• racial differences;
  • doctor-patient communication;
  • treatment


  1. Top of page
  2. Abstract


Whether doctor-patient communication differs by race was investigated in patients with pulmonary nodules or lung cancer.


Eligible patients (n = 137) had pulmonary nodules or lung cancer and were seen in thoracic surgery or oncology clinics for initial treatment recommendations at a large southern Veterans Affairs Medical Center from 2001–2004. Doctor-patient consultations were audiotaped. Audiotapes were transcribed, unitized into utterances, and utterances were coded as doctors' information-giving or patients' and companions' active participation (asking questions, expressing concerns, and making assertions). Data were compared by patient race and doctor-patient racial concordance using t-tests or chi-square tests as appropriate. Mixed linear regression was used to determine the independent predictors of doctor's information-giving after controlling for clustering of patients by doctor.


Patient age, gender, marital status, clinical site, and health status were similar by race (P > .20), but black patients were somewhat less likely to have education beyond high school and to bring a companion to the visit (P = .06) than white patients. Black patients and their companions received significantly less information from doctors (49.3 vs. 87.3 mean utterances; P < .001) and produced significantly fewer active participation utterances (21.4 vs. 37.2; P < .001) than white patients. In mixed regression analyses, after adjusting for patients' and companions' participation, clustering by doctor, and other factors, race no longer predicted information-giving (P = .54). Patients in racially discordant interactions received significantly less information and were significantly less active participants (P < .001) when compared with patients in racially concordant interactions, and after controlling for patients' participation and other factors using mixed regression, racial discordance did not predict information-giving.


The results indicate a pattern of communication that may perpetuate patient passivity and limited information exchange where black patients and patients in discordant interactions do less to prompt doctors for information and doctors in turn provide less information to these patients. Cancer 2006. © 2006 American Cancer Society.

Black patients receive potentially curative surgical resection for early-stage lung cancer less often and also receive chemotherapy for advanced lung cancer less often than white patients.1–3 Racial differences in treatment are widely reported for other cancers and other conditions.4, 5 Studies suggest that racial differences in care are unlikely to be explained by biological differences in black and white patients5 and may be due at least in part to nonclinical factors. For example, treatment differences by race could be due to differences in patients' preferences,4 but studies in noncancer conditions suggest that racial differences in patients' preferences are small and not large enough to explain the magnitude of racial disparities in health care.6–8 Differences in treatment recommendations could also be explained by doctors' attitudes toward patients, but whether doctors' recommendations are racially biased and are the cause of disparities is difficult to study.9

Alternatively, racial disparities in health care may stem, in part, from problems in doctor-patient communication.10 Doctor-patient communication is the primary process by which medical decision-making occurs and numerous studies have shown that the communicative features of the consultation (e.g., information exchange, shared decision-making) can influence outcomes.11 Moreover, studies indicate communication in medical interactions sometimes varies by race. Although some report few racial differences in doctor-patient communication,12 others report that doctors are perceived as less informative13 and use less supportive communication14 when interacting with nonwhite than with white patients. Others studies show that white patients tend to be more active participants in medical consultations compared with nonwhite patients.14–18 Also, in medical interactions that are racially concordant (i.e., doctor and patient are the same race), patients perceived that they had better care19, 20 and had doctors engaging in more participatory decision-making21 than patients in discordant interactions, perhaps because of better communication. Racial disparities in doctor-patient communication could lead to less information exchange, less patient involvement in care, and less informed medical decisions.

The purpose of this investigation was to examine racial variation (black patients compared with white patients) in doctors' information-giving in lung cancer consultations. We focus on information-giving because this is a fundamental feature of medical consultations and, importantly, the foundation upon which medical decision-making occurs.22 How communication unfolds in medical encounters depends on a number of factors, such as the doctors' and patients' beliefs and goals, styles of communicating, perceptions of each other, and how each adapts and accommodates the communication of the other.23 Disparities in doctor-patient information-giving may occur for at least 2 reasons. First, doctors may believe some patients (e.g., middle-aged, more educated, white) are more interested in, more capable of understanding, or in need of more information than are other patients.24 If this was the case, we would expect variation in information-giving to be uniquely related to the patient's race. On the other hand, doctors also give more information to patients who are more active participants in the consultation.25, 26 Patients who ask questions, express concerns, and assert their preferences generally receive more information from doctors because these behaviors alert the doctor to patients' needs as well as raise expectations for the doctor to respond (e.g., answers are expected to follow questions).27 If this is the case, then racial variation in doctors' information-giving would not be uniquely related to race per se, but to the tendency for patients of 1 racial group to be less actively involved than patients from another group.


