When is a shared decision not (quite) a shared decision?Negotiating preferences in a general practice encounter

Richard Gwyna,*, Glyn Elwynb

aHealth Communication Research Centre, School of English, Communication and Philosophy, P.O. Box 94, Cardi� University, Cardi�

CF1 3XB, UKbDepartment of Postgraduate Education for General Practice, University of Wales College of Medicine, Cardi� CF4 4XN, UK

Abstract

We consider whether there are situations in which `shared decision making' in primary care is inherently

problematic, such as in the demand for antibiotics to treat viral disorders. In such an instance there might be a lackof the equipoise necessary for a decision-making context in which apparent choices are genuine options. Using thetechniques of discourse analysis on the transcript of a consultation with the parents of an infant with tonsillitis, we

illustrate how a general practitioner's (GP's) e�orts to reach a `shared decision' come unstuck through acombination of the embedded power imbalance and the con¯ict between the GP's own prescription preferences andthose of the parent. # 1999 Elsevier Science Ltd. All rights reserved.

Keywords: Shared decision-making; Primary care; Treatment preferences

Introduction

Shared decision-making is receiving increasing atten-tion in professional circles, particularly within medi-cine, bastion for so long of the `doctor knows best'style of consulting. Consumerism (Lupton, 1997),

fuelled by an exponential increase in public access toon-line information, is changing clinical practice (Laineand Davido�, 1996). There is a growing awareness

that participation in decisions results in greater clientsatisfaction and improved outcomes, at least asmeasured by decision acceptance and adherence to

agreed action plan (Elwyn, in press). However, `shared

decisions' are not made in a vacuum, and it might besuggested that a situation of equipoise, that is, one in

which options really are options, must exist in orderfor a shared decision to successfully take place andthereby justify the term. When there is not equipoise,for example, in the demand for antibiotics to treat

viral disorders, problems may arise for both doctorand patient, rendering the successful negotiation of agenuinely shared decision di�cult. In such a case

`shared decision-making' might be a misnomer(Charles et al., 1997) and although a `shared decision'is reached, it would be more accurately described as an

informed decision engineered according to doctor pre-ference. In this paper we would like to examine indetail a case study in which such a decision is reachedwith the parents of a small boy su�ering from tonsilli-

tis.

Social Science & Medicine 49 (1999) 437±447

0277-9536/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.

PII: S0277-9536(99 )00067-2

www.elsevier.com/locate/socscimed

* Corresponding author. Fax: +44-1222-874-242.

E-mail address: gwyn@cf.ac.uk (R. Gwyn)

Decision-making

Decisions are made in one of three ways. You can

decide for yourself, weighing the options, as an auton-omous individual. You can be told, or advised, what'sbest to do, guided, one hopes, by superior wisdom, ex-

perience or expertise. Or, two or more individuals,considering the risks and bene®ts of the availableoptions, can share decisions. However, this simpli®ed

decision-making outline does not do justice to thecomplexities of arriving at judgements. Do we reallymake decisions alone? Do we make use of rational in-formation or intuition? Often the information available

is biased and/or incomplete. Cognitive psychologistsnote that we simplify complex information, and use`heuristic' methods (rules of thumb) to arrive at de-

cisions. The human brain is poor at integrating prob-abilities mathematically; it makes large errors whenmore than two or three probabilities have to be con-

sidered simultaneously and so (using probabilistic cri-teria) makes `bad' decisions.How do these issues impinge on medical practice

and primary care in particular? There is little doubt

that the encouragement of `patient choice' during thesame period as the emphasis on evidence-based prac-tice (Hope, 1996; Muir Gray, 1996) has concentrated

attention on e�ective decision-making in healthcare(Coulter, 1997) and how this can be achieved within(or perhaps outside) consultations with clinicians.

Charles et al. (1997) have reviewed the models of de-cision-making and ®nd that the most prominent arethe paternalistic model, the informed choice model and

the shared decision-making model.The paternalistic model is derived from Parsons'

(1951) conceptualisation of the sick role. This role(granted by the physician) excuses patients from nor-

mal duties but also obliges them to try to get well, toseek expert help and to comply with the medical regi-men. It's a consulting style where the physician does

what he thinks is best for the patient, without eliciting

the latter's preferences. Clearly this model does not fa-cilitate `shared decisions'.The informed choice model arises out of the recog-

nition that there is an information imbalance betweenthe patient and the doctor: ``technical knowledge residesin one party, the physician, while preferences reside in

the other, the patient'' (Charles et al., 1997, p. 683). Theinformed choice decision-making model describes a pro-cess whereby patients receive (usually from physicians)

information about the choices they have to make. Intheory, decisions need not be `shared' as the patient nowhas both components (information and preferences)necessary to reach a decision. In fact, it has been noted

that the physician is proscribed from giving a treatmentrecommendation for fear of imposing his or her will onthe patient and thereby competing for the decision-mak-

ing control that has been given to the patient (Emanueland Emanuel, 1992). Concern has been raised that the`informed choice' model, where control over decision-

making is vested entirely with the patient, may lead toincreased anxiety; a predicament termed by Quill aspatient `abandonment' (Quill and Cassel, 1995; Quilland Brody, 1996). Charles et al. (1997, p. 683) point out

that none of the models ``explicitly describe a process inwhich both physicians and patients share in decision-making, no matter how much information they share''.

As Ong et al. (1995) have noted, sharing informationand sharing decisions are not synonymous. They areseparate goals within the consultation.

