lable at ScienceDirect

Social Science & Medicine 70 (2010) 763–768

Contents lists avai

Social Science & Medicine

journal homepage: www.elsevier .com/locate/socscimed

Interpersonal perception in the context of doctor–patient relationships: A dyadicanalysis of doctor–patient communication

David A. Kenny a, Wemke Veldhuijzen b, Trudy van der Weijden b, Annie LeBlanc c, Jocelyn Lockyer d,e,France Legare f,*, Craig Campbell g

a Department of Psychology, University of Connecticut, Storrs, CT, United Statesb Department of General Practice, School for Public Health and Primary Care (CAPHRI), Maastricht University, Maastricht, The Netherlandsc Research Center, Hopital Saint-François d’Assise, Centre Hospitalier Universitaire de Quebec, Quebec, Canadad Continuing Medical Education & Professional Development, University of Calgary, Calgary, Alberta, Canadae Department of Community Health Sciences, University of Calgary, Calgary, Alberta, Canadaf Universite Laval, Department of Family Medicine and Emergency Medicine, Faculty of medicine, Campus Universitaire, Quebec City, Quebec, G1K 7P4, Canadag Royal College of Physicians and Surgeons of Canada, Ottawa, Ontario, Canada

a r t i c l e i n f o

Article history:Available online 11 December 2009

Keywords:Shared decision makingRelationship-centered careDyadic data analysisDoctor–patient relationshipDoctor–patient communication

* Corresponding author. Tel.: þ1 418 525 4437; faxE-mail address: france.legare@mfa.ulaval.ca (F. Leg

0277-9536/$ – see front matter � 2009 Elsevier Ltd.doi:10.1016/j.socscimed.2009.10.065

a b s t r a c t

Doctor–patient communication is an interpersonal process and essential to relationship-centered care.However, in many studies, doctors and patients are studied as if living in separate worlds. This studyassessed whether: 1) doctors’ perception of their communication skills is congruent with their patients’perception; and 2) patients of a specific doctor agree with each other about their doctor’s communicationskills. A cross-sectional study was conducted in three provinces in Canada with 91 doctors and their 1749patients. Doctors and patients independently completed questions on the doctor’s communication skills(content and process) after a consultation. Multilevel modeling provided an estimate of the patient anddoctor variance components at both the dyad-level and the doctor-level. We computed correlationsbetween patients’ and doctors’ perceptions at both levels to assess how congruent they were. Consensusamong patients of a specific doctor was assessed using intraclass correlation coefficient (ICC). The meanscore of the rating of doctor’s skills according to patients was 4.58, and according to doctors was 4.37. Thedyad-level variance for the patient was .38 and for the doctor was .06. The doctor-level variance forthe patient ratings was .01 and for the doctor ratings, .18. The correlation between both the patients’ andthe doctors’ skills’ ratings scores at the dyad-level was weak. At the doctor-level, the correlation was notstatistically significant. The ICC for patients’ ratings was .03 and for the doctors’ ratings .76. Overall, thisstudy suggests that doctors and their patients have a very different perspective of the doctors’communication skills occurring during routine clinical encounters.

� 2009 Elsevier Ltd. All rights reserved.

Introduction

If the focus of the 21st-century health care system is to be thepatient, then the value of the doctor–patient relationship must bepromoted (Frist, 2005). In this context doctor–patient communi-cation is regaining interest because it is expected to lead toimproved patient outcomes (Stewart, 1995) and fewer complaintsfrom patients regarding medical practice (Tamblyn et al., 2007).However, researchers are still exploring both what makescommunication effective and the underlying mechanisms by whichpatients’ and providers’ outcomes are affected (Street, Makoul,Arora, & Epstein, 2009). Although doctor–patient communication is

: þ1 418 525 4194.are).

All rights reserved.

considered an interpersonal process (Street, O’Malley, Cooper, &Haidet, 2008) and an essential component of relationship-centeredcare (Pew-Fetzer Task Force on Advancing Psychosocial HealthEducation, 1994), it tends to be operationalized as a set of behaviorsenacted by only one member of the dyad (Bensing & Dronkers,1992; Street, 1992). If indeed doctor–patient communication isa process by which a transmitter and a receiver of messages interactin a defined social context, then gaining more insight into thisinterpersonal process would be desirable (Bennett, 1976).

