Sharing Vs. Caring—the Relative Impact Of Sharing Decisions Versus Managing Emotions On Patient Outcomes

Transcript

  See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/224615196 Sharing vs. caring-The relative impact of sharing decisions versus managing emotionson patient outcomes  Article   in  Patient Education and Counseling · April 2010 DOI: 10.1016/j.pec.2010.04.001 · Source: PubMed CITATIONS 36 READS 53 8 authors , including: Some of the authors of this publication are also working on these related projects: Pathways for diagnosis to lung cancer: qualitative interviews with patient and GPs View projectAdvance care planning in incurable cancer patients: a randomised controlled trial   View projectAllan 'Ben' SmithUniversity of Sydney 22   PUBLICATIONS   135   CITATIONS   SEE PROFILE Ilona JuraskovaUniversity of Sydney 90   PUBLICATIONS   896   CITATIONS   SEE PROFILE Phyllis N ButowUniversity of Sydney 297   PUBLICATIONS   9,617   CITATIONS   SEE PROFILE Richard F BrownVirginia Commonwealth University 74   PUBLICATIONS   1,694   CITATIONS   SEE PROFILE All content following this page was uploaded by Ilona Juraskova on 26 December 2016. The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the srcinal documentand are linked to publications on ResearchGate, letting you access and read them immediately.  Medical Decision Making Sharing vs. caring—The relative impact of sharing decisions versus managingemotions on patient outcomes Allan Smith a, *, Ilona Juraskova a , Phyllis Butow a , Caroline Miguel a , Anna-Lena Lopez a ,Sarah Chang a , Richard Brown b , Jurg Bernhard c a University of Sydney, Australia b Department of Social and Behavioural Health, Virginia Commonwealth University, USA c International Breast Cancer Study Group Coordinating Centre, Bern, Switzerland 1. Introduction In recent times shared decision making (SDM) has become thefavoured model of treatment decision making, largely due to anincreased emphasis on informed choice driven by the consumerrights movement, and changes in the nature of contemporarymedical care [1]. Shared decision making refers to the mutual involvement of the doctor and patient in all aspects of decisionmaking: information exchange, deliberation regarding treatmentoptions, and choosing which treatment to implement [2]. Theelicitation of patients’ fears, concerns and hopes has also beenidentifiedasacriticalaspectofthisprocess,throughinterviewswithkey informants regarding the essential components of SDM [3]. In addition, Stewart [4] noted that SDM could not occur outside of acaring, respectful and empowering doctor–patient relationship.Further to granting patients’ greater autonomy and control overtheir care, there is some evidence that SDM improves patientoutcomes,suchasdecisionalconflict,satisfaction,andpsychologicaladjustment[5,6].However,researchindicatesthatSDMaccountsfor onlyasmallamountofvarianceinsomeoftheseoutcomes[7],andis notalwayspractical,ordesiredbypatients[8,9].Further,thespecific componentsofSDMthatproducebeneficialoutcomeshavenotbeenclearly delineated. In particular, it is unclear whether it is thecognitive aspects of SDM (information exchange, deliberation andchoice), or the relationship that is established between the doctorandpatientduringthisprocess,whichleadtopositiveoutcomes.Forexample, Fallowfield et al. [10] found that after SDM in breastsurgery consultations, it was not having choice that producedimprovedpsychologicaladjustmentperse,butratherbeingseenbya surgeon who routinely offered choice.Otherresearch has highlightedthe impactof emotionalaspectsof SDM such as emotional relating (facilitation of emotional Patient Education and Counseling 82 (2011) 233–239 A R T I C L E I N F O  Article history: Received 29 May 2009 Received in revised form 28 February 2010 Accepted 2 April 2010 Keywords: Doctor–patient communicationShared decision makingEmotional relatingPatient outcomesOncology A B S T R A C T Objective:  To assess the relative impact of cognitive and emotional aspects of shared decision making(SDM) on patient outcomes. Methods:  Cognitive and emotional