Are Physicians and Patients in Agreement? Exploring Dyadic Concordance

Abstract

Dyadic concordance in physician–patient interactions can be defined as the extent of agreement between physicians and patients in their perceptions of the clinical encounter. The current research specifically examined two types of concordance: informational concordance—the extent of agreement in physician and patient responses regarding patient information (education, self-rated health, pain); and interactional concordance—the extent of physician–patient agreement regarding the patient’s level of confidence and trust in the physician and the perceived quality of explanations concerning diagnosis and treatment. Using a convenience sample of physicians and patients (N = 50 dyads), a paired survey method was tested, which measured and compared physician and patient reports to identify informational and interactional concordances. Factors potentially related to dyadic concordance were also measured, including demographic characteristics (patient race, gender, age, and education) and clinical factors (whether this was a first visit and physician specialty in family medicine or oncology). The paired survey showed informational discordances, as physicians tended to underestimate patients’ pain and overestimate patient education. Interactional discordances included overestimating patients’ understanding of diagnosis and treatment explanations and patients’ level of confidence and trust. Discordances were linked to patient dissatisfaction with physician listening, having unanswered questions, and feeling the physician had not spent enough time. The paired survey method effectively identified physician–patient discordances that may interfere with effective medical practice; this method may be used in various settings to identify potential areas of improvement in health communication and education.

Keywords

dyadic concordance dyadic measurement physician–patient communication

The physician–patient relationship can be defined as co-constructed: Both clinicians and patients have an interest in and responsibility for the process and outcomes of their communication (Beach & Thomas, 2006). Dyadic concordance, or the extent of agreement between patients and physicians, reflects how well they have achieved mutual understanding through communication. Positive health outcomes have been linked with greater physician–patient agreement (or concordance) about diagnosis, specific health problems, health care priorities (Liaw, Young, & Farish, 1996), and health beliefs (Street & Haidet, 2011), but less attention has been paid to other potential disagreements in clinical interactions.

The purpose of this exploratory study was to develop a paired survey method to assess how effectively clinical information about patients was conveyed (i.e., informational concordance) and the extent of shared perceptions of the medical encounter (i.e., interactional concordance). The paired survey method was used to elicit and compare physician and patient responses after the medical consultation and explore factors related to discordances.

Dyadic Concordance in Health Care Research

The current research draws on relational theory in communication research, which emphasizes the co-construction of relationships through interpersonal communication. Relationships between doctors and patients are created, maintained, and continually modified through interaction as the actors interpret and respond to the situation (Walker, Arnold, Miller-Day, & Webb, 2002). Effective co-construction depends on accurate perceptions and clear communication; dyadic concordance reflects the successful exchange of information as well as mutual construction of shared perceptions of the interaction, particularly with regard to trust, confidence, and comprehension (see Arnold & Arnold, 1999). Discordances may shape what is (or is not) said or asked in the encounter, influence nonverbal behavior and interpretations of nonverbal cues, and interfere with effective communication, yet physicians are often unaware of their limitations or how they are perceived by patients (Kenny et al., 2010).

The concept of concordance has been used in various ways in health research. Most basically, concordance has referred to matching demographic characteristics between physicians and patients (Street, O’Malley, Cooper, & Haidet, 2008). For example, gender discordances have been linked to clinician behaviors and styles (Bertakis, Franks, & Epstein, 2009; Gross, McNeill, Davis, Lay-Yee, Jatrana, & Crampton, 2008), and racial-ethnic (and language) discordances have been linked to barriers in care (e.g., August, Nguyen, Ngo-Metzger, & Sorkin, 2011; Gany et al., 2011). Demographic concordance is assumed to play a role in physician–patient communication, patient-centered approaches, and health outcomes (Schnittker & Liang, 2006; Schoenthaler, Allegrante, Chaplin, & Ogedegbe, 2012; Street et al., 2008), but other scholars have noted that evidence of positive associations between race-concordance and health outcomes is inconclusive (Meghani et al., 2009).

Other studies have examined concordance in doctor–patient agreement in the clinical process or aspects of shared decision making (e.g., involvement in treatment decisions, patient preferences, Janz et al., 2004). For example, doctor–patient agreement about diagnosis, specific health problems, and health care priorities have been linked with more positive health outcomes (Freidin, Goldman, & Cecil, 1980; Liaw, Young, & Farish, 1996), and concordance in beliefs and expectations is linked with greater patient satisfaction (Jahng, Martin, Golin, & DiMatteo, 2005; Krupat et al., 2000). More recent studies have examined concordance between patients’ preferences and doctors’ clinical styles (i.e., physician- vs. patient-centered), but this research has been limited by inconsistent definitions and measures (Sepucha & Ozanne, 2010).