In the last decade, a vast body of research has advanced ourunderstanding of how individuals influence each other in thecontext of relationships (Roter & Hall, 2006). Even when themeeting is brief, as in medical consultations, individuals involved insuch relationships have the potential to influence each other’scognitions, emotions and behaviors in a reciprocal way (Kenny &Cook, 1999; LeBlanc, Kenny, O’Connor, & Legare, 2009; Legare et al.,

D.A. Kenny et al. / Social Science & Medicine 70 (2010) 763–768764

2008). Nonetheless, many doctor–patient communication studieshave assessed patients or doctors separately but not the interper-sonal dynamics. If the doctor–patient interaction is an interper-sonal system (Street, Gordon, & Haidet, 2007; Street et al., 2008),then the two participants need to be considered simultaneously(Kenny, 1994). Consequently, this paper addresses the value ofusing a dyadic data analysis approach to the study of doctor–patient communication (Kenny, Kashy, & Cook, 2006).

Interpersonal perception in the context of doctor–patientrelationships

In recent years, there has been increased interest in addressinggaps in mutual understanding of knowledge and values betweenphysicians and their patients (O’Connor et al., 1998; Towle & God-olphin, 1999). An examination of agreement between the patientperspective and the provider perspective identifies these gaps. Thisis important because a shared perspective is positively associatedwith resolution of problems and symptoms (Bass et al., 1986;Cedraschi et al., 1996; Gabbay et al., 2003; Starfield et al., 1981),satisfaction with physician (Krupat et al., 2000) and with the clin-ical encounter (Fagerberg, Kragstrup, Stovring, & Rasmussen, 1999),trust in and endorsement of the physician’s recommendations(Krupat, Bell, Kravitz, Thom, & Azari, 2001), adherence to treatment(Sewitch et al., 2003), and the patient’s assessment of self-management and self-efficacy when faced with a chronic disease(Heisler et al., 2003).

Some factors influencing patient–provider agreement havebeen identified as modifiable, some of which could be the target ofintervention. These include: the number and nature of problemsassessed during the clinical encounter (Freidin, Goldman, & Cecil,1980), education of patient (Taylor, Burdette, Camp, & Edwards,1980), number of prescription medications (Freidin et al., 1980),ordering of tests (Freidin et al., 1980), and type of medication(Bikowski, Ripsin, & Lorraine, 2001). In addition, a lack of trust andagreement between patients and physicians contributes to frus-trating visits from the physicians’ perspective and thus couldhamper the quality of care they provide (Levinson, Stiles, Inui, &Engle, 1993). This potential for disagreement is consistent with datafrom a study in general internal medicine suggesting that patientsand physicians form their opinion about a consultation in differentways (Zandbelt, Smets, Oort, Godfried, & de Haes, 2004).

Nonetheless, there are several challenges in advancing knowl-edge in this area. Conceptualization and operationalization ofeffective communication as an interpersonal and interdependentprocess (i.e., when those involved influence each other) betweendoctors and patients have important consequences. First, itemphasizes the need for concept definitions that are congruentwith an interpersonal and interdependent process. In particular,consensus among individuals refers to whether two or more indi-viduals agree on their assessment of a common target. For example,do two patients who are treated by the same physician agree witheach other about their interactions with the physician? If there isstrong consensus in patients’ perceptions, then interventions needto be targeted at doctors who have poor outcomes. If there is littleor no consensus, then we need to understand what leads a doctor tosometimes have good outcomes and sometimes have pooroutcomes. Alternatively, self-other agreement (Kenny, 1994) refersto the correspondence between how a person sees him or herself ininteraction with another and how others see that person. Forexample, if a doctor thinks he or she has good communication skillswith a given patient, does the patient agree with that assessment?Particularly important here would be determining whethera doctor thought he or she had skills, but in fact his or her patientsdisagreed with this assessment. This distinction is important

because it suggests that different underlying relationship processesare operating to form these two different perspectives.

From a methodological point of view, the measurement ofeffective communication as an interpersonal and interdependentprocess requires dyadic measurements and analytical methods.Both patients and physicians would be required to assess the sameconsultation or skills. In a review of instruments that assess theperception of physicians on decision making in specific clinicalencounters, eleven instruments were identified (Legare, Moher,Elwyn, LeBlanc, & Gravel, 2007). Five of the six most recentlydeveloped instruments measured both doctors’ and patients’perceptions of the same phenomenon, suggesting that a dyadicapproach to the clinical encounter is gaining in popularity. Inter-estingly, two of the dyadic instruments addressed communicationskills (Campbell, Lockyer, Laidlaw, & Macleod, 2007; Cegala, Cole-man, & Turner, 1998).

Research questions

Using data from a previous study that had evaluated thepsychometrics of a dyadic assessment tool of doctors’ communi-cation skills, we examined the level of consensus between doctorsand their patients in routine clinical encounters and amongpatients of the same doctor (Campbell et al., 2007). More specifi-cally, our research questions were: 1) Do patients of a specificdoctor agree on this doctor’s communication skills? 2) Are doctors’self-perception of their communication skills congruent withpatients’ perceptions of them?

