(Mis)understanding in patient-health care provider communication about total knee replacement

Authors

  • Richard L. Street JR.,

    Corresponding author
    1. Texas A&M University, College Station, and The Houston Center for Quality of Care and Utilization Studies and Baylor College of Medicine, Houston, Texas
    • Department of Communication, Texas A&M University, College Station, TX 77843-4234
    Search for more papers by this author
  • Marsha N. Richardson,

    1. The University of Texas M. D. Anderson Cancer Center, Houston
    Search for more papers by this author
  • Vanessa Cox,

    1. The University of Texas M. D. Anderson Cancer Center, Houston
    Search for more papers by this author
  • Maria E. Suarez-Almazor

    1. The University of Texas M. D. Anderson Cancer Center, Houston
    Search for more papers by this author
    • Dr. Suarez-Almazor is the Director of the Houston Center for Education and Research on Therapeutics, which is funded by Agency for Healthcare Research and Quality (grant U18-HS16093), and holds a K24 award from the National Institute of Arthritis and Musculoskeletal and Skin Diseases.

    • Dr. Suarez-Almazor has received consultancies, speaking fees, and/or honoraria (less than $10,000) from the Bristol-Myers Squibb Speakers Bureau.


Abstract

Objective

To examine whether communication factors affect health care provider and patient agreement on the need for, risks of, and benefits of joint replacement, and also whether degree of agreement predicts patient satisfaction and intent to follow treatment recommendations.

Methods

Health care providers (n = 27) and patients (n = 74) with severe osteoarthritis (OA) were recruited from clinics in Houston, Texas. Patients completed a baseline survey prior to the consultation. After the visit, patients and providers completed measures of the severity of the patient's OA, the expected benefits of total knee replacement (TKR), and concern about surgical complications. Patients also completed satisfaction and intent to adhere measures. Provider communication and patient participation were measured by patient self-report and by observers' codings of audiorecordings of the consultations.

Results

Provider-patient agreement was modest to poor regarding severity of the patient's OA and the expected benefits and risks of TKR. Providers and patients were more aligned on the patient's OA severity when providers used more partnership building but spent less time simply giving information. Differences between providers' and patients' concerns about surgery were greater when patients were less participatory, African American, or expressed lower trust in their doctors. Patient satisfaction and intent to adhere were predicted by provider-patient agreement on the benefits of TKR.

Conclusion

Patients and providers often differ in their beliefs about the need for, risks of, and benefits of TKR, and these differences can affect patient satisfaction and commitment to treatment. Facilitating active patient participation might contribute to greater physician–patient agreement on the patient's concerns about OA and surgical interventions.

INTRODUCTION

Osteoarthritis (OA) affects more than 20 million Americans and disables approximately 1 of 10 people >60 years of age (1). For end-stage OA, total joint replacement (TJR) offers potential for significant functional improvement and pain relief (2). The surgery is not without risks, however. Patients and physicians must carefully assess the potential benefits and harm of TJR in the context of the patient's lifestyle and values. Although health care providers and patients should reach a shared understanding of the risks and benefits of treatment (3), any number of factors may detract from this understanding, including providers' inability to present information in a way that patients can comprehend, patients' perceptions of need not corresponding with radiologic data (4), and the wide variability in patients' concerns about surgery and recovery (5).

The investigation reported here examined whether active patient participation and provider information giving influenced concordance between patient and provider beliefs about the need for, risks of, and benefits of total knee replacement (TKR) and, in turn, whether belief concordance predicted patient satisfaction and intent to follow treatment recommendations. The study is significant for several reasons. First, despite the increased emphasis on informed decision making, there is little evidence in the literature related to communication factors affecting decision making in orthopedic surgery (6). Second, a large body of research has shown that certain features of provider-patient communication (e.g., patient question asking and involvement in decision making, and provider information giving and partnership building) are often correlated with health-related outcomes following the consultation (e.g., adherence, satisfaction, improved health) (3, 7–11). Finally, a growing number of studies have discovered that patient satisfaction and commitment to treatment are usually higher when the physician and the patient achieve agreement on such matters as the patient's role in decision making, the meaning of diagnostic and prognostic information, and the treatment plan (12–15). This investigation examines one pathway through which communication may contribute to better outcomes: shared understanding between clinicians and patients about diagnosis and treatment.