aspects of SDM in consultations between 20 oncologists and 55earlybreast cancer patients were coded using the Observing Patient Involvement (OPTION) scale and theResponse to Emotional Cues and Concerns (RECC) coding system, plus blocking and facilitatingbehaviour scales. Patient outcomes including anxiety, decisional conflict, and satisfaction with: (i) thedecision,(ii)theconsultation,and(iii)doctorSDMskills,wereassessed.Relationshipsbetweencognitiveand emotional aspects of SDM, and patient outcomes were examined using hierarchical regression. Results:  The OPTION score predicted satisfaction with doctor SDM skills 2 weeks post-consultation(  p  = .010), and with the treatment decision 4 months post-consultation (  p  = .004). Emotional blockingpredicted decisional conflict (  p  = .039), while the number of emotional cues emitted (  p  = .003), and thedegree of empathy provided (  p  = .011), predicted post-consultation anxiety. Conclusion:  Cognitive and emotional aspects of SDM in oncology consultations have different effects onvarious patient outcomes. Practice implications:  It is important that doctors focus on both sharing decisions and managingemotions in consultations. Communication skills training addressing both these areas may be aneffective way to improve diverse patient outcomes.   2010 Elsevier Ireland Ltd. All rights reserved. * Corresponding author at: CeMPED, School of Psychology, Brennan MacCallumBuilding (A18), University of Sydney NSW 2006, Australia. Tel.: +61 2 9351 4518;fax: +61 2 9036 5292. E-mail address:  [email protected] (A. Smith). Contents lists available at ScienceDirect Patient Education and Counseling journal homepage: www.elsevier.com/locate/pateducou 0738-3991/$ – see front matter    2010 Elsevier Ireland Ltd. All rights reserved.doi:10.1016/j.pec.2010.04.001  expression and provision of empathy), on patient outcomesincluding satisfaction and psychological adjustment [11]. For example, a survey of 454 patients attending an oncologyoutpatient clinic showed that physician attentiveness and empa-thy were associated with greater patient satisfaction, increasedself-efficacy, and reduced emotional distress [12]. Street et al. [13] found that patient participatory behaviours were associated withdoctor supportive talk, further demonstrating the inter-related-ness of these behaviours. In a later paper, Street et al. [14] notedthat communication might affect health in a number of ways,including indirectly via proximal outcomes of the interaction (e.g.satisfaction withcare). Theysuggested that future researchshouldinvestigate relationships between specific aspects of communica-tion and particular patient outcomes, and hypothesise potentialpathways for any associations.One pathway by which specific aspects of communication,namely the cognitive and emotional aspects of SDM, may affectpatient outcomes is outlined in Leventhal’s parallel processingmodel[15].Accordingtothismodel,peopleadapttohealththreats via cognitive and emotional processing of illness information [15].Both representations influence the patient’s appraisal of threat,which in turn influences their behaviour. Thus, doctors need toconsider both their cognitive and emotional interactions withpatients in order to optimise copingSystems for coding SDM in medical consultations have tendedto focus on the cognitive aspects of SDM. The Observed PatientInvolvement (OPTION) scale for instance, codes whether thedoctor defines the problem, explains the existence of differenttreatment options, describes the outcomes associated with theseoptions, and initiates the decision-making process [16]. Emo-tional relating is usually coded outside the context of an over-arching communication framework like SDM. For example, theResponse to Emotional Cues and Concerns (RECC) coding systemdescribed below, was developed to characterise doctors’responses to emotional expressions; however, it capturesemotional relating behaviours described by Ford et al. [3] asessential components of SDM.The aim of the current study was therefore to assess therelative impact of cognitive and emotional aspects of SDM onpatient outcomes in a single dataset. It was hypothesised thatdoctor behaviours targeting cognitive processing of information(as measured by the OPTION scale) would account for morevariance in outcomes related to the decision itself, such asdecisional conflict and decisional satisfaction. In contrast, it washypothesised that doctor behaviours focused on eliciting andresponding to emotion (as measured by the RECC coding system)would account for more variance in outcomes linked withemotional processing, such as anxiety and satisfaction with theconsultation. 