act of speaking, but are frequently agnostic in their stance towards its content. In conversations outside medical visits, they treat the speaker as unfinished, and normally function to invite speaker continuation (Schegloff 1982). Our data suggest that they also perform this function in the medical visit. Minimal acknowledgments imply, rather than display, understanding of the prior turn’s content (Robinson 2014), and do not register the prior turn as ‘news’ (Heritage & Sefi 1992; Gardner 2001). In general, minimal acknowledgements are also a weak form of response to diagnostic information (Gardner 1997), and treat it as incomplete and 'on the way' to something else.. iii. Accepting responses (e.g. I see, Okay) acknowledge a DU by treating it as possibly complete and by accepting its content and its import as an action (Beach 1993; Schegloff 2007). Accepting responses of this sort tend to exhibit a preparedness to shift to a new topic or activity and hence may index the patient’s willingness to proceed from diagnosis to treatment. However, while accepting responses acknowledge diagnostic information, they do not acknowledge it as ‘news’ per se. iv. News acknowledgments (e.g. Oh, Really, Is it?) register previous utterances as news, or at least newly informative (Jefferson 1981; Heritage 1984; Schegloff 2007). However, in the context of medical consultations, news acknowledgments may convey that a diagnosis was unexpected, or even index a surprised or disbelieving stance towards it. Such a stance may implicate resistance to the diagnosis itself, and can function as an initial step towards more substantial resistance (Heath 1992; Stivers 2007). 24 v. Finally, there are expanded and expansive responses, such as assessments (That’s good), commentaries (That explains why I’m limping like this), and questions that invite expansion of the news in some particular direction (Does this take a long time to get better?) (Maynard 2003; Terasaki 2004). These topicalize prior information as an object of further discussion and may in some cases be associated with patient resistance to the diagnosis (Stivers 2007; Ijäs-Kallio, Ruusuvuori & Peräkylä 2010; McArthur 2018). The distribution of patient responses to diagnostic utterances in our data is presented in Table 6. Table 6 Distribution of patient responses to diagnostic utterances Response Type Frequency Percent No Response 16 15% Nod (Only) 12 11% Minimal Response 35 33% Okay/ Accepting Response 19 18% Oh/ Newsmark 10 9% Expanded Response 14 13% Total 106 99% *Percentage totals subject to rounding effects Table 6 shows that, in addition to the 26% of diagnoses that did not receive a verbal response, 33% received just a minimal, 'continuative' verbal response (most frequently mm hm), while 40% received a more 'engaged', accepting, or otherwise more substantive response. 25 DISCUSSION We began this study with a conception of the diagnostic moment comprising two primary elements: the delivery of a diagnostic utterance in the diagnostic slot, and a response by the patient, ideally indicating reception and understanding of the diagnosis. When we consider our overall results, we observe that in cases where a treatment recommendation is offered the diagnostic moment so defined is comparatively infrequent in US primary care data: physicians offered diagnoses in only about 50% of cases. This outcome should be considered in relation to the circumstances of primary care. This is a context where, as previously discussed, it may often be difficult to arrive at determinate diagnoses. Moreover, because many of the ailments presented are minor and self-limiting in character, symptomatic treatment may not require full evaluation of an underlying diagnosis. Indeed, our empirical findings of the infrequency and mitigated nature of diagnoses in this context are remarkably consistent with a small number of studies from physician researchers suggesting that diagnosis may have a different status in primary care than in other emergency or secondary care contexts, and that reaching a determinate diagnosis may not actually be the central goal of general practice, relative to ruling out serious illness and providing symptomatic comfort (Green & Holden 2003; Heneghan et. al 2009). Thus a variety of practical clinical considerations, that reach back to the 1970s in Byrne and Long's (1976) finding that that patients did not receive