Patient Education and Counseling 50 (2003) 33–38

Manufactured but not imported: new directions for researchin shared decision making support and skills

Adrian Edwards∗, Rhodri Evans, Glyn ElwynDepartment of Primary Care, Swansea Clinical School, University of Wales, Singleton Park, Swansea, Wales SA2 8PP, UK

Abstract

Significant conceptual work on shared decision making has taken place but there are still significant challenges in achieving it in routineclinical practice. This paper outlines what research has identified to date that may promote shared decision making, and the further researchthat is required to enable continuing progress. Greater understanding of the models of decision making and instruments to identify them inpractice are still required. Specifying consumer competences, developing instruments to assess these and interventions to enhance them mayalso be important. Clarifying all these aspects may enable those charged with training professionals to improve the content of professionaldevelopment programmes. This may be particularly important in the field of cancer treatments where the stakes are high-patients usuallydesire much information but their desire for involvement in decision making is more variable. The consequences of getting this balanceright or wrong are significant with much to be gained or lost. Continued development and evaluation of decision aids and decision explorersthat use interactive technology will also be important in identifying how to progress with consumer involvement. If we can learn theselessons, then wider implementation of shared decision making or consumer involvement may become a nearer prospect.© 2003 Elsevier Science Ireland Ltd. All rights reserved.

Keywords: Shared decision making; Consumer involvement; Decision aids

1. Introduction

The movement towards greater consumer involvementin decisions about their treatment or health care options—often referred to as ‘shared decision making (SDM)’—hasreached an important and interesting juncture. It is over 5years since the seminal paper by Charles et al.[1] was pub-lished subsequently supported by many other publicationsfrom the same and other key authors in the field[2–4]. Onecan probably summarise developments in the field by sayingthat the initiative to promote SDM has spread across manyof the academic institutions, including both researchers andeducationalists, and is well supported likewise by consumerrepresentative and advocate organisations. However, in thereality of ordinary health care practice, away from these‘ivory towers’, SDM remains a relative rarity or at best a nov-elty whose place is uncertain[5,6]. Attention has thereforebegun to focus on the reasons for this apparent gap betweentheory, policy and education and the reality of health carepractice. Considerable advances have been made to date inresearch in this field and a number of important areas for fur-

∗ Corresponding author. Tel.:+44-1792-513062;fax: +44-1792-513423.E-mail address: adriangkedwards@btinternet.com (A. Edwards).

ther research are being identified which may help to under-stand issues relating to implementation. In doing so they mayalso take forward the understanding of SDM conceptually.

2. Models of decision making

The first area of research concerns models of decisionmaking. Charles et al. identified three broad models of de-cision making—paternalism, shared decision making andinformed choice[1]—seeFig. 1. The distinction betweenthe latter two is where much difficulty arises. From the the-oretical perspective, the informed choice model comprisesgiving information to consumers who, with their preferencesand utilities then have the required elements with which tomake an informed decision[7]. The professional has nothowever contributed his or her opinions or recommenda-tions. In the SDM model, both professionals and consumerscontribute information and opinions so that decisions takeaccount of both recommendations and preferences, and theprocess is characterised by interaction[7].

Despite this delineation of three models it appears hard todistinguish them in practice, particularly between SDM andinformed choice. One explanation for this may be that thethree models are not compartments but part of a spectrum:

0738-3991/03/$ – see front matter © 2003 Elsevier Science Ireland Ltd. All rights reserved.doi:10.1016/S0738-3991(03)00077-6

34 A. Edwards et al. / Patient Education and Counseling 50 (2003) 33–38

Fig. 1. Models of decision making[7].

from paternalism through SDM to informed choice. Wherean individual consultation sits on this continuum is a sub-jective view (at least with current research methods). Theseviews inevitably vary between people assessing the consul-tation. A large part of this variation probably stems fromthe fact that the assessment must be based on observationsof the verbal and non-verbal communication between con-sumer and professional. The level of involvement, though, isa function of the perceptions of the participating consumerand professional—an altogether harder aspect to gauge inassessing consultations.