Informational and Interactional Concordance Between Doctors and Patients

The current research defines dyadic concordance within the clinical interaction as the extent of agreement between physicians and patients with regard to direct information about the patient (i.e., informational concordance) and their perceptions about what happens during the interaction (i.e., interactional concordance). Informational concordance is a direct measure of how effectively important facts have been conveyed, including the patient’s education level, self-rated health, and pain. The need for informational concordance is taken for granted in medical practice: Diagnosis depends on clinicians eliciting complete information and patients fully conveying symptoms, history, and health concerns. For example, assessment of pain is important for accurate diagnosis and pain management, where underestimates of pain may lead to inadequate treatment (Benyamini, Leventhal, & Leventhal, 1999; Idler & Benyamini, 1997; Jylha, 2009).

Interactional concordance, or shared perceptions of the medical encounter, offers an indirect measure of relational processes between physician and patient. Specifically, we examine how much physicians and patients agree in their perceptions of two areas: the patient’s level of confidence and trust in the physician, and the quality of the physician’s explanations concerning diagnosis and treatment. Interactional discordances may reflect miscommunication or incorrect assumptions in physician–patient interactions. For example, if a physician overestimates the patient’s confidence and trust, there may be less emphasis and effort placed on communicating effectively and building trust in the medical interaction. We believe that dyadic concordance may represent an important mediator in the clinical communication process: It may operate as a mechanism through which potential challenges to physician–patient communication (e.g., first visits, demographic differences) result in discordances that in turn influence health care satisfaction and other potential outcomes.

Research Questions and Hypotheses

The current research focused on the following questions: (a) Can a paired survey instrument for physicians and patients effectively identify informational and interactional discordances? (b) To what extent are discordances related to patient demographic characteristics (race, gender, age, and education) or clinical aspects of the interaction (first visit and physician specialty in family medicine or oncology)? (c) To what extent are discordances predictive of patient dissatisfaction with the physician and medical encounter?

Demographic characteristics, such as race, gender, age, and education, have been linked with health communication and outcomes. Prior research has shown that physicians deliver less information and less supportive talk to racial-ethnic minority patients (Ross, Mirowsky, & Duff, 1982; Waitzkin, 1985), and patients report less involvement, less partnership with physicians, and lower levels of satisfaction (Cooper-Patrick et al., 1999). Female patients have been found to get more information with less medical jargon (Waitzkin, 1985), ask more questions and provide more information compared with male patients (Elderkin-Thompson & Waizkin, 1999; Strong et al., 2009), and have their pain more accurately diagnosed than men’s (Bertakis, Azari, & Callahan, 2004). We therefore expected to find more discordances (informational and interactional) in dyads with racial minority or male patients. Research on patient age, however, has been more mixed. Some studies have described older patients as more passive, spending less time with physicians, and as less likely to have their decision preferences assessed (Janz et al., 2004), yet younger patients reported greater dissatisfaction with physicians (DeVoe, Wallace, & Fryer, 2009). Therefore, we did not have a priori hypotheses about age and physician–patient discordance. Patients with higher education levels have been found to be more proactive, involved, and report a sense of shared responsibility, compared with less educated patients (Smith, Dixon, Trevena, Nutbeam, & McCaffery, 2009). We therefore expected to find more discordances when patient education was lower.

We also expected that aspects of the clinical encounter, specifically, whether this was a first patient visit and physician’s specialty (family medicine or oncology), would be related to discordances. First visits are a critical time for collecting information, building rapport, and getting acquainted; they generally require more time than subsequent visits and include more systematic collection of medical and social histories. For oncologists, the first visit may involve discussing a new diagnosis and treatment plan (Back, Arnold, Baile, Tulsky, & Fryer-Edwards, 2005). On the other hand, greater familiarity may enhance communication and understanding. Therefore, we tested the relationship between first visits and concordance, but without a priori expectations. With regard to medical specialty, both primary care (e.g., family medicine) and oncology have emphasized the importance of communication-related research and practice (cf. Back et al., 2005; Bertakis & Azari, 2012; Rodriguez et al., 2011), so we expected high concordance in both groups but did not have specific hypotheses about their comparison.

Finally, we tested the extent to which discordances would be predictive of patients’ satisfaction with their physician. We expected that discordances would be linked to negative patient assessments of the physician and medical encounter, including physician listening, the amount of time spent, having unanswered questions, and feeling that the physician did not understand their background and values.