diagnostic information in about 35% of consultations, may shape the relevance, frequency, and design of diagnosis in primary care. Other evidence points to the conclusion that, from the point of view of providers, the communication of diagnosis may not be a primary consideration in the primary care visit. First, the diagnostic slot is frequently occupied by treatment recommendations. While these recommendations may imply a diagnosis, especially when the treatment involves antibiotics, they nonetheless index a prioritization of treatment relative to diagnosis - a prioritization that 26 may be evident as well in patients' communications in problem presentation and elsewhere (Stivers 2007). Second, as we have seen, the preponderance of diagnostic remarks that occur outside the diagnostic slot are embedded in, or subordinated to, other courses of action. This may, in turn, reduce their import as actions, and downplay the relevance of patient response to the diagnostic information that is conveyed. Third, when physicians deliver diagnoses without gazing at patients, they weaken both the sense that the diagnosis is directed to the patient, and that the patient should respond to it. For these reasons, the absence of gaze can also convey a downgrading of the significance of diagnosis as an action within the medical consultation. Doctors may communicate the reduced significance of diagnosis in yet another way. It will be recalled that two thirds of diagnoses are presented in mitigated or uncertain terms. Considered in a historical context, this represents a significant shift in the culture of medical practice from the days in which physicians were trained to express diagnoses as definite matters of fact (Fox 1957), in ways that helped them to maintain professional authority (Light 1979), manage patient expectations (Waitzkin & Waterman 1974), or supposedly shield patients from the complexities of uncertainty (Katz 1984). These forms of diagnosis delivery may actually embody a more realistic clinical stance toward diagnostic uncertainty, and may also form a more grounded basis for doctor and patient to evaluate next steps. However they can shade into a stance that de-emphasizes the significance of a determinate diagnosis. Suggestions that a treatment recommendation is akin to a field experiment ("Lets try you on X, and see if that helps"), and "as if" statements (e.g., “I will treat you as having a sinus infection” or "I'm gonna treat it as strep") both downplay diagnosis as the precondition for treatment, and communicate a privileging of treatment over diagnosis as the primary goal of the encounter (Stivers et al. 2018). Finally, in contexts where patients made minimal responses to diagnoses, physicians frequently proceeded to treatment recommendations rather than expanding on the diagnosis itself 27 in pursuit of more 'engaged' responses indicating comprehension. These sequences suggest that both parties were treating the diagnostic moment as a ritualized rite of passage on the way to treatment, at least from an interactional perspective. Looking at the communicated significance of diagnosis from the patients’ perspective, our data show that the majority of patients’ responses to diagnoses tend to be minimal. Indeed only 40% of responses conveyed any kind of substantive stance. This observation is the common finding of a number of studies of diagnosis during the past thirty years (Byrne & Long 1976; Heath 1992; Peräkylä 1998, 2002, 2006). It is at least arguable that these predominantly minimal responses are grounded in the epistemics of the doctor-patient relationship. Diagnosis arguably represents a primary locus for the expression of epistemic authority in medicine. Grounded in the physician’s expert knowledge and experience in its implementation, and flanked by the apparatus of professional accreditation that remains the foundation of the 'sovereign profession of medicine' (Starr 1982), the diagnostic moment is a summit in the expression of medical authority during the primary care visit, and the foundation of the deontic authority exerted in the treatment recommendation (Stevanovic & Peräkylä 2012; Lindström & Weatherall 2015; Stivers et al. 2018). If, as Starr (1982:12) remarks, drawing on Lukes (1978), medical authority involves the patient in a ‘surrender of private judgment,’ the weight of epistemic authority in the moment of diagnosis may tend to