  1. Top of page
  2. Abstract


Eligible patients had biopsy-confirmed lung cancer or a pulmonary nodule or mass requiring surgical diagnosis at presentation to thoracic surgery or oncology clinics at a large southern Veterans Affairs Medical Center (thus, resulting in a predominantly male study sample) between April 2001 and March 2004. Patients with lung cancer were identified at diagnosis from pathology reports and patients with suspicious pulmonary nodules were identified at a weekly multidisciplinary clinical conference. At the conference, consultants in pulmonary medicine, thoracic surgery, oncology, radiation oncology, pathology, and radiology reviewed each case and recommended treatment or further diagnostic evaluation. The Institutional Review Board approved the study and all participating patients and doctors provided informed consent.

We identified 252 eligible patients and 175 consented to participate. Of these, 23 patients were not audiotaped because of logistical difficulties (n = 3), or because the doctor refused to be audiotaped (n = 20), leaving 152 consultations that were audiotaped. We excluded 13 patients because of incomplete or inadequate recordings and 2 Hispanic patients, leaving 137 consultations for analysis. Patients in the analysis did not differ significantly from excluded patients by age, gender, or race (P > .05). Patients consulted with 1 of 15 medical doctors (5 attendings and 10 fellows) or 2 physician assistants (2 were white Hispanic, 8 were white non-Hispanic, 2 were black, and 5 were Asian; 4 were in oncology and 13 in thoracic surgery).


Patient demographics, including race, educational attainment (high school or less, or more than high school), and marital status were determined by patient self-report. We assessed patient health status using the Short Form-12 version 1 (SF-12).28 Medical encounters were audio-recorded using a standard cassette recorder placed on doctors' desks during the interaction.

Verbal behavior coding

Doctors' and patients' communication behaviors were coded using a system developed by Gordon et al.18 and Street and Millay.29 Doctors' information-giving statements included information in the form of diagnoses, prognoses, explanations, rationales, instructions, risks, benefits, and treatment options. Doctors' information-giving was further coded into 2 categories according to whether the utterance was “self-initiated” by the doctor or was “prompted” by the pa tient's communication (e.g., asking a question, offering an opinion).

Patients' participation was coded for 3 types of verbal communication behaviors that have the potential to influence doctors' behavior, perceptions of the patient, and treatment decisions.29 These include asking questions, being assertive, and expressing concerns. Questions are utterances in interrogative form that ask for information and clarification. Assertions are utterances where the patient interjects his or her beliefs, preferences, and perspective into the consultation. Examples of assertiveness include offering an opinion about health or treatment, making a recommendation, disagreeing with the doctor, making a request, and introducing new topics for discussion. Concerns are statements of negative affect (worry, frustration, anger, fear) and may be signaled by word choice (e.g., “worried,” “afraid,” “upset”) or by an emotional tone of voice. These were summed to create a composite index of active pa tient participation. Because more than half the patients brought a companion to the consultation, active participation was coded separately for patient and companion.

Two individuals, who did not interact with study participants, were trained to code the verbal behaviors. Coders were not informed of the purpose of the study nor of patients' characteristics. Transcripts of the tape recordings were unitized into utterances, the units of analysis for coding the different types of behaviors into the communication categories. An utterance is the oral analog of a simple sentence and may be in the form of a complete sentence, independent clause, nonrestrictive dependent clause, multiple predicate, or evaluation.30, 31 Coders then followed the transcript while listening to the tape, and then identified the targeted behaviors whenever they occurred. Thus, the coding scheme was not exhaustive but did allow comparison of the proportion of coded behaviors to the total number of utterances. Reliability was established by having both coders code a subset of 15 doctor-patient consultations independently of one another. Coders achieved 90% agreement on unitizing utterances. Reliabilities, calculated using Cohen kappa,32 for doctor information-giving and active patient participation were 0.83 and 0.78, respectively. The remaining doctor-patient consultations were divided into 2 groups and coding was completed after giving 1 group to each coder.


