Background
The role of patient race and ethnicity in treatment decision-making has met considerable debate [
1,
2]. On one hand, clinicians sometimes use patients' race to predict disease risk and to determine appropriate drug therapy [
3‐
5]. On the other hand, consideration of patients' race in clinical settings is also associated with bias and stereotyping, which can contribute to racial and ethnic healthcare disparities [
6‐
9]. What we know about healthcare bias and race is further complicated by minority patients' perceptions of frequent discrimination. In both the United States and Europe, racial and ethnic minority patients report more discrimination and poorer quality of clinical care compared to non-minority whites [
10‐
12].
Given the complexities of understanding medical and social contexts associated with race and racial bias in healthcare, it is critical to understand physicians' perspectives about when it is and is not appropriate to consider race when making clinical decisions for patients. Physicians' perspectives about race are particularly important to understand as the tangled web linking race and health continues to evolve, and as scientists continue to develop models of how race can be used to predict disease risk. What, then, do physicians think is the importance of race in treatment decision-making? Moreover, how do physicians consciously utilize race to make treatment decisions?
The purpose of this study is to explore United States primary care physicians' views about the importance of race in treatment decision-making by explicitly asking physicians if, and how race would factor into their decisions about treatment for a hypothetical patient. Using physicians' responses, we examined the importance of race, social, and medical factors in treatment decision-making, and explored differences in beliefs between black and white physicians.
Methods
Study Design
This investigation was part of the Physicians' Understanding of Human Genetic Variation (PUHGV) Study at the Social and Behavioral Research Branch of the National Human Genome Research Institute, National Institutes of Health (NIH). Aims of the PUHGV project were to acquire information regarding physicians' knowledge of human genetic variation [
13], their beliefs about the relationships between genetics, race and medicine [
14], and their use of race in clinical practice. In the present study, we consider the last PUHGV project aim - physicians' attitudes regarding the relevance of patients' race and other social attributes (beliefs, cultural practices, and socioeconomic status) in treatment decision-making.
Sampling and Recruitment
A list of eligible physicians was generated using the American Medical Association Physician Profile Database of general internists, as well as identification of practicing physicians at Departments of Internal Medicine for the medical schools in each of the five study locations in the United States (Atlanta, GA; Detroit, MI; Los Angeles, CA; Philadelphia, PA; and Baltimore, MD). Using this list, general internists were recruited through invitation letters sent by mail or email. Because of underrepresentation of black general internists in the United States, we also notified local chapters of the National Medical Association (the largest and oldest national organization representing African-American physicians and their patients in the United States) in each of the five metropolitan areas of the study, and requested names of potentially eligible physicians; Snowball sampling was used to recruit additional black physicians, until project staff was able to gain a minimum of 12 physicians per focus group. The goal was to recruit 12 physicians per focus group in order to seat at least 8 to 10 participants, expecting 2 - 4 "no shows". Additional information regarding the PUHGV study, recruitment, and data collection are available in previously published literature [
13,
14].
Human Participant Protection
Informed consent was obtained from each focus group participant before the start of the study, and a monetary incentive of $250.00 was provided. The study was approved by the institutional review board of the National Human Genome Research Institute of the National Institutes of Health.
Data Collection
Ten focus groups were conducted with self-identified black or African American and self-identified white or Caucasian, board-eligible or board-certified general internists between October 2005 and March 2006. The focus groups were stratified by physician race. Two general internists who were experienced in focus group research methods served as racially concordant moderators for the focus groups; the same moderators were used in each of the five study locations.
Vignette
To elicit discussion about physicians' attitudes about the use of race in treatment decision-making, physicians were presented with a brief clinical vignette describing a hypothetical patient. The patient's race was not provided to the physicians as part of the vignette. The clinical vignette read:
"Michelle is a 57-year-old patient who is new to your practice. Upon review of her medical records from an outside physician, you note that she has a history of untreated hypertension. Per the medical records, her blood pressure ranged 145 over 90 and 175 over 105 on two occasions. She has a history of type 2 diabetes; she smokes cigarettes, about a pack per day. Her current HDL is 35 and her LDL is 162. Michelle has health insurance through her employer, General Motors."
Probes
First, physicians were asked to discuss what information they would want to have to make decisions about treating the patient. Then, the moderator of each focus group explicitly asked the physicians about their opinions on the relevance of race to the patient's treatment. Some questions included: What further information would you like to have to treat Michelle? How might race and ethnicity play a role in treatment of Michelle? How is race used in your clinical practice? When do you think race is medically relevant? Using the responses to these questions, we examined physicians' attitudes about the use of race in clinical practice, as well as of patient social attributes that physicians used to inform their decision-making.
Analysis
Audio tapes of all focus group sessions were transcribed verbatim. Notes taken by trained observers at the focus groups sessions were also used to ensure accuracy of the transcripts. Transcription and coding for the initial PUHGV study is fully described in a previous publication [
13].