compel at least acquiescence, if not actual surrender. In a context of radical epistemic imbalance between doctor and patient, any response that goes beyond acquiescence – for example, Oh really? Is it? Are you sure? – may risk being understood as an expression of surprise, doubt, or resistance (Stivers 2007) and avoided accordingly. Thus, in a situation of generally minor and self-limiting complaints, patients may find themselves in an ‘epistemic vise,’ in which anything more than bland acknowledgment may be understood as questioning the diagnosis, and hence as a challenge to medical authority. As a 28 result of this vise, outside of more serious diagnoses, patients may be less inclined to ask questions or otherwise pursue as-yet-unspoken diagnostic intricacies. Thus, regardless of their actual interest in diagnostic conclusions – an interest that may be expressed in questioning of the diagnosis later in the visit, and in pressure for medical tests (McArthur in prep; Kassirer 1989; Gallagher et al 1997; O'Sullivan et al 2018) – patients’ minimal responses to diagnostic utterances communicate a de-prioritization of diagnosis paralleling the stance conveyed by physicians, in what is potentially a mutually reinforcing process. Diagnoses, of course, are also a way station in the progression of visits towards treatment recommendations, and may be understood for the most part in terms of their treatment implications. Writing in 1970, Eliot Freidson emphasized the valorization of treatment as the superordinate goal of medical care, observing that "the work of the doctor is first of all concretely directed to the solution of a practical problem" (Freidson 1970:13), and adding that "the patient's request is "Doctor, do something," not "Doctor, tell me if this is true or not" (ibid:22, quoted in Robinson 2003:31). While patients throughout the ages have always come to doctors in search of solutions, it is possible that the growing availability of effective treatments has led to a generally reduced focus on diagnoses by patients and doctors alike (Shorter 1986). This trend may be particularly marked in societies like the United States characterized by a relatively intense focus on treatment efficacy (Bergen et. al 2018). It is perhaps significant in this context that, as previously noted, treatment recommendations commonly occur in the place where diagnosis would ordinarily be delivered. By these actions, physicians themselves directly convey a prioritization of treatment over diagnosis, and an orientation to the patient as having similar priorities. As noted earlier, a majority of treatment recommendations occurring in the diagnostic slot were for antibiotics – a treatment preference that is the object of extensive patient signaling from the very beginning of 29 the visit onwards, and that has been shown to influence physician perceptions and treatment outcomes (Mangione-Smith et al. 1999, 2004, 2006, 2015; Stivers 2002, 2007; Stivers et al. 2003). In cases of this kind, physician prioritization of treatment recommendations over diagnoses may be a realistic response to reliable patient signals. CONCLUSION This is the first study that, to our knowledge, attempts a systematic inventory of the practice of diagnosis in US primary care. We have shown that, regardless of their private and professional concerns, both physicians and their patients exhibit behaviors that interactionally de-valorize the significance of diagnosis in primary care visits. The upshot is that diagnosis is markedly less frequent and specific in primary care than a perspective adopted from the study of secondary care might have anticipated. We have argued that, heretofore, sociologists of diagnosis have examined the formation of diagnostic categories and the social consequences of their implementation, but have devoted less attention to the empirical realization of Jutel’s (2014) ‘diagnostic moment’ in everyday medical visits. In this paper, we have contributed a preliminary overview of the nature and distribution of diagnostic moments in data drawn from the context of US primary care. Our study is strongly compatible with Heneghan et. al’s (2009) study of diagnostic reasoning in UK primary care, which also points to the absence or inconclusiveness of many diagnostic outcomes in that context. Both studies propose a disparity between the importance that the medical profession attaches to diagnosis in principle, and the practical implementation of diagnosis ‘on the ground,’ in everyday primary care interactions. Our findings suggest that there is much to be learned both from the distribution of diagnostic moments, and from an analysis of the specific details of their realization in the actions of doctors and their patients. Indeed our study, and the 30 small sequence of conversation analytic studies on diagnosis we have discussed, represent only the beginning of this undertaking. Expansion of this research focus is perhaps overdue. 