Objectives. We examined the associations of clinicians’ implicit attitudes about race with visit communication and patient ratings of care.

Methods. In a cross-sectional study of 40 primary care clinicians and 269 patients in urban community-based practices, we measured clinicians’ implicit general race bias and race and compliance stereotyping with 2 implicit association tests and related them to audiotape measures of visit communication and patient ratings.

Results. Among Black patients, general race bias was associated with more clinician verbal dominance, lower patient positive affect, and poorer ratings of interpersonal care; race and compliance stereotyping was associated with longer visits, slower speech, less patient centeredness, and poorer ratings of interpersonal care. Among White patients, bias was associated with more verbal dominance and better ratings of interpersonal care; race and compliance stereotyping was associated with less verbal dominance, shorter visits, faster speech, more patient centeredness, higher clinician positive affect, and lower ratings of some aspects of interpersonal care.

Conclusions. Clinician implicit race bias and race and compliance stereotyping are associated with markers of poor visit communication and poor ratings of care, particularly among Black patients.

Racial/ethnic disparities in healthcare are documented across conditions, settings, diagnostic and treatment modalities, and dimensions of technical quality,1 and ethnic minorities rate interpersonal quality of care from physicians more negatively than do Whites.2–6 Minorities experience poorer communication with physicians,7,8 particularly in race-discordant patient–clinician relationships.9,10 The Institute of Medicine’s report “Unequal Treatment,” suggests disparities in healthcare emerge from

bias (or prejudice) against minorities; greater clinical uncertainty when interacting with minority patients; and beliefs (or stereotypes) held by the provider about the behavior or health of minorities.1(p9)

Biases may reflect explicit (conscious) biases or implicit (unconscious) biases. There is evidence that physicians have more negative explicit attitudes toward Blacks than towards Whites, including stereotypes about nonadherence,11–13 and that negative explicit attitudes are associated with and mediate racial disparities in physicians’ treatment decisions.13,14 Physicians’ implicit racial biases have been linked to some treatment decisions in clinical vignettes.15–17 Yet, little work has examined how clinicians’ implicit racial attitudes affect communication and patient experiences in actual medical encounters.18

We examined 2 implicit attitudes about race among clinicians. The first relates to general racial bias; the second is specific to the medical context, assessing racial bias regarding stereotyping patient compliance. We explored these distinct measures of implicit bias because they represent potentially different pathways to medical care (Figure A, available as a supplement to the online version of this article at http://www.ajph.org). We hypothesized that clinicians’ implicit race bias is associated with nonspecific aspects of communication (e.g., socioemotional and stylistic, but not necessarily medically, focused). However, the pathway by which race-based compliance stereotyping affects communication may be different from that proposed for general racial bias because it may relate to physicians’ professional obligations to enhance patient adherence. Two implicit association studies have documented a physician pro-White bias regarding the concept of the compliant patient.15,16 We reasoned that a race and compliance stereotype would have 2 communication-related consequences: less positive emotional tone, reflecting frustration with a patient perceived as nonadherent; and heightened attention to providing medical information. Given evidence that Black patients are sensitive to pro-White bias,19,20 we hypothesized that both implicit measures are associated with more negative ratings of interpersonal care in the visits of Black, but not White, patients.

Secondary data came from enrollment visits of clinicians and patients who participated in 2 randomized clinical trials of interventions to enhance patient–provider communication and outcomes for patients with hypertension (study 1) and depression (study 2).21,22 In each study, clinicians randomized to a communication skills training program (not focused on race) received detailed individualized feedback on their communication behaviors measured from a video-recorded encounter with a Black standardized patient. Information collected from participants included a self-administered clinician baseline survey (study 1: January 2002–January 2003; study 2: June 2004–March 2006) and patient interviews upon study enrollment (study 1: September 2003–August 2005; study 2: October 2005–August 2006). We audiotaped patients in both studies with their clinician at enrollment, and patients completed a postvisit questionnaire.

We collected primary data after the clinician interventions via an Internet survey that included 2 cognitive tests of implicit racial attitudes and stereotyping derived from the Implicit Association Test (IAT). The median time between the patient’s baseline interview and the clinician completing the online survey was 9.4 months (range = −3.2–30.2 months).

Study Variables

Independent variables were 2 measures of implicit attitudes about race from clinicians’ survey responses. Dependent variables, collected in the earlier 2 studies, included patient–clinician communication measures from audiotapes of medical visits and patient perceptions of the clinician from postvisit surveys.

Measures of implicit attitudes about race.

The IAT is a computer-based, indirect measure of social cognition widely used in social psychology to measure implicit attitudes and stereotypes about race and other sociocultural phenomena.23 It measures the relative association strength between a pair of target concepts such as race (White vs Black) and a pair of attribute categories such as good versus bad.24–26 The IAT’s measure is predicated on the assumption that concepts that the test taker readily associates will be sorted together more quickly than are concepts that are weakly associated. Whereas the race attitude IAT examines how much generic positive terms are associated with Black versus White faces, the race and medical compliance IAT examines a more complex association of a race and treatment adherence stereotype.15,16

On the basis of previous IAT research, the general race attitude IAT used words such as joy, wonderful, and laughter to represent the concept of good and agony, evil, and hurt to represent the concept of bad (Figure B, available as a supplement to the online version of this article at http://www.ajph.org). We constructed the race and compliance patient IAT to measure the association between race and the concept of a “compliant patient.”16 The category of compliant patient is represented by, for example, willing, reliable, and helpful. Words to represent the reluctant patient category are, for example, reluctant, apathetic, and lax (Figure C, available as a supplement to the online version of this article at http://www.ajph.org). The IAT score is derived from the difference in average response time on the 2 sorting tasks.24 We scored the IATs according to published guidelines26; possible scores ranged from −2 to +2, and scores of zero indicated no greater preference for White relative to Black or no greater association of White than Black with the compliant patient concept. Higher positive scores indicated greater preference for White or greater association of White with compliant patient. We gave participants the race attitude IAT and the race and compliance IAT16 in randomly assigned order. The IAT is a stable measure of implicit cognition when tested over time.27

Medical visit audio recordings.