Bivariate associations of demographic characteristics, functional status, and perceptual measures with patient race were examined using the t-test, Wilcoxon, or χ2 test. Sociodemographic characteristics included age (continuous), gender (male or female), educational level (high school up to graduation vs. beyond high school), and marital status (not married or married). Functional status measures consisted of the SF-12 mental and physical component summary scores (continuous). To identify factors associated with doctors' information-giving, we used multivariable mixed linear regression. We modeled the doctor as a random effect to adjust for potential clustering of patients by doctor; other independent variables were modeled as fixed effects. Independent variables included patient demographic characteristics (age, race, educational level) and functional status measures. Variables were entered in groups by race, other patient and visit characteristics, and communication behavior. Race was the only independent variable included in the first model (Model 1). For Model 2, sociodemographic characteristics, functional status, and presence of a companion were added to Model 1. In the final model (Model 3), patients' active communication behavior was added to Model 2. Similarly, we conducted secondary analyses to examine the association of doctor-patient racial concordance and racial discordance with doctors' information-giving. Analyses were conducted using SAS statistical software, v. 9.0 (SAS Institute, Cary, NC). We considered a P-value of 0.05 as statistically significant.


  1. Top of page
  2. Abstract

Black and white patients were similar (P > .05) in mean age, gender, proportion visiting the oncology or thoracic surgery clinic, and in self-reported mental and physical health status. Fewer black patients had attended college, or brought a companion to the medical consultation when compared with white patients, although the differences were not statistically significant (P = .06; Table 1). As measured by the total number of utterances, visits with black patients were shorter than visits with white patients (254 vs. 403 mean total utterances; P < .001).

Table 1. Characteristics of Patients
 Black (n = 30)White (n = 107)P
Age >65 y50.0%51.4%.89
Some college (n = 133)27.6%47.1%.06
Married (%) (n = 131)39.3%50.5%.29
Participating companion/spouse46.7%65.4%.06
Clinic site (% oncology visits)50.0%41.1%.39
Mental health (SF-12) (n = 135)62.664.0.32
Physical health (SF-12) (n = 135)37.439.9.31

The frequency of doctors' information-giving and patients' and companions' active participation in all visits are shown in Table 2. Compared with white patients, consultations with black patients, on average, had fewer doctors' information-giving utterances (49.3 vs. 87.3 mean information-giving utterances; P < .001) and fewer acts of patient and companion participation (21.4 vs. 37.2 mean patient participation utterances; P < .001). We also performed stratified analyses (Table 2) according to whether consultations were dyadic (i.e., patient and doctor) or patients brought a companion to the consultation (triads). In dyads, doctors' information-giving was not statistically different among consultations with black and white patients, but black patients were significantly less active participants in the interaction than white patients. In triads, black patients and their companions received significantly less information-giving statements from their doctor and were significantly less active participants when compared with white patients and their companions (Table 2).

Table 2. Mean Doctor's Information-Giving and Patient (and Companion) Active Participation in All Visits, Dyads and Triads
All visits(n = 30)(n = 107) 
 Doctor's information-giving (total)49.387.3<.001
 Doctor information (prompted by patient)14.534.6<.001
 Doctor information (self-initiated by doctor)34.852.6.02
 Patient and companion participation21.437.2<.001
Dyads(n = 16)(n = 37) 
 Doctor's information-giving56.473.5.23
 Patient participation20.540.1.002
Triads(n = 14)(n = 70) 
 Doctor's information-giving41.194.6.001
 Patient and companion participation22.435.8.05

In additional analyses of racial differences, we examined whether doctors' information-giving was self-initiated by the doctor or prompted by the patient. We found that both prompted and self-initiated doctors' information-giving statements were less frequent in consultations with black patients compared with white patients (P < .05; Table 2).

In addition to the analyses of frequency of communication behaviors, analyses were also conducted to examine whether proportional measures of these communication behaviors differed by race. The proportion of doctors' information-giving to the total number of doctors' utterances was statistically similar, on average, among visits with black patients compared with white patients (32% vs. 38%; P = .09). When examining the substratum of information-giving that was prompted by the patient, black patients received a lower proportion of prompted information than white patients (10 vs. 15%; P = .005), but the proportion of self-initiated information-giving from doctors did not differ by race.