31 Compliance with Ethical Standards Conflict of Interest JH and AM declare that they have no conflicts of interest. Human Subjects and Informed Consent Informed consent was obtained from all subjects in the study. Original data collection for this study was funded by the Agency for Healthcare Research and Quality (Grant No: 5R01HS010922-03) and the study protocol was approved by the UCLA Human Subjects Protection Committee, which also approved the re-use of the recorded data for the present study. Acknowledgments: An earlier version of this paper was presented to the Conversation Analysis Working Group at UCLA. We thank the members of that group and, in particular, Tanya Stivers and Steve Clayman for their comments on that occasion. We also thank three anonymous reviewers for this journal for very valuable feedback and suggestions. Finally, we thank Saskia Maltz, Lisa Moreno, Melissa Tay, and Elin Wrammerfors for their outstanding research assistance on this project. 32 References Abbott, A. (1988). The system of professions: An essay on the division of expert labor. Chicago, IL: The University of Chicago Press. Argyle, M., Ingham, R., Alkema, F., & McCallin, M. (1973). The different functions of gaze. Semiotica, 7(1), 19–32. Armstrong, D. (2011). Diagnosis and nosology in primary care. Social Science & Medicine, 73, 801– 807. Armstrong, N., & Hilton, P. (2014). Doing diagnosis: Whether and how clinicians use a diagnostic tool of uncertain clinical utility. Social Science & Medicine, 120, 208–214. Beach, W. A. (1993). Transitional regularities for 'casual' "Okay" usages. Journal of Pragmatics, 19, 325–352. Bergen, C., Stivers, T., Barnes, R.K., Heritage, J., McCabe, R., Thompson, L., et al. (2018). Closing the deal: A cross-cultural comparison of treatment resistance. Health Communication, 33, 13771388. Blaxter, M. (1978). Diagnosis as category and process: The case of alcoholism. Social Science & Medicine, 12, 9–17. Brown, P. (1990). The name game: Toward a sociology of diagnosis. The Journal of Mind and Behaviour, 11(3–4), 385–406. Brown, P., & Levinson, S. C. (1987). Politeness: Some universals in language usage. Cambridge: Cambridge University Press. Byrne, P. S., & Long, B. E. L. (1976). Doctors talking to patients: A study of the verbal behaviours of doctors in the consultation. London, England: Her Majesty's Stationery Office. Conrad, P. (1975). The discovery of hyperkinesis: Notes on the medicalization of deviant behavior. Social Problems, 23. Cornillie, B. (2009). Evidentiality and epistemic modality: On the close relationship between two different categories. Functions of Language, 16, 44–62. Cox, K. (in prep). Contingency planning: Negotiating the window of recovery. Unpublished Ms, UCLA. de Haan, F. (1999). Evidentiality and epistemic modality: Setting boundaries. Southwest Journal of Linguistics, 18, 83–101. Ford, C.E., & Thompson, S.A. (1996). Interactional units in conversation: syntactic, intonational and pragmatic resources for the management of turns. In E. Ochs, E.A. Schegloff, & S.A. Thompson (Eds.), Interaction and Grammar (pp. 134-184). Cambridge: Cambridge University Press. Fox, R.C. (1957). Training for uncertainty. In R.K. Merton, G.G. Reader, & P.L. Kendall (Eds.), The Student-Physician (pp.207-241). Cambridge MA: Harvard University Press. Freidson, E. (1970). Profession of medicine: A study of the sociology of applied knowledge. New York, NY: Atherton. Gallagher, T.H., Lo, B., Chesney, M., & Christensen, K. (1997). How do physicians respond to patients’ requests for costly, unindicated services? Journal of general internal medicine, 12, 663-668. 