What then are the characteristics of shared decision-making? They are summarised in Fig. 1.Contained within each step are areas that can be

expanded and discussed. However, for the purpose ofthis paper, for shared treatment decision-making tooccur, there needs to be a two-way exchange, not onlyof information but also of treatment preferences. This

process cannot, by de®nition, occur within a doctor-centred consultation and the process of shared de-cision-making is integral to the wider concept of

patient-centred consulting (Stewart et al., 1995).

Fig. 1. Features of shared decision making.

R. Gwyn, G. Elwyn / Social Science & Medicine 49 (1999) 437±447438

Power asymmetry and consensus

Discourse and conversational analytic studies of theclinical encounter have consistently emphasised theasymmetry of power in the consultation (e.g. Taussig,

1980; Paget, 1983; Mishler, 1984; Treichler et al., 1984;Silverman, 1987; ten Have, 1991; Heath, 1992;Maynard, 1992; Coupland et al., 1994; Fisher, 1995).

The issue of power asymmetry in the medical consul-tation is not under dispute, and is, in any case welldocumented, along with concomitant interactional pro-

cesses of coercion and exclusion (Atkinson, 1995).What is important, we would argue, is that strugglesover power asymmetry, which are in any case inescap-able, do not obscure for us the positive interactional

work that can take place within consultations to helpengender shared decision-making. Opposition to thisprocess of shared decision-making, o�ered both by the

embedded dynamic of power asymmetry and bypatient preference con¯icting with that of the doctor'sprescribing choice, forms the basis for this paper.

The most astute commentary on the embedded qual-ity of power within clinical interaction is that given bySilverman (1987), who argues that despite their pleas

for humanism and equality, the proponents of thepatient-centred approach are unwittingly reinforcing acentral strategy of power in the doctor±patient re-lationship. The e�ects of power are most e�cacious

when least visible, i.e. when people themselves de®neand organise their behaviour according to normativerules and codes. These rules and codes are invariably

the ones held by professionally-de®ned bodies ofknowledge. To argue against asymmetry in the doctor±patient relationship therefore misses the point since

any attempted adoption of equality would only be asimulation and would leave intact the essential natureof the power imbalance, which is based upon special-ised knowledge. Silverman, pursuing this line of argu-

ment, claims that the movement for `patient power',which began to assert itself in the 1970s, to a largeextent re¯ects a burgeoning consumerism. However

understandable such opposition to established clinicalprocedure might be, by focusing on notions of asym-metry, Silverman contends, the `patient power' lobby

misses the point for at least two reasons.Firstly, professional expertise predetermines an

imbalance in the doctor±patient relationship. The prin-

ciple that a doctor should disclose relevant informationto a patient in order to help that patient come to anintelligent understanding of his or her course of treat-ment is itself asymmetrical. We can add to this the

axiom, driven home by the `evidence based medicine'movement (Sackett et al., 1997), that there are rela-tively few clinical treatments which have the bene®t of

`evidence', but yet no shortage of `expert' views. Theconsultation is therefore forever threatened with

becoming the domain of `other experts' in which anyprogress to be made in terms of interactional symmetry

will have to begin again from scratch.Secondly, Silverman maintains, many patients want

to keep the asymmetry of the relationship and would

feel uncomfortable without it. They want the doctor to`know best' and are likely to resent any shifts in de-cision-making onto themselves. Again, we might add

to Silverman's argument by suggesting that perhaps,too, patients accept the discourse strategies that domi-nate doctor±patient interviews because they are seen as

an implicit part of `the treatment' and that by ques-tioning the authoritative voice of medicine (by speak-ing out of turn, interrupting, initiating topics, etc.)they might be seen to be symbolically challenging the

status quo of medical knowledge, thereby causing cov-ert damage to their chances of recovery. Studies whichhave investigated patient preferences for involvement

in decision-making have found wide variation betweendi�erent groups as distinguished by age, educationalstatus, disease type and severity. As a general guide,

the older the patients and the more severe their ill-nesses, the less they wanted to be involved in decision-making (cf. Coupland et al., 1991).

If there is real consensus between patient and doctoror parents of a patient and a doctor during the second,decision-taking part of the consultation, then pro-fessional dominance is surely not contentious. Merely

employing criteria of patient-centredness is not su�-cient in analysis of a medical interview in which thedominance of the professional is maintained by con-

sensus. Maynard (1991) too has argued that asymme-try in the consultation is `interactively achieved'. Thegreat insight of this argument, is that it makes use of a

consensual notion of power in relation to the doctor±patient encounter, and thereby undermines the simplis-tic dichotomy of ``powerful doctor versus powerlesspatient''. Heath (1992) too, argues convincingly that

interactional asymmetry is achieved through the talkof both patient and doctor rather than simply imposedfrom above by the doctor.