Method

Survey Questions and Measures

We created a paired survey, using parallel questions asked of both physicians and patients, to measure dyadic concordance. Physician and patient questionnaires included the following six paired items: patient’s education, level of pain, and self-rated health (informational items), and ratings of the physician’s explanations of diagnosis (if discussed), physician’s explanation of treatment, and patient’s confidence and trust in the physician (interactional items). Paired items used parallel content and response scales to compare physician and patient responses. Patient’s education was measured with six categories, from high school incomplete to postgraduate training or professional schooling after college, corresponding to standard measures of educational attainment. Pain was measured with a 10-point scale ranging from 1 (low) to 10 (high). Self-rated health used a single item that asked, “In general, how would you (your patient) describe your (his/her) own health?” (excellent, very good, good, only fair, poor). The measures of pain and self-rated health reflected standard single-item measures used in prior studies with high face validity (Idler & Benyamini, 1997; McCaffery & Beebe, 1993). Perceptions of the physician’s explanations of diagnosis and treatment were measured using 10-point scales from 1 (not well) to 10 (very well). Perception of the patient’s confidence and trust was measured using a four-point scale from a great deal to not at all. Physicians could respond with don’t know on self-rated health and patient’s confidence and trust. The interactional measures were developed for this study and expected to have an intuitive meaning for patients and physicians and therefore high face validity.

Patient questionnaires included additional, nonpaired questions regarding perceptions of the physician and medical encounter: perception of physician listening, whether the patient had unanswered questions, satisfaction with amount of time physician spent with patient, and perception that the physician understood the patient’s background and values. In addition, demographic data, including gender, racial-ethnic background, and age, were collected from both physicians and patients. Patients were also asked how long they had seen the physician (coded as first visit or not).

Procedure and Sample

This study used a convenience sample of physicians and patients in North-Central Florida. Sixteen physicians were contacted in Family Medicine and Oncology, and 10 agreed to participate: six male and four female physicians, of whom seven were White, one was African American, one Hispanic, and one Asian. Physicians were 31 to 59 years old (median = 42 years), with seven younger than 50. Physicians’ postresidency medical experience ranged from less than 5 years to more than 20 years.

Patients were approached in waiting rooms prior to appointments, where informed consent was obtained. In accordance with institutional review board–approved procedures and the Health Insurance Portability and Accountability Act of 1996 (HIPPA) regulations, patients were not prescreened and all were older than 18 years. Five patients were surveyed for each physician, resulting in 50 physician–patient dyads, with equal numbers enrolled from Family Medicine and Oncology clinics. All patients who were approached by the investigator agreed to participate (possibly reflecting the brevity of the survey and interest in the topic). Patient participants included 29 males (58%) and 21 females (42%). Nineteen (38%) were of racial-ethnic minority groups, with 10 African Americans (20%), 6 Hispanics (12%), and 3 Asians (6%). Ages varied from 19 to 92 years (median = 54), and more than half were older than 50 years. All had received at least a high school diploma, and 64% completed at least some college. Thirty-two patients (64%) reported that this was their first visit.

Parallel data were collected from patients and physicians after the medical encounter. Physicians’ surveys were filled out immediately after the interaction. To protect patient anonymity, physicians were never informed about enrolled patients (the procedure included surveys that were completed for random, nonenrolled patients). Patient surveys were administered after the medical encounter and only after the physician had left the examination room. Patient surveys were completed in 10 to 15 minutes.

Analysis

Dyadic concordance was assessed by comparing ratings from the physician and patient on parallel questions. Cross-tabulations were used to identify exact matches, matches within one unit, and larger differences. Based on preliminary analyses, we determined that the exact-match criterion was appropriate for patient’s education, where assessments were verifiable and even small differences represented meaningful misunderstandings. For ratings of pain, self-rated health, and interactional assessments, we used a less strict criterion that defined concordance as agreement within one unit to allow for greater confidence in identifying meaningful discordances (where exact matches were less common, and differences of one unit still represented relatively close agreement on questions that were more subjective). Dyadic concordances were then analyzed with regard to patient’s demographic characteristics and clinical factors as well as measures of patient satisfaction. Descriptive analyses included cross-tabulations and Fisher exact tests as well as t tests, comparison of means, and logistic regression models predicting discordance. Regression models tested various combinations of dummy variables for patient demographic characteristics (racial minority, female, older than 50 years, college education) and clinical factors (first visit, specialty). Because of the nested nature of the sample, with patients matched to a smaller number of physicians, we used Kruskal–Wallis tests to assess the extent of independence among the 10 physicians’ responses. The tests were not statistically significant (p > .05), which justified pooling the data across physicians despite the potential for dependency within physician responses. Due to small sample size, this preliminary study focused on patient characteristics in testing the paired survey method and analyzing correlates of discordances.

Results

Informational Concordance: Patient Characteristics

Patient characteristics and physician–patient perceptions of patient’s education, self-rated health, and pain are summarized in Table 1. Informational concordance in patient’s education was observed in 29 dyads (58%), with physicians overestimating education in 13 cases (26%); seven cases involved physician reports of college-level education for patients with only a high school diploma. Education was more likely to be mislabeled for patients without a college degree (p = .025), which remained statistically significant after controlling for other factors (p < .05).