To determine the independent relation among several potential predictors of doctors' information-giving and to adjust for clustering of patients by doctor, we used mixed effects multivariable regression. Variables were added to the models in 3 groups (Table 3). In Model 1, doctors' information-giving was significantly lower by 24.3 mean utterances among consultations with black compared with white patients. After adding other patient and visit characteristics and whether the patient brought a companion to the visit (Model 2), doctors' information-giving was 21.7 mean utterances lower. When the number of patient (and companion) active participation utterances were added (Model 3), doctors' information-giving did not differ significantly by patient race, but each act of active patient (and companion) participation was associated with an increase of 1.1 statements of information from the doctor and visits with a companion present were associated with 18.2 more statements of information from the doctor (Table 3). In additional stratified analyses according to whether information was prompted or self-initiated (Table 4), doctors' self-initiated information-giving was not statistically different by race, after controlling for patient and visit characteristics, presence of a companion, and clustering of patients by doctor. In contrast, when examining information-giving that was prompted by the patient, black patients received, on average, 16.5 fewer information-giving statements from their doctor, even after controlling for several potential covariates.

Table 3. Independent Predictors of Doctors' Information-Giving*
 Model 1 (n = 137)Model 2 (n = 131)Model 3 (n = 131)
Estimate (SE)PEstimate (SE)PEstimate (SE)P
  • SE, standard error.

  • *

    Controlling for clustering of patients by physician.

Black−24.3 (9.3)0.01−21.7 (9.6)0.03−5.0 (8.0).54
Age >65 y  −2.5 (7.7)0.75−1.9 (6.3).77
Some college  −2.9 (7.9)0.72−1.4 (6.4).83
Oncology clinic  8.3 (21.7)0.718.7 (18.9).65
Mental health  −1.0 (0.5)0.06−0.7 (0.4).09
Physical health  0.6 (0.4)0.080.6 (0.3).05
Companion present (triad/dyad)  15.5 (8.0)0.0618.2 (6.5)0.01
Patient + companion participation    1.1 (0.1)<.001
Table 4. Independent Predictors of Doctors' Information-Giving Stratified by Whether Information Was Prompted or Self-Initiated*
 Prompted information-giving (n = 131)Self-initiated information-giving (n = 131)
Estimate (SE)PEstimate (SE)P
  • SE, standard error.

  • *

    Controlling for clustering of patients by physician.

Black−16.5 (5.6)0.004−7.1 (6.2).25
Age >65 y2.5 (4.5)0.59−5.2 (5.1).30
Some college−0.8 (4.6)0.87−2.0 (5.2).70
Oncology clinic1.2 (6.8)0.874.8 (14.9).76
Mental health status−0.5 (0.3)0.11−0.5 (0.4).19
Physical health status0.4 (0.2)0.060.2 (0.2).32
Companion present (triad / dyad)6.1 (4.7)0.198.9 (5.2).09

We conducted secondary analyses to examine whether doctor-patient communication was associated with doctor-patient racial concordance. Consultations were racially concordant (n = 62) or racially discordant (n = 75) when the doctor and patient were the same or different race, respectively. Patients (and companions) in racially concordant visits received significantly more information from their doctor (99.4 vs. 62.1 mean utterances; P < .001) and were significantly more active participants (41.7 vs. 27.2 mean utterances; P < .001) when compared with patients in racially discordant visits. We used mixed regression to examine the independent relation of concordance with information-giving and to adjust among several predictors of doctor's information-giving and for clustering of patients by doctor. After adjusting for patients' active participation and other factors, doctors' information-giving was not significantly different among racially concordant and racially discordant consultations (9.4 mean utterances, P = .37). In stratified analyses according to whether information was prompted or self-initiated, doctors' self-initiated information-giving was not statistically different by doctor-patient racial concordance after controlling for patient and visit characteristics, the presence of a companion, and clustering of patients by doctor. When examining-information giving that was prompted by the patient, patients in racially concordant consultations received, on average, 18.5 more information-giving statements (P = .003) from their doctor, even after controlling for several potential covariates.