This study had 3 objectives. First, we assessed the degree to which health care providers' and patients' beliefs were concordant with respect to the severity of the patient's OA, the potential benefits of TKR, the importance of the patient having TKR, and concern over surgical complications. Second, we examined communication, relationship, and patient demographic factors that potentially affect provider-patient agreement on these issues. In accordance with models of informed decision making (16), agreement should be greater when providers give more information about the patient's condition, treatment options, and potential risks and benefits of these options, and when patients actively participate in the encounter by expressing their opinions, beliefs, questions, concerns, and preferences (17, 18). Demographic variability in the utilization of TKR, particularly with respect to race and sex (19), also may be related to differences among patients' expectations and concerns about TKR (5, 20). This in turn could affect the degree to which providers and patients agree on the risks and benefits of TKR. With respect to the quality of the provider-patient relationship, belief concordance may be greater when patients have more trust in their doctors, given that trust is associated with providers' informativeness and sensitivity to patient's concerns (21–23). Finally, we looked at whether concordance in provider-patient understandings contributes to the immediate outcomes of the consultation: the patient's satisfaction with care and intent to follow treatment recommendations.

SUBJECTS AND METHODS

Research setting and subjects

Patients and health care providers were recruited from the Michael E. DeBakey Veterans Affairs Medical Center Orthopedic Outpatient Clinic and the Ben Taub General Hospital Orthopedic Outpatient Clinic. These institutions are staffed by fellows, residents, and physician's assistants under the supervision of attending physicians. The Institutional Review Board at each institution approved the study. All participating providers and patients provided informed consent.

Provider and patient recruitment occurred between May 2004 and October 2005. With the physician's consent, the medical records of patients attending the participating clinics were reviewed in advance in order to identify patients with knee OA who were possibly eligible for TKR. Inclusion criteria were physician or radiologic diagnosis of knee OA, clinical symptoms including pain (moderate or severe on a 4-point scale) and disability (mild, moderate, or severe on a 4-point scale), symptoms lasting 1 year or longer with ≥3 months of medical treatment, age ≥40 years, and adequate mental status and English language proficiency (determined by the interviewer). Patients were recruited and screened for eligibility either by telephone in advance of their scheduled appointment or while waiting for their clinic visit.

A total of 27 health care providers agreed to participate in the study, including 4 orthopedic physicians (2 fellows and 2 staff physicians), 1 rheumatologist staff physician, 17 medical residents (5 primary care/internal medicine and 12 orthopedics), and 5 orthopedic physician assistants. The providers were mostly men (74%) but were ethnically diverse (41% white, 11% African American, 4% Hispanic, 37% Asian, and 7% unknown).

Procedures

Research participants were told that the purpose of the study was to examine the beliefs different people have about arthritis and to study different ways in which patients with arthritis and doctors communicate. After consenting to participate in the study, patients completed a battery of self-report measures that included demographic information and patient trust (24), as well as baseline measures assessing their beliefs about the severity of their OA, the importance of undergoing TKR, how concerned they were about surgery, and their expected benefit of TKR. Patients then attended their clinical visit, which was audiorecorded. After the visit, both the patient and the health care provider completed questionnaires (one patient mailed in the postvisit questionnaire and another completed it via telephone). The patients' survey included a postvisit assessment of beliefs and expectations about TKR (same measures as baseline), their perceptions of their own and the provider's communication, whether the provider recommended TKR, and measures of their satisfaction with care and their intent to adhere to provider recommendations. After the visit, providers also completed the belief and expectation measures and reported whether they recommended TKR.

