dcyphr | The Effects of Oncologist Implicit Racial Bias in Racially Discordant Oncology Interactions


Implicit bias negatively affects doctor-patient relationships and communications. However, its effect on interactions between oncologists and their patients is unknown. In this study, the researchers examined how implicit bias influences oncology interactions. They also examined whether implicit bias in oncologists negatively affect patients confidence in and perceived difficulty of suggested cancer treatments. 18 non-black physicians and 112 black patients participated. Oncologists completed a implicit racial bias test weeks before treatment discussions with new patients, which were recorded. People watched the recordings and rated oncologist communication and the length of discussion between patients and oncologists. After their discussion, patients were asked questions about patient-centeredness, how well they remembered their discussion, distress, trust, and perceptions of recommended treatments. The researchers found that oncologists with more implicit bias had shorter interactions with patients, who rated oncologist communication as less patient-centered and less supportive. More implicit bias was also associated with patient difficulty of remembering the discussion, less patient confidence in recommended treatments, and greater perceived difficulty of completing recommended treatments. In conclusion, implicit bias negatively affects oncologist-patient communication and may be a source of health disparities that must be addressed in oncology.


The researchers wanted to understand how implicit bias influences interactions between oncologists and their patients.


Black patients often receive worse care than white patients across a variety of diseases. This disparity is heavily present in cancer treatment. In addition, difficulties in doctor-patient communication are linked to poorer health outcomes, and are more prevalent in racially discordant interactions, where the doctor and patient are of two different races/ethnicities. Racially discordant interactions make up 80% of black patients’ interaction with physicians. In this study, the researchers investigated implicit bias as a cause of communication difficulties in racially discordant interactions.

Implicit racial bias refers to negative thoughts and feelings about racial minorities that operate at a subconscious level. Non-black healthcare providers have substantial levels of implicit bias toward black patients.

Implicit racial bias in physicians has been previously researched. Previous studies have focused on how physician implicit bias influences communication with patients in real-world interactions. These studies found that physician implicit bias negative affects communication and black patients’ perception of how the interaction went. Medical areas in which these studies were conducted included primary care, spinal cord injuries, and genetic counseling.

However, the effects of physician implicit bias in racially discordant oncology interactions are largely unknown. So, in this study, the researchers in this study wanted to understand if implicit bias had a similar effect on racially discordant oncologist-patient interactions that was observed in previous studies in other areas. They also added an investigation about how implicit bias influences patient perceptions of recommended treatments.

The researchers examined the following outcomes that could be affected by implicit bias: interaction length, verbal dominance, extent of patient involvement in treatment decisions, patient perceptions of provider patient-centeredness, patients’ trust of their physician, third-party observers’ ratings of interactions, patient difficulty in remembering contents of their interaction, and patient distress.

The researchers also examined the influence of physician racial bias on patients’ perceptions of recommended treatments, specifically their confidence in the treatment working and how difficult they thought going through the treatment would be.


Participant Characteristics

56% of the oncologists were men and had been practicing for an average of 7 years. 91% of patients were women and breast cancer was the most common kind of cancer. Oncologists mean and median IAT scores showed a small to moderate level of implicit racial bias, which was lower than national averages for physicians but consistent with the average for their geographic region.


As oncologist implicit bias increased, observers’ ratings of oncologists’ supportive communication  and interaction length decreased.


Patients who interacted with more implicitly biased physicians thought their interaction was less-patient centered, and had more difficulty with remembering the contents of their interaction. In addition, oncologist implicit bias significantly affected patients’ confidence in their oncologist’s recommended treatments and how difficult they thought the treatment would be. Figure 1A, B, and C show indirect pathways from oncologist implicit bias to patients’ perceptions of treatment difficulty, confidence in treatment, and the patient-centeredness of their interactions.


This study replicated previous findings of negative associations between physician implicit racial bias and quality of communication/patient interactions in oncology, and also identified new associations. Especially interesting was that third-party observers also perceived lower-quality communication when oncologists had more implicit bias. In addition, this studied showed that oncologist implicit bias significantly affects black patients’ perception of their recommended treatments, which had not been shown before.


Design and Participants

Data was collected over 2 years in two cancer hospitals in Detroit, MI. Oncologists were eligible if they were not black and treated cancer patients. Patients were eligible if they were black and had a confirmed diagnosis of breast, colorectal, or lung cancer. 18 oncologists and 112 patients were enrolled in the study.


Several weeks before scheduled visits, participating oncologists took an assessment of implicit racial bias. After, participating patients had a clinical interaction with a participating oncologist. The interaction was an initial discussion of cancer treatment. 96 of 112 interactions were video recorded. After the interaction, oncologists answered questions about patient participation in treatment decisions and patients answered questions about what they thought about the interaction and their oncologist. A week later, patients were called and asked more questions about the interaction and about their trust in their oncologist.


The Implicit Association Test (IAT) was used to assess implicit racial bias. It is the most widely used measure of implicit bias. Higher scores indicate for pro-white/anti-black implicit bias.

Four observers watched the 96 video recordings and rated the quality of communication between oncologists and their patients. These observers were blinded and did not know the study hypothesis.

Interaction length and the amount of time each participant spoke were measured. Verbal dominance was assessed using a ratio of these two measures.

Patients used a four-point scale to rate how their oncologist displayed 14 patient-centered behaviors. They used a five-point scale to rate their difficulty remembering the contents of their interactions, and an eleven-point scale to rate distress. Finally, they used a five-point scale to rate their confidence in and perceived difficulty of their oncologists’ recommended treatment options.


The goal of oncology is to treat high-stakes, life-threatening diseases. Although implicit bias and racial discordance in patient interactions might seem to play an insignificant role in comparison, the findings of this study state otherwise. Racial bias is not the only source of racial disparities in cancer treatment--however, reducing implicit bias may be a way to reduce these disparities and should be further explored as an intervention to reduce disparities in cancer treatment.