  1. Top of page
  2. Abstract

In this study, black patients received less information from their doctors, were less active participants, and were less likely to bring a companion to consultations for lung cancer or a pulmonary nodule. However, we did not find significant differences in the frequency of doctors' information-giving by patient race after adjusting for differences in the frequency of active participation behaviors of black and white patients and their companions. Moreover, in adjusted, stratified analyses, we found no differences by race in the frequency of information provided by doctors that was self-initiated, but information provided by doctors that was prompted by patients was significantly less frequent among visits with black compared with white patients. Similar results were found for patients in racially discordant compared with racially concordant consultations. In other words, black patients, and patients in racially discordant consultations, in this study received less information overall because they less often engaged in communication behaviors (e.g., questions, concerns, assertions) that typically elicit more information from doctors. These findings are notable because, while not directly negating the possibility that racial disparities in care are due to doctor bias or patient preferences, they suggest that disparities in medical care are related in part to the communicative dynamics of the encounter, particularly the degree to which patients are actively involved.

Our study adds to other studies suggesting that some medical consultations with black patients and some racially discordant consultations have barriers to less effective doctor-patient communication in part because of less active participation behaviors.17, 18 Communication may be more difficult for black patients than for white patients for a number of reasons. First, some black patients may have difficulty connecting with doctors if their explanatory models about health and illness are dramatically different.10, 33 For example, black patients were more likely to believe that exposure of lung cancer to air at surgery would cause tumor spread, and as a result of this belief would be more likely to decline a recommendation for potentially curative surgery.33 Second, communication may be more difficult for black patients due to racial differences in trust in doctors,13 to fears among blacks of exploitation by the medical profession due to knowledge of prior discrimination against blacks in health care.34 Third, those doctors that do have less favorable attitudes toward black patients35 may subtly communicate distance or less interest in ways that make patients become more cautious. These issues may be most prominent in racially discordant interactions (physician and patient of different race), which have less patient satisfaction and less participatory decision-making.19, 21


Potential racial variation in doctor-patient communication becomes an issue of concern especially when considering a growing body of research that links patterns of communication to outcomes of care. Studies indicate that when doctors are less informative and more controlling, patients are less likely to gain adequate understanding of their health condition and treatment options,36 are less likely to adhere to the doctor's recommendations,37 and may experience poorer health after the consultation.11 Moreover, when patients assume a passive role in the interaction, doctors may not get sufficient information for making appropriate treatment decisions,38 and patients may be less committed and less satisfied with those recommendations.16, 39

Fortunately, communication is a skill that can be taught. Patients' communication with their doctors can be improved with training to increase patients' participation in medical encounters. For example, patients who were coached to ask questions, negotiate medical decisions, and to overcome barriers when talking to doctors were more active participants compared with patients who were educated with disease-specific information.40 In addition, communication interventions using a pamphlet to prompt patients to write down questions before the visit41 or using patient education videotapes to role-model active behaviors (asking questions, initiating a discussion of concerns)42 increased patients' active participation in medical consultations when compared with control groups who received information alone. In addition, training of medical students and doctors about racial disparities in doctor-patient interactions may improve communication with patients from different cultural backgrounds.43, 44 Furthermore, encouraging patients to bring a companion to important consultations, such as discussion of potential cancer diagnosis and treatment and including patients' companions in communication training, may improve communication in medical interactions.


Our results should be considered in the context of several limitations. First, our study was small and is based on data from 2 clinics in 1 hospital with 17 providers and 137 mostly male patients, 30 of whom were black. Thus, our results may have limited generalizability to women and to other racial and ethnic groups, practice settings, and geographic locations. Second, black patients may be less willing to participate in research.45 In our study, we did not find statistical differences in rates of participation by race; however, our results could be biased if patients who did not participate were systematically more or less active participants than participating patients. In addition, our study evaluated a single visit and does not evaluate potential improvements in communication across multiple visits and with ongoing continuity of care.


Information is an important resource that doctors provide to patients, is a feature of patient-centered care, and may help develop the doctor-patient relationship. Medical consultations with less communication may have less information exchange, less patient involvement in care, and could lead to less informed medical decisions. Our findings raise concern for a pattern of communication that may perpetuate pa tient passivity and limited information-exchange where black patients, when compared with white patients, do less to prompt the doctor for information and the doctor, in turn, provides less information to black patients.


  1. Top of page
  2. Abstract