Belief and expectations measures

Based on input from patient focus groups and physician interviews, we created 4 single-item measures of beliefs and expectations about TKR (with wording modified depending on whether the respondent was a doctor or a patient): 1) the severity of the patient's OA (“How would you rate the severity of your/the patient's arthritis?”), 2) the importance of having TKR (“How important is it for you/the patient to undergo joint replacement?”), 3) expected benefit from TKR (“How likely are you/is the patient to benefit from joint replacement?”), and 4) concern about surgical complications (“How concerned are you about potential complications during surgery?”). Patients and providers responded to each item on a 10-point scale.

Communication measures

Both behavioral and patient self-report measures of communication were used. Behavioral measures typically have coders listen to audiorecordings and then classify the provider's and patient's utterances into different categories (e.g., question, information-giving statement, partnership building). Subsequently, a quantitative measure (e.g., frequency) of that aspect of communication is computed for the interactant for that interaction. Patient self-report measures tap into the perceptions of their own and the provider's communication. Researchers advocate using both sets of assessments in order to gauge what kind of communication occurred and how patients perceived the interaction (25–27).

Behavioral measures

Providers' information giving and partnership building and patients' active participation were coded using a previously validated coding system (28–31). Providers' information-giving statements included utterances giving information about diagnoses, treatment procedures, prognoses, risks, and benefits, and were typically in the form of explanations, descriptions, and instructions. These features of providers' communication represent key features of informed decision making (16). Provider partnership building consisted of utterances in which the provider solicited or accommodated patient participation in the consultation and in decision making (e.g., inviting the patient to talk about health concerns, asking for the patient's opinion about surgery, agreeing with the patient's perspective).

Active patient participation consisted of 3 types of verbal communication: asking questions, assertiveness (offering opinions, stating preferences, making a request), and expressing concerns (worries, fears, negative feelings). These behaviors are considered active because they can influence a provider's behavior, perceptions of the patient, and treatment decisions (18).

Two undergraduate majors in communication were trained to code the verbal behaviors. Coders were not aware of the purpose of the study. On transcripts of the audiorecordings, coders unitized the talk into utterances, the units of analysis for coding the different types of communication behaviors. An utterance is the oral analog of a simple sentence and may be in the form of a complete sentence, independent clause, or multiple predicate. Coders then followed the transcript while listening to the recording and placed targeted behaviors into appropriate verbal categories. Reliability was established by having both coders code a subset of 12 consultations independently of one another. Coders achieved 92% agreement on unitizing utterances. Reliabilities (Cohen's kappa) for provider information giving, partnership building, and active patient participation were 0.85, 0.65, and 0.77, respectively. The remaining consultations were divided between the 2 coders, who coded them independently.

Self-report measures

Patients reported their perceptions of their own involvement and of the providers' partnership building using items adapted from the patient participation and doctor facilitation subscales of Lerman and colleagues' Patients' Perceived Involvement in Care Scale (32). For patient involvement, 5 items were used: “I asked the doctor to explain the procedure to me in great detail,” “I went into great detail about my symptoms,” “I asked the doctors a lot of questions,” “I expressed doubts about joint replacement surgery,” and “I gave my opinions (agreement or disagreement) about joint replacement.” Three items measured provider partnership building: “The doctor encouraged me to talk about personal concerns related to joint replacement,” “The doctors asked me whether I agreed with their recommendations,” and “The doctors encouraged me to give my opinion about joint replacement.” The 5 questions that measured patients' perceptions of provider information giving, developed from input from the patient focus groups and physician interviews, were: “The doctors clearly explained my treatment options,” “The doctors told me what my surgery would do,” “I did not understand my treatment plan,” “The doctors clearly explained the treatment procedure,” and “The doctors provided me with as much information as I wanted.” All items were scored on a 5-point Likert scale with response options ranging from strongly agree to strongly disagree.

Satisfaction and adherence measures

Patient satisfaction was measured with 6 items from the Patient Visit Rating Questionnaire from the Medical Outcomes Study (33) using 5-point Likert scales. Intent to adhere was a single-item measure: “How likely are you to accept the doctor's recommendation?” Response options were very likely, somewhat likely, not sure, somewhat unlikely, and not at all.