Alternatively, it may be that SDM and informed choiceare not identifiably separate models, but are more likely tobe dictated by the context of the consultation. Efforts toachieve greater consumer involvement in decision makingmay intrinsically veer more towards the informed choice insome settings. Examples include family planning decisions,choices about antenatal screening tests, or treatments forlong-term conditions which are important but generally notlife threatening, such as menorrhagia, menopausal symp-toms, and prostatism. In these scenarios, the consumer ismost likely to be making the decision, once provided withrelevant information. As personal choices, the professional’sview is likely to be less significant than in other contexts.These others include, for example making decisions aboutchronic treatments that affect future risk of disease, suchas hypertension and ischaemic heart disease, or consider-ing screening tests such as mammography or Papanicolausmears. In these the harm and benefit profile of the availabletreatments or tests may be important in the decision makingand weighed by both consumer and professional[8].

Still further, there may be other contexts where patientsare more likely to prefer the balance in decision mak-ing to be with the professional. Examples include acuteor emergency care, or—and particularly relevant to thispublication—treatment decisions in major illnesses such ascancer, in which the significance of the illness, treatmentoptions and outcomes is so great. In these scenarios, some

patients still actively prefer to lead and to take decisions,but many others prefer to be led by professionals, even ifthey might have been more involved in the types of deci-sions noted above (personal and lifestyle choices, etc.). Thismay be explained by the fact that a significant ‘side-effect’of the decision making process can be regret experienced ifone’s decisions lead to adverse outcomes—this is relativelyspared when someone else has taken on the responsibilityof decision making.

However, these variations are not easily distinguished inclinical practice or research. It may be that separating theSDM and informed choice models is not actually justified.The context may make one model automatically more likely.The important issue is whether the apparent distinction ofthese models in the first few years of the consumer involve-ment movement has led to confusion among health care pro-fessionals[9]. Confusion could give rise to disillusionmentas professionals try unsuccessfully to implement an inap-propriate model, such as shared decision making in familyplanning contexts where the ‘natural’ model of involvingpatients effectively is informed choice.

A third possible explanation for the difficulty in distin-guishing between SDM and informed choice may lie inthe weaknesses of instruments available to assess decisionmaking models in consultations. The range of instrumentsavailable has been reviewed elsewhere[10]. A new instru-ment specifically to assess levels of involvement was devel-oped by Elwyn et al.[11]. Whilst more specifically address-ing the shared decision making competences than the pre-vious instruments, it also focuses on observed behaviours(by professionals). In doing so it synthesises these com-petences into a summary level of involvement but doesnot distinguish between shared and informed choice mod-els. There may still be value in developing an instrumentthat can identify and distinguish models of decision makingin practice.

2.1. Instruments to assess decision making models

Theoretical and empirical research to date has examinedmodels of decision making and has been a key element inthe promotion of greater consumer involvement. Furtherresearch should explore whether it is feasible to develop avalid and reliable instrument to assess the global decisionmaking models operating in consultations. If it is feasiblethen it may be possible to achieve greater conceptual clarityabout whether certain models are intrinsically more likelyin different contexts or whether a range of decision makingmodels is evident. If the former, this could feed into thedevelopment of professional development programmes sothat the goals of implementing ‘consumer involvement’ areclearer and simplified for professionals, i.e. which mod-els to apply in which circumstances. This would be verypertinent in health care contexts such as cancer treatment,where it is imperative for professionals to have the skillsto apply, and to be able to apply them appropriately for the

A. Edwards et al. / Patient Education and Counseling 50 (2003) 33–38 35

individual concerned. Such an instrument may, however,prove impossible to achieve, with persistent and unaccept-able inter-rater differences in interpretation, perhaps reflect-ing the same issue of a gap between observed behavioursand perceptions inside the participants’ minds. In this sce-nario, the question about context influencing likely decisionmodels may need to be resolved by mainly theoretical ratherthan empirical means. Further detailed qualitative analysis,such as with conversation or discourse analytic methods,may, however, shed valuable light on the issues, and willbe important. It may, though, confirm that quantificationof such interactive process issues, which are essentiallybased on the perceptions of and between professional andconsumer, is always likely to remain illusory!