33 Gardner, J., Dew, K., Stubbe, M., Dowell, T., & Macdonald, L. (2011). Patchwork diagnoses: The production of coherence, uncertainty, and manageable bodies. Social Science & Medicine, 73, 843–850. Gardner, R. (1997). The conversation object Mm: A weak and variable acknowledging token. Research on Language and Social Interaction, 30(2), 131–156. Gardner, R. (2001). When listeners talk: Response tokens and listener stance. Amsterdam: Benjamins. Glaser, B.G., & Strauss, A.L. (1967). The Discovery of Grounded Theory: Strategies for Qualitative Research. Chicago: Aldine. Goodwin, C. (1979). The interactive construction of a sentence in natural conversation. In G. Psathas (Ed.), Everyday language: Studies in ethnomethodology (pp. 97–121). New York, NY: Irvington Publishers. Green, C., & Holden, J. (2003). Diagnostic uncertainty in general practice. Eur J Gen Pract, 9(1), 13– 15. Heath, C. (1986). Body movement and speech in medical interaction. Cambridge: Cambridge University Press. Heath, C. (1992). The delivery and reception of diagnosis in the general practice consultation. In P. Drew & J. Heritage (Eds.), Talk at work. Cambridge: Cambridge University Press. Heneghan, C., Glasziou, P., Thompson, M., Rose, P., Balla, J., Lasserson, D., . . . Perera, R. (2009). Diagnosis in general practice: Diagnostic strategies used in primary care. British Medical Journal, 338(7701), 1003–1006. Heritage, J. (1984). A change-of-state token and aspects of its sequential placement. In J. M. Atkinson & J. Heritage (Eds.), Structures of social action: Studies in conversation analysis (pp. 299–345). Cambridge: Cambridge University Press. Heritage, J. (2011). The interaction order and clinical practice: Some observations on dysfunctions and action steps. Patient Education and Counseling, 84, 338–343. Heritage, J., & Clayman, S. E. (2010). Talk in action: Interactions, identities and institutions. Malden, MA: Wiley Blackwell. Heritage, J., & Maynard, D. M. (Eds.). (2006a). Communication in medical care: Interaction between primary care physicians and patients. Cambridge: Cambridge University Press. Heritage, J., & Maynard, D. M. (2006b). Problems and prospects in the study of physician-patient interaction: 30 years of research. Annual Review of Sociology, 32, 351–374. Heritage, J., & Sefi, S. (1992). Dilemmas of advice: Aspects of the delivery and reception of advice in interactions between health visitors and first time mothers. In P. Drew & J. Heritage (Eds.), Talk at Work (pp. 359–417). Cambridge: Cambridge University Press. Ijäs-Kallio, T., Ruusuvuori, J., & Peräkylä, A. (2010). Patient resistance towards diagnosis in primary care: Implications for concordance. Health, 14(5), 505–522. Institute of Medicine (2015). Improving diagnosis in health care. Washington, DC: National Academies Press. 34 Jefferson, G. (1981). The abominable "Ne?": A working paper exploring the phenomenon of postresponse pursuit of response. . Paper presented at the Occasional Paper No. 6, University of Manchester, Department of Sociology. Jefferson, G., & Lee, J. R. E. (1981). The rejection of advice: Managing the problematic convergence of a 'troubles-telling' and a 'service encounter'. Journal of Pragmatics, 5, 399–422. Jutel, A. (2014). When the penny drops: Diagnosis and the transformative moment. In A. Jutel & K. Dew (Eds.), Social issues in diagnosis: An introduction for students and clinicians (pp. 78–92). Baltimore, MD: Johns Hopkins University Press. Jutel, A., Nettleton, S., & (Eds.). (2011). Sociology of Diagnosis. Social Science & Medicine, 73(6). Kärkkäinen, E. (2003). Epistemic stance in English conversation. Amsterdam: John Benjamins. Kassirer JP. Our stubborn quest for diagnostic certainty: A cause of excessive testing. N Engl J Med. 1989;320:1489–91. Katz, J. (1984). The silent world of doctor and patient. Baltimore, MD: The Johns Hopkins University Press. Kendon, A. (1967). Some functions of gaze-direction in social interaction. Acta Psychologica, 26, 22– 63. Koenig, C. J. (2011). Patient resistance as agency in treatment decisions. Social Science & Medicine, 72, 1105–1114. Lerner, G. H. (2003). Selecting next speaker: The context-sensitive operation of a context-free organization. Language in Society, 32, 177–201. Light, D. W. (1979). Uncertainty and control in professional training. Journal of Health and Social Behavior, 20(4), 310–322. Lindström, A., & Weatherall, A. (2015). Orientations to epistemics and deontics in treatment discussions. Journal of Pragmatics, 78, 39–53. Lukes, S. (1978). Power and authority. In T. Bottomore, & R.A. Nisbet (Eds.), A history of sociological analysis (pp. 633-676). New