Statistical analysis

Two methods were used to assess concordance among providers' and patients' beliefs about the severity of the patient's OA, the importance of having TKR, the expected benefit of TKR, and concerns about surgery. First, intraclass correlation coefficients (ICCs) were computed comparing provider and patient scores on each belief measure. The ICC is more appropriate than Pearson's correlation coefficient because the ICC takes into account both the distribution of the scores as well as the similarity between the 2 scores (34). Second, for each belief measure, t-tests were used to determine whether providers' scores significantly differed from patients' scores, and in what direction.

To examine whether communication and demographic factors influenced the degree of concordance between provider and patient beliefs, the absolute difference between provider and patient beliefs about the severity of the patient's OA, the potential benefits of TKR, the importance of the patient having TKR, and concerns about surgery each served as dependent measures in multiple variable models that included the patient's race (African American versus white), age (≤55 versus ≥56 years, based on median split), sex, education (high school or less versus some college or more), patient participation, and the provider's information giving and partnership building as predictors.

Finally, to examine whether belief concordance predicted outcomes, multivariable regression analyses were conducted using patient satisfaction and intent to follow provider recommendations as dependent measures and the patients' demographic characteristics and the absolute difference between provider and patient scores on each belief measure as predictors. All analyses controlled for patients nested within providers and treated the individual provider as a random effect.

RESULTS

A total of 166 patients were approached, and 35 of these missed their scheduled appointment. Of the remaining 131 patients, 46 refused participation in the study, leaving a total of 85 participants. Four Hispanic patients were excluded because the sample was too small to draw meaningful conclusions attributable to this ethnic group. Seven patients were excluded for the following reasons: the patient's spouse contributed to a large portion of the communication in the visit, recording malfunctions, the patient only expressed anger about a previous doctor and did not engage in the expected discussions about OA, and the participant's visit was for a joint injection only. The final sample consisted of 74 patients.

Patient characteristics are shown in Table 1. The sample was balanced with respect to race (African American and white patients) and level of education. However, only 7 women volunteered for the study, which is in part an artifact of having the Veterans Affairs Medical Center serve as the primary study site, from which 93% of the participants were recruited (the remaining were enrolled from the local hospital district). Thus, our analyses controlled for sex effects but we do not report these findings. Provider ethnicity, sex, and age were not significantly related to the dependent measures, and thus were not included in the multivariable analyses. However, all analyses treated the individual provider as a random effect to control for differences related to patients nested within providers.

Table 1. Characteristics of the patient sample (n = 74)*
 Value
  • *

    Values are the percentage unless otherwise indicated.

  • Range: 1 = very poor, 10 = excellent.

Age, mean ± SD years58.5 ± 9.29
Men91
Ethnicity 
 African American46
 White54
Education 
 High school or less27
 Some college or more73
Marital status 
 Married52.7
 Single9.5
 Divorced/separated35.1
 Widowed2.7
Employment status 
 Employed full-time23
 Retired25.7
 Disabled31.1
 Unemployed12.2
 Other8
Self-reported health status, mean ± SD6.93 ± 1.89

Agreement between provider and patient beliefs

Provider-patient agreement on whether the provider recommended joint replacement was 82%, but did not reach statistical significance (κ = 0.54, P = 0.11), Forty-six provider-patient pairs agreed that the provider did not recommend joint replacement, 13 pairs agreed that the doctor recommended it, and 13 pairs disagreed on whether joint replacement was recommended. Degree of concordance was modest, but statistically significant, regarding provider-patient beliefs about the benefits of TKR for the patient (ICC 0.47, P < 0.001) and the importance of having the patient undergo TKR (ICC 0.46, P < 0.001). A weaker association was observed for beliefs about the severity of the patient's OA (ICC 0.35, P < 0.09). Providers generally believed the patient's arthritis was less severe than patients themselves thought (Table 2). Concerns about surgical complications were not correlated (ICC 0.02, not significant), in large part because the providers generally had less concern than the patients had. Providers and patients did not significantly differ in their beliefs about the potential benefits of TKR and the importance of the patient having TKR (Table 2).