Method

Data source and participants

Data originated from a cross-sectional study carried outbetween January and May 2005 and for which the overall goal wasto develop, test and psychometrically assess a dyadic instrument onthe process and the content of communication from the perspec-tives of both doctors and patients (Campbell et al., 2007). Briefly,family doctors and specialists (n ¼ 91) from three provinces inCanada and their patients (n¼ 1749) completed a post-consultationquestionnaire that included the Matched-Pair Instrument (MPI).Doctors were recruited through an invitation letter from theCollege of Family Physicians of Canada and the Royal College ofPhysicians and Surgeons of Canada. Patients were then recruited bytheir own participating physician. Both doctors and patientsparticipated voluntarily and did not receive any financial incen-tives. The number of patients per doctor was a little more than 19,ranging from 2 to 25. Ethical approval was provided by theUniversity of Ottawa Ethics Board.

Data collected

Basic socio-demographic characteristics were collected for bothdoctors (i.e., gender and medical specialty) and patients (i.e., gender).Doctors’ and patients’ perceptions on doctors’ communication skillswere measured using the MPI, a dyadic instrument that wasdeveloped based on several existing instruments and expertise ofdevelopers in communication skills (Campbell et al., 2007). The MPIis comprised of 19 items that assess both the content (4 items) andthe process (15 items) of doctors’ communication skills. A copy of theMPI is included in Appendix 1. The doctor’s version assesses self-perception while the patient’s version assesses his or her perceptionof the doctor’s communication skills. Each item of the MPI is scoredon a 5-point Likert scale (1¼ strongly disagree – 5¼ strongly agree).The overall score for the MPI is the mean of the 19 ratings. Higher

D.A. Kenny et al. / Social Science & Medicine 70 (2010) 763–768 765

scores refer to higher communication skills (i.e., the physician wasperceived or perceived him or herself as more competent incommunication). Internal consistency coefficients (Cronbach alpha)for the MPI are .69 for the patient version and .70 for the doctorversion (Campbell et al., 2007). Both doctors and patients completedtheir own version of the MPI immediately after a consultation andwere blinded to each other’s answers. Therefore, doctors rated manyconsultations (i.e., for each patient) but patients only rated oneconsultation, their own. (There was no audiotape of the consulta-tion.) Detailed information regarding the original study may befound elsewhere (Campbell et al., 2007).

Data analysis

The existing dataset was by nature dyadic and, more specifically,represented a one-with-many (i.e., many patients nested withinone doctor), reciprocal design (i.e., doctors and patients aremeasured) with distinguishable members (i.e., members of thedyad can be distinguished as patient and doctor) (Kenny et al.,2006). A multilevel modeling framework was used to estimate thedifferent variance components (Kenny et al., 2006). This frameworkpartitions variance into two levels: 1) dyadic-level (within vari-ance), also known as the relationship-level; and 2) the doctor-level(between variance). Therefore, to answer our first and secondresearch questions, the variance components were estimated fromthe dyad-level (within variance) and the doctor-level (betweenvariance) for both doctor and patient ratings. At the dyad-level, thetwo variance components (patient and doctor) represent theassessment of the communication skills as seen by patients ordoctors removing the mean for doctors. At this level, the patientvariance represents how much variability there is on the MPI scoresof patients nested within doctors (i.e., higher values indicate thatpatients’ MPI scores vary considerably from one patient to anotherpatient for one doctor). Likewise, the variance in doctor ratingsrepresents how much variability there is on the MPI score ofa doctor with their patients (i.e., higher values indicate that doctors’MPI scores vary considerably from one patient to another patientwithin one doctor). At the doctor-level, the variance of patientratings represents how much variability there is on the MPI scoresof patients between doctors, that is, from one doctor to anotherdoctor (i.e., higher values indicate that patients’ MPI scores varyconsiderably from one doctor to another doctor and so indicatepatient consensus about their doctors). At this level, the variance indoctor ratings represents how much variability there is on the MPIscore of doctors from one doctor to another doctor (i.e., highervalues indicate that doctors’ MPI scores vary considerably from onedoctor to another doctor).

Table 1Estimates of variance and correlation parameters.

Parameter level Definition

Dyad (withinvariance)

How much variability there is on the MPI scores of patients withinpatients’ MPI scores vary considerably from one patient to anothe

How much variability there is on the MPI score of doctors withindoctors’ MPI scores do vary considerably from one patient to anot

Dyadic reciprocity: Within a doctor, are doctors’ MPI scores relate

Person (betweenvariance)

How much variability there is on the MPI scores of patients betweanother doctor (i.e., higher values indicate that patients’ MPI scoranother doctor).