Table 1.

Patient Characteristics, and Perceptions of Both Patients and Physicians (N = 50)

Characteristic	n	Percentage
Patient’s education: Distribution of patient’s report
High school graduate	18	36
Some college or business/vocational/technical	7	14
College graduate	15	30
Postcollege education	10	20
Patient’s education: Distribution of physician’s report
High school graduate	11	22
Some college or business/vocational/technical	5	10
College graduate	29	58
Postcollege education	5	10
Concordance of patient’s education
Concordant (±1 point)	29	58
Discordant	21	42
Self-rated health status: Distribution of patient responses
Excellent or very good	8	16
Good	8	16
Only fair	14	28
Poor	20	40
Self-rated health status rating^a
Mean (SD) patient’s rating, total	2.10 (1.15)
Mean (SD) physician’s rating, if gave response (n = 34)	2.32 (1.47)
Mean (SD) difference between patient’s and physician’s rating, if physician gave response (n = 34)	−0.21 (1.01)
Self-reported health status rating concordance
Concordant (±1 point)	29	58
Discordant (includes physicians’ “don’t know” responses)	21	42
Pain
Patient reported at least some pain	38	74
Mean (SD) patient’s pain rating^b if reported pain (n = 38)	5.50 (2.29)
Pain rating^b (N = 50)
Mean (SD) patient’s rating	4.18 (3.10)
Mean (SD) physician’s rating of patient’s pain	3.48 (2.86)
Pain concordance (±1 point)
Concordant	33	66
Discordant	17	34

Note. Percentages are rounded to the nearest whole percentage.

Self-reported health was rated on a scale from 1 = poor to 5 = excellent.

Ratings are reported on a scale from 1 = low to 10 = high. Patients reporting no pain were assigned a rating of 1.

Two thirds of patients reported poor or only fair self-reported health (mean = 2.10). The mean physician rating of 2.32 was highly correlated with the patient’s responses (r = .74; p < .001) and not significantly different (t test, p = .24). Sixteen physicians (32%) responded with don’t know, mostly when patients were male (75%, p ≤ .09) or in first visits (69%, p < .01). Ratings of self-rated health were concordant within one level in 58% of dyads (n = 29). Discordance was not related to patient demographic characteristics, but first visits had consistently more discordant responses in bivariate (p = .016) and regression analyses.

Nearly three fourths of patients (38 out of 50) reported at least some pain, with an average rating of 5.50 out of 10 (4.18 when no pain was included). Physician ratings of patient pain had an average of 3.48, which was highly correlated with patient responses (r = .71, p < .001) but significantly lower than the patients’ mean level (p = .002). Responses were concordant in 33 dyads (66%), 8 of whom agreed that there was no pain. Physicians overestimated patient pain in 4 cases but underestimated in 13 cases; 11 cases diverged by three levels or more and most involved significant underestimations of patients’ pain. Large discordances were more prevalent for patients aged 50 years and younger compared with older patients (41% compared with 11%), which remained statistically significant after controlling for demographic and clinical factors.

Interactional Concordance: Physician–Patient Perceptions

Table 2 summarizes the perceptions of the medical encounter. Diagnoses were discussed and rated by 39 dyads. Patients rated their physician’s explanation of the diagnosis with a mean score of 7.95 compared with the physicians’ average of 9.57 (p < .001). Nearly half (49%) of the dyads were concordant, but nearly all discordances involved physicians rating their explanations higher than patients. Physicians rated themselves with a 9 or 10 in all but three cases. There were no gender, age, or education differences, but racial-ethnic minority patients were less likely to have concordant ratings (in bivariate and regression analyses). There were no significant patterns related to first visit or medical specialty.

Table 2.

Patient’s and Physician’s Perceptions of Medical Encounter (N = 50)

Characteristic	n	Percentage
Discussion of diagnosis
Interactions in which diagnosis was discussed	39	78
Diagnosis rating^a (n = 37)
Mean (SD) patient’s rating if diagnosis was discussed	7.95 (2.18)
Mean (SD) physician’s rating if diagnosis was discussed	9.57 (0.728)
Concordance of diagnosis ratings (±1 point)
Concordant	19	49
Discordant	20	51
Discussion of treatment
Interactions in which treatment was discussed	42	84
Treatment rating^a
Mean (SD) patient’s rating if treatment was discussed	8.00 (2.18)
Mean (SD) physician’s rating if treatment was discussed	9.77 (0.571)
Concordance of treatment discussion ratings (±1 point)
Concordant	22	52
Discordant	20	48
Confidence and trust in physician: Patient’s report
Great deal	24	48
A fair amount	19	38
Not too much	5	10
None	2	4
Confidence and trust in physician: Physician’s report^b
Great deal	35	70
A fair amount	4	8
Not too much/none	0	0
Don’t know	11	22
Concordance of confidence and trust in physician
Concordant (exact match of responses)	18	36
Discordant (including physician “don’t know” responses)	32	64

Note. Percentages in table are rounded to nearest whole percentage.