Table 2. Concordance between physician and patient beliefs*
 No.Physician perceptionPatient perceptionDifferenceP (95% CI)
  • *

    95% CI = 95% confidence interval; TKR = total knee replacement.

  • Range: 1 = very mild, 10 = very severe.

  • Range: 1 = not at all, 10 = very important.

  • §

    Range: 1 = not at all, 10 = very likely.

  • Range: 1 = not at all, 10 = very concerned.

Severity of arthritis736.637.51−0.880.015 (−1.58, −0.18)
Importance of having TKR735.305.53−0.230.67 (−1.27, 0.82)
Benefit from TKR§716.456.93−0.480.35 (−1.49, 0.54)
Concerned about surgery735.566.78−1.220.01 (−2.14, −0.29)

Factors affecting degree of shared understanding

The results of factors predicting the degree to which providers and patients agreed on the severity of the patient's arthritis, the benefits of TKR, and the potential complications from surgery are presented in Table 3. Agreement on the importance of the patient having TKR was not predicted by patient characteristics or provider-patient communication variables and thus was excluded from Table 3. The providers' communication had the strongest effect on provider-patient agreement regarding the severity of the patient's OA. Specifically, analyses using either behavioral or patient self-report communication measures indicated greater agreement when providers more often engaged in partnership building. However, contrary to expectations, there was less agreement when providers gave more information and were perceived by patients as more informative.

Table 3. Predictors of agreement between provider and patient beliefs*
MeasuresSeverity of arthritis (n = 72)Benefits of TJR (n = 70)Concern about surgery (n = 72)
Estimate ± SEMPEstimate ± SEMPEstimate ± SEMP
  • *, †

    A negative coefficient indicates less discrepancy (more agreement) between provider-patient beliefs. TJR = total joint replacement.

  • Reference = white.

  • Reference = some college or more.

Behavioral      
 Patient      
  African American race−0.34 ± 0.450.45−0.39 ± 0.550.481.08 ± 0.550.059
  Age−0.01 ± 0.030.61−0.02 ± 0.030.57−0.06 ± 0.030.072
  Education, high school or less−0.02 ± 0.500.970.25 ± 0.590.670.81 ± 0.610.19
  Trust in the provider0.01 ± 0.010.690.01 ± 0.020.44−0.04 ± 0.020.025
  Active participation−0.02 ± 0.030.59−0.01 ± 0.040.81−0.14 ± 0.040.001
 Physician      
  Partnership building−0.15 ± 0.080.049−0.02 ± 0.090.810.11 ± 0.090.23
  Information giving0.03 ± 0.010.0020.01 ± 0.010.570.01 ± 0.010.31
Patient self-report      
 African American race−0.33 ± 0.440.46−0.63 ± 0.480.191.38 ± 0.600.026
 Age−0.00 ± 0.030.85−0.01 ± 0.030.72−0.04 ± 0.030.24
 Education, high school or less0.59 ± 0.510.260.40 ± 0.530.450.76 ± 0.670.27
 Trust in the provider0.02 ± 0.010.270.01 ± 0.020.36−0.03 ± 0.020.087
 Active participation−0.06 ± 0.090.520.12 ± 0.090.21−0.09 ± 0.120.43
 Physician's partnership building−0.32 ± 0.100.004−0.06 ± 0.110.58−0.01 ± 0.140.92
 Physician's information giving0.16 ± 0.060.015−0.19 ± 0.070.0070.07 ± 0.080.42

Regarding concerns about surgical complications, providers and patients were more similar in their views when patients more actively participated in the encounter based on coder assessment. However, agreement was not related to patients' perceptions of their involvement. Two other variables predicting similarity between provider and patient concerns about surgery were the patient's race and previsit trust in the clinician. Specifically, there was greater discrepancy between providers' and African American patients' concerns about surgical complications than there was between providers and white patients' concerns, and there was more concordance between provider and patient concerns about surgery when patients reported more trust in their clinicians. Finally, provider-patient agreement on how likely the patient was to benefit from TKR was greater when patients perceived their providers as more informative, but was not related to the sheer amount of information providers gave (Table 3).