How much variability there is on the MPI score of doctor from onvalues indicate that doctors’ MPI scores vary considerably from on

Whether the mean doctor’s mean MPI scores is related to the avera

From the multilevel modeling, we computed two intraclasscorrelation coefficients (ICC): one for the patient data and one forthe doctor data. The ICC for the patient represents the proportionof the variance due to doctor on the MPI score of patients. A highervalue indicates high consistency among the patients of a doctor,i.e., the within doctor variation is relatively smaller than thebetween doctor variation on the MPI score of patients; thus, thereis less variability on the patients’ MPI score from one patient toanother patient of the same doctor than from one patient of onedoctor to another patient of another doctor. The ICC for the doctorrepresents the proportion of the variance due to doctor for thedoctors’ MPI score. A higher value indicates that there is morevariability from one doctor to another doctor on the MPI scoresthan from one MPI score of one doctor to another MPI score of thesame doctor.

To answer question 2 we correlated each of the componentsacross doctors and patients. We could then determine if doctorswho generally felt they had good communication skill were ratedas such by patients who felt the same way. We could also assesswhether a doctor who felt he or she had good communicationwith one patient and not with another was rated as such by thepatients.

We conducted analyses on different factors of the MPI, and wefound essentially the same results; thus, only the composite score isanalyzed here. Also, in the dataset that was used, there wasminimal information in the dataset about patient and doctorcharacteristics. One variable was whether physicians were generalpractitioners or specialists. However, we found only small differ-ences between the two.

Results

Characteristics of participants

Ninety-one physicians contributed 1749 patients of whom 1059were female and 594 male. There were 58 general practitioners (28females and 17 males) and 43 medical specialists (8 females and 29males). The mean score of the MPI for patients was 4.58 (CI 95%4.54–4.61) and for physicians, 4.37 (CI 95% 4.28–4.46), thus sug-gesting that overall patients tended to rate doctors higher thandoctors rated themselves. Details regarding means for individualitems may be found elsewhere (Campbell et al., 2007).

Do patients of a specific doctor reach a consensus on this doctor’communication skills?

Table 1 shows the different variance components that werecomputed at both the dyad-level and the doctor-level. The first two

Estimate Standard error p value

doctors (i.e., higher values indicate thatr patient within a doctor).

.382 .013 <.001

doctors (i.e., higher values indicate thather patient within a doctor).

.056 .002 <.001

d to patients’ MPI scores?. .131 .024 <.001

en doctors, that is, from one doctor toes vary considerably from one doctor to

.014 .005 <.001

e doctor to another doctor (i.e., highere doctor to another doctor).

.180 .027 <.001

ge score from his/her own patients’ MPI. .123 .167 .460

D.A. Kenny et al. / Social Science & Medicine 70 (2010) 763–768766

dyad-level variance terms refer to relationship variance andrepresent the assessment of the communication skills as seen bythe patient or doctor, removing the average rating of that doctor.Note that there is substantially more relationship variance for thepatient than for the doctor (.382 versus .056), which implies thatpatients’ MPI scores vary considerably from one patient to anotherpatient within doctors. In contrast, the doctors’ MPI scores do notvary much from one patient to another patient within doctors. Inother words, doctors perceived themselves to be more consistentfrom one patient to another patient than patients perceived doctorsto be.

Next we consider variation in the mean perception of doctors.The variance in these means for patients’ MPI is much smaller thanthe variance for doctors (.014 versus .180). This implies thatpatients’ MPI scores do not vary much from one doctor to anotherdoctor. In contrast, doctors’ MPI scores vary considerably from onedoctor to another doctor. In other words, some doctors think thatthey are generally good at communicating with their patients andother doctors think that they are generally quite poor.

The ICC for patients is only .034. Thus, patients of the samedoctor do not agree very much with each other about the level ofcommunication skills of their physician as revealed during theconsultation. The ICC for the doctor is very large: .763. This meansthat overall doctors perceive their level of communication skills tobe the same with all of their patients.

Are doctors’ self-perceptions of their communication skillscongruent with how their patients perceive them?

If we remove the effect of the doctor (i.e., examine the rela-tionship-level), the correlations between the patient MPI scoreand the doctor MPI score is weak (r ¼ .13, p < .001). Thus, despitesome agreement, patients and doctors have different views aboutdoctors’ communication skills during their interaction. Thecorrelation between the patient MPI score and the doctor MPIscore at the doctor-level is also weak and not significantlydifferent from zero (r ¼ .123, p ¼ .460). Therefore given thesesmall correlations, we found physicians did not agree with theirpatients about the level of their own communication skills. Forexample, providers who think they are especially good atcommunicating with their patients are not necessarily seen thatway by their patients. The same held true when providers’perceptions were negative.