Treatment preferences and the `charm' of medicines

As we have proposed, shared decision-making is aninteresting blend of the humanistic medical philosophy

of patient-centredness and a newly resurgent and ever-growing consumerism. There are other forces at workas well. It is as if doctors have accepted that the `evi-dence' for the bene®t of their treatments is, in many

circumstances, slim or non-existent and that they have,therefore, an obligation to help patients arrive at more`informed' decisions. Add to that the need to avoid liti-

gation and an awareness that patients search electroni-cally for data in previously inaccessible clinical

R. Gwyn, G. Elwyn / Social Science & Medicine 49 (1999) 437±447 439

domains and the scene is set for an increased opennessabout treatment choices. So the situation is not asclearly demarcated as it once was between the forces

of `good' (the caring sharing healer) and `evil' (thepower-crazed medic). Patients, it is argued, have chan-ged in their expectations of the medical consultation,

as well as in what they bring to the surgery in terms ofpreconceptions and received information from a widen-ing range of sources.Patients may have views at odds with those of clini-

cians attempting to employ a shared decision strategyand will call on their own `evidence', the recall of priorexperiences, anecdotes received from relatives and

friends, the stream of often contradictory informationfrom newspapers and television, and are often wary ofa decision to withhold therapy lest it be in¯uenced by

an eye on cost. This plurality of value systems is, itwould seem, an inescapable characteristic of late mod-ernity. The individual `subject' is constructed as an ac-cumulation of responses to a continual bombardment

of `facts', whether `lay' or `scienti®c', `valid' or `unreli-able'. How then do we achieve shared decisions inhealthcare which neither defy clinical logic, waste

resources, or disregard our patients' ideas, concernsand expectations?Struggling (as most people must) through a bewil-

dering and constant display of (mis)information, tele-vised images and health scares posing as news items,the emphatically consumable solution of a prescribed

medicine often appears to have almost talismanicpowers. It has been argued that the very concreteness,the tangibility of a prescribed treatment such as anti-biotics is perceived by patients as an essential antidote

to the physicality or concreteness of the illness itself.Van der Geest and Whyte (1989, p. 345) refer to the`charm' of medicines, the key to their charm being in

their concreteness: in them healing is objecti®ed. Thus,``medicines are commodities which pass from one con-text of meaning to another. As substances they are

`good to think with'''. Essentially then, ``pharmaceuti-cals predicate a graspable world of healing for the suf-ferer, giving the imagined `itness' of the disease the

countering `itness' of the medicine and vice versa''. Inthis respect, the medicine itself might be perceived as ametaphor for `getting well' and what is being requested

when a patient insists on a speci®c concrete therapy isthis substance, or commodity, which is `good to thinkwith'.

An example from clinical practice

Perhaps the dilemma can be illustrated by means ofthe following example, taken from clinical practice. Init, a general practitioner who regularly employs shared

decision-making strategies is visited by a couple andtheir small son, Ali. It is the ®rst meeting between thedoctor and this family. The child is su�ering from ahigh temperature and vomiting. The father, a non-

native speaker, does the talking. Ali's mother makesoccasional background comments, and only partici-pates in the interaction during the episode where the

doctor attempts to `share' the decision. However,within that brief interaction she seems to have a deci-sive in¯uence on the ®nal decision taken. She is also

vocal in the ritual (phatic) language of departure, the`sealing o�' of the interview, and, in the context of thisparticular consultation, this is signi®cant.

The example given here does not pretend to resolvethe issue of shared decision-making in a context whereequipoise does not obtain, nor can the di�erent cul-tural backgrounds of doctor and patient, typical of an

inner-city practice such as this, do anything butemphasise the issue of doctor±patient asymmetry.However, as a ®rst study in a series of shared decision-

making scenarios which will, among other things, serveto illustrate the problematics of `equipoise' (e.g. Elwyn,Gwyn et al., in press), it will present facets of how a

Fig. 2. Key to transcript symbols.

R. Gwyn, G. Elwyn / Social Science & Medicine 49 (1999) 437±447440

doctor's decision-sharing strategies can be disruptedboth by the inherent power imbalance of the consul-

tation and by the doctor's own preferences for treat-ment.Details of transcription symbols are described in

Fig. 2.Turning to the consultation itself, after a very open

invitation from the clinician (line 1), the father (F)

begins with an outline of his son's condition, (lines 2±9). D: Doctor, F: Ali's father. M: Ali's mother.

001 D: okay how can I help?002 F: yeah (.) he has a high fever (.)003 since eight o'clock yesterday night004 and he vomited yesterday night

005 and in the morning again (.)006 he:: doesn't eat anything (.)007 refuses everything (.)

008 and (.) he's weak (.)009 since (.) two days

This information is immediately re-processed by thedoctor, item by item (lines 12±26). Although this kindof repetition is a common-sense technique for con®rm-

ing information which has all been given already (tem-perature, rather than the unspeci®c and dramatic `highfever', vomiting, not eating). However this is also anexchange between native and non-native speakers, and

in line 29, D's apparent mistrust of F's capacity tounderstand is signalled, where the doctor simulates acoughing sound in order to aid patient understanding,

an event which in other circumstances (and quite poss-ibly here) might be interpreted by P as eccentric orpatronising. This precipitates a bizarre sequence of

chorusing wherein a phrase is picked up and repeatedby all three participants. The repetition comes aboutas the result of a second language learner's mistake (Fsaying `no at all' instead of `not at all') which D

appears to `correct' to `not at all', prompting F, inturn to correct himself, and his wife (as if in echo) torepeat the utterance. After D closes the episode,

indeed, indicates a shift with the discourse marker`okay' (line 34) and a pause, F takes it up again,repeating `not at all' twice, as though attempting to

memorise the idiom:

010 D: weak yes

011 F: yeah012 D: (.) has he um had a temperature?013 F: [M assenting] yes uh thirty eight014 D: thirty eight

015 F: yeah016 D: for how long?017 (2.0)

018 F: yesterday night since yesterday night019 M: [almost inaudible: all day]

020 F: [quietly] okay021 D: has he carried on eating or not eating

022 F: not eating anything[

023 M: [quietly] not eating

024 D: vomiting?025 F: vomiting that's right yeah=026 D: =yeah? (.) any diarrhoea?

027 F: no diarrhoea at all[

028 M: no

029 D: any coughing? [simulates sound of coughing]030 F: no at all031 D: not at all032 F: not at all

033 M: [quietly] not at all034 D: okay035 (.)

036 F: not at all not at all

The doctor again initiates the shift into the next

phase of the consultation with the `okay' (line 38)used as a discourse marker to indicate change oftopic. D does smalltalk with the child, suggesting ten-

tatively that the parents help undress him (maybe ifwe take the jersey o�) followed by the permission-seeking ``I'll listen to the chest ®rst if I may''). Thisextreme politeness indicates sensitivity to the delicacy

of consulting with a small child. The examinationitself is preceded by D asking Ali's age, a questionwhose tone of delivery is `conversational' rather than

`professional'. This change of speech style correspondsto the doctor moving literally into a di�erent space(close to Ali) as well as securing a di�erent interac-

tional positioning, by seeking to gain the child's con-®dence.

037 D: okay (.) I'd like to just have a look at him

038 F: okay039 D: right you keep him on your lap there and I'll (.)040 [to Ali] hullo (.) what's your name? (.) Ali yeah?

041 (.) maybe if we take the jersey o� ®rst of all[

042 F: yeah

043 D: (3.0) how old is Ali now?044 F: (.) two years and (.) eight months045 D: two years and eight months okay

046 I'll listen to the chest ®rst if I may (2.0)047 F: okay [three syllables to A]

The examination

The examination prompts a single D assessment(``yes he's quite hot'') which con®rms the father's

R. Gwyn, G. Elwyn / Social Science & Medicine 49 (1999) 437±447 441

account of Ali's high temperature. F's `yeah' thereforeis almost dismissive (he's given that information

already) and he immediately changes the subject toremark on the size of Ali's tonsils. This change oftopic is in turn ignored by D, who, being engaged

in the professional act of examination, may nothave heard it (Baron, 1985). The fact that Dcontinues with his professional commentary, without

responding to F's comment, would indicate thathe is still `listening', literally with a stethoscope,or at least engaged in a physical examination and

that he is not yet ready to return to the consultationmode:

048 (18.0) [D examines Ali's chest]

049 D: yes he's quite hot050 F: yeah uh do you know I've realised that uh051 he has uh very big tonsils?

052 (5.0)053 D: the other side (.)054 let's have a look in this ear as well

055 (7.0) [D unwraps tongue depressor]

Returning to the consultation, the doctor states his

diagnosis (line 64 ``he's got a sore throat'') and then,strangely, the lay description is re-formulated (line66): ``in medical terms we call it tonsillitis''. F repeats

the word, after which D continues evaluating thepatient's condition, to which F responds (line 71)with a formulaic expression of thanks:

056 F: open your mouth ah:: ah:: ah::057 D: very good058 F: good

059 D: thank you060 (2.0)061 right (3.0) it's very clear what he's got

062 he's got some white spots on his tonsils=063 F: =I see064 D: yeah? so he's got um a sore throat=

065 F: =I see066 D: in medical terms we call it tonsillitis067 F: yeah tonsillitis068 D: =okay (.) his ears are ®ne

069 F: yeah070 D: chest is ®ne071 F: okay thanks very much=

072 D: =so that's why he's got a high temperatureokay?=

073 F: =okay thank you

074 (.)

Heath (1992) provides an example in which a similar

recycling of the clinical term takes place before thedoctor, encouraged by this minimal act of con®r-mation, begins to discuss the diagnosis (Heath, 1992,

p. 242):

Dr.: That's shingles.(1.2)

Dr.: < that's what it is:

(.2)P: Shingles.Dr.: Yes

The procedure of `naming' illness in this way mightbe regarded either as the sharing of professional

knowledge between doctor and patient, or else as anintrusion of that professional discourse (``in medicalterms we call it tonsillitis'') and serving no valid pur-

pose other than reinforcing professional dominance ofthe encounter. This issue will be taken up again in oursummary. In our own data, this section of consultation

is clearly marked o� by F's `thank you' and we movestraight into the key sequence: elicitation of the shareddecision.

`Sharing' a decision?

D begins this section again with a clear discoursemarker: `now' followed by a long (two second) pause.

He is preparing for the process of ``doing shared de-cision-making'' with F. This is F's `turn' to expressany ideas he might have had about Ali's condition

before he came to the surgery. This might be con-sidered an unusual move in the normal script or for-mula of the doctor±patient consultation, but if so, F

does not suggest as much. He responds to this invita-tion without surprise and gives a brief account of asimilar event that took place some time previously

with another son. By doing this he claims prior experi-ence of the situation and is therefore locating himselfas one with a certain limited knowledge. This is a par-ticular kind of strategy known as a `category entitle-

ment', by which individuals' experience entitles themto special knowledge about a topic (Potter, 1996).However, the only course of actual treatment that F

suggests, is that the child should be kept from ``colddrinks something like that'', the partial disclaimer indi-cating that he is not expert in any real knowledge on

this account, a position reinforced by his next utter-ance: ``I don't know any more'' (line 85).This uncertainty suggests that F is treating D's invi-

tation to contribute as rhetorical, as though whatever

he (F) says, he knows that D is the real purveyor ofknowledge in an interaction of asymmetrical power,and therefore he chooses to say little and quali®es that

with a disclaimer, even though he has previous knowl-edge of the condition, or one very similar with anotherchild. This reluctance would seem to undermine D's

attempt to negotiate a shared decision from the verystart and provides an illustration of Heath's claim thatpatients exhibit ``a central concern to avoid any re-

R. Gwyn, G. Elwyn / Social Science & Medicine 49 (1999) 437±447442

sponse which could serve to imply that the participantsversions and assessment of the conditions had an

equivalent status'' and, furthermore, supports theMaynard (1992, p. 362) suggestion that asymmetry isinteractionally achieved rather than imposed and ten

Have's (1991) notion of ``interactional submission'' (cf.Pilnick, 1998).