Patient satisfaction and intent to adhere

Patients were more satisfied with care and expressed stronger intent to follow provider recommendations the more providers and patients agreed on whether the patient would benefit from having TKR (Table 4). Also, there was a trend (P = 0.1) indicating that patient satisfaction was higher when the provider and patient were more aligned on concerns about surgical complications.

Table 4. Predictors of patient satisfaction and intent to adhere to recommendations (n = 68)*
PredictorSatisfactionIntent to adhere
Estimate ± SEMPEstimate ± SEMP
  • *

    A negative coefficient indicates that the outcome was associated with less discrepancy (more agreement) between provider-patient beliefs. OA = osteoarthritis; TJR = total joint replacement.

  • Reference = white.

  • Reference = some college or more.

Patient characteristic    
 African American race−1.88 ± 1.410.19−0.26 ± 0.230.26
 Age0.05 ± 0.080.54−0.02 ± 0.010.09
 Education, high school or less2.07 ± 1.540.180.01 ± 0.260.97
Provider-patient agreement    
 Severity of patient's OA0.68 ± 0.420.120.10 ± 0.070.17
 Importance of having TJR0.12 ± 0.340.710.04 ± 0.060.51
 Patient would benefit from TJR−0.99 ± 0.360.009−0.23 ± 0.060.001
 Concern about surgical complications−0.49 ± 0.290.10−0.03 ± 0.050.57

DISCUSSION

To our knowledge, the way that health care providers and patients make decisions about joint replacement surgery has received scant research attention. An informed decision will be one in which patients are actively involved in the process and in which they, along with their doctors, achieve consensus on the need for, risks of, and benefits of surgery (6, 16). This investigation examined communication factors affecting the degree of provider-patient agreement on these issues regarding whether the patient should have TKR. Several findings were noteworthy and have important implications for future research and clinical practice.

First, provider-patient agreement overall was modest, and sometimes quite disparate. In almost 1 in 5 encounters (18%), providers and patients differed on whether they thought TKR was even recommended or not. While providers and patients were mostly aligned on whether it was important for the patient to have TKR and its potential benefits, there were considerable differences related to beliefs about the severity of the patient's OA and concerns about surgical complications. Providers generally saw these as less serious issues than patients did. Although this may simply reflect differences in patient and provider opinions, our findings are comparable with the findings of studies in other settings that doctors are not particularly good judges of patient concerns and preferences (12, 35), that they underestimate patient's fears and preferences (36–38), and that they fail to pick up on cues to patient's emotions or concerns (39, 40).

Discrepancies in provider-patient beliefs about the risks of, benefits of, and need for TKR are not only a barrier to informed decision making, but such differences also can affect postconsultation outcomes. Several studies have shown that the less physicians understand patients' beliefs about health and preferences for care, the lower the patient's satisfaction with care, satisfaction with the decision, and adherence will be (12–15). This study found some evidence of this, in that patients were more satisfied with care and more inclined to follow treatment recommendations when they and the clinician were more closely aligned on how likely the patient was to benefit from TKR.

Providers and patients can improve decision making about joint replacement by employing communication practices that lead to a shared understanding of the patient's health status and the risks and benefits of surgery. Our results indicate that certain aspects of communication can contribute to provider-patient belief concordance, including more active patient participation (greater agreement on concerns about surgery), provider partnership building (greater agreement on the severity of the patient's arthritis), and patients' perceptions of the provider's informativeness (more agreement on the benefits of TKR). Active patient participation and provider partnership building represent efforts to bring the patient's perspective (beliefs, values, and concerns) into the encounter. By so doing, providers and patients have an opportunity to discover common ground in their viewpoints and, importantly, to reconcile differences (3).