For this analysis, transcripts were coded and reviewed by the vignette study core research team (authors S.A.S, A.O.O., J.C.F, and V.L.B). Several steps were taken to ensure careful, systematic qualitative analysis. First, two team members independently coded the vignette portion of each focus group transcript line by line using "open coding" [
15‐
17]. "Open Coding" consisted of using salient key words and phrases that emerged from the focus group transcripts to formally identify categories and concepts relevant to the primary research questions: What do physicians think is the importance of race in treatment decision-making? How do physicians consciously utilize race to make treatment decisions?
From the open coding, a comprehensive set of codes was created. The core vignette study research team then developed a condensed list of the most significant codes relevant to the research question. The condensed list of codes was agreed upon by the entire research team. Coding differences were reconciled over the course of several meetings of team members, held to handle discrepancies in coding and to revise the coding scheme as needed. Finally, the vignette core research team compared codes and responses until consensus was reached for all transcripts.
NVivo (QSR International, v7) was used to code and qualitatively analyze the vignette transcripts. Using NVivo, the vignette core research team ran coding reports by physician race to create two data sets - one for black physicians and the other for white physicians, which were then used to compile comprehensive tables that highlighted the number of coding references (words, phrases, or statements) associated with each code. Repeated codes of the same participant and within the same broad category were treated as one instance. Next, data were analyzed by physician age, gender, number of years in practice, and practice setting (academic vs. non-academic) for each race-based data set. Coding reports were then created based on tables of all queries in the race-based data sets and then for the complete dataset with all physicians. Analysis was an iterative process, reviewing coding reports of the queried texts, reading related transcript paragraphs, and re-reading all transcripts. Our analysis also involved paying careful attention to negative cases (e.g., race not important), which were addressed by examination and re-examination of every case discussion around the medical relevance of race. Negative cases were resolved by the vignette core research team through comparison of all code differences to see whether the emergent themes were applicable to the majority of cases. Once negative cases were noted, we re-examined our codebook adding codes for negative cases. For example, codes were created for discussion around when race is not considered medically relevant. Re-coding of negative cases was conducted by the analysis team until properly coded. This iterative process continued until it was determined that there were no negative or unresolved cases.
Differences in discussions by physician race were determined by number of coding references, content, density, and breadth of discussion. Responses with the most coding references and most extensive discussion were listed as major themes.
Discussion
Our study explored beliefs that black and white physicians have about the importance of patient race for treatment decision-making. Although patient medical history was important to both groups of physicians, we found noteworthy differences in how black and white physicians viewed the importance of patient race. Specifically, we discovered that black physicians in our sample viewed race as important for treatment decision-making, while only four white physicians considered race medically relevant.
Our data suggest that black physicians view race as important when making decisions about their patients' treatment. In contrast to most white physicians, many black physicians viewed race as important for choosing medications and assessing risk of disease. It should be noted, however, that both groups - black and white physicians alike - viewed medical history as most important for medical decision-making. This finding has important implications, since probability-based models suggest that race should be a primary proxy for establishing that any given patient of a particular racial or ethnic group will experience a health problem. For example, since African Americans in the United States suffer from increased risk of hypertension and diabetes [
18], disorders outlined in our clinical vignette, some reasoning suggests that population-based probability of disease should accompany the decision-making process. However, we found that white physicians in our study did not rely on race as a determinant for treatment decision-making. Other studies agree with this, finding that among a largely white physician population, doctors rarely mentioned race and ethnicity to determine clinical assessments [
19]. Moreover, while black physicians indicated using race as a proxy for disease risk, black physicians held nuanced and complex views about the appropriate context in which race should be used (e.g. to determine appropriate medication and to understand social determinants of health linked with stress and health disparities).
Black physicians also linked race to the hypothetical patient's social determinants of health such as socioeconomic status, as well as their own ability to deliver culturally appropriate care (i.e., being knowledgeable, as well as responsive to the patient's cultural beliefs about illness). Other studies are consistent with our findings. For example, a national survey of doctors suggests that black physicians have greater knowledge and awareness of health and healthcare disparities that affect minority populations [
20]. Our findings might be further explained by evidence that minority physicians provide care in underserved areas for minority patients at higher rates than non-minority physicians. In our study, 36% of white physicians served a minority patient population, while 73% of black physicians served a minority patient population. This difference may have influenced the discussion about race and social determinants of health among physicians in the study, who may have filled in details about the clinical vignette based on their own experiences.
In depth analysis of black physicians' discussion of social determinants of health revealed differences based on physician-level attributes. When compared to black physicians who did not work in an academic setting, those whose medical practice setting was linked to an academic institution more frequently expressed views about the importance of the patient's social determinants of health. Black academic physician participants articulated that race was associated with cultural values related to health, and especially emphasized access to consistent, affordable care.