York: Basic Books. Mangione-Smith, R., Elliott, M. N., Stivers, T., McDonald, L. L., & Heritage, J. (2006). Ruling out the need for antibiotics: Are we sending the right message? Archives of Pediatric and Adolescent Medicine, 160, 945–952. Mangione-Smith, R., Elliott, M. N., Stivers, T., McDonald, L. L., Heritage, J., & McGlynn, E. A. (2004). Racial/ethnic variation in parent expectations for antibiotics: Implications for public health campaigns. Pediatrics, 113(5), e385–3394. Mangione-Smith, R., McGlynn, E., Elliott, M. N., Krogstad, P., & Brook, R. H. (1999). The relationship between perceived parental expectations and pediatrician antimicrobial prescribing behavior. Pediatrics, 103, 711–718. Mangione-Smith, R., Zhou, C., Robinson, J. D., Taylor, J. A., Elliott, M. N., & Heritage, J. (2015). Communication practices and antibiotic use for acute respiratory tract infections in children. Annals of Family Medicine, 13, 221–227. 35 Maynard, D. M. (1991). The perspective display series and the delivery and receipt of diagnostic news. In D. Boden & D. H. Zimmerman (Eds.), Talk and social structure (pp. 164–192). Berkeley: University of California Press. Maynard, D. M. (1992). On clinicians co-implicating recipients' perspective in the delivery of diagnostic news. In P. Drew & J. Heritage (Eds.), Talk at work (pp. 331–359). New York: Cambridge University Press. Maynard, D. M. (2003). Bad news, good news: Conversational order in everyday talk and clinical settings. Chicago, IL: University of Chicago Press. Maynard, D. M. (2004). On predicating a diagnosis as an attribute of a person. Discourse Studies, 6(1), 53–76. Maynard, D. M., & Frankel, R. R. (2006). On diagnostic rationality: Bad news, good news, and the symptom residue. In J. Heritage & D. M. Maynard (Eds.), Communication in medical care: interaction in primary care physicians and patients. Cambridge: Cambridge University Press. McArthur, A. (2018). Diagnostic indeterminacy about musculoskeletal problems in primary care. Paper presented at the International Conference on Conversation Analysis, Loughborough, England. McArthur, A. (under review). Communicating uncertain diagnoses in primary care: The impact of wording on patient confidence. Patient Education and Counseling. McArthur, A. (in prep). Diagnosis in primary care. PhD dissertation in progress, Department of Sociology, UCLA. McGann, P., & Hutson, D. J. (Eds.). (2011). Sociology of diagnosis. Bingley, UK: Emerald Group Publishing Limited. Nettleton, S., Kitzinger, J., & Kitzinger, C. (2014). A diagnostic illusory? The case of distinguishing between 'vegetative' and 'minimally conscious' states. Social Science & Medicine, 116, 134–141. O’Sullivan, J.W., Albasri, A., Nicholson, B.D., Perera, R., Aronson, J.K., Roberts, N., et al. (2018). Overtesting and undertesting in primary care: a systematic review and meta-analysis. BMJ open, 8, e018557. Palmer, F. R. (2001). Mood and modality (2nd ed.). Cambridge: Cambridge University Press. Parsons, T. (1951). The social system. London, UK: Routledge. Peräkylä, A. (1998). Authority and accountability: The delivery of diagnosis in primary health care. Social Psychology Quarterly, 61(4), 301–320. Peräkylä, A. (2002). Agency and authority: Extended responses to diagnostic statements in primary care encounters. Research on Language and Social Interaction, 35(2), 219–247. Peräkylä, A. (2006). Communicating and responding to diagnosis. In J. Heritage & D. M. Maynard (Eds.), Communication in medical care (pp. 214–247). Cambridge: Cambridge University Press. Robinson, J. D. (2003). An interactional structure of medical activities during acute visits and its implications for patients' participation. Health Communication, 15(1), 27–59. Robinson, J. D. (2014). What 'what?' tells us about how conversationalists manage intersubjectivity. Research on Language and Social Interaction, 47(2), 109–129. 