While the patients' perceptions of the providers' informativeness were related to agreement on the benefits of TKR, both the behavioral and patient self-report measures of provider information giving were associated with greater belief discrepancy regarding the severity of the patient's OA. While one should not over-interpret this single finding, it does remind us that effectively informing patients is multifaceted. As Braddock et al note (16), informed decision making requires information about disease, treatments, options, and risks, as well as having that information presented in a way the patient comprehends. In the context of OA, there is some evidence indicating that options for joint replacement are sometimes not offered to patients who might otherwise be eligible (41), perhaps because doctors and patients cannot reach agreement on the severity of the patient's condition and their need for surgery. In this study, providers and patients may have spent considerable time discussing the nature of the surgery, including expected benefits, without having adequately talked about the need for TKR. This might explain why the provider's perceived information giving was related to greater provider-patient agreement on the possible benefits of having TKR, but also to more discrepancy in beliefs about the severity of the patient's OA. More research is needed on effective information giving so that all the requirements of informed decision making (e.g., severity of the condition, what the procedures entails, risks of treatment) are met.

Finally, there is widespread variation in the use of TKR with respect to socioeconomic status and race (19). Although these differences are related to a number of factors, there is increasing recognition that patient–doctor communication may play a role in disparities in care (30, 42). Our data showed greater discordance between African American patients and their providers regarding concerns about surgery, which may in part explain why TKR is underutilized in ethnic minorities. Teaching clinicians more patient-centered communication skills has been suggested as one means to address the communication needs for working with minority patients (43, 44), an idea supported by research showing that racial differences in trust and satisfaction are attenuated when physicians are more informative and supportive in their communication (21–23).

Our study had several limitations. First, although the sample was balanced with respect to African American and white patients, generalizability is limited because the sample size was relatively small (n = 74) and had few women. Also, although we found relationships between some communication behaviors and provider-patient agreement, there were also null results suggesting limited statistical power and the need for better assessment of communication, especially as they relate to different components of informed decision making. Third, we excluded 3 consultations because a companion contributed to a substantial portion of the conversation in the consultation. Because many TKR decisions are made with a companion present and participating, future research should examine the role these companions play, especially given that companions at times may be surrogates for or share floor time with patients during consultations (45). Fourth, communication behaviors may vary as a function of the providers' level of training (e.g., staff physicians versus trainees) and specialization. In this study, the distribution of providers precluded a test of these possibilities, which should be issues to explore in future research. Finally, future research will likely need to distinguish between misunderstanding between providers and patients (e.g., congruence between the patient's beliefs and how the provider thought the patient believed) from simply disagreeing (e.g., provider and patient differed in their opinion of the severity of the patient's OA).

Provider-patient agreement on the patient's health status and the most appropriate treatment is not only a prerequisite for informed and shared decision making; it also may contribute to better outcomes. We found some support for this notion in that patients were more satisfied with care and had a stronger intent to follow treatment recommendations when they and the provider were more in agreement on the whether the patient would benefit from TKR. However, in this study patients and providers often differed in their beliefs about the need for, risks of, and benefits of TKR. Active patient participation contributed to greater provider-patient agreement on concerns about surgery, perhaps because the provider gained a better understanding of the patient's perspective. Eliciting the patient's beliefs about the severity of their OA may also explain why provider partnership building was associated with more mutual understanding. More research is needed on how providers can be both comprehensive and comprehensible in the provision of clinical information and in discovering patient's beliefs and concerns.

AUTHOR CONTRIBUTIONS

Dr. Street had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Street, Suarez-Almazor.

Acquisition of data. Street, Richardson, Suarez-Almazor.

Analysis and interpretation of data. Street, Cox, Suarez-Almazor.

Manuscript preparation. Street, Richardson, Cox, Suarez-Almazor.

Statistical analysis. Street, Cox, Suarez-Almazor.

Ancillary