We analyzed audiotapes of primary care visits with the widely used Roter Interaction Analysis System, a coding system with demonstrated reliability and predictive validity.28–30 This system assigns each thought the patient and clinician express to mutually exclusive and exhaustive codes that can be combined to reflect categories of exchange, including functions of the medical interview.31 We measured visit length (in minutes), speech speed (the number of statements per minute), clinician verbal dominance (ratio of clinician to patient statements), and patient centeredness (ratio of the sum of psychosocial, rapport-building, and facilitative behaviors by clinicians and patients representing the patient’s agenda to the sum of biomedical questions, information giving, and closed-ended questions representing the clinician’s agenda).7,9 Verbal dominance is an indicator of the level of participation of the clinician relative to the patient in the dialogue, with scores greater than 1 meaning the clinician verbally dominated the dialogue. Patient centeredness and verbal dominance have demonstrated concurrent and predictive validity and have been linked to patient satisfaction and reported rapport with clinicians.28,30,32

Coders also rated the emotional tone of the dialogue on a 6-point scale (1 = low or none to 6 = high) on global dimensions of patient positive affect (e.g., responsiveness) and clinician positive affect (e.g., friendliness). We assessed intercoder reliability (the Pearson r) on a 10% random sample of double coded tapes; reliability averaged 0.90 over clinician and 0.86 over patient verbal categories. Coder agreement within 1 point on affect scales was 88% to 100%. Except affective tone, which was only socioemotional, communication measures contained socioemotional and task-focused medical elements (Figure A, available as a supplement to the online version of this article at http://www.ajph.org).

Patient perceptions of clinicians measured by postvisit survey.

We used measures of interpersonal care related to continuity of care and patient adherence in other studies33–40 and shown to differ by patient race3,5 and race concordance with clinicians.9 Patients rated their primary care clinician’s attitude toward them as (1) “My doctor likes me,”33 and (2) “My doctor has a great deal of respect for me.”34 Patients then rated how they felt about their primary care clinician as (1) “I would recommend this physician to a friend,” (2) “I like this doctor,” (3) “I trust this doctor to look out for my best interests,” and (4) “I have confidence in this doctor’s knowledge and skills.”9,33,35 Finally, we asked patients, if there were a choice between treatments, how often would this doctor ask you to help make the decision?36 Five-point Likert scales were skewed toward positive responses. We dichotomized responses as the top category (e.g., strongly agree) versus all others.

Other variables.

From clinicians we collected age, gender, race, type of clinician (physician vs nurse practitioner), board certification status, location of training (United States or not), number of years in practice, and political identity (self-report on a 6-point scale from strongly conservative to strongly liberal). We collected self-reported measures from clinicians designed to parallel the targeted concepts in the IATs, including preferences or feelings toward and perceived cooperativeness of Whites and Blacks.15–17 From patients, we collected age, gender, race, educational attainment, employment status, insurance coverage, annual household income, living arrangements (alone or not), and physical and mental health status (the Medical Outcomes Short Form 12).41

Statistical Analyses

We used linear and logistic regression with generalized estimating equations to assess the strength of associations between clinician pro-White bias or stereotyping and our dependent variables while accounting for the nesting of patients within clinician.42,43 We calculated confidence intervals using robust empirical SE estimates. We identified covariates for multivariate analysis from their known associations with patient–physician communication and included clinician gender; patient age, gender, and education; and the mental component of the Medical Outcomes Short Form 12. We evaluated the effect of clinician race on outcomes with separate analyses stratified by clinician race. We analyzed the data with all patients in the models, including the IAT measure, patient race, and the interaction of the IAT measure and patient race as covariates.

We stratified the results by patient race only. For communication measures, which are continuous and were analyzed with linear models using generalized estimating equations, the estimate for “implicit bias” is the difference in the outcome associated with an increase in bias score of 0.5 (considered a moderate level of bias). For patient perceptions of clinicians, which were dichotomous and analyzed with logistic models using generalized estimating equations, we estimated the predicted probabilities of having each outcome for a hypothetical patient whose clinician exhibited no implicit bias (a score of 0.0) and a similar hypothetical patient whose clinician exhibited a moderate level of implicit bias (a score of 0.5). We performed analyses using SAS version 9.2 (SAS Institute, Cary, NC).

Forty of the 63 clinicians (63%) with eligible patients (self-reported race Black or White) participated. There were no differences in demographic and professional characteristics or intervention assignments between clinician participants and nonparticipants. Table 1 summarizes clinician characteristics and scores on the 2 IAT measures. The mean scores (range) for participating clinicians were 0.26 (−1.23 to 1.32) on the race attitude IAT and 0.29 (−0.60 to 1.39) on the race and compliance IAT. The Cohen d, an effect size measure,44 indicated a moderate implicit bias (0.54) for the race attitude IAT and a slightly stronger association of White race with compliance (0.70) for the race and compliance IAT. The relationship between the 2 IAT measures was weaker among non-Black than Black clinicians (the Pearson ρ = 0.29 and 0.70, respectively). Clinicians who considered themselves more politically conservative had higher implicit race attitude IAT scores, and White clinicians had higher implicit race and compliance IAT scores. Clinician exposure to communication skills training was not associated with lower IAT scores. On explicit measures, clinicians perceived Whites as more cooperative patients than Blacks, but race attitudes (e.g., warmth, preference) were neutral or favorable toward Blacks (Table A, available as a supplement to the online version of this article at http://www.ajph.org).

Table

TABLE 1— Clinicians’ Characteristics and Implicit Association Test (IAT) Scores: The Race and Relationship-Centered Care Study, Baltimore, MD, January 2002–August 2006

TABLE 1— Clinicians’ Characteristics and Implicit Association Test (IAT) Scores: The Race and Relationship-Centered Care Study, Baltimore, MD, January 2002–August 2006