The progress in identifying professional competences forSDM was alluded to earlier[12,13]. Perhaps arising fromthe lack of implementation of SDM by professionals, thereis now increasing attention to the role of consumers inachieving SDM[9]. Research is required to identify the im-portant consumer competences as these may be the key tounderstanding how SDM can be achieved in consultationsand encouraged more widely. The emphasis in trainingprofessionals may thus shift from encouraging profession-als to demonstrate the competences of SDM to trainingprofessionals more specifically to facilitate the SDM com-petences of patients. Other methods of enhancing consumerparticipation may also be highly relevant, particularly in thecancer treatment field, such as via the information materi-als that are plentifully produced. By prompting people toidentify and then ask their questions and explore issues indiscussion, the competences of SDM may be more likelyto happen. In formal terms thedecision aids (see later)that are produced for many treatment decisions, includingcancer choices such as for prostate cancer, can specificallyaddress these issues of patient involvement and SDM. Suchdevelopments should be evaluated in terms of whether theyachieve their desired goals. Consequently, as with profes-sional competences, there will be a need for valid instru-ments to assess patient competences, so that interventionsdirected more towards consumers can be evaluated.

Instruments for consumer competences may also haveadded value when returning to consider the issue of deci-sion making models. We noted above that attempts shouldbe made to develop a valid instrument to identify decisionmaking models in consultations, or to revisit the theoreticaland qualitative empirical understanding of the models. Analternative approach would be to examine how the instru-ments for consumer and professional competences might beintegrated with theoretical or conceptual understanding. Thiscould identify the patterns of competences (behaviours) thatmay constitute or be a proxy for the specific decision makingmodels, even though these may not be identifiable by ratersas a single concept or model in action. The value of any ofthese possible research developments is that they may firstlyenhance our understanding, but secondly we may be able tobuild this into professional development programmes, and

make progress towards greater consumer involvement in rou-tine health care encounters.

3. What is a decision aid?

In addition to the models of decision making and pro-fessional communication skills to achieve consumer in-volvement in practice, attention also centres on the supportmaterials that may facilitate it. These are often termed ‘de-cision aids’. Decision aids are defined as “interventionsto help people make deliberative choices from health careoptions by providing information relevant to a person’shealth status”[14]. They may be used before, during orafter consultations, according to when it is most appropri-ate or helpful to offer such information and support. Manydecision aids have been developed and evaluated, and havebeen extensively reviewed by O’Connor et al.[14,15] andothers[16]. Most focus on information provision and arenot theoretically driven. However, some are based on cur-rent psychological theories of decision making that addressthe cognitive processes that operate on information to ar-rive at a decision. They therefore seek not only to provideinformation but also to facilitate the cognitive processesthat people use on the information to make the decisionitself [17].

Despite the advances in understanding and developmentof decision aids, there is debate about what exactly con-stitutes a decision aid. At the low technological end of aspectrum there are simple health educational leaflets or onesthat seek to enhance participation in health promotion orpublic health orientated programmes (e.g. screening tests).There is debate about whether these can truly be regardedas decision aids. Increasingly, there are also many varia-tions at the high technological end of the spectrum also.We distinguishdecision aids as more standard products thatprovide information for decision making. They may includeformal clarification of users’ values and preferences. How-ever,decision explorers use interactive multimedia formatsto let users select different types of information (e.g. narra-tive, numerical, graphical), different levels of depth in thisinformation and different levels of support in decision mak-ing [18]. Each has its place, according to context and users’needs[19].

It seems likely that many of the treatment decisions incancer conditions will be well supported by decision aids orparticularly explorers. This is because the decision makingin practice is frequently influence or even constrained by thecontext—busy clinics, the shock of a diagnosis, informationoverload and anxiety generated (seeFig. 2)—all of whichmake decision making involving patient impractical. Deci-sion aids and explorers offer people the opportunity to takethe relevant information away, together with, for example atape-recording of the consultation with the specialist if thisis provided, in order to reflect on the diagnosis, its impli-cations and treatment options. The patient can then explore

36 A. Edwards et al. / Patient Education and Counseling 50 (2003) 33–38

Fig. 2. Information overload?

different areas of information and recommendations as theyneed or desire, and at a pace that they feel comfortable with.