Ratings were reported on a scale from 1 = low to 10 = high. ^bRatings from physicians were reported with the following response categories: great deal, a fair amount, not too much, none, and don’t know. No physicians reported not too much or none.

Forty-two dyads reported discussing treatment options. Within these dyads, physicians rated their explanations with a mean score of 9.77 out of 10 (98% reported levels of 9 or 10). Patients’ mean rating was 8.00 (difference, p < .001); only half reported ratings of 9 or 10, and one third reported ratings of 7 or lower. Responses were concordant in half (52%) of the dyads and did not vary by patient characteristics or clinical variables in bivariate and regression analyses.

Perceptions of patient confidence and trust were defined as concordant when there was an exact match in patient and physician responses (on a 4-point scale); concordance was identified in 18 dyads (36%). Physicians tended to overestimate patient confidence and trust, reporting “a great deal” of confidence and trust in 70% of dyads compared with only 48% of patients. Discordances did not vary by patient characteristics or physician specialty but were more likely in first visits (89% vs. 50%, p = .007; significant after controlling for other factors).

Patient Perceptions of Physician and Clinical Encounter

When asked of their perceptions of the physician and medical encounter, between 22% and 38% of patients reported dissatisfaction. Twenty-two percent of patients reported that the physician had not listened to everything that they had said, though these perceptions were unrelated to patient demographic characteristics. About one third (36%) reported that they had unanswered questions, especially racial minority patients (53%, p = .073) and those with less than a college degree (52%, p = .038); the education difference remained significant after controlling for other factors. About one third (36%) reported that the physician had not spent enough time, which was again higher among racial minority groups (53%, p = .073) and less educated patients (56%, p = .007); the education difference remained significant after adding controls. Thirty-eight percent of patients reported that they did not feel that the physician had understood their values and background, but this proportion rose to 74% among racial minority patients (compared with 16% for nonminority; p < .001). This racial difference was consistently strong across all models controlling for other factors. There were no differences related to first visits, but patients seen in family medicine were more likely than those in oncology to report unanswered questions and that the physician had not spent enough time (p = .038 for both); controlling for other factors, these differences remained significant.

Relations between physician–patient concordance and patient perceptions of the encounter are summarized in Table 3. In general, patients in discordant dyads were more likely to report that their doctor had not listened to everything and that they had unanswered questions (except with regard to self-rated health and pain). Informational discordance on patient’s education was significant in bivariate analyses but not after controlling for other factors. Interactional discordances in ratings of diagnosis discussions were linked to greater dissatisfaction with physician listening, and discordance in treatment discussions were associated with dissatisfaction in listening, unanswered questions, and not spending enough time (all significant after controlling for other factors). Discordances in patient confidence and trust were linked to more negative patient perceptions on each item (both bivariate and regression analyses).

Table 3.

Proportions of Dyads With Concordant or Discordant Physician–Patient Responses With Patients Who Reported Negative Perceptions of the Physician or Medical Encounter (N = 50)

Patient’s Education Level^a	Concordant Percentage (n = 36)	Discordant Percentage (n = 14)
Doctor listened to everything I had to say: Strongly disagree or disagree	14	43*
Patient had unanswered questions after medical interaction	25	64*
Patient reported that physician did not spend enough time	25	64*
Physician understood values: Strongly disagree or disagree	33	50
Ratings of Diagnosis Discussion^b	Percentage (n = 19)	Percentage (n = 18)
Doctor listened to everything I had to say: Strongly disagree or disagree	5	33*^d
Patient had unanswered questions after medical interaction	21	50+
Patient reported that physician did not spend enough time	26	44
Physician understood values: Strongly disagree or disagree	37	44
Ratings of Treatment Discussion^b	Percentage (n = 22)	Percentage (n = 20)
Doctor listened to everything I had to say: Strongly disagree or disagree	0	40**^d
Patient had unanswered questions after medical interaction	18	50*^d
Patient reported that physician did not spend enough time	18	50*^d
Physician understood values: Strongly disagree or disagree	27	40
Confidence and Trust in Physician^c	Percentage (n = 18)	Percentage (n = 32)
Doctor listened to everything I had to say: Strongly disagree or disagree	0	34**^d
Patient had unanswered questions after medical interaction	11	50**^d
Patient reported that physician did not spend enough time	17	47^† d
Physician understood values: strongly disagree or disagree	17	50*^d

Note. Percentages in table are rounded to nearest whole percentage.