CharacteristicMean ±SD, No. (%), or Median (Range)Race Attitude IAT Score,a Mean ±SDPbRace and Compliant Patient IAT Score,a Mean (SD)Pb
All clinicians40 (100)0.26 ±0.49.0020.29 ±0.41< .001
Age, y.82.13
 27–3915 (38)0.20 ±0.460.23 ±0.43
 40–4914 (36)0.27 ±0.670.47 ±0.43
 50–6210 (26)0.32 ±0.300.13 ±0.33
Gender.43.12
 Female25 (62)0.22 ±0.520.21 ±0.41
 Male15 (38)0.35 ±0.450.42 ±0.40
Race.07.01
 White19 (48)0.32 ±0.490.47 ±0.33
 Black9 (22)–0.05 ±0.45–0.01 ±0.39
 Asian, including Indian subcontinent12 (30)0.41 ±0.450.20 ±0.43
Type of clinician.5.93
 Physician36 (90)0.28 ±0.510.29 ±0.42
 Nurse practitioner4 (10)0.10 ±0.380.27 ±0.45
US medical graduate.06.47
 Yes34 (85)0.20 ±0.470.27 ±0.43
 No6 (15)0.61 ±0.520.40 ±0.33
Board certified.53.45
 Yes36 (90)0.28 ±0.510.30 ±0.42
 No4 (10)0.11 ±0.390.11 ±0.36
Specialty.34.22
 Internal medicine31 (78)0.31 ±0.500.33 ±0.41
 Family medicine9 (22)0.13 ±0.460.13 ±0.39
Years since completing residency13.4 ±7.3
Political identity.09.37
 Conservative10 (29)0.53 ±0.540.41 ±0.40
 Liberal25 (71)0.26 ±0.350.27 ±0.41
Number of patients6.5 (1–16)
Communication skills intervention assignment.19.15
 Yes18 (45)0.38 ±0.510.39 ±0.43
 No22 (55)0.17 ±0.470.20 ±0.38

Note. Although 40 clinicians participated, 1 did not provide age data, 5 did not complete the political identity question, 1 did not complete the race IAT, and 1 did not complete the race and compliant patient IAT.

aThe IAT scores do not have units. They are the standardized difference in response time to the 2 tasks (the difference in latency measures [measured in milliseconds] divided by the standard deviation).

bThe P values for the IAT scores for all clinicians compare the mean scores to a score of zero using a 2-sided 1 sample t-test. Other P values are from t-tests or analysis of variance (if more than 2 groups) to compare mean scores across physician characteristics.

Patient Characteristics

Of 445 potentially eligible patients, 269 (60%) had clinicians who completed the IAT survey. Greater proportions of patients included in the analysis were high school graduates, were insured, and had an annual household income greater than $35 000 than did those excluded. Smaller proportions were assigned to or had clinicians who were assigned to patient-centered interventions. Similar proportions of included and excluded patients were audio taped at their enrollment visit. The sample was largely middle-aged women. More than two thirds were known to the clinician. Blacks were younger, were less likely to be married, were more likely to have Medicaid insurance coverage, had poorer mental health status, and were less likely to be in race-concordant relationships with clinicians than were Whites (Table 2).

Table

TABLE 2— Patient Characteristics, Overall and by Race: The Race and Relationship-Centered Care Study, Baltimore, MD, January 2002–August 2006

TABLE 2— Patient Characteristics, Overall and by Race: The Race and Relationship-Centered Care Study, Baltimore, MD, January 2002–August 2006

Characteristic, Frequency, %All Patients (n = 269), No. (%) or Mean ±SDBlack Patients (n = 213), No. (%) or Mean ±SDWhite Patients (n = 56), No. (%) or Mean ±SDPa
Age, y56.2 ±13.254.5 ±13.362.7 ±10.4< .001
Female gender192 (71)156 (73)36 (64).19
High school graduate210/260 (81)165/204 (81)45 (80).99
Married107/267 (40)70/211 (33)37 (66)< .001
Lives alone58 (22)52 (24)6 (11).03
Health care insurance
 Has any health insurance247 (92)194 (91)53 (95).58
 Has Medicaid63/259 (24)55/203 (27)8 (14).05
 Has Medicare87/256 (34)67/200 (34)20 (36).75
 Has private health insurance163/257 (63)125/201 (62)38 (68).53
Annual household income < $35 000150/254 (59)122/203 (60)28/51 (55).53
Employment status, working111/267 (42)95/211 (45)16 (29).03
Mental component score of the Medical Outcomes Short Form 12b47.0 ±12.646.1 ±13.250.4 ±9.2.006
Physical component score of the Medical Outcomes Short Form 12b41.8 ±12.941.7 ±13.042.2 ±12.5.83
Moderately well known by clinician155/224 (69)113/173 (65)42/51 (82).02
Seen by race-concordant clinician66 (25)42 (20)24 (43)< .001

Note. If there are any missing data for the characteristic, the number of patients with data is specified.

aP value from the Fisher exact test or 2-sample t-test.

bThere were 266 patients with data for the Medical Outcomes Short Form 12 (211 Blacks and 55 Whites).

General Race Bias and Interpersonal Care

More implicit bias on the race attitude IAT was linked to 2 measures of communication process: more clinician verbal dominance in the visits of Blacks (9% greater) and Whites (11% greater) and a 0.10-point lower patient positive affect score in the visits of Blacks (Table 3). For Black patients, higher levels of implicit bias on the race attitude IAT were also linked to lower predicted probabilities of perceiving respect from the clinician (15% lower), liking the clinician themselves (14% lower), having confidence in the clinician (9% lower), and recommending the clinician to others (13% lower; Table 4).

Table

TABLE 3— Implicit General Race Bias and Race and Medical Compliance Stereotyping as Predictors of Patient–Clinician Communication by Patient Race: The Race and Relationship-Centered Care Study, Baltimore, MD, January 2002–August 2006

TABLE 3— Implicit General Race Bias and Race and Medical Compliance Stereotyping as Predictors of Patient–Clinician Communication by Patient Race: The Race and Relationship-Centered Care Study, Baltimore, MD, January 2002–August 2006