2. Methods The audiotaped consultations and patient outcome dataanalysed in this study come from the International Breast CancerStudyGroup(IBCSG)Trial33-03,whichisevaluatingtheefficacyof a communication skills training program.  2.1. Sample 2.1.1. Clinicians Medical, radiation and surgical oncologists from Australian,New Zealand and European centers participating in IBCSG clinicaltrials were invited to take part in the communication study.However, to avoid cross-cultural issues (such as differences inemotional expression, clinic organization and services) cloudingthe results, only data from the Australian and New Zealandparticipants were included in the current analysis. Australia andNew Zealand both have a public health system, although inAustralia there are tax incentives to join a private health fund.  2.1.2. Patients Eligible patients were: (i) over the age of 18, (ii) newlydiagnosed with early stage breast cancer, (iii) competent atcommunicating in English, and (iv) mentally and physicallycapable of participating in the study.  2.2. Procedure and study design EthicsapprovalwasobtainedfromtheUniversityofSydneyandarea health service ethics committees linked to participatingcancer centres ( n  = 10). Participating doctors were asked to inviteconsecutive patients with whom treatment options would bediscussed to take part in the study. Patients completed a baselineassessment prior to their initial consultation eliciting demo-graphics, information and involvement preferences, and currentanxietylevels.Doctorswereaskedtoaudiotapetheirconsultationswith a subset of the patient cohort (on average three to four perdoctor). In the srcinal study these were used to provide feedbackto doctors about their consultation skills. Two weeks after theirinitialconsultation,patientsweremailedaquestionnaireassessingtheir: (i) level of decisional conflict, (ii) satisfaction with theirtreatment decision, (iii) satisfaction with the consultation, (iv)satisfaction with doctor SDM skills, and (v) current anxiety levels.Patients were mailed another questionnaire assessing satisfactionwith the decision 4 months after the consultation.Quantitative and qualitative data were available from 70audiotaped consultations. After removing cases with incompletetranscripts(duetorecordingproblems),orinsufficientpatientdata(due to non-return of questionnaires), a total of 55 consultationsfrom 20 clinicians were included in the current study. Nodifferences in demographics or patient outcomes were foundbetween the 55 complete cases and the 15 cases with incompletedata (data not shown).  2.3. Coding  Audiotaped consultations were transcribed verbatim. TheOPTION scale [16], and the RECC coding system [17,18], plus scales assessing blocking and facilitating behaviours [19], were applied to the transcripts by different coders to minimise bias.Coders weretrainedonfiveincompletetranscripts(excludedfromthe main analysis) by an experienced independent coder until ahighlevelofagreementwasestablished.Codersreadthehardcopyof the transcript while listening to the audiotape to pick up para-verbal cues. When coding was complete, each coder re-coded 10%of the transcripts coded by a second coder to establish inter-raterreliability.  