Discussion

This study measured and analyzed the interpersonal percep-tions and the interdependent processes that occur betweendoctors and patients during routine clinical encounters. Overall,they suggest that doctors’ self-perceptions of their communicationskills are not congruent with how their patients perceive them.More specifically, they suggest that: 1) doctors tend to perceivethemselves to be more alike from one patient to another patientthan patients perceive doctors to be; 2) from one doctor to anotherdoctor, doctors see themselves quite differently but patients donot see them so differently; 3) patients of the same doctor do notagree very much with each other about the level of communica-tion skills of their doctors; and 4) some doctors’ scores did notcorrelate better with their patients’ scores of their communicationskills than with scores received by other doctors. There are severalexplanations that can help interpret what has been observed inthis study.

Doctors’ and patients’ perceptions of the medical encounterhave been characterized as being so different that they appear to befrom different worlds (Mishler, 1984). During a consultation neither

the doctor nor patient are likely to focus on doctors’ communica-tion skills. The patient is likely to be concerned about presenting hisor her complaints, the nature of his or her diagnosis, the treatmentoptions that may be available and the prognosis of his or hercondition. The doctor is likely to be concerned about making theright diagnosis, finding the right treatment and explaining thecourse of action to the patient. As pointed out by Hall, Murphy, andMast (2007), ‘‘monitoring one’s own nonverbal behavior may bea difficult task because a person is simultaneously a sender anda receiver, and therefore has to plan and execute his or hernonverbal and verbal behavior, process the other person’snonverbal and verbal behavior, and engage in metacognitiveactivity such as asking oneself how the interaction is going’’ (Hallet al., 2007, p.1675). Therefore, the perception of both doctors andpatients about the doctor’s communication skills are likely todiverge (DePaulo, Kenny, Hoover, Webb, & Oliver, 1987) or to belimited to issues such as establishing whether the doctor has beenable to clarify information or to reassure a patient. This may raisesome concern because a shared perspective between doctors andpatients has been said to be positively associated with patientoutcomes (Krupat et al., 2001; Krupat et al., 2000; Sewitch et al.,2003; Starfield et al., 1981).

The results of this study are consistent with what has beenreported in the literature. Although doctors and patients agree onthe core competencies regarding the physicians’ communicationskills, they usually do not agree on the actual presence of theseskills in consultations (Cegala, McNeilis, McGee, & Jonas, 1995).Moreover, patients’ assessment of doctors’ communication skillsdiffer from assessments made by experts. More importantly, bothpatients’ assessment and experts’ assessment of doctors’ commu-nication skills differ from assessments made by other doctors. Thisis worrisome because patient-centered communication appears toinfluence patients’ outcomes through patients’ perceptions and notthrough perceptions of others, including that of a third observer(Stewart et al., 2000). Therefore, it would be important to keepmonitoring the experience of patients regarding their consultationwith a doctor.

We also found that there was little agreement across patientsfrom the same physician on the level of communication skills oftheir doctor, although doctors themselves estimate their level ofcommunication skills to be almost similar for every patient.Patients of the same doctor showed little agreement with otherpatients of that doctor about his or her level of communicationskills. Few factors combine to determine consensus among a groupof individuals (Kenny, 1994). One of these is observation of thesame consultation. In this study, each patient had assessed his orher doctor following his or her own specific consultation. Anotherpotential factor that might have contributed to low consensus issimilarity of meaning systems or the extent to which differentpatients understand the interaction. Low consistency (i.e., thedegree to which the doctor exhibited very different communica-tive behaviors between his/her patients) might have also contrib-uted since physicians may adapt their behavior to patients’specificities.

In contrast, we observed that doctors’ self-ratings were quiteconsistent across a number of consultations with different patients.As reported by Kenny (1994), ‘‘(S)elf-ratings carry a great deal ofexcess baggage; that is, they measure other things besides how theperson truly is’’ (p. 202). In this study, although doctors in generalrated themselves more severely than did their patients, they mayhave overestimated the consistency of their communication fromone patient to another one. This would be in line with a review ofICCs from several datasets of primary and secondary care studies inthe U.K. (Campbell, Grimshaw, & Steen, 2000). This review reportedthat overall, ICCs at the level of the doctors for process variables in

Patients’ items Physicians’ items

This doctor: I:Greeted me in a way that made me

feel comfortableGreeted the patient in a way thatmade them feel comfortable

Discussed my reason(s) for comingtoday.

Discussed the patients reason(s) forcoming today.

Encouraged me to express mythoughts concerning myhealth problems.

Encouraged the patient to expresstheir thoughts concerning theirhealth problem.

Listened carefully to what I hadto say.