Implicit General Race Bias, Black Patients (n = 131)a
Implicit General Race Bias, White Patients (n = 48)
Implicit Race and Medical Compliance Stereotyping, Black Patients (n = 135)a
Implicit Race and Medical Compliance Stereotyping, White Patients (n = 48)
Communication BehaviorMean Estimateb (95% CI)PMean Estimateb (95% CI)PMean Estimateb (95% CI)PMean Estimateb (95% CI)P
Verbal dominance ratio.05.01.14.02
 No implicit bias1.47 (1.29, 1.66)1.27 (1.09, 1.49)1.48 (1.28, 1.71)1.56 (1.26, 1.94)
 Implicit bias9% (0, 19)11% (2, 21)10% (−3, 24)−25% (−42, −4)
Visit length, minutes.36.14.02.001
 No implicit bias14.0 (11.7, 16.7)17.1 (14.6, 19.9)13.3 (11.2, 15.7)19.5 (16.0, 23.8)
 Implicit bias7% (−7, 22)8% (−17, 3)20% (3, 40)−21% (−31, –9)
Speech speed, statements per minute.23.83.02.001
 No implicit bias25.5 (24.1, 26.9)23.7 (21.6, 25.8)25.8 (24.6, 27.0)21.6 (19.4, 23.8)
 Implicit bias–0.76 (−2.02, 0.50)0.25 (−2.05, 2.55)–1.75 (−3.25, −0.25)3.9 (1.6, 6.3)
Patient centeredness ratio.63.37.06.02
 No implicit bias1.66 (0.95, 2.37)0.70 (0.58, 0.82)1.97 (1.00, 2.94)0.60 (0.45, 0.74)
 Implicit bias–0.10 (−0.51, 0.31)–0.05 (−0.17, 0.07)–0.93 (−1.91, 0.04)0.15 (0.02, 0.28)
Clinician positive affect.14.78.35.02
 No implicit bias3.63 (3.51, 3.75)3.38 (3.31, 3.45)3.60 (3.49, 3.72)3.30 (3.20, 3.40)
 Implicit bias–0.10 (−0.23, 0.03)–0.01 (−0.07, 0.05)–0.06 (−0.19, 0.07)0.12 (0.02, 0.21)
Patient positive affect.04.87.53.09
 No implicit bias3.39 (3.30, 3.49)3.31 (3.19, 3.43)3.36 (3.26, 3.45)3.24 (3.10, 3.38)
 Implicit bias–0.10 (−0.19, –0.00)0.01 (−0.09, 0.10)–0.04 (−0.16, 0.08)0.11 (−0.02, 0.24)

Note. CI = confidence interval. Adjusted for clinician gender and patient age, gender, education, and the mental component of the Medical Outcomes Short Form 12.

aWe excluded 1 observation from speech speed and 1 from patient centeredness because they were extreme outliers.

bWe estimated the means while holding all other covariates at their means. The estimate for “no pro-White bias” for verbal dominance and visit length is the geometric mean from the generalized estimating equations (GEE) model for a bias score of zero; the estimate for “implicit bias” is the percentage change in verbal dominance and visit length associated with a 0.5-point increase in the bias score. For all other variables, the estimate for “no implicit bias” is the mean outcome score from the GEE model for a bias score of zero; the estimate for “implicit bias” is the change in the outcome associated with a change in bias score of 0.5 (considered a moderate level of bias).

Table

TABLE 4— Implicit General Race Bias and Race and Medical Compliance Stereotyping as Predictors of Patient Perceptions of Clinician, by Patient Race: The Race and Relationship-Centered Care Study, Baltimore, MD, January 2002–August 2006

TABLE 4— Implicit General Race Bias and Race and Medical Compliance Stereotyping as Predictors of Patient Perceptions of Clinician, by Patient Race: The Race and Relationship-Centered Care Study, Baltimore, MD, January 2002–August 2006

Implicit General Race Bias, Black Patients (n = 191)a
Implicit General Race Bias, White Patients (n = 55)
Implicit Race and Medical Compliance Stereotyping, Black Patients (n = 197)a
Implicit Race and Medical Compliance Stereotyping, White Patients (n = 55)
Patient PerceptionsPredicted Probabilitiesb (95% CI)PPredicted Probabilitiesb (95% CI)PPredicted Probabilitiesb (95% CI)PPredicted Probabilitiesb (95% CI)P
Clinician would ask patient to help decide treatment.48.002.02.04
 No implicit bias31.3 (22.8, 41.2)33.2 (17.9, 53.3)35.2 (27.8, 43.4)46.3 (26.5, 67.3)
 Implicit bias28.3 (22.3, 35.1)16.6 (7.3, 33.5)23.0 (16.8, 30.7)19.4 (6.8, 44.1)
Clinician respects him or her.001< .001.15.47
 No implicit bias50.2 (38.6, 61.7)14.2 (7.0, 26.5)44.1 (33.7, 55.1)27.9 (15.5, 44.9)
 Implicit bias34.9 (27.2, 43.5)26.5 (18.7, 36.0)37.0 (28.7, 46.2)21.6 (8.3, 45.5)
Clinician likes him or her.25.003.41.23
 No implicit bias28.5 (20.4, 38.2)2.4 (0.5, 9.9)26.5 (18.8, 35.9)15.1 (5.0, 37.8)
 Implicit bias23.8 (17.9, 31.0)8.0 (3.0, 19.5)23.2 (16.7, 31.3)6.6 (1.8, 21.4)
Likes clinician< .001.12.75.12
 No implicit bias46.6 (37.5, 55.9)22.8 (11.9, 39.1)39.8 (31.0, 49.3)42.1 (27.1, 58.7)
 Implicit bias32.7 (26.2, 39.9)31.4 (20.8, 44.5)38.4 (31.7, 45.5)31.6 (22.5, 42.5)
Trusts clinician.29.49.02.17
 No implicit bias71.2 (64.3, 77.3)81.3 (61.7, 92.2)74.4 (67.3, 80.5)83.1 (64.2, 93.1)
 Implicit bias67.6 (60.6, 73.8)76.9 (68.9, 83.4)64.0 (56.5, 70.9)73.9 (64.2, 81.7)
Has confidence in clinician.007.25.05.15
 No implicit bias80.1 (72.9, 85.8)83.6 (69.0, 92.1)79.3 (72.2, 85.0)83.9 (66.4, 93.2)
 Implicit bias71.4 (64.1, 77.7)79.0 (70.7, 85.4)71.2 (63.8, 77.7)74.9 (66.0, 82.0)
Would recommend clinician.001.51.18.03
 No implicit bias47.3 (38.6, 56.1)32.3 (17.4, 51.8)42.4 (32.8, 52.7)43.8 (24.0, 65.8)
 Implicit bias34.4 (27.1, 42.5)29.1 (19.4, 41.2)36.3 (29.4, 43.8)23.7 (15.5, 34.5)

Note. CI = confidence interval. Adjusted for clinician gender and patient age, gender, education, and the mental component of the Medical Outcomes Short Form 12.

aA few patients did not respond to some of the outcomes (minimum n = 186 for Black and n = 54 for White patients).

bWe estimated the predicted probabilities while holding all other covariates in the model constant at their mean. These are the probabilities of having the outcome for a hypothetical patient whose clinician exhibited no implicit bias (a score of 0.0) and for a similar hypothetical patient whose clinician exhibited a moderate level of implicit bias (a score of 0.5).

For White patients, having a clinician with higher levels of general race bias was linked to higher likelihoods of perceiving respect from the clinician (12% higher) and believing they are liked by the clinician (6% higher) but a lower likelihood of perceiving the clinician as participatory (17% lower; Table 4).