075 D: now (2.0)076 did you have any ideas as to how we should077 deal with this (.) problem?

078 F: actually I have a (.) other son [D: mmm] (.)079 six and a half years old [D: mmm] (.) he had080 lots of problem (.) about his tonsils (.)081 the same problem (.) actually he [all come?] now

082 he ®nished this problem (1.0) he's coming to ageseven

083 (.) so (.) I think it is better to keep the child

from cold084 (.) no cold drinks? something like that (.)085 I don't know any more

D's reaction is to `normalise' the condition by

emphasising its regularity (Potter, 1996). He does thisby reassuring the parents that ``this kind of problem''is something that ``comes and goes''. He also takes the

opportunity to establish that it is a viral infection or avirus, a term which he immediately de®nes in terms ofits treatment (``which means that..''). This explanatory

sequence is followed by prescription and advice, butadvice which directly contradicts F's only active contri-bution to the decision-making process thus far, namely

that it is better not to give the child cold drinks:

086 D: okay (.) the the ways we deal with tonsillitis(.) um (.)

087 it's quite normal for children to have this kindof problem

088 (.) yeah? d'ya?

[089 F: yes=090 D: =it comes and goes it's usually a viral infec-

tion091 a virus okay? (.)092 which means that (1.0) I would like you to

u::se (.)

093 either Disprol or Calpol to keep the tempera-ture down

094 F: I see=

095 D: =and I would like you to use that everyfour to six hours (.)

096 regularly (.) plenty to drink (.)

097 an' it doesn't matter what really098 so long as it's (.) cold not too much sugar099 F: I see

The doctor provides F with this last information asthough oblivious to F's observation on the avoidanceof the cold and of cold drinks. He then explains why

he doesn't want to prescribe antibiotics on this oc-casion (``they just (.) are no use''), a position he re-inforces by mentioning harmful side-e�ects (``diarrhoea

and vomiting'') as well as ``problems for the future''.After this bleak presentation of the possible harmfule�ects of antibiotics, the question ``which way would

you like to deal with the problem'' (line 109) wouldseem to be a loaded one - but F too has a clear standon the issue of antibiotics, one gained from his ownexperience (the same problem, line 81) of watching his

children `going down' (the metaphor of downwardmovement is here accompanied by a slower utterance)and he is not so easy to coerce into the decision that

D evidently wants him to take. Indeed, this passagewould seem most emphatically to question the under-lying motivation of the `shared decision' procedure,

which on this occasion can be seen as the attemptedimposition of a clearly preferred D option. D goesalong with F's explanation, appearing (as he must) tocollapse his argument (``sure I understand'', line 118,

then `sure', line 122) in deference to F's forthcomingstatement of preference (line 123):

100 D: right? (.) now (.) some people then (.) like touse (.)

101 antibiotics as well (.)

102 but (.) I'm not so keen because103 antibiotics don't deal with viruses (.)104 they just (.) are no use (1.0)

105 and they also cause some problems (.)106 they sometimes cause diarrhoea and vomiting (.)

um (.)107 and it means that you have (.) problems for the

future (1.0)108 so (.) those are the kind of possibilities (1.0)109 which (.) which way would you like to deal with

the problem?110 (1.0)111 F: actually if I use antibiotics for my children (.)

112 the problem (.) is ending in a short time (.)113 which I ha ob observe (.) but the the another

way (.)114 some paracetamol or things yeah (1.0)

115 it will end but a little bit more than the uh (.)116 D: yes take a bit longer=117 F: =yeah take longer

118 D: sure I understand ((yeah))119 (1.0)120 F: so it's it's uh (.) family I mean the uh parents

we don't (1.0)121 want to see our children (.) going down I mean

getting weak

R. Gwyn, G. Elwyn / Social Science & Medicine 49 (1999) 437±447 443

122 D: [quietly] sure=123 F: =so we want to take some (.) antibiotics

Making a choice

This provides Ali's father with his choice, and itwould have been possible to terminate the consultation

here. However, D is not ready to ®nish. There is anoteworthy pause after F's utterance (line 124), afterwhich D pushes for a re-statement of this choice. F

repeats this forcefully, again (line 126), D questions it(line 127) and F repeats (a third time) his preferencefor antibiotics (line 128). He then launches into anunintelligible defence of his decision (lines 128±130),

which D appears to concur with (`sure', line 131),before F closes this turn with the ambiguous remark``we cannot lie'' (line 132). However, D is not satis®ed

with this outcome, and attempts to change F's mind.This he does with an appeal to preventing future sick-ness, by the lowering of resistance (a technical term, or

more precisely a metaphor, for which, this time, heo�ers no explanation). This is the `®rmest' positionthat D has taken so far and it would have been inter-

