Do Physicians' Implicit Views of African Americans Affect Clinical Decision Making?

M. Norman Oliver, MD, MA; Kristen M. Wells, MPH, PhD; Jennifer A. Joy-Gaba, PhD; Carlee Beth Hawkins, MA; Brian A. Nosek, PhD


J Am Board Fam Med. 2014;27(2):177-188. 

In This Article

Abstract and Introduction


Background: Total knee replacement (TKR) is a cost-effective treatment option for severe osteoarthritis (OA). While prevalence of OA is higher among blacks than whites, TKR rates are lower among blacks. Physicians' implicit preferences might explain racial differences in TKR recommendation. The objective of this study was to evaluate whether the magnitude of implicit racial bias predicts physician recommendation of TKR for black and white patients with OA and to assess the effectiveness of a web-based instrument as an intervention to decrease the effect of implicit racial bias on physician recommendation of TKR.

Methods: In this web-based study, 543 family and internal medicine physicians were given a scenario describing either a black or white patient with severe OA refractory to medical treatment. Questionnaires evaluating the likelihood of recommending TKR, perceived medical cooperativeness, and measures of implicit racial bias were administered. The main outcome measures included TKR recommendation, implicit racial preference, and medical cooperativeness stereotypes measured with implicit association tests.

Results: Subjects displayed a strong implicit preference for whites over blacks (P < .0001) and associated "medically cooperative" with whites over blacks (P < .0001). Physicians reported significantly greater liking for whites over blacks (P < .0001) and reported believing whites were more medically cooperative than blacks (P < .0001). Participants reported providing similar care for white and black patients (P = .10) but agreed that subconscious biases could influence their treatment decisions (P < .0001). There was no significant difference in the rate of recommendation for TKR when the patient was black (47%) versus white (38%) (P = .439), and neither implicit nor explicit racial biases predicted differential treatment recommendations by race (all P > .06). Although participants were more likely to recommend TKR when completing the implicit association test before the decision, patient race was not significant in the association (P = .960).

Conclusions: Physicians possessed explicit and implicit racial biases, but those biases did not predict treatment recommendations. Clinicians' biases about the medical cooperativeness of blacks versus whites, however, may have influenced treatment decisions.


Osteoarthritis (OA) is a leading cause of physical disability among older adults in the United States, often leading to significant pain, swelling, and reduced motion at the joint. Total knee replacement (TKR) is a cost-effective treatment option for moderate to severe knee OA,[1–3] relieving pain and improving quality of life. Despite these benefits, marked disparities in TKR utilization in patients with OA exist by age, sex, geographic location, and racial category. While the rate of diagnosed OA is generally higher among African Americans than whites,[4,5] African Americans receive TKR less than half as often as whites.[6–11] These racial and ethnic disparities remain after controlling for health insurance status,[12] overall health,[13] and disease severity.[13]

Other factors that may account for this disparity include patient preferences and physician recommendation bias. Physicians' treatment planning may be unintentionally influenced by race or without them being aware of the bias. Implicit racial biases favoring whites compared with blacks are pervasive in the general population[14] and among physicians.[15] Racial biases of health care providers may play a role in the unequal treatment of minority populations and subsequent health disparities.[16–24] Racial bias has been identified as a factor in the delivery of care for cardiovascular disease,[25] pain management,[26] and mental health care.[27]

Implicit racial biases are related, but distinct from, explicit, self-reported racial biases,[28] and both implicit and explicit biases predict behavior.[29] Only 2 studies have been conducted that are of direct relevance to this report. The first was an Internet-based study of medical residents that evaluated the association between implicit racial bias and thrombolytic therapy of black and white patients presenting with symptoms of myocardial infarction. The differential likelihood of treating white patients and not treating black patients with appropriate thrombolytic therapy was positively related to the physicians' implicit racial biases.[30] The second, by Haider et al,[31] failed to replicate this result, finding that medical students' preference toward white patients did not result in a statistically significant variation in treatment decisions in a similar scenario.

There have been no objective assessments of the role of racial bias in the recommendation of patients for TKR. This study used a web-based survey instrument to evaluate whether implicit and explicit racial bias predicts recommendation of TKR for white and black patients with severe OA and assessed the effectiveness of the instrument as an intervention to decrease racial bias in treatment.