Patient’s education was measured with the following categories: high school graduate, some college or business/vocational/technical, college graduate, postcollege education. Concordance is defined as matching within one point. ^bRatings were reported on a scale from 1 = low to 10 = high. Concordance is defined as matching within one point. ^cConcordance for confidence and trust is defined as an exact match in physician and patient responses. ^dDifference is statistically significant after controlling for other factors.

†

p <. 09. *p < .05. **p < .01. (Statistical significance based on chi-square statistics.)

Discussion

The paired survey method identified both informational and interactional discordances between physicians and patients. Although more than half of dyads were concordant on most items, significant discordances raised important questions for medical practice and education. Ratings of self-rated health and patient education were discordant in more than 40% of dyads, and pain ratings were discordant in about one third, signaling potential problems with the communication of key facts. Interactional discordances in perceptions of diagnostic and treatment explanations and patients’ trust and confidence were observed in 51%, 48%, and 64% of dyads, respectively. These results highlight substantial levels of physician–patient discordance in a small pilot study and point to the value of paired survey methods for identifying potential miscommunications.

Our findings revealed several important patterns in informational and interactional discordances. Physicians tended to overestimate their patients’ education, echoing earlier findings (Kelly & Haidet, 2007) and raising questions about the sources of such misperceptions. Do patients’ questions and conversations suggest a higher educational level, or are physicians generalizing from their contexts and experiences? Educational discordances may be less explicitly problematic than disagreements on clinical issues, such as pain, but overestimating education may contribute to assumptions of greater understanding and mutual comprehension than is warranted, leading to greater patient dissatisfaction with the physician and interaction.

Physicians’ tendency to overestimate patients’ self-reported health and underestimate pain, on the other hand, is potentially more troubling. These findings concur with other observations of physicians’ misjudgments of patient self-rated health (Desalvo & Muntner, 2011) and pain in primary care (Bertakis et al., 2004) and oncology (Mäntyselkä, Kumpusalo, Takala, & Ahonen, 2001; Radowsky et al., 2012). These findings raise serious questions about diagnosis and pain management, and they reveal important challenges in dealing with fundamentally subjective assessments. Attempts to measure and quantify pain are complicated by differences in perception (“scale bias”), wherein the frames of reference may be quite different for physicians and patients: Physicians are likely to compare their observations with other cases and clinical experiences, whereas patients assess their pain within the context of personal history, current discomfort, or efforts to attain a desired outcome from their medical encounter. As a result, physicians may expect less pain from younger patients (based on clinical experience), yet these patients themselves may regard the pain as disruptive or distressing (within the context of their own life experiences and expectations). Achieving complete concordance may not be possible or even entirely necessary, but identifying and anticipating common discordances may help physicians and patients to better communicate in ways that result in more effective medical practice and greater patient satisfaction.

Interactional discordances consistently showed positive assessments from physicians that were not matched by patients and were reflected in greater patient dissatisfaction. Again, different frames of reference may be at play. A physician’s professional role requires a high level of training and competence as well as confidence in one’s skills and abilities. Patients are assumed to have trust in seeking out the expertise and assistance of the physician. Patients, on the other hand, are likely to assess each physician and medical encounter based on specific performance in the current encounter and comparisons with prior experiences and their own expectations. A physician’s overconfidence with diagnostic or treatment explanations may decrease the use of helpful communication techniques, whereas assumptions about implicit trust may reduce physicians’ efforts to build rapport (Stevenson & Scambler, 2005), in either case undermining the physician’s performance and the patient’s satisfaction.

Our preliminary findings suggest few links, between discordances and demographic characteristics or clinical factors. There were no differences across social groups in discordances regarding self-rated health or perceptions of confidence and trust. Underestimations of pain were more likely with younger patients, and misreports of patient education were more common for less-educated patients. Racial-minority patients had more discordance in perceptions of diagnosis discussions, but there were no other demographic differences in interactional discordance. First visits were linked to greater discordance in self-rated health and perceptions of confidence and trust, but there were no consistent differences by specialty. Thus, our hypotheses were only weakly supported.

However, several important limitations should be noted. First, although the small sample size allowed us to test the paired survey method, it limited our statistical power and ability to test our hypotheses. Second, the findings cannot be generalized as they come from a nonrepresentative, local sample of physicians and patients. Third, although the nested design of matching patients with physicians is a promising area for research (e.g., see Fagerberg, Kragstrup, Støvring, & Rasmussen, 1999), future work should use a larger sample and statistical techniques that take into account the potential nonindependence of responses. A larger sample of physicians would also allow for exploration of physician characteristics that may be related to informational or interactional discordances. Finally, although our measures of discordance seem to offer a high level of face validity, future work should go beyond single-item measures to construct and test more reliable scales to measure different facets of physician–patient concordance.