esting to see what might have happened had F becomemore insistent in his request for antibiotics at thisstage. He appears to back down, however, conceding``I see yeah I understand'' (lines 139±140). D accom-

modates to this concession in F's stance by o�ering acompromise, stating that he is prepared to do a`delayed prescription'. In this way he has actually pre-

sented three choices. They are (1) paracetamol only;(2) paracetamol and antibiotics; (3) paracetamol andthe possibility of antibiotics in a few days. However F

seems to consider only a straight choice between para-cetamol and antibiotics, a reduction of three choices totwo which is consistent with other research ®ndingsthat patients will seek to rede®ne more complex

options as a dualistic choice when possible (Parsonsand Atkinson, 1992; Charles et al., 1998). Thus, D'squestion (line 152): ``which of these possibilities would

you like to do?'' becomes translated in F's version tohis wife as ``which one paracetamol or (.) antibiotics?''he then repeats (with stress) his preferred choice `anti-

biotics?' before M responds in their own language(unfortunately inaudible on tape).

124 (1.0)125 D: you would like to do that would you?

[126 F: yeah

127 D: yeah?128 F: yeah (.) it is too di�cult to to explain but

(2.0)

129 if we can uh (2.0) can be encouraged by doctorsyeah

130 we can do some uh paracetamol131 D: sure=

132 F: =[we cannot lie]133 (.)134 D: my own feeling is that

135 you're probably better to use paracetamol and¯uids

136 rather than use antibiotics

137 because you can cause sickness138 and also resistance for the future

[

139 F: I see140 yeah I understand141 D: um (.) but if you feel strongly142 that you would like to definitely have an anti-

biotic143 we can do that as well (.)144 um the other possibility's for me to give you

145 a prescription for an antibiotic146 and for you to wait147 F: I see (.) yeah

[148 D: and and only use it149 if things get worse

150 you can give me a telephone call or something151 F: yeah (.)152 D: so which one of these possibilities would you

like to do?

153 (1.0)154 F: okay [slight laughter in voice] let me ask my

wife

155 [to M] which one paracetamol or (.) antibiotics?156 (.) antibiotics?

The interchange between parents is brief, beginningwith M's short laugh, but the outcome is (in the con-text of F's previous stand) startling. In one short utter-ance (line 157) F states his new preference and (while

his wife continues to speak to him in a quiet voice)o�ers no further contribution whatsoever to the shareddecision, only giving his son's age, the family's address,

some minimal feedback and a farewell. It is as thoughthe entire preceding discussion has been wiped out. Hiswife meantime is busy thanking the doctor and bidding

him good bye.

Parents' decision and closure

[After a subdued and brief laugh, M responds to Fat some length in their own language, quietly and insis-

tently.]

157 F: yeah paracetamol this time please [M still

talking quietly to F ]158 D: okay (2.0) Disprol or Calpol?

R. Gwyn, G. Elwyn / Social Science & Medicine 49 (1999) 437±447444

159 F: yeah160 D: which one? doesn't matter

161 F: I see uh Calpol is uh eh better than paraceta-mol or euh which one? [M whispers to F through-out ]

162 D: children like it a bit better than most stu�[laughing ]

163 M: yeah=

164 F: =okay165 D: Ali? [writing]166 F: [gives family name and spells it]

167 D: and how old is he?168 F: uh two years and eight months169 D: okay and the address?170 F: gives address

171 D: okay (.) plenty to drink172 F: okay173 D: Calpol every four hours please

174 F: okay thanks175 M: thank you very much176 D: no problem and he's you know he'll be

healthy ®ne177 F: okay178 D: okay no problem

179 M: thanks very much180 D: bye bye now181 F: bye bye

[

182 M: bye

The preference for antibiotics expressed by F in lines

111±124 can be seen in the light of Van der Geest andWhyte's argument for medicines as a metaphor for`getting well'. The choice is borne out by F's previousexperience with another child in a similar situation,

which we have described as a `category entitlement'.This medicine is known to work and therefore argu-ments against its prescription can only be secondary

arguments, that is, they are working against the grainof personal experience and shared knowledge. Whatthe doctor achieves is to present the parents with a

choice in the issue of prescription, a choice which hethen e�ectively undermines by arguing against thefather's preferred alternative. The father o�ers to con-fer with his wife, only to ®nd his own choice over-

turned, a potentially face-threatening outcome.However, the hasty concession: ``yeah paracetamol thistime please'' is, to a certain degree, face-saving, because

F quali®es the decision as being appropriate to thisparticular occasion.In many ways this consultation di�ers very little

from typical doctor±patient interviews based on apatient-centred model (Byrne and Long, 1976; Stewartet al., 1995). However, in accordance with the criteria

of formal interviews, turn-taking is very much con-trolled by the doctor, who, moreover, employs a reper-

toire of discourse markers throughout the consultationas a way of initiating new topics. In order to analyseexactly what goes on in the initial introduction of a

shared decision by the doctor, we need to look at whattakes place in the process of interaction immediatelyafter the medical examination takes place.