3.1. Effective decision making

There is still a need to evaluate these types of materials.The evaluations will address the effects on decision mak-ing, in terms of the choices made, resources used and eitheractual health outcomes or modelling the effects of differentchoices on potential future health outcomes. The evaluationswill also assess whether these information and decision sup-port materials have achieved ‘effective decision making’. Aswith other concepts described above, there is debate aboutexactly what this constitutes. In this case it centres on thecomponents of effective decision making, and a distinctionbetween rational and reasoned choice models. In the rea-soned choice models, the emphasis is on the way individualsmake decisions. An informed choice is described as one thatis based on (a) an accurate assessment of the relevant infor-mation about alternatives, (b) an assessment of their likeli-hood and desirability in accord with the individual’s beliefs,and (c) a ‘trade-off’ between these factors[20].

In contrast, rational choice models view decisions as ef-fective if the actual choices are consistent with the theoret-ically derived maximum expected utility. The latter modelsare criticised on the grounds that individuals rarely have orare indeed able to use all available information to assess ex-pected utilities. Decisions may be consistent with maximumexpected utility but may not have been achieved by assessingit. There are different views about this[20–26]. It is impor-tant to resolve this debate, as developments in the field arelikely to be dependent on it. With clearer outcome measuresthe scope for meta-analysis will become wider, and impor-tantly it will be possible then to examine for the principaleffect modifiers. That is, it will be possible to examine formore robust evidence than we currently have about whichare the most effective strategies, contexts, characteristics andcontent of decision aids to achieve effective decision mak-ing in different settings.

3.2. Barriers and opportunities for using decision aids

There is an important further issue, however. It relatesto all aspects of consumer involvement, but is particularlypertinent when considering decision aids. Despite increas-ing evidence about their apparent effectiveness, there is alsomuch evidence that implementation—especially outsidethe academic environments of most research—is limited.Holmes-Rovner has explored the barriers and opportunitiesfor implementing shared decision making and has notedthat apparent professional approval for innovations does notequate with participation[6]. She notes also that some ofthe barriers may relate to physical locations of technical in-novations (away from where professionals and patients areconsulting). Time pressures are frequently cited as the mainproblem, as well as costs of implementation (including pro-fessional time). There is a need for further research into thebarriers and opportunities for implementing SDM and us-ing decision aids, and particularly the possible differentialimpacts of decision aids available outside or within consul-tations. That is, the influence of the decision aid-informedconsumer on the subsequent interaction with professionalsmay be highly significant. The availability of appropriatedecision aids to consumers before they visit health care pro-fessionals may then prove to be a key step towards not justuse of decision aids, but wider involvement of consumersin decisions about their treatment or care.

Alongside this, however, the research should examine thediscourse of using decision aids with patients, or discussingthe information and issues raised by them. There is more to itthan simply making decision aids available, and we need tounderstand more about the challenges and successful tech-niques, so that they can be integrated into the emerging ‘NewMedical Conversation’[27]. Greater understanding about theeffective skills and techniques as well as availability of deci-sion aids are likely to be necessary as a means to facilitatingimplementation. Imparting this understanding to profession-als in training programmes is likely to lead to more positiveexperiences and positive reinforcement. This may encourageprofessional uptake and widespread use of decision aids.

Perhaps more outside consultations, the consumer healthinformatics (or ‘e-health’[28]) developments are also im-portant in this sphere of decision aids, or as we have termedthem here decision explorers. Interactive technology allowsincreasingly complex but flexible and tailored packages tobe produced. Currently, patients with cancer are perhapsamong the most involved and informed of all patients, dueto the significance of their condition compared to others.People search at length for information but its quality maybe unreliable, or the support it offers to patients somewhatquestionable. With increasing internet access, however, goodquality information and decision support can be made avail-able to all types of patients, and this is rapidly increasingtoo. Perhaps this will prove to be the most natural and real-istic context for SDM—away from the pressures of clinicalpractice.