36 Robinson, J. D., & Heritage, J. (2005). The structure of patients' presenting concerns: The completion relevance of current symptoms. Social Science & Medicine, 61, 481–493. Robinson, J. D., & Heritage, J. (2006). Physicians' opening questions and patients' satisfaction. Patient Education and Counseling, 60, 279–285. Robinson, J. D., & Stivers, T. (2001). Achieving activity transitions in physician-patient encounters: From history taking to physical examination. Human Communication Research, 27(2), 253–298. Rossano, F. (2012a). Gaze in conversation. In J. Sidnell & T. Stivers (Eds.), The Handbook of Conversation Analysis (pp. 308–329). Malden, MA: Wiley-Blackwell. Rossano, F. (2012b). Gaze behaviour in face-to-face interaction. (Doctoral Dissertation), Radboud University Nijmegen, Nijmegen, The Netherlands. Roter, D. L., & Hall, J. A. (2006). Doctors talking with patients/ Patients talking with doctors: Improving communication in medical visits. Westport Conn: Praeger. Sacks, H. (1992). Lectures on conversation: Volumes I & II. Malden, Ma: Blackwell Publishing. Sacks, H., Schegloff, E. A., & Jefferson, G. (1974). A simplest systematics for the organization of turntaking for conversation. Language, 50(4), 696–735. Schaepe, K. S. (2011). Bad news and first impressions: Patient and family caregiver accounts of learning the cancer diagnosis. Social Science & Medicine, 73, 912–921. Schegloff, E. A. (1982). Discourse as an interactional achievement: Some uses of 'uh huh' and other things that come between sentences. In D. Tannen (Ed.), Analyzing Discourse: Text and Talk (pp. 71–93). Washington, DC: Georgetown University Press. Schegloff, E. A. (2007). Sequence organization in interaction: A primer in conversation analysis. Cambridge: Cambridge University Press. Shorter, E. (1985). Bedside manners: The troubled history of doctors and patients. New York, NY: Simon and Schuster. Silverston, P. (2016). Teaching patient-centered safety-netting in primary care. Journal of Community Medicine & Health Education, 6(3). Starr, P. (1982). The social transformation of American medicine. New York, NY: Basic Books. Stevanovic, M., & Peräkylä, A. (2012). Deontic authority in interaction: The right to announce, propose, and decide. Research on Language and Social Interaction, 45(3), 297–321. Stivers, T. (2002). Presenting the problem in pediatric encounters: 'Symptoms only' versus 'candidate diagnosis' presentations. Health Communication, 14(3), 299–338. Stivers, T. (2007). Prescribing under pressure: Physician-parent conversations and antibiotics. Oxford: Oxford University Press. Stivers, T. (2008). Stance, alignment and affiliation during story telling: When nodding is a token of affiliation. Research on Language and Social Interaction, 41, 31–57. Stivers, T. (2015). Coding social interaction: A heretical approach in conversation analysis? Research on Language and Social Interaction, 48(1), 119. 37 Stivers, T., Enfield, N.J., Brown, P., Englert, C., Hayashi, M., Heinemann, T., et al. (2009). Universals and cultural variation in turn-taking in conversation. Proceedings of the National Academy of Sciences, 106, 10587-10592. Stivers, T., Heritage, J., Barnes, R.K., McCabe, R., Thompson, L., & Toerien, M. (2018). Treatment recommendations as actions. Health Communication, 33, 1335-1344. Stivers, T., Mangione-Smith, R., Elliott, M. N., McDonald, L. L., & Heritage, J. (2003). Why do physicians think parents expect antibiotics? What parents report vs what physicians perceive. The Journal of Family Practice, 52(2), 140–148. Stivers, T., & Rossano, F. (2010). Mobilizing response. Research on Language and Social Interaction, 43(1), 3–31. Stivers, T., & Timmermans, S. (2016). Negotiating the diagnostic uncertainty of genomic test results. Social Psychology Quarterly, 79(3), 199–221. Stivers, T., & Timmermans, S. (2017). Always Look on the Bright Side of Life: Making Bad News Bivalent. Research on Language and Social Interaction, 50(4), 404-418. Tarn, D. M., Heritage, J., Paterniti, D. A., Hays, R. D., Kravitz, R. L., & Wenger, N. S. (2006). Physician communication when prescribing new medications. Archives of Internal Medicine, 166, 1855–1862. Terasaki, A. K. (2004). Pre-announcement sequences in conversation. In G. H. Lerner (Ed.), Conversation analysis: Studies from the first generation (pp. 171–223). Amsterdam: John Benjamins. Trundle, C. (2011). Biopolitical endpoints: Diagnosing a deserving British nuclear test veteran. Social Science & Medicine, 73, 882–888. Waitzkin, H., & Waterman, B. (1974). The exploitation of illness in capitalist society. Indianapolis: Bobbs-Merrill. Whalen, J., Zimmerman, D. H., & Whalen, M. R. (1988). When words fail: A single case analysis. Social Problems, 35(4), 335–362. Zama, A., & Robinson, J.D. (2016). A relevance rule organizing responsive behavior during one type of institutional extended telling. Research on Language and Social Interaction, 49, 220-237.