The `black box'

At the end of the examination, D tells F that his sonhas ``some white spots on his tonsils'' (line 62) and

then goes on to explain what that means in non-clini-cal terms: ``so he's got um a sore throat'' (line 64).However, D's next utterance, ``in medical terms we call

it tonsillitis'' (line 66), while presented as a piece of in-formation which D `shares' with F and his family is, interms of shared decision-making, quite unnecessary. A

paradox of using clinical terms with patients is thatdoctors are, on the one hand, including the patientwithin their professional discourse and, on the other,possibly excluding them. Whatever the intentions of

this particular doctor and whether or not in other cir-cumstances the `naming' of illness might serve a kindof patient empowerment, within the context of the in-

teraction under discussion, F is, we would suggest, dis-advantaged by the introduction of the new term.After giving the explanation for Ali's sore throat

and temperature, D asks ``now (2.0) did you have anyideas as to how we should deal with this (.) problem?''(lines 75±77), inviting the parent to respond to his

(D's) presentation of the diagnosis. This, we have con-tended, is the initial stage of eliciting a shared decision,and here F is able to state his previous experience ofthe condition and his account of treating his other son.

We ascertain that D and F are therefore both speakingfrom positions of shared knowledge about what consti-tutes `tonsillitis', however di�erently that knowledge

has been formulated. However, D has already re-framed Ali's illness as `tonsillitis' and is thereforespeaking from a position of professional authority,

whereas F (who doesn't use the term tonsillitis, exceptto repeat it after D), speaks of his other son havinghad not only ``lots of problems.. about his tonsils'',but ``the same problem''. Furthermore, his hesitant re-

sponse to the question ``how we should deal with the...problem'' (lines 78±85) suggests that, rather than nothaving anything to say on the topic (he has, after all,

had the `same' problem with his other son's tonsils), hesimply does not want to say something foolish. Inother words he is treating D's question as rhetorical,

one that he knows that D knows the answer to. Hisclosing words in this turn: ``I don't know any more''would appear to con®rm this. Such a confession of ig-

R. Gwyn, G. Elwyn / Social Science & Medicine 49 (1999) 437±447 445

norance presumes a superior knowledge on the doc-tor's part and serves to illustrate how the patient's con-

tribution preserves the contrasting status of the tworoles, emphasising the lay or subjective standpoint ofthe patient (cf. Heath, 1992).

By introducing new terminology, initiating the turn-taking sequences and controlling the topic choice, D ise�ectively sabotaging the chances of a genuinely nego-

tiated shared decision from the outset. That F treatsD's `decision-sharing' question as rhetorical onlyemphasises that an embedded acceptance of power

asymmetry in this consultation will work against anyshared decision even in a situation (like this one)where F already has views, gained from previous ex-perience, on how his son's complaint might best be

treated.

Concluding comments

The clinical discourse analysed is recognised to beone of the most `uncomfortable' situations in whichgeneral practitioners ®nd themselves (Bradley, 1992).The professional stance, backed up by logic and evi-

dence (Little et al., 1997) is that prescribing antibioticsfor self-limiting viral upper respiratory tract infectionsis counter-productive. Microbiological resistance is

stimulated, side-e�ects are commonplace and, arguablythe most salient issue, rewarding attendance by provid-ing expensive, unnecessary and ine�ective magic bullets

rather than promoting self-reliance is a rod for theback of future clinical practice (Stott, 1983). It is inevi-table therefore, if the doctor is doing his job properly,

that the professional opinion is made available, even ifthis undermines the opportunity to arrive at a `shared'but essentially `incorrect' decision. This does notnegate the shared decision-making process, only

enriches it by admitting an element of responsibility(rather than paternalism) and re¯exivity to the doctor'scontribution.

This critique of shared decision-making identi®es theimportant contextual framework which facilitates the`sharing' of decisions between clinician and patient.

The presence of professional `equipoise' about the out-comes of decisions is a key factor: the patient is free tochoose whatever option they prefer, given that theyhave received the necessary information on which to

base their choice. Many decisions in medicine have thisquality and with the increasing emphasis on detailedinformed consent, it is likely that clinicians will be

increasingly asked to maintain `equipoise' in matters ofdecision-making, and to be readier to explain the risksand bene®ts of the treatments at their discretion. It is

merely an acceptance that patient `autonomy' is placedmore to the fore than in previous clinical encounters.However, as this example illustrates, professionals can-

not maintain equipoise on all issues: there will be timeswhen `demand' will cut across `reason', when patient

preference will contradict professional wisdom. Shareddecision-making has limitations, but does it lose allcredibility for having its weaknesses exposed?

The analysis of this consultation exposes a funda-mental problem of any decision-making process carriedout between professional and patient, namely, how to

account for the way in which the embedded powerasymmetry operates so as to facilitate a decision thatthe doctor feels is in the best interests of the patient.

Whether or not this process can legitimately be termeda `shared decision' is therefore open to question, itmight be better quali®ed as a shared decision madeunder the proviso that the patient's preferences are at

least commensurable with their own best interests.What would be of greater complexity, both discursivelyand in terms of patient-centredness, would be analyses

of situations where the principle of equipoise obtainsand while the problem of power asymmetry would notbe `solved' in any way, doctors would be expected to

negotiate a shared decision regardless of their own pre-scribing preference. In such cases, as we hope to haveshown here, a discourse analytic approach can provide

an appropriate method for examining the rhetoricalstrategies employed by both doctors and patients inarriving at a shared, or even a partially-shared de-cision.

References

Atkinson, P., 1995. Medical Talk and Medical Work. Sage,

Londen.

Baron, R., 1985. An introduction to medical phenomenology.

Annals of Internal Medicine 105, 606±611.

Bradley, C., 1992. Uncomfortable prescribing decisions: a

critical incident study. BMJ 304, 294±296.

Byrne, P., Long, B., 1976. Doctors talking to Patients.

HMSO, London.