Listened carefully to what thepatient had to say.

Understood what I had to say. Understood what the patient hadto say.

If a physician examination wasrequired for your healthconcerns, the doctor fullyexplained what

If a physical examination was required,explained what was done and why.

Explained the lab tests needed(e.g., Blood, X-rays, ultrasound,etc).

Explained the lab tests needed (e.g.,Blood, X-rays, ultrasound, etc.) toexplore the patient’s problem’s.

Discussed treatment optionswith me.

Discussed treatment optionswith patient.

Gave me as much informationas I wanted.

Gave the patient as much informationas they wanted.

Checked to see if the treatmentplan(s) was acceptable to me.

Checked with patient to see if thetreatment plan(s) was acceptable.

Explained medications, if any,including possible side effects.

Explained medications, if any, includingpossible side effects.

Encouraged me to ask questions. Encouraged the patient to ask questions.

Responded to my questions andconcerns.

Responded to the patient’s questionsand concerns.

Involved me in decisions as muchas I wanted.

Involved the patient in decisions asmuch as they wanted.

Discussed next steps includingany follow-up plans.

Discussed next steps including anyfollow-up plans.

Checked to be sure I understoodeverything.

Checked to be sure the patientunderstood everything.

Showed care and concern aboutme as a person.

Showed care and concern aboutthe patient as a person.

Spent the right amount of timewith me.

Spent the right amount of timewith the patient.

Overall, I was satisfied with myvisit to the doctor today.

Overall, I was satisfied with thisconsultation today.

D.A. Kenny et al. / Social Science & Medicine 70 (2010) 763–768 767

primary care as observed by a third observer were of the order of.05–.15, whilst those in secondary care were of the order of .30 andthus much lower than the observed .76 in this study. Also, patientsmay have ‘‘missed’’ some of the consistency in the doctor’s behaviorbecause they were too focused on their own needs (DePaulo et al.,1987; Hall et al., 2007).

Lastly, we also observed that some doctors’ scores did notcorrelate with their patients’ scores of their communication skills.This suggests that the misperception that we have found is chronicand endemic in doctor–patient interactions. This is of someconcern because the doctor–patient relationship is one of thepillars of patient-centered care (Stewart, 1995). Because a rela-tionship would imply a minimum of interdependence betweendoctors and their patients, it is somehow surprising that in thisstudy, all doctors have little overlap with their patients’ percep-tions. As discussed above, it is possible that this ‘‘relationship’’ ispresent in other components of the consultation or, in the worst-case scenario, non-existent.

This study has some limitations. First, this is a secondary anal-ysis of an existing dataset from a cross-sectional study for whichthe overall goal was to develop, test and psychometrically assessa dyadic instrument on the communication skills of doctors(Campbell et al., 2007). However, given the paucity of large dyadicreciprocal datasets in the field of doctor–patient communication,we believe that this study sheds some interesting light that willhelp in the design of future studies. Second, the means of the MPIscores of the doctors and the patients are quite high, being close tothe maximum possible value which may have affected our results.However, high scores have been reported before in doctor–patientcommunication research (Makoul, Krupat, & Chang, 2007). Third,physicians recruited patients and might have been more inclined todo so with patients they felt most comfortable with. Therefore, theresults from this study would be an overestimation of the realagreement found in clinical encounters. However, our results arequite consistent with those reported in a previous study in whichpatients were recruited by the office staff (Makoul et al., 2007).Another limitation of this study might reside in the diversity of theitems of the MPI. For example, the item ‘‘Discussed treatmentoptions with the patient (doctor)/Discussed treatment options withme (patient)’’ seems to measure agreement on the occurrence of anevent. Deficiencies in memory of events, similar to those describedby witnesses, might account for dissimilarities here even thoughthe questionnaire was administered immediately after theconsultation. Also, the consultation was not audiotaped, thushampering our ability to assess how the patients’ and the physi-cians’ perspectives correlated with a third observer perspective(Saba et al., 2006).

Overall, this study suggests that doctors and their patients havea very different perspective of doctors’ communication skillsoccurring during routine clinical encounters. Moreover, we foundvery little or no interdependence between doctors and their patientson these aspects. Given the importance of the doctor–patient rela-tionship in patient-centered care, future studies will need to assess ifmutuality is present or not on other components of the consultation.

Acknowledgements

We thank Toni Laidlaw, Chair and other members of the steeringcommittee that designed and oversaw the administration of thedata collection in the original study, including Heather Macleod,Francois Goulet, Suzanne Kurtz, Gordon Page, and David Blackmore.We explicitly acknowledge the financial support from the MedicalCouncil of Canada and the Canada Research Chair in Implementa-tion of Shared Decision Making in Primary Care. There are nopotential conflicts of interest.