Our research highlights the value and unique insights that can be gained from a dyadic approach. As a research tool, paired survey methods can be used along with observations to examine how particular communication skills may reduce informational and interactional discordances. Further research should also examine further social factors related to discordances as well as the roles of other parties (e.g., clinical personnel, family members, caregivers) in the medical encounter (see Roter, 2003). As an indirect measure of physician–patient interaction, paired survey methods may help identify avenues for improving communication and medical training, especially in circumstances where direct observation may be problematic (Campbell, Lockyer, Laidlaw, & MacLeod, 2007; Roter, 2003). Informational discordances on key clinical factors (i.e., self-rated health, pain) might be reduced by confirming information through close listening, asking strategic questions, and paraphrasing, as well as encouraging patients to confirm and correct key facts. Measuring and examining physician–patient concordance may help in the development of educational programs (for both physicians and patients) to enhance awareness of common discordances and strategies for reducing them in medical encounters.

Footnotes

Acknowledgements

We thank Dr. Jesse Arnold for his statistical advice and suggestions.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

References

Arnold

C. L.

Arnold

J. C.

(1999). Managing conflict in decision-making groups: When does mediation and coorientational accuracy make a difference. Florida Communication Journal, 26, 11-30.

August

K. J.

Nguyen

Ngo-Metzger

Sorkin

D. H.

(2011). Language concordance and patient-physician communication regarding mental health needs. Journal of the American Geriatrics Society, 59, 2356-2362.

Back

A. L.

Arnold

R. M.

Baile

W. F.

Tulsky

J. A.

Fryer-Edwards

(2005). Approaching difficult communication tasks in oncology. CA: A Cancer Journal for Clinicians, 55, 164-177.

Beach

M. C.

Thomas

(2006). Relationship-centered care: A constructive reframing. Journal of General Internal Medicine, 21, S3-S8.

Benyamini

Leventhal

E. A.

Leventhal

(1999). Self-assessments of health: What do people know that predicts their mortality? Research on Aging, 21, 477-500.

Bertakis

K. D.

Azari

(2012). Patient-centered care: The influence of patient and resident physician gender and gender concordance in primary care. Journal of Women’s Health, 21, 326-333.

Bertakis

K. D.

Azari

Callahan

E. J.

(2004). Patient pain in primary care: Factors that influence physician diagnosis. Annals of Family Medicine, 2, 224-230.

Bertakis

K. D.

Franks

Epstein

R. M.

(2009). Patient-centered communication in primary care: Physician and patient gender and gender concordance. Journal of Women’s Health, 18, 539-545.

Campbell

Lockyer

Laidlaw

MacLeod

(2007). Assessment of a matched-pair instrument to examine doctor-patient communication skills in practicing doctors. Medical Education, 41, 123-129.

10.

Cooper-Patrick

Gallo

J. J.

Gonzales

J. J.

H. T.

Powe

N. R.

Nelson

Ford

D. E.

(1999). Race, gender, and partnership in the patient-physician relationship. Journal of the American Medical Association, 282, 583-589.

11.

Desalvo

K. B.

Muntner

(2011). Discordance between physician and patient self-rated health and all-cause mortality. Ochsner Journal, 11, 232-240.

12.

DeVoe

J. E.

Wallace

L. S.

Fryer

G. E.

Jr. (2009). Patient age influences perceptions about health care communication. Family Medicine, 41, 126-133.

13.

Elderkin-Thompson

Waizkin

(1999). Differences in clinical communication by gender. Journal of General Internal Medicine, 14, 112-120.

14.

Fagerberg

C. R.

Kragstrup

Støvring

Rasmussen

N. K.

(1999). How well do patient and general practitioner agree about the content of consultations? Scandinavian Journal of Primary Health Care, 17, 149-152.

15.

Freidin

R. B.

Goldman

Cecil

R. R.

(1980). Patient-physician concordance in problem identification in the primary care setting. Annals of Internal Medicine, 93, 491-493.

16.

Gany

Leng

J. C.

Winston

Tseng

C. H.

Qin

Changrani

(2011). Diabetes care and language discordance among Chinese and Latino primary care patients. Ethnicity & Disease, 21, 473-479.

17.

Gross

MCNeill

Davis

Lay-Yee

Jatrana

Crampton

(2008). The association of gender concordance and primary care physicians’ perceptions of their patients. Women & Health, 48, 123-144.

18.

Idler

E. L.

Benyamini

(1997). Self-rated health and mortality: A review of twenty-seven community studies. Journal of Health and Social Behavior, 38, 21-37.

19.

Jahng

K. H.

Martin

L. R.

Golin

C. E.

DiMatteo

R. M.

(2005). Preferences for medical collaboration: Patient-physician congruence and patient outcomes. Patient Education and Counseling, 57, 308-314.

20.