Charles, C., Gafni, A., et al., 1997. Shared decision-making in

the medical encounter: what does it mean? (Or it takes at

least two to tango). Social Science and Medicine 44, 681±

692.

Charles, C., Redko, C., Whelan, T., Gafni, A., Reyno, L.,

1998. Doing nothing is no choice: lay constructions of

treatment decision-making among women with early stage

breast cancer. Sociology of Health and Illness 20 (1), 71±

95.

Coulter, A., 1997. Partnerships with patients: the pros and

cons of shared clinical decision-making. Journal of Health

Services Research and Policy 2, 112±121.

Coupland, N., Coupland, J., Giles, H., 1991. Language,

Society and the Elderly. Blackwell, Oxford.

Coupland, J., Robinson, J., Coupland, N., 1994. Frame nego-

tiation in doctor±elderly patient consultations. Discourse

and Society 5 (1), 89±124.

Elwyn, G.J., in press. Shared decision-making: the neglected

R. Gwyn, G. Elwyn / Social Science & Medicine 49 (1999) 437±447446

second half of the consultation. British Journal of General

Practice.

Elwyn, G.J., Gwyn, R., Edwards, A., Grol, R., in press.

Is shared decision making feasible in a consultation

for a viral upper respiratory tract function. Health

Expectations.

Emanuel, E.J., Emanuel, L.L., 1992. Four models of the phys-

ician±patient relationship. JAMA 267, 2221.

Fisher, S., 1995. Nursing Wounds. Rutgers University Press,

New Brunswick, NJ.

Heath, C., 1992. The delivery and reception of diagnosis in

the general-practice consultation. In: Drew, P., Heritage, J.

(Eds.), Talk at Work: Interaction in Institutional Settings.

C.U.P, Cambridge.

Hope, T., 1996. Evidence-based Patient Choice. King's Fund

Publishing, London.

Laine, C., Davido�, F., 1996. Patient-centered medicine: a

professional evolution. JAMA 275, 152±156.

Little, P., Williamson, I. et al., 1997. Open randomised trial

of prescribing strategies in managing sore throats. BMJ

314, 722±728.

Lupton, D., 1997. Consumerism, re¯exivity and the medical

encounter. Social Science and Medicine 45, 373±381.

Maynard, D., 1991. Interaction and asymmetry in clinical dis-

course. American Journal of Sociology 97 (2), 448±495.

Maynard, D., 1992. On clinicians co-implicating recipients'

perspective in the delivery of diagnostic news. In: Drew,

P., Heritage, J. (Eds.), Talk at Work: Interaction in

Institutional Settings. C.U.P, Cambridge.

Mishler, E., 1984. The Discourse of Medicine: Dialectics of

Medical Interviews. Ablex, Norwood, NJ.

Muir Gray, J.A., 1996. Evidence-based Healthcare. Churchill

Livingstone, New York.

Ong, L., de Haes, J., et al., 1995. Doctor±patient communi-

cation: a review of the literature. Social Science and

Medicine 40, 903.

Paget, M., 1983. On the work of talk: studies in misunder-

standings. In: Fisher, S., Todd, A. (Eds.), The Social

Organization of Doctor±Patient Communication. Centre

for Applied Linguistics, Washington, DC.

Parsons, T., 1951. The Social System. The Free Press,

Glencoe.

Parsons, E., Atkinson, P., 1992. Lay constructions of genetic

risk. Sociology of Health and Illness 14, 437±455.

Pilnick, A., 1998. ``Why didn't you just say that?'' Dealing

with issues of asymmetry, knowledge and competence in

the pharmacist/client encounter. Sociology of Health and

Illness 20 (1), 29±51.

Potter, J., 1996. Representing Reality. Sage, London.

Quill, T., Brody, H., 1996. Physician recommendations and

patient autonomy: ®nding a balance between physician

power and patient choice. Annals of Internal Medicine

125, 763±769.

Quill, T., Cassel, C., 1995. Nonabandonment: a central obli-

gation for physicians. Annals of Internal Medicine 122,

368±374.

Sackett, D., Scott Richardson, W., et al., 1997. Evidence-

based Medicine: How to Practice and Teach EBM.

Churchill Livingstone, New York.

Silverman, D., 1987. Communication and Medical Practice:

Social Relations and the Clinic. Sage Publications,

London.

Stewart, M., Brown, J., et al., 1995. Patient Centred

Medicine: Transforming the Clinical Method. Sage

Publications, Thousand Oaks, CA.

Stott, N., 1983. Primary Health Care: Bridging the Gap

between Theory and Practice. Springer-Verlag, Berlin.

Taussig, M., 1980. Rei®cation and the consciousness of the

patient. Social Science and Medicine 14B, 3±13.

ten Have, P., 1991. Talk and institution: a reconsideration of

the `asymmetry' of doctor±patient interaction. In: Boden,

D., Zimmerman, D. (Eds.), Talk and Social Structure:

Studies in Ethnomethodology and Conversation Analysis.

University of California Press, Berkeley.

Treichler, P., Frankel, R., et al., 1984. Problems and pro-

blems: power relationships in a medical encounter. In:

Kramarae, C., Schulz, M., O'Barr, W. (Eds.), Language

and Power. Sage, London.

Van der Geest, S., Whyte, S., 1989. The charm of medicines:

metaphors and metonyms. Medical Anthropology

Quarterly 3 (4), 345±367.

R. Gwyn, G. Elwyn / Social Science & Medicine 49 (1999) 437±447 447