Race and Compliance Stereotyping and Interpersonal Care

Greater implicit stereotyping on the race and compliance IAT was linked to communication process: for Blacks, 20% longer visits and slower pace of the dialogue; for Whites, 21% shorter visits with a more rapid pace of the dialogue, 25% less clinician verbal dominance, and higher clinician positive affect. Additionally, the race and compliance stereotype was associated with communication content. In Black patient visits, it was associated with less, and in White patient visits more, patient-centered dialogue (Table 3). Higher levels of the race and compliance IAT were linked to lower predicted probabilities of Blacks (12% lower) and Whites (27% lower) perceiving the clinician as involving them in decisions, lower likelihoods of Blacks having trust (10% lower) and confidence (8% lower) in the clinician, and lower likelihood of Whites recommending the clinician (20% lower; Table 4).

In analyses stratified by clinicians’ race, the associations of implicit bias or stereotyping with most communication measures were similar regardless of the race of the clinician. However, the associations of implicit bias or stereotyping with patient ratings were attenuated among Black patients seeing Black clinicians.

The interaction of clinician implicit bias with patient race was not significant for any of the communication behaviors (although there was suggestive evidence for patient positive affect, P = .06). However, this interaction was statistically significant for the following patient perceptions: clinician respect, liking, and participatory style, and patient liking and recommending the clinician. The interaction of clinician implicit compliance stereotyping with patient race was significant for visit length, speech speed, verbal dominance, and clinician positive affect but not for any patient perceptions.

Clinicians, like everyone else, hold varying attitudes toward members of racial/ethnic minorities. However, clinicians differ from others regarding their professional role and code of conduct. We hypothesized that general implicit racial bias would be related to socioemotional and stylistic patterns of communication—those common to social interactions regardless of professional setting—whereas implicit race and compliance stereotyping would be related to both medically focused communication and normative patterns of interaction. We expected more implicit bias on both measures to have a negative influence on Black, but not White, patients’ experiences. We were generally correct in hypothesizing distinct communication pathways. As in previous work,15–17,45 the correlation between the 2 IAT measures was relatively weak for non-Black clinicians, suggesting that although related, the measures may reflect different cognitive processes related to race. Moreover, the measures were associated with different indicators of visit communication.

As expected, the IAT measures were consistently associated with Blacks’ poor ratings of patient care. Unexpectedly, we observed both positive and negative associations between implicit racial attitudes and White patient ratings of patient care. Although 1 study has shown an association of implicit bias with Blacks’ negative ratings of care,19 this is the first study, to our knowledge, to demonstrate that both implicit bias and stereotyping are associated with directly observed medical visit communication and patient perceptions of care.

The negative effect of implicit race bias for Black patients is evident in communication indicators (e.g., more clinician-dominated visit dialogue and lower coder ratings of patient positive affect during the visit) and a broad array of negative patient ratings. The effect of implicit stereotyping is also negative for Black patients. It is associated with lower levels of patient-centered dialogue and lower patient ratings of trust and confidence in the clinician. These findings are consistent with other studies demonstrating that Blacks are at greater risk than are Whites for narrowly biomedically focused visits with restricted patient input in the psychosocial and lifestyle realm.7,28

Patient-centered communication is associated with greater patient trust,10,46,47 which is in turn associated with adherence and continuity of care.37,38,40 Thus, the differences in communication behaviors and patient ratings of care in this study may have implications for health outcomes. The communication markers associated with implicit race attitudes may be a proxy for a more pervasive pattern of communication, including nonverbal behaviors not detected by our coders, that contributes to poorer patient perceptions. Social psychology studies show that implicit attitudes “leak” during interactions through the inadvertent display of negative nonverbal behaviors48 and sentiments,49 even when individuals consciously endorse racial equality and are averse to any suggestion of racial bias.50 In previous studies, IAT scores reflecting more racial bias predicted less speaking time, less smiling, fewer social comments, less speech fluency, and more speech errors among participants interacting with Black (than White) experimenters.49,51 Blacks may be sensitive to these cues and use them in drawing conclusions about poor interpersonal treatment.20 Less positive affect among Black patients may also reflect “stereotype threat,” a phenomenon whereby cues in one’s surroundings accentuate negative stereotypes associated with one’s group and activate physiological and psychological processes with negative influences on behavior.52,53

The associations of implicit stereotyping with longer visits and slower speech might suggest a more conscientious and thorough handling of the concerns of Black patients.54–58 Clinicians with more implicit compliance stereotyping may be trying to compensate for what they perceive as greater mistrust from Black patients, or they may be making well-intentioned efforts to promote patient adherence or to not appear prejudiced. Conversely, the combination of slow speech and low patient centeredness may convey an authoritative tone that creates an overall negative impression on patients. Associations of slower speech with more biomedically focused and information-dense visits, less patient centeredness, and poorer interpersonal care ratings have been reported in previous studies.59,60

In contrast to consistent negative findings for Blacks, the effect of implicit race attitudes for White patients is largely, although not exclusively, positive. As implicit race bias increases, White patients report being more respected and liked, and as implicit compliance stereotyping increases, coders rate the communication in their visits as more patient centered, less verbally dominant, and higher in clinician positive affect. Some negative influences of implicit bias are common to both Blacks and Whites. Clinicians with general race bias are more verbally dominant in the visits of all patients, and both patient groups perceive clinicians with race and compliance stereotyping as less likely to involve patients in treatment decisions. A more conservative political orientation is a correlate of implicit bias in our sample, and previous work suggests that politically conservative doctors work in more traditional settings and may be less open to partnership relationships with their patients.61–63 To explore whether the association of implicit bias and stereotyping with clinician verbal dominance might be a consequence of a more traditional, authoritarian approach to the doctor–patient relationship, we adjusted for clinicians’ political ideology and found it partially explained this finding, but only among Black patients, perhaps because we had limited statistical power in our smaller sample of White patients.

Limitations

The study has some limitations. The sampling of clinicians and patients was not random but contingent on participation in one of the earlier studies. Participating clinicians may be more motivated about communication skills and caring for minority patients than are other physicians; they are also from an urban area with a high representation of Blacks, where interracial relationships may differ from other areas. The study sample included mostly Black patients seeing White clinicians; we might have seen different results with a larger sample of White patients. However, because most ethnic minorities in the United States receive care from White physicians, these findings may be relevant to a large proportion of ethnic minority primary care patients. Knowledge that the visit was being recorded may have biased clinician performance. However, studies of performance bias conclude that recordings have little systematic effect on performance.64–67 Patients were well known by their clinicians and may have given socially desirable responses to survey items; however, it is unlikely this would have differed according to the implicit attitudes of clinicians.