Appendix 1. Patients’ and physicians’ questionnaires.

References

Bass, M. J., Buck, C., Turner, L., Dickie, G., Pratt, G., & Robinson, H. C. (1986). Thephysician’s actions and the outcome of illness in family practice. Journal ofFamily Practice, 23(1), 43–47.

Bennett, A. E. E. (1976). Communication between doctors and patients. 200 MadisonAve., New York, New York 10016: Oxford University Press.

Bensing, J. M., & Dronkers, J. (1992). Instrumental and affective aspects of physicianbehavior. Medical Care, 30(4), 283–298.

Bikowski, R. M., Ripsin, C. M., & Lorraine, V. L. (2001). Physician–patient congruenceregarding medication regimens. The Journal of the American Geriatrics Society,49(10), 1353–1357.

Campbell, M., Grimshaw, J., & Steen, N. (2000). Sample size calculations for clusterrandomised trials. Changing professional practice in Europe group (EU BIOMEDII Concerted Action). Journal of Health Services Research and Policy, 5(1), 12–16.

Campbell, C., Lockyer, J., Laidlaw, T., & Macleod, H. (2007). Assessment of a matched-pair instrument to examine doctor–patient communication skills in practisingdoctors. Medical Education, 41(2), 123–129.

D.A. Kenny et al. / Social Science & Medicine 70 (2010) 763–768768

Cedraschi, C., Robert, J., Perrin, E., Fischer, W., Goerg, D., & Vischer, T. L. (1996). Therole of congruence between patient and therapist in chronic low back painpatients. Journal of Manipulative Physiological Therapeutics, 19(4), 244–249.

Cegala, D. J., Coleman, M. T., & Turner, J. W. (1998). The development and partialassessment of the medical communication competence scale. Health Commu-nication, 10(3), 261–288.

Cegala, D. J., McNeilis, K. S., McGee, D. S., & Jonas, A. P. (1995). A study of doctors’and patients’ perceptions of information processing and communicationcompetence during the medical interview. Health Communication, 7(3),179–203.

DePaulo, B. M., Kenny, D. A., Hoover, C. W., Webb, W., & Oliver, P. V. (1987). Accuracyof person perception: do people know what kinds of impressions they convey?Journal of Personality and Social Psychology, 52, 303–315.

Fagerberg, C. R., Kragstrup, J., Stovring, H., & Rasmussen, N. K. (1999). How well dopatient and general practitioner agree about the content of consultations?Scandinavian Journal of Primary Health Care, 17(3), 149–152.

Freidin, R. B., Goldman, L., & Cecil, R. R. (1980). Patient–physician concordance inproblem identification in the primary care setting. Annals of Internal Medicine,93(3), 490–493.

Frist, W. H. (2005). Shattuck lecture: health care in the 21st century. New EnglandJournal of Medicine, 352(3), 267–272.

Gabbay, M., Shiels, C., Bower, P., Sibbald, B., King, M., & Ward, E. (2003). Patient–practitioner agreement: does it matter? Psychological Medicine, 33(2), 241–251.

Hall, J. A., Murphy, N. A., & Mast, M. S. (2007). Nonverbal self-accuracy in inter-personal interaction. Personality and Social Psychology Bulletin, 33(12),1675–1685.

Heisler, M., Vijan, S., Anderson, R. M., Ubel, P. A., Bernstein, S. J., & Hofer, T. P. (2003).When do patients and their physicians agree on diabetes treatment goals andstrategies, and what difference does it make? Journal of General InternalMedicine, 18(11), 893–902.

Kenny, D. A. (1994). Interpersonal perception: A social relations analysis. New York:Guilford Press.

Kenny, D. A., & Cook, W. (1999). Partner effects in relationship research: conceptualissues, analytic difficulties, and illustrations. Personal Relationships, 6, 433–448.

Kenny, D. A., Kashy, D. A., & Cook, W. (2006). Dyadic data analysis. New York: TheGuilford Press.

Krupat, E., Bell, R. A., Kravitz, R. L., Thom, D., & Azari, R. (2001). When physicians andpatients think alike: patient-centered beliefs and their impact on satisfactionand trust. Journal of Family Practice, 50(12), 1057–1062.

Krupat, E., Rosenkranz, S. L., Yeager, C. M., Barnard, K., Putnam, S. M., & Inui, T. S.(2000). The practice orientations of physicians and patients: the effect ofdoctor–patient congruence on satisfaction. Patient Education and Counseling,39(1), 49–59.

LeBlanc, A., Kenny, D. A., O’Connor, A. M., & Legare, F. (2009). Decisional conflict inpatients and their physicians: a dyadic approach to shared decision making.Medical Decision Making, 29(1), 61–68.