Janz

N. K.

Wren

P. A

Copeland

L. A.

Lowery

J. C.

Goldfarb

S. L.

Wilkins

E. G.

(2004). Patient-physician concordance: Preferences, perceptions, and factors influencing, the breast cancer surgical decision. Journal of Clinical Oncology, 22, 3091-3098.

21.

Jylha

(2009). What is self-rated health and why does it predict mortality? Towards a unified conceptual model. Social Science & Medicine, 69, 307-316.

22.

Kelly

P. A.

Haidet

(2007). Physician overestimation of patient literacy: A potential source of health care disparities. Patient Education and Counseling, 66, 119-122.

23.

Kenny

D. A.

Veldhuijzen

van der Weijden

LeBlanc

Lockyer

Légaré

Campbell

(2010). Interpersonal perception in the context of doctor-patient relationships: A dyadic analysis of doctor-patient communication. Social Science & Medicine, 70, 763-768.

24.

Krupat

Rosenkranz

S. L.

Yeager

C. M.

Barnard

Putnam

S. M.

Inui

T. S.

(2000). The practice orientations of physicians and patients: The effect of doctor-patient congruence on satisfaction. Patient Education and Counseling, 39, 49-59.

25.

Liaw

S. T.

Young

Farish

(1996). Improving patient-doctor concordance: An intervention study in general practice. Family Practice, 13, 427-431.

26.

Mäntyselkä

Kumpusalo

Takala

Ahonen

(2001). Variability in prescribing for musculoskeletal pain in Finnish primary health care. Pharmacy World & Science, 23, 232-236.

27.

McCaffery

Beebe

(1993). Pain: Clinical manual for nursing practice. Baltimore: V.V. Mosby.

28.

Meghani

S. H.

Brooks

J. M.

Gipson-Jones

Waite

Whitfield-Harris

Deatrick

J. A.

(2009). Patient-provider race-concordance: Does it matter in improving minority patients’ health outcomes? Ethnicity & Health, 14, 107-130.

29.

Radowsky

J. S.

Baines

Howard

R. S.

Shriver

C. D.

Buckenmaier

C. C.

Stojadinovic

(2012). Pain ratings by patients and their providers of radionucleotide injection for breast cancer lymphatic mapping. Pain Medicine, 13, 670-676. doi:10.1111/j.1526-4637.2012.01374.x

30.

Rodriguez

K. L.

Bayliss

N. K.

Alexander

S. C.

Jeffreys

A. S.

Olsen

M. K.

Pollak

K. I.

Arnold

R. M.

(2011). Effect of patient and patient-oncologist relationship characteristics on communication about health-related quality of life. Psycho-Oncology, 20, 935-942.

31.

Ross

C. E.

Mirowsky

Duff

R. S.

(1982). Physician status characteristics and client satisfaction in two types of medical practice. Journal of Health and Social Behavior, 23, 317-329.

32.

Roter

D. L.

(2003). Observations on methodological and measurement challenges in the assessment of communication during medical exchanges. Patient Education and Counseling, 50, 17-21.

33.

Schnittker

Liang

(2006). The promise and limits of racial/ethnic concordance in physician-patient interaction. Journal of Health Politics, Policy, and Law, 31, 811-838.

34.

Schoenthaler

Allegrante

J. P.

Chaplin

Ogedegbe

(2012). The effect of patient-provider communication on medication adherence in hypertensive black patients: Does race concordance matter? Annals of Behavioral Medicine, 43, 372-382.

35.

Sepucha

Ozanne

E. M.

(2010). How to define and measure concordance between patients’ preferences and medical treatments: A systematic review of approaches and recommendations for standardization. Patient Education and Counseling, 78, 12-23.

36.

Smith

S. K.

Dixon

Trevena

Nutbeam

McCaffery

K. J.

(2009). Exploring patient involvement in healthcare decision making across different education and functional health literacy groups. Social Science & Medicine, 69, 1805-1812.

37.

Stevenson

Scambler

(2005). The relationship between medicine and the public: The challenge of concordance. Health, 9, 5-21.

38.

Street

R. L.

Haidet

(2011). How well do doctors know their patients? Factors affecting physician understanding of patients’ health beliefs. Journal of General Internal Medicine, 26, 21-27.

39.

Street

R. L.

O’Malley

K. J.

Cooper

L. A.

Haidet

(2008). Understanding concordance in patient-physician relationships: Personal and ethnic dimensions of shared identity. Annals of Family Medicine, 6, 198-205.

40.

Strong

Mathews

Sussex

New

Hoey

Mitchell

(2009). Pain language and gender differences when describing a past pain event. Pain, 145, 86-95.

41.

Waitzkin

(1985). Information giving in medical care. Journal of Health and Social Behavior, 26, 81-101.

42.

Walker

Arnold

Miller-Day

Webb

(2002). Investigating the physician-patient relationship: Examining emerging themes. Health Communication, 14, 45-68.