2.4. Coding systems 2.4.1. The observing patient involvement (OPTION) scale The 12-item OPTION scale was designed by Europeanresearcherstoassesskeycompetencesofdoctors’SDMbehaviours,namely:definingtheproblem,explainingtheexistenceofdifferenttreatment options, describing the outcomes associated with theseoptions, and initiating the decision-making process [16]. Shared decisionmakingisassessedonafive-pointscale,rangingfrom‘‘thebehaviour is not observed’’ (0) to ‘‘the behaviour is exhibited to avery high standard’’ (4). Raw scores for the OPTION scale, rangingfrom0 to 48,weretransformedto a scoreoutof 100.Higherscoresindicate a high level of behaviour exhibiting the competences of SDM.  A. Smith et al./Patient Education and Counseling 82 (2011) 233–239 234  Evaluation of the OPTION scale in the general practice settinghas demonstrated its’ construct validity through associations withfactors thought to affect involvement in decision making, such asage and the presence of equipoise in consultations [16]. It has also shown the ability to reliably detect differences between practi-tioners and the extent to which they involve patients in decisionmaking [16]. Inter-rater reliability was confirmed for the 12 statements with a mean kappa score of .66 [16]. In the current study the mean inter-rater kappa score was .58 indicatingacceptable agreement after correcting for chance.  2.4.2. The response to emotional cues and concerns (RECC)coding system TheRECCsystem [17,18] isbasedon generalcounseling theory,which emphasises the importance of facilitating emotionalexpression and responding in an empathic manner [20]. It was developed to code emotional expressions, specifically cues,concerns, and psychosocial issues raised by cancer patients, anddoctor responses to these. This system has proven inter-raterreliability [17,18], and has demonstrated associations withpatient outcomessuchasdepression[18].Ithasbeenfurtherdevelopedby theVeronaNetworkonSequenceAnalysisgroup[21],althoughthe srcinal version of the coding system (with some minor modifica-tions) is used here. A cue was defined as a verbal or non-verbalexpression by the patient that contained an implicit reference tonegative emotional content (e.g. ‘‘I just can’t seem to relax’’). Aconcern was defined as an explicit expression of a negativeemotion relatedto an issue of importanceto the patient (e.g. ‘‘I amso worried about this constant pain’’). A psychosocial issue wasdefined as a reference to a lifestyle issue, without any associatednegative emotion (e.g. ‘‘Will I be able to work during mychemotherapy?’’).The specific features of each cue and concern, such as theinitiator,andtheintensityoftheexpressedemotionwerecodedfordescriptive purposes. Cues and concerns were coded as beinginitiated by the doctor if they occurred immediately after thedoctor had said something with psychosocial content (e.g. ‘‘Howhave you been feeling?’’ or ‘‘That must have been very hard foryou.’’). Cues and concerns were coded as initiated by the patient if they occurred without such a prompt. Cues and concerns werecoded as  weak  if the implicit or explicit emotion was of lowintensity (e.g. ‘‘It has been difficult lately’’). Cues and concernswere coded as  strong   if the implicit or explicit emotion was of highintensity(e.g.‘‘I’vebeenextremelyworried’’or‘‘It’sbeenterrible’’).The doctor’s level of empathy in response patient cues orconcerns was coded as follows:Level0:Ignores,changesthesubjectoroffersfalsereassurance;Level 1: Responds with minimal encouragers (e.g. ‘‘Mmmm’’,‘‘Yes’’, Indeed’’)Level2: Respondsto contentorfeeling(e.g.‘‘Itmustbehardforyou . . . ’’)Level 3: Responds to feeling and invites elaboration (e.g. ‘‘I cansee you are worried . . . can you tell me a bit more about whatscares you?’’)Thetotalnumberofcuesandconcernsemitted,andtheaverageempathy level (0–3) of doctors’ responses across all cues andconcerns were calculated. The initiator and the intensity of eachcue and concern were not taken into account when calculatingdoctors’ empathy levels.  