A. Edwards et al. / Patient Education and Counseling 50 (2003) 33–38 37

These decision explorers require full evaluation, includingwhether they promote effective decision making (howeverdefined), different choices and resource use. However, inaddition to these existing areas of outcomes for patients andthe health care system, further data are potentially available.The technology can track an individual’s pathway throughthe package, identifying the nature, sequence and complexityof information accessed by the user. This may add furtherimportant empirical data about decision making processes,and add to our conceptual and theoretical understanding.Again, such developments will be able to feed into servicedevelopments and training developments for professionalstrying to achieve wider implementation of SDM.

4. Levers for change

The issue about wider implementation is also broaderthan this. So far the issues we have described have mostlycentred on single strategies, usually to professionals toencourage implementation. However, there may be otherlevers for change, and a fundamental question arises: Howis SDM or patient involvement to be promoted most ef-fectively? We have commented on training interventionsto professionals and possible interventions to enhance con-sumer competences in the health care setting. But widerinfluences may be important too. These include, for exam-ple, general or societal influences on consumers such asthe live and written media, including traditional publicityand e-health information, and consumer advocate and pa-tient representative groups[29]. Other levers on the healthcare system include financial control[6]. Payers for healthcare may be able to demand SDM or evidence of patientinvolvement as quality measures to achieve higher fundingor accreditation rates. Some of these could be more easilyimplemented as a defined intervention, but there is a needto evaluate the differential effects of these different strate-gies. This does not suggest that one strategy is likely to beidentified as most effective and then become the favouredapproach. But it may be valuable to understand whichare the most influential and where resources need to bededicated to try to achieve wider implementation of SDM.

5. Conclusion

Considerable achievements have been made in the fieldof research on shared decision making in the last 5 years.Significant conceptual work on decision making models andinstruments to identify them in practice are still required.Specifying consumer competences, developing instrumentsto assess these, and interventions to enhance them may bean important facet of such work. Clarifying all these aspectsmay enable those charged with training professionals to im-prove the content of professional development programmes.This may be particularly important in the field of cancer

treatments where the stakes are high—patients usually de-sire much information but their desire for involvement indecision making is more variable. The consequences of get-ting this balance right or wrong are significant with much tobe gained but also much to be lost. Allied to continued workon development and evaluation of decision aids, and deci-sion explorers that use interactive technology, these lessonswill be important in identifying how to make progress inthe field of consumer involvement. If we can learn theselessons, then wider implementation of SDM or consumerinvolvement may become a nearer prospect.

References

[1] Charles C, Gafni A, Whelan T. Shared decision making in themedical encounter: what does it mean? (Or it takes at least two totango). Soc Sci Med 1997;44:681–92.

[2] Charles C, Gafni A, Whelan T. Decision making in theprofessional–patient encounter: revisiting the shared treatmentdecision-making model. Soc Sci Med 1999;49:651–61.

[3] Guadagnoli E, Ward P. Patient participation in decision making. SocSci Med 1998;47:329–39.

[4] Coulter A. Partnerships with patients: the pros and cons of sharedclinical decision making. J Health Services Res Policy 1997;2:112–21.

[5] Elwyn GJ, Edwards A, Kinnersley P. Shared decision making inprimary care: the neglected second half of the consultation. Br JGen Prac 1999;49:477–82.

[6] Holmes-Rovner M, Valade D, Orlowski C, Draus C, Nabozny-ValerioB, Keiser S. Implementing shared decision making in routine practice:barriers and opportunities. Health Expect 2000;3:182–91.

[7] Elwyn G, Charles C. Shared decision making: the principles andthe competences. In: Edwards A, Elwyn G, editors. Evidence-basedpatient choice: inevitable or impossible? first ed. Oxford: OxfordUniversity Press; 2001. p. 118–43.

[8] Gwyn R, Elwyn G. When is a shared decision not (quite) a shareddecision? Negotiating preferences in a general practice encounter.Soc Sci Med 1999;49:437–47.