Although the coders achieved high reliability and identified some racial differences in patient and clinician affect linked to implicit attitudes, both are White women and may be less sensitive to biased clinicians’ subtle negative nonverbal vocal cues that might be evident to a Black coder.20 Measuring the IAT some months after the patient interview makes it conceivable that the patient interview influenced IAT measures, rather than the reverse. However, this alternative interpretation is rendered implausible by, in combination, the known stability of IAT measures,27 their limited malleability,45,68 and the lack of evidence for effects of order of administration of IAT measures on relevant behaviors.18 We have not included an analysis of the association of clinicians’ self-reported race attitudes with communication behaviors or ratings of care. However, IAT measures have greater validity than do self-report measures in predicting stereotyping behavior.18,24 Because the study included multiple comparisons, the possibility of statistical type I error exists; however, this is unlikely because analyses were conceptually driven and grounded in previous literature, most of the observed associations are in the expected directions, and findings across related measures are consistent.

Conclusions

Notwithstanding these limitations, this study informs future research and interventions targeting health professionals to reduce healthcare disparities. Theoretically based intervention strategies—which increase clinicians’ awareness and understanding of the basis of bias and help them develop cultural sensitivity, patient-centered communication, and partnership-building in the patient–clinician relationship—will enable clinicians to reduce their reliance on social categories when clinically irrelevant.69 Interpersonal bias in healthcare is only 1 of the manifestations of racial discrimination in our society; however, health professionals can serve as influential advocates for social justice70 by encouraging open discourse about the existence of bias in healthcare and upholding the elimination of healthcare disparities as a local, national, and global priority.

Acknowledgments

This work was supported in part by the Fetzer Foundation Relationship-Centered Care Research Network (grant 2005), the National Heart, Lung, and Blood Institute (grants R01HL69403 and K24HL083113), the Agency for Healthcare Research and Quality (grant R01HS013645), and the Implicit and Unconscious Cognition Research Fund at University of Washington. L. A. Cooper also gratefully acknowledges the support of the MacArthur Foundation Fellows Program.

Human Participant Protection

The Johns Hopkins institutional review board approved all study protocols. All patients provided written informed consent. All clinicians provided either oral or written informed consent.