2.5. Blocking and facilitating behaviour scales These scales were developed to assess doctor blocking andfacilitating behaviours as part of a previous study evaluatingcommunication skills training in eliciting and responding toemotional cues [19]. Specifically, the 10-item blocking behaviour scale identified behaviours known to hinder discussion of emotional issues (e.g. interrupting, monopolising, changing thesubject). Conversely, the 9-item facilitating behaviour scaleincluded behaviours that facilitate such discussions via activelistening and conveying basic empathy (e.g. appropriate use of questions, use of words with emotional content, listening withoutinterrupting). The subscales demonstrated good inter-rateragreement (.68–0.91). Blocking and facilitating behaviours wereidentifiedasperthescales,andthenasubjectiveoverallratingwasgiventothedegreeofblockingandfacilitatingbehaviourexhibitedby the doctor.Ten percent of consultations were double-coded to determineinter-rater reliability. The average inter-rater kappa score for theRECC coding system and the blocking and facilitating scales was.54indicatingacceptableinter-rateragreementaftercorrectingforchance.  2.6. Demographics and outcome measures Decisional Conflict was assessed using two independentlyvalidated subscales of the Decisional Conflict Scale (DCS) [22],measuring decisional uncertainty (3 items), and factors contribut-ing to uncertainty (9 items). Subscale scores were summed toproduce a total score (12–60). Higher scores (reversed forconsistency with other measures) indicate lower decisionalconflict. In the current sample the Cronbach’s alpha coefficientwas .71.Anxietylevelsweremeasuredusingthestate scaleoftheState-TraitAnxietyInventory[23],whichconsistsof20itemsmeasuringcurrentlevels of anxiety.Scores range from 20 to 80. Higherscoresindicate greater levels of anxiety. Internal consistency has beenfound to be very high in several normative samples, withCronbach’s alphas above .90 [23].Satisfaction with the decision was assessed using the satisfac-tionwithdecisionscale[24].Thisscaleconsistsof6itemsandtotal scores range from 6 to 30. Higher scores are indicative of greatersatisfaction with the decision. In the current study the Cronbach’salpha coefficient was .91.Satisfaction with the consultation was assessed using a 25-item Likert scale adapted from Roter [25] and Korsch et al. [26]. Total scores range from 25 to 125. The measure includes itemsassessing satisfaction with the: (i) amount and quality of information received, (ii) communication skills of the clinician,and (iii) level of patient participation throughout the consulta-tion. Higher scores indicate greater satisfaction with theconsultation. In the current sample, the Cronbach’s alphacoefficient was .93.  2.7. Satisfaction with doctor SDM skills PatientsatisfactionwithdoctorSDMskillswasassessedusinga12-item purpose-designed measure. Total scores range from 12 to60. Higher scores represent greater satisfaction with doctorSDM skills. The Cronbach’s alpha coefficient for the current studywas .79.  2.8. Demographic characteristics Patients’ age, gender, marital status, education, occupation,place of birth, first language spoken, and relevant health trainingwere assessed. Doctor characteristics including age, gender,specialty, communication skills training, years of practice, yearsspent working with cancer patients, and hours per week in directpatient contact were also assessed.  A. Smith et al./Patient Education and Counseling 82 (2011) 233–239  235   2.9. Data analysis DataanalyseswereconductedusingtheStatisticalPackagesforSocial Sciences (SPSS) Version 16.0 [SPSS Inc., Chicago, IL]. TheOPTION total score, average empathy level over cues/concerns,overall level of blocking behaviour, and overall level of facilitatingbehaviour were calculated for each consultation. BivariatePearson’s correlations were conducted as a preliminary assess-ment of the hypothesised relationships between each of thesevariables and patient outcomes.A series of hierarchical linear regressions were then performedwith each of the patient outcomes entered as the dependentvariable,andanydemographicorconsultationbehaviourvariablesthat were correlated at   .3 entered as independent variables.TheseincludedtheOPTIONtotalscore,levelofblockingbehaviour,and average empathy score. 