We also offer possible explanations regarding the fewer occurrences of discussion regarding the use of patient race in treatment decisions among white physicians in our study. Although multiple responses from the same individual were only counted once, it is possible that the number of mentions of particular codes regarding the use of race reflect how talkative physicians were in some focus groups versus others. However, when we assessed differences in the coding across all focus groups, there were marked thematic differences between black and white physicians. All five focus groups among black physicians contained discussion about the importance of race for medical decision-making, regardless of length of discussion. Conversely, all focus groups among white physicians indicated that race is not especially medically relevant.
The differences in black and white physicians' responses about race for treatment decision-making may also indicate that, compared to black physicians, white physicians were less comfortable discussing issues related to race. Other evidence supports this as a possible factor. For example, Littleford et al (2005) found that whites experience greater discomfort than minorities when discussing issues of race and ethnicity in a group setting [
21]. In our study, white physicians voiced concerns about the potential negative effects of incorporating race into decision-making for their patients. White physicians in our study specifically cautioned one another to be careful about stereotyping patients based on race, or about minority patients being offended by the topic of race being raised.
Although white physicians may have felt less comfortable discussing race compared to black physicians, our focus group protocol was organized to encourage free-flowing discussion of race. The study design was careful to include same race participants and race-concordant physician moderators since choosing moderators with similar characteristics to focus group participants is an effective strategy toward reducing discomfort when discussing sensitive issues [
22,
23]. In addition, the physician moderators followed identical protocols and guides in both black and white focus groups, which included the same questions and probes. Nonetheless, some white physicians may have been cautious about discussing issues of race.
The differences we observed in discussions held by black and white physicians about the importance of race in treatment decision-making may also reflect that, in some cases, white physicians may not consciously view race as an important factor for treatment decision-making. Still, some evidence indicates that while white physicians may consciously state that race is not of importance for treatment decision-making they subconsciously use race to make decisions. For example, a study by Green et al. [
7] indicates that while most white physicians do not admit having different feelings toward and perceptions of blacks and whites explicitly, their implicit measures show some degree of unconscious use of race in the form of bias in treatment decisions, favoring whites over blacks. Our study did not measure racial bias among physicians, nor do our results suggest that white physicians in our sample display racial bias in delivery of care.
We note, finally, that perceptions of black physicians regarding the importance of patient race may have also been influenced by their own experiences as members of a racial minority group [
24]. Evidence suggests that black physicians report seeing, or experiencing discrimination ascribed to their racial assignment, while whites think of overt expressions of racism as rare occurrences in our society [
25,
26]. Nunez-Smith et al. (2008) reported that black physicians sometimes find it their place to protect their minority patients from discrimination in healthcare settings [
27]. Therefore, it is not surprising that black physicians in our study view race as an important determinant of health to be considered in treatment decisions. Consistent with other research [
23‐
26], our findings suggests that the role of physician race in reducing health disparities cannot be ignored, and that racial and ethnic U.S. trained minority physicians may play a particularly important role for understanding racial and ethnic minority patients' social experiences in the United States.
Strengths and Limitations
This study is one of few to probe how and why physicians might consider race when treating patients with complex medical problems (i.e., chronic disease, high risk health behavior such as smoking, etc.). Still, some limitations of this work should be considered. First, participants were not randomly selected and were limited to black and white physicians from particular geographic regions in the United States. Second, focus group studies are generally characterized by small sample sizes and may not be representative of the population. Moreover, our results are exploratory in nature and quantitative methods were not used to compare differences between groups. These design and sampling strategies may reduce the external validity of our study.
Also, our exploration of clinicians' intended use of race used a clinical vignette with self-report of intent rather than observation of actual clinician behavior. It is important to note that our clinical vignette focused on a hypothetical patient with Type 2 diabetes and untreated hypertension, who was also a current smoker. Thus, physicians' responses about the relevance of race for this patient's treatment may not represent physicians' responses for other types of clinical cases or illnesses. Finally, while we were able to elicit physicians' attitudes regarding the use of race in clinical settings, perceptions about race influence decisions both consciously and sub-consciously, and we did not measure subconscious attitudes about race. Our findings must be interpreted in light of these limitations.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
SAS led the writing team in the manuscript preparation, as well as contributed to all aspects of the qualitative analysis including coding, grouping codes by attribute, and developing themes from the data. SLS contributed significantly to paper development and study design. AOO participated in study design and preliminary analysis and paper for content. LAC contributed significantly to paper development and revised critical sections of the paper for content. JCF contributed to all aspects of preliminary qualitative analysis and early drafts of the paper. VLB conceived of the study, participated in design and coordination of the analysis, and assisted in all drafts of the manuscript. All authors approved the final manuscript.