References

1. Smedley BD, Stith AY, Nelson AR, eds. Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care. Washington, DC: National Academies Press; 2003. Google Scholar
2. Saha S, Arbelaez JJ, Cooper LA. Patient–physician relationships and racial disparities in the quality of health care. Am J Public Health. 2003;93(10):17131719. LinkGoogle Scholar
3. Doescher MP, Saver BG, Franks P, Fiscella K. Racial and ethnic disparities in perceptions of physician style and trust. Arch Fam Med. 2000;9(10):11561163. Crossref, MedlineGoogle Scholar
4. Malat J. Social distance and patients’ rating of healthcare providers. J Health Soc Behav. 2001;42(4):360372. Crossref, MedlineGoogle Scholar
5. Boulware LE, Cooper LA, Ratner LE, LaVeist TA, Powe NR. Race and trust in the health care system. Public Health Rep. 2003;118(4):358365. Crossref, MedlineGoogle Scholar
6. Ngo-Metzger Q, Legedza AT, Phillips RS. Asian Americans’ reports of their health care experiences. Results of a national survey. J Gen Intern Med. 2004;19(2):111119. Crossref, MedlineGoogle Scholar
7. Johnson RL, Roter D, Powe NR, Cooper LA. Patient race/ethnicity and quality of patient–physician communication during medical visits. Am J Public Health. 2004;94(12):20842090. LinkGoogle Scholar
8. Street RL Jr, Gordon H, Haidet P. Physicians’ communication and perceptions of patients: is it how they look, how they talk, or is it just the doctor?Soc Sci Med. 2007;65(3):586598. Crossref, MedlineGoogle Scholar
9. Cooper LA, Roter D, Johnson R, Ford D, Steinwach DM, Powe NR. Patient-centered communication, ratings of care, and concordance of patient and physician race. Ann Intern Med. 2003;139(11):907915. Crossref, MedlineGoogle Scholar
10. Street RL Jr, O’Malley KJ, Cooper LA, Haidet P. Understanding concordance in patient–physician relationships: personal and ethnic dimensions of shared identity. Ann Fam Med. 2008;6(3):198205. Crossref, MedlineGoogle Scholar
11. van Ryn M, Burke J. The effect of patient race and socio-economic status on physicians’ perceptions of patients. Soc Sci Med. 2000;50(6):813828. Crossref, MedlineGoogle Scholar
12. Lutfey KE, Ketcham JD. Patient and provider assessments of adherence and the sources of disparities: evidence from diabetes care. Health Serv Res. 2005;40(6 pt 1):18031817. Crossref, MedlineGoogle Scholar
13. Bogart LM, Catz SL, Kelley JA, et al. Factors influencing physicians’ judgments of adherence and treatment decisions for patients with HIV disease. Med Decis Making. 2001;21(1):2836. Crossref, MedlineGoogle Scholar
14. van Ryn M, Burgess D, Malat J, Griffin J. Physicians’ perceptions of patients’ social and behavioral characteristics and race disparities in treatment recommendations for men with coronary artery disease. Am J Public Health. 2006;96:351357. LinkGoogle Scholar
15. Green AR, Carney DR, Pallin DJ, et al. Implicit bias among physicians and its prediction of thrombolysis decisions for Black and White patients. J Gen Intern Med. 2007;22(9):12311238. Crossref, MedlineGoogle Scholar
16. Sabin JA, Rivara FP, Greenwald AG. Physician implicit attitudes and stereotypes about race and quality of medical care. Med Care. 2008;46(7):678685. Crossref, MedlineGoogle Scholar
17. Sabin J, Nosek BA, Greenwald A, Rivara FP. Physicians’ implicit and explicit attitudes about race by MD race, ethnicity, and gender. J Health Care Poor Underserved. 2009;20(3):896913. Crossref, MedlineGoogle Scholar
18. Greenwald AG, Poehlman TA, Uhlmann E, Banaji MR. Understanding and using the Implicit Association Test: III. Meta-analysis of predictive validity. J Pers Soc Psychol. 2009;97(1):1741. Crossref, MedlineGoogle Scholar
19. Penner LA, Dovidio JF, West TV, et al. Aversive racism and medical interactions with Black patients: a field study. J Exp Soc Psychol. 2010;46(2):436440. Crossref, MedlineGoogle Scholar
20. Livingston RW, Drwecki BB. Why are some individuals not racially biased? Susceptibility to affective conditioning predicts nonprejudice toward Blacks. Psychol Sci. 2007;18(9):816823. Crossref, MedlineGoogle Scholar
21. Cooper LA, Roter DL, Bone LR, et al. A randomized controlled trial of interventions to enhance patient–physician partnership, patient adherence and high blood pressure control among ethnic minorities and poor persons: study protocol NCT00123045. Implement Sci. 2009;4:7. Crossref, MedlineGoogle Scholar
22. Cooper LA, Ford DE, Ghods BK, et al. A cluster randomized trial of standard quality improvement versus patient-centered interventions to enhance depression care for African Americans in the primary care setting: study protocol NCT00243425. Implement Sci. 2010;5:18. Crossref, MedlineGoogle Scholar
23. Greenwald AG, McGhee DE, Schwartz JL. Measuring individual differences in implicit cognition: the Implicit Association Test. J Pers Soc Psychol. 1998;74(6):14641480. Crossref, MedlineGoogle Scholar
24. Nosek BA, Greenwald AG, Banaji MR. Understanding and using the Implicit Association Test: II. Method variables and construct validity. Pers Soc Psychol Bull. 2005;31(2):166180. Crossref, MedlineGoogle Scholar
25. Nosek BA, Smyth FL. A multitrait-multimethod validation of the Implicit Association Test: implicit and explicit attitudes are related but distinct constructs. Exp Psychol. 2007;54(1):1429. Crossref, MedlineGoogle Scholar
26. Greenwald AG, Nosek BA, Banaji MR. Understanding and using the Implicit Association Test: I. An improved scoring algorithm. J Pers Soc Psychol. 2003;85(2):197216. Crossref, MedlineGoogle Scholar
27. Egloff B, Schwerdtfeger A, Schmukle SC. Temporal stability of the Implicit Association Test—anxiety. J Pers Assess. 2005;84(1):8288. Crossref, MedlineGoogle Scholar
28. Roter DL, Stewart M, Putnam SM, Lipkin M Jr, Stiles W, Inui TS. Communication patterns of primary care physicians. JAMA. 1997;277(4):350356. Crossref, MedlineGoogle Scholar
29. Roter DL, Larson S. The relationship between residents’ and attending physicians’ communication during primary care visits: an illustrative use of the Roter Interaction Analysis System. Health Commun. 2001;13(1):3348. Crossref, MedlineGoogle Scholar
30. Mead N, Bower P. Measuring patient-centredness: a comparison of three observation-based instruments. Patient Educ Couns. 2000;39(1):7180. Crossref, MedlineGoogle Scholar
31. Lazare A, Putnam SM, Lipkin MJ. Three functions of the medical interview. In: Lipkin M, Putnam S, Lazare A, eds. The Medical Interview: Clinical Care, Education, and Research. New York: Springer-Verlag; 1995:319. CrossrefGoogle Scholar
32. Paasche-Orlow M, Roter D. The communication patterns of internal medicine and family practice physicians. J Am Board Fam Pract. 2003;16(6):485493. Crossref, MedlineGoogle Scholar
33. Hall JA, Horgan TG, Stein TS, Roter DL. Liking in the physician–patient relationship. Patient Educ Couns. 2002;48(1):6977. Crossref, MedlineGoogle Scholar
34. Beach MC, Roter DL, Wang NY, Duggan PS, Cooper LA. Are physicians’ attitudes of respect accurately perceived by patients and associated with more positive communication behaviors?Patient Educ Couns. 2006;62(3):347354. Crossref, MedlineGoogle Scholar
35. Kao AC, Green DC, Davis NA, Koplan JP, Cleary PD. Patients’ trust in their physicians: effects of choice, continuity, and payment method. J Gen Intern Med. 1998;13(10):681686. Crossref, MedlineGoogle Scholar
36. Kaplan SH, Gandek B, Greenfield S, Rogers W, Ware JE. Patient and visit characteristics related to physicians’ participatory decision-making style. Results from the Medical Outcomes Study. Med Care. 1995;33(12):11761187. Crossref, MedlineGoogle Scholar
37. Schneider J, Kaplan SH, Greenfield S, Li W, Wilson IB. Better physician–patient relationships are associated with higher reported adherence to antiretroviral therapy in patients with HIV infection. J Gen Intern Med. 2004;19(11):10961103. Crossref, MedlineGoogle Scholar
38. McGinnis B, Olson KL, Magid D, et al. Factors related to adherence to statin therapy. Ann Pharmacother. 2007;41(11):18051811. Crossref, MedlineGoogle Scholar
39. O’Malley AS, Sheppard VB, Schwarts M, Mandelblatt J. The role of trust in use of preventive services among low-income African-American women. Prev Med. 2004;38(6):777785. Crossref, MedlineGoogle Scholar
40. Safran DG, Montgomery JE, Chang H, Murphy J, Rogers WH. Switching doctors: predictors of voluntary disenrollment from a primary physician’s practice. J Fam Pract. 2001;50(2):130136. MedlineGoogle Scholar
41. Ware J Jr, Kosinski M, Keller SD. A 12-item short-form health survey. Construction of scales and preliminary tests of reliability and validity. Med Care. 1996;34(3):220233. Crossref, MedlineGoogle Scholar
42. Zeger SL, Liang K. Longitudinal data analysis for discrete and continuous outcomes. Biometrics. 1986;42(1):121130. Crossref, MedlineGoogle Scholar
43. Liang K-Y, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika. 1986;73(1):1322. CrossrefGoogle Scholar
44. Cohen J. Statistical Power Analysis for the Behavioral Sciences. 2nd ed.Hillsdale, NJ: Erlbaum; 1988. Google Scholar
45. Nosek BA, Smyth FL, Jansen JJ, et al. Pervasiveness and correlates of implicit attitudes and stereotypes. Eur Rev Soc Psychol. 2007;18(1):3688. CrossrefGoogle Scholar
46. Zolnierek KB, Dimatteo MR. Physician communication and patient adherence to treatment: a meta-analysis. Med Care. 2009;47(8):826834. Crossref, MedlineGoogle Scholar
47. Fiscella K, Meldrum S, Franks P, et al. Patient trust: is it related to patient-centered behavior of primary care physicians?Med Care. 2004;42(11):10491055. Crossref, MedlineGoogle Scholar
48. Dovidio JF, Kawakami K, Gaertner SL. Implicit and explicit prejudice and interracial interaction. J Pers Soc Psychol. 2002;82(1):6268. Crossref, MedlineGoogle Scholar
49. Richeson JA, Shelton JN. Brief report: thin slices of racial bias. J Nonverbal Behav. 2005;29(1):7586. CrossrefGoogle Scholar
50. Dovidio JF, Penner LA, Albrecht TL, Norton WE, Gaertner SL, Shelton JN. Disparities and distrust: the implications of psychological processes for understanding racial disparities in health and health care. Soc Sci Med. 2008;67(3):478486. Crossref, MedlineGoogle Scholar
51. McConnell AR, Leibold JM. Relations among the Implicit Association Test, discriminatory behavior, and explicit measures of racial attitudes. J Exp Soc Psychol. 2001;37(5):435442. CrossrefGoogle Scholar
52. Steele CM, Aronson J. Stereotype threat and the intellectual test performance of African Americans. J Pers Soc Psychol. 1995;69(5):797811. Crossref, MedlineGoogle Scholar
53. Burgess DJ, Warren J, Phelan S, Dovidio J, van Ryn M. Stereotype threat and health disparities: what medical educators and future physicians need to know. J Gen Intern Med. 2010;25(suppl 2):S169S177. Crossref, MedlineGoogle Scholar
54. Wiggers JH, Sanson-Fisher R. Duration of general practice consultations: association with patient occupational and educational status. Soc Sci Med. 1997;44(7):925934. Crossref, MedlineGoogle Scholar
55. Gross DA, Zyzanski SJ, Borawski EA, Cebul RD, Stange KC. Patient satisfaction with time spent with their physician. J Fam Pract. 1998;47(2):133137. MedlineGoogle Scholar
56. Wilson IB, Kaplan S. Physician–patient communication in HIV disease: the importance of patient, physician, and visit characteristics. J Acquir Immune Defic Syndr. 2000;25(5):417425. Crossref, MedlineGoogle Scholar
57. Howie JG, Porter AM, Heaney DJ, Hopton JL. Long to short consultation ratio: a proxy measure of quality of care for general practice. Br J Gen Pract. 1991;41(343):4854. MedlineGoogle Scholar
58. Morrell DC, Evans ME, Morris RW, Roland MO. The “five minute” consultation: effect of time constraint on clinical content and patient satisfaction. Br Med J (Clin Res Ed). 1986;292(6524):870873. Crossref, MedlineGoogle Scholar
59. Erby LH, Roter D, Larson S, Cho J. The rapid estimate of adult literacy in genetics (REAL-G): a means to assess literacy deficits in the context of genetics. Am J Med Genet A. 2008;146A(2):174181. Crossref, MedlineGoogle Scholar
60. Roter DL, Larson SM, Beach MC, Cooper LA. Interactive and evaluative correlates of dialogue sequence: a simulation study applying the RIAS to turn taking structures. Patient Educ Couns. 2008;71(1):2633. Crossref, MedlineGoogle Scholar
61. Mechanic D. Politics, Medicine, and Social Science. New York: John Wiley & Sons; 1974. Google Scholar
62. Whitney SN, Brown BW Jr, Brody H, Alcser KH, Bachman JG, Greely HT. Views of United States physicians and members of the American Medical Association House of Delegates on physician-assisted suicide. J Gen Intern Med. 2001;16(5):290296. Crossref, MedlineGoogle Scholar
63. Waitzkin H. Information-giving in medical care. J Health Soc Behav. 1985;26:81101. Crossref, MedlineGoogle Scholar
64. Inui TS, Carter WB, Kukull WA, Haigh VH. Outcome based doctor–patient interaction analysis: I. Comparison of techniques. Med Care. 1982;20(6):535549. Crossref, MedlineGoogle Scholar
65. Redman S, Dickinson JA, Cockburn J, Hennrikus D, Sanson-Fisher RW. The assessment of reactivity in direct observation studies of doctor–patient interactions. Psychol Health. 1989;3(1):1728. CrossrefGoogle Scholar
66. Pringle M, Stewart-Evans C. Does awareness of being video recorded affect doctors’ consultation behavior?Br J Gen Pract. 1990;40(340):455458. MedlineGoogle Scholar
67. Coleman T, Manku-Scott T. Comparison of video-recorded consultations with those in which patients’ consent is withheld. Br J Gen Pract. 1998;48(427):971974. MedlineGoogle Scholar
68. Joy-Gaba JA, Nosek BA. The surprisingly limited malleability of implicit racial evaluations. Soc Psychol. 2010;41(3):137146. CrossrefGoogle Scholar
69. Burgess D, van Ryn M, Dovidio J, Saha S. Reducing racial bias among health professionals: lessons from social-cognitive psychology. J Gen Intern Med. 2007;22(6):882887. Crossref, MedlineGoogle Scholar
70. Medical professionalism in the new millennium: a physician charter. Ann Intern Med. 2002;136(3):243246. Crossref, MedlineGoogle Scholar