[9] Kim Y, Kols A, Putjuk F, Heerey M, Rinehart W, Elwyn G, et al.Participation by clients and nurse midwives in family planning de-cision making in Indonesia. Patient Educ Counsel 2003, in press.

[10] Elwyn G, Edwards A, Mowle S, Wensing M, Wilkinson C, KinnersleyP, et al. Measuring the involvement of patients in shared decisionmaking: a systematic review of instruments. Patient Educ Couns2001;43:5–22.

[11] Elwyn G, Edwards A, Wensing M, Hood K, Atwell C, Grol R.Fleeting glimpses: measuring shared decision making in primary careusing the OPTION instrument. Qual Safety Health Care 2002, inpress.

[12] Elwyn G, Edwards A, Kinnersley P, Grol R. Shared decision makingand the concept of equipoise: defining the ‘competences’ of involvingpatients in health care choices. Br J Gen Prac 2000;50:892–9.

[13] Towle A, Godolphin W. Framework for teaching and learning in-formed shared decision making. Br Med J 1999;319:766–71.

[14] O’Connor AM, Rostom A, Fiset V, Tetroe J, Entwistle V,Llewelyn-Thomas H, et al. Decision aids for patients facinghealth treatment or screening decisions: systematic review. BMJ1999;319:731–4.

[15] O’Connor A, Fiset V, Rosto A, Tetroe J, Entwistle V,Llewellyn-Thomas H, et al. Decision aids for people facing healthtreatment or screening decisions [Cochrane Review]. Cochrane Li-brary, Issue 1. Update Software; 2002.

38 A. Edwards et al. / Patient Education and Counseling 50 (2003) 33–38

[16] Molenaar S, Spranger sM, Postma-Shhuit C, Rutgers E, NoorlanderJ, et al. Feasibility and effects of decision aids. Med Decision Making2000;20:112–27.

[17] Feldman-Stewart D, Brundage MD, van Manen L. A decision aidfor men with early stage prostate cancer: theoretical basis and a testby surrogate patients. Health Expect 2001;4:221–34.

[18] Wright P. Document-based decision-making. In: Rouet JF, LevonenA, Biardeau A, editors. Multimedia learning: cognitive and instruc-tional issues. Amsterdam: Pergamon Press; 2001. p. 31–43.

[19] Edwards A, Elwyn G. The potential benefits of decision aids inclinical medicine. J Am Med Assoc 1999;282:779–80.

[20] Bekker H. Genetic testing: facilitating informed choices. In: Ency-clopaedia of the human genome. Nature Publishing Group; 2002.

[21] Bekker H. Why decision analysis facilitates decision making: link-ing process and outcome measures. In: Proceedings of the 1st In-ternational Conference on Shared Decision Making in Health Care.Oxford; 2001.

[22] Marteau TM, Dormandy E, Michie S. A multi-dimensional measureof informed choice for use in Down’s Syndrome screening. HealthExpect 2001;4:99–108.

[23] Entwistle VA, Sowden AJ, Watt IS. Evaluating interventions to pro-mote patient involvement in decision making: by what criteria shouldeffectiveness be judged? J Health Serv Res Policy 1998;3:100–7.

[24] O’Connor AM. Validation of a decisional conflict scale. Med Deci-sion Making 1995;15:25–30.

[25] Edwards A, Elwyn G. How should ‘effectiveness’ of risk communi-cation to aid patients’ decisions be judged? A review of the literature.Med Decision Making 1999;19:428–34.

[26] Edwards A, Elwyn G, Hood K, Robling M, Atwell C, Russell I, et al.The development of COMRADE: a patient-based outcome measureto evaluate the effectiveness of risk communication and treatmentdecision-making in consultations. Patient Educ Counsel 2003, inpress.

[27] Mazur D. The new medical conversation. Boulder, Connecticut:Rowman and Littlefield; 2002.

[28] Eysenbach G. Recent advances: consumer health informatics. BrMed J 2000;320:1713–6.

[29] Edwards A, Elwyn G. Understanding risk, and lessons for clinicalrisk communication about treatment preferences. Qual Health Care2001;10(Suppl I):i9–13.