Legare, F., Elwyn, G., Fishbein, M., Fremont, P., Frosch, D., Kenny, D. A., et al. (2008).Translating shared decision-making into health care clinical practices: proof ofconcepts. Implementation Science, 3(1), 2.

Legare, F., Moher, D., Elwyn, G., LeBlanc, A., & Gravel, K. (2007). Instruments toassess the perception of physicians in the decision-making process of specificclinical encounters: a systematic review. BMC Medical Informatics and DecisionMaking, 7(1), 30.

Levinson, W., Stiles, W. B., Inui, T. S., & Engle, R. (1993). Physician frustration incommunicating with patients. Medical Care, 31(4), 285–295.

Makoul, G., Krupat, E., & Chang, C. H. (2007). Measuring patient views of physiciancommunication skills: development and testing of the communication assess-ment tool. Patient Education and Counseling, 67(3), 333–342.

Mishler, E. G. (1984). The discourse of medicine: Dialectics of medical interviews. NewJersey: Ablex Publishing Corporation.

O’Connor, A. M., Tugwell, P., Wells, G. A., Elmslie, T., Jolly, E., Hollingworth, G., et al.(1998). A decision aid for women considering hormone therapy after meno-pause: decision support framework and evaluation. Patient Education andCounseling, 33(3), 267–279.

Pew-Fetzer Task Force on Advancing Psychosocial Health Education. (1994). InC. Tresolini (Ed.), Health professions education and relationship–Centered care.San Francisco, CA: Pew-Health Professions Commission Fetzer Institute.

Roter, D., & Hall, J. (2006). Doctors talking with patients/patients talking with doctors:Improving communication in medical visits. Westport, CT: Praeger Publishers.

Saba, G. W., Wong, S. T., Schillinger, D., Fernandez, A., Somkin, C. P., Wilson, C. C., etal. (2006). Shared decision making and the experience of partnership inprimary care. Annals of Family Medicine, 4(1), 54–62.

Sewitch, M. J., Abrahamowicz, M., Barkun, A., Bitton, A., Wild, G. E., Cohen, A., et al.(2003). Patient nonadherence to medication in inflammatory bowel disease.American Journal of Gastroenterology, 98(7), 1535–1544.

Starfield, B., Wray, C., Hess, K., Gross, R., Birk, P. S., & D’Lugoff, B. C. (1981). Theinfluence of patient–practitioner agreement on outcome of care. AmericanJournal of Public Health, 71(2), 127–131.

Stewart, M. (1995). Effective physician–patient communication and healthoutcomes: a review. Canadian Medical Association Journal, 152(9), 1423–1433.

Stewart, M., Brown, J. B., Donner, A., McWhinney, I. R., Oates, J., Weston, W. W., et al.(2000). The impact of patient-centered care on outcomes. Journal of FamilyPractice, 49(9), 796–804.

Street, R. L. (1992). Communicative styles and adaptations in physician–parentconsultations. Social Science & Medicine, 34(10), 1155–1163.

Street, R. L., Gordon, H., & Haidet, P. (2007). Physicians’ communication andperceptions of patients: is it how they look, how they talk, or is it just thedoctor? Social Science & Medicine, 65(3), 586–598.

Street, R. L., Jr., Makoul, G., Arora, N. K., & Epstein, R. M. (2009). How doescommunication heal? Pathways linking clinician–patient communication tohealth outcomes. Patient Education and Counseling, 74(3), 295–301.

Street, R. L., O’Malley, K. J., Cooper, L. A., & Haidet, P. (2008). Understandingconcordance in patient–physician relationships: personal and ethnic dimen-sions of shared identity. Annals of Family Medicine, 6(3), 198–205.

Tamblyn, R., Abrahamowicz, M., Dauphinee, D., Wenghofer, E., Jacques, A., Klass, D.,et al. (2007). Physician scores on a national clinical skills examination aspredictors of complaints to medical regulatory authorities. Journal of theAmerican Medical Association, 298(9), 993–1001.

Taylor, R. B., Burdette, J. A., Camp, L., & Edwards, J. (1980). Purpose of the medicalencounter: identification and influence on process and outcome in 200 encountersin a model family practice center. Journal of Family Practice, 10(3), 495–500.

Towle, A., & Godolphin, W. (1999). Framework for teaching and learning informedshared decision making. British Medical Journal, 319(7212), 766–771.

Zandbelt, L. C., Smets, E. M., Oort, F. J., Godfried, M. H., & de Haes, H. C. (2004).Satisfaction with the outpatient encounter: a comparison of patients’ andphysicians’ views. Journal of General Internal Medicine, 19(11), 1088–1095.