3. Results Ten centres participated, with a median of 1.5 doctors percentre (range: 1–6), and 3 patients per doctor (range: 1–5). Thedoctor sample consisted of 11 male and 9 female doctors, with amean ageof 47 years(SD = 8.5, range 33–62).Elevenwere medicaloncologists, 6 were radiation oncologists, and 3 were surgicaloncologists with an average of 20 years in practice (SD = 10.3,range 2–37). Participating patients ( n  = 55) had a mean age of 52.5years (SD = 12.5, range 31–81), and almost two-thirds (62%) werein a married or de facto relationship. Nearly three quarters (72.7%)were born in Australia or New Zealand, and close to 89% listedEnglish as their first language. The highest level of educationreachedbymostofthewomen(62%)wassecondaryschooling,and61% were working in a non-professional capacity. Less than aquarter of all patients reported some form of healthcare training(18%).  3.1. SDM in consultations The mean total score on the OPTION scale was 23.44 (SD = 9.1,range 10–44). Although this mean score is higher than reported inprevious samples [16], in absolute terms it still indicates a lowlevel of SDM behaviours, with most items being rated as notobserved or minimally exhibited.  3.2. Emotional relating in consultations A total of 51 cues, concerns and psychosocial issues wereidentified,withamedianof5perconsultation,andarangeof0–14(additional details can be found in Table 1). Emotional expressionswere initiated by patients (48%) and doctors (48%) equally, mostcommonly in the form of a verbal cue (52%), with the intensity of emotionbeingweakinthemajorityofcases(63%).Mostemotionalexpressions were related to existential issues (48%), and almosttwo-third (63%) were responded to with no empathy. As Table 2indicates, whilst more than two-thirds of the consultations (69%)contained a medium to high level of facilitating behaviour, half of the consultations (50%) also contained a medium to high level of blocking behaviour.  3.3. Patient outcomes Descriptive statistics for all patient outcomes are shown inTable 3. Satisfaction with the treatment decision remained highand stable over time. Satisfaction with the consultation was alsohigh, but there was more variability in satisfaction with doctorSDM skills. Anxiety and decisional conflict were relatively low inthis sample.  3.4. Relationships between consultation behaviours and patient outcomes Correlations between scores on each coding system andoutcome measure are displayed in Table 4. Table 5 shows the results of the 6 linear regressions conducted with each of thepatient outcomes entered as the dependent variable, andthe number of cues, average empathy, level of blocking behaviour,and OPTION total score entered as independent variables. As  Table 1 Features of cues, concerns and psychosocial issues as coded by the RECC codingsystem.Feature Category Count PercentageThe initiator Dr prompt 5 10.4Dr related question 3 6.3Dr unrelated question 15 31.3Patient 23 47.9Kin 2 4.2Expression pattern Non-verbal cue 1 2.1Verbal cue 25 52.1Verbal cue (question) 4 8.3Concern 11 22.9Psychosocial issue 6 12.5Psychosocial issue(request)1 2.1Intensity of emotion Weak 31 63.3Strong 11 22.4Content Social 3 6.3Existential 23 47.9Intimacy 1 2.1Finances 3 6.3Treatment 1 2.1Other 3 6.3Dr Response No empathy 30 62.5Empathy level 1(minimal encourager)8 16.7Empathy level 2(responds to contentor feeling)7 14.6Empathy level 3(explores feeling)3 6.3  Table 2 Blocking and facilitating behaviour.Category Count PercentageDr facilitating behavioursLow 15 31.3Medium 19 39.6High 14 29.2Dr blocking behavioursLow 24 50.0Medium 15 31.3High 9 18.8  Table 3 Descriptive statistics for patient outcome measures.TheoreticalrangeActualrange M   SDPost-consultation anxiety (T1) 20–80 20–66 38.0 13.7Decisional conflict (T1) 12–60 12–45 23.9 7.2Satisfaction with decision (T1) 6–30 15–30 26.7 3.4Satisfaction with consultation (T1) 25–125 82–125 105.7 11.1Satisfaction with Dr SDM Skills (T1) 12–60 33–59 44.1 7.0Satisfaction with decision (T2) 6–30 21–30 26.5 3.0M: mean; SD: Standard Deviation; T1: assessment 2 weeks after the consultation;T2: assessment 4 months after the consultation.  A. Smith et al./Patient Education and Counseling 82 (2011) 233–239 236