Related

No related items

TOOLS

SHARE

ARTICLE CITATION

Lisa A. Cooper, MD, MPH, Debra L. Roter, DrPH, Kathryn A. Carson, ScM, Mary Catherine Beach, MD, MPH, Janice A. Sabin, PhD, MSW, Anthony G. Greenwald, PhD, and Thomas S. Inui, MDLisa A. Cooper and Mary Catherine Beach are with the Department of Medicine, Johns Hopkins University School of Medicine, Baltimore, MD. Debra L. Roter is with the Department of Health Behavior and Society at the Johns Hopkins Bloomberg School of Public Health, Baltimore, MD. Kathryn A. Carson is with the Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health. Janice A. Sabin is with the Department of Medical Education and Biomedical Informatics, Center for Clinical and Epidemiological Research, University of Washington, Seattle. Anthony G. Greenwald is with the Department of Psychology, University of Washington. Thomas S. Inui is with the Regenstrief Institute at Indiana University School of Medicine, Indianapolis. “The Associations of Clinicians’ Implicit Attitudes About Race With Medical Visit Communication and Patient Ratings of Interpersonal Care”, American Journal of Public Health 102, no. 5 (May 1, 2012): pp. 979-987.

https://doi.org/10.2105/AJPH.2011.300558

PMID: 22420787