INTRODUCTION

Although racial groups are not genetically distinct,1 physicians are encouraged to consider race when providing medical care. For instance, guidelines advise using race to select antihypertensive medications2 and risk calculators use race to estimate the risk of atherosclerotic cardiovascular events.3 However, physicians have varied perceptions of race (e.g., as a socially assigned identity insufficiently correlated with human genetic variation4 or a biological trait5 representing shared genetic ancestry) and must reconcile these perceptions with their use of race when providing care. How physicians perceive race affects patients. One study found that viewing race as a biological trait was associated with an increased use of race in medical decision-making (race-based care),6 which can result in poorer outcomes for racially minoritized patients.7,8 Moreover, physicians’ perceptions of race may affect how they understand and interpret racial health disparities and the solutions they offer to solve them. Attributing disease risk to race obscures the impact of structural racism on health outcomes,9 as it relates disease risk with patients’ social identities instead of the society-specific privileges and harms associated with their status in the racial hierarchy. More research is needed to evaluate how physicians perceive race and use it in clinical care.

Only one systematic review has evaluated how physicians value race in clinical care. That review—a qualitative metasynthesis10 focused on physician decision-making for African American and Hispanic patients—incorporated two studies assessing the clinical value of race. The study authors concluded that physicians had varied beliefs regarding the utility of race in clinical care. This metasynthesis did not address how physicians perceived race, how their perceptions were related to their use of race in care, nor physician factors associated with race-based practice. Furthermore, as the review was a qualitative metasynthesis, quantitative studies evaluating race-based care were excluded, limiting the scope of the analysis.

A mixed-methods systematic review, which includes both quantitative and qualitative studies, can offer greater insight into physicians’ perceptions of race and physician characteristics—including racial beliefs—related to the use of race in clinical care. Quantitative studies test qualitative discoveries, while qualitative research provides context for quantitative work. We conducted a mixed-methods systemic review and narrative synthesis evaluating (1) how physicians conceptualize and use race in clinical decision-making and (2) physician characteristics and beliefs associated with the use of race in clinical care.

METHODS

Citation Index Database and Search

The search was conducted under the guidance of a research librarian and reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Search terms,11 consisting of combinations of key words and Medical Subject Headings terms synonymous with “race,” “clinical decision-making,” and “physicians,” were trialed and refined in PubMed and translated for other databases using the Polyglot Search Translator.12 See Supplement 1 for the complete search strategy. A search of PubMed, CINAHL, EMBASE, and Scopus was conducted on 11/24/2020 and updated on 12/10/2021. The research team used Google Scholar to conduct a citation search on all included studies resulting from the 11/24/2020 search.

Selection Criteria

Table 1 describes the search inclusion and exclusion criteria. Included studies were published in peer-reviewed journals; written in English; and evaluated United States (US) physicians’ perceptions of race or the association between physician characteristics and use of race in clinical care. Study designs were either cross-sectional, controlled, or qualitative. Studies including non-physician clinicians were included if physicians represented >75% of the sample or if physician responses could be analyzed separately. This study was limited to physicians to allow comparison among a sample with similar educational and training backgrounds. Excluded studies were those not published in peer-reviewed journals; non-English language; evaluated non-US population; and did not address physicians’ perceptions of race or the association between physician characteristics and use of race in clinical care. Case reports, cohort, and case-control study designs were excluded. The date of publication was not used to determine study inclusion.

Table 1 Inclusion and Exclusion Criteria
Full size table

Data Extraction

Abstract, title, and full-text review were conducted independently by the lead investigators (EO and LG). The researchers resolved discrepancies through discussion or engaging a third author (GD) as a tiebreaker. Extractions were completed by EO and reviewed by TG. The research team extracted data on study year, aim, design, population, and outcomes from each study.

Quality Assessment

Quality assessment was determined independently by the two lead authors using the Mixed Methods Appraisal Tool (MMAT), version 18.13 MMAT (see Table 2) evaluates the quality of multiple study designs and has been used in previous mixed-methods systematic reviews.14,15 The MMAT assesses each study design using five characteristics unique to that study type. The scores for each characteristic are not summed but provide a measure of study quality that gives context to study findings. MMAT scores were not used to determine study inclusion.

Table 2 Mixed Methods Appraisal Tool Scores
Full size table

Analysis and Synthesis

Qualitative studies were evaluated using thematic analysis with an inductive approach.16 A preliminary list of codes, representing ideas found in the studies and relevant to the study question, was developed after a cursory read of the Result sections of each study. Codes were refined through a thorough line-by-line evaluation of all aspects of the Results sections (i.e., quotes, theme headings, and thematic summaries) related to race-based clinical care. Self-reflexivity was accomplished through discussions regarding perceptions of race-based care prior to the initiation of coding and through writing positionality memos. All qualitative texts were double coded by EO and LG in Atlas.ti (Version 8.4.26.0). Consensus meetings, at least four, were focused on reconciling codes and selecting illustrative quotes. Following these meetings, memos were written summarizing changes to the coding process or aspects of the text that required additional discussion. Findings were discussed, codes were condensed, and themes—representing a collection of codes—were developed. The entire research team reviewed, discussed, and modified codes and themes. Due to the heterogeneity of quantitative studies, a meta-analysis could not be conducted. Statistically significant associations (i.e., p value < 0.05) found in the original research articles or quantitative findings related to qualitative themes were described using a narrative synthesis.

RESULTS

Study Selection

A total of 1149 articles were reviewed (Fig. 1); 1093 and 42 were excluded in abstract and full-text review, respectively. The most common reason for exclusion in the full-text review was lacking evaluation of physicians’ perceptions of race or the association between physician characteristics and the use of race in clinical care. Fourteen relevant articles representing 9 (4 qualitative17,18,19,20,21,22 and 5 quantitative)6,23,24,25,26,27,28,29 unique studies (see Table 3) were included in the narrative synthesis. All studies were published between 2008 and 2021. The combined study population totaled 2450 physicians (228 in qualitative studies; 2222 in quantitative studies) and 198 other healthcare workers (20 in qualitative studies; 178 in quantitative studies). Inter-rater agreement between reviewers was 95% for abstract/title review, 91% for full-text review, and 93% for quality assessment. Five themes emerged from qualitative studies (see Table 4).

Figure 1
figure 1

Flow diagram for study inclusion.

Full size image
Table 3 Characteristics of Included Studies
Full size table
Table 4 Codes, Themes, and Example Quotes from Qualitative Studies
Full size table

Theme 1: Conceptualizations of Race as a Biological Phenomenon

Physicians’ perceptions of race ranged from describing it as a strictly social construct17,19,21 to a mixture of biological and social components.17,18,19,20,21 Two studies17,18 noted that clinicians discussed race almost exclusively as a combination of biological and cultural factors. Notably, one study17 found that clinicians used medical research as evidence that justified their belief in race as a biological trait.

Quantitative studies that directly assessed physicians’ perceptions of race found the majority believed race had biological components. In national survey studies, 82% of internists25 defined race as primarily a biological or genetic ancestry group, and 97% of family physicians23 believed genetic differences played a role in racial health disparities. In a randomized controlled study evaluating BiDil (hydralazine and isosorbide dinitrate) prescribing,24 almost half (24 of 49; 48.9%) of internists selected pharmacogenetic and physiological differences to explain racial differences in BiDil efficacy.

Theme 2: Race Used to Contextualize Patients’ Health

Physicians used racial narratives to explain the poor health of racial minorities. Physicians associated race with culture,17,18,19,22 health behaviors,17,18,22 socioeconomic factors,17,19,22 and exposure to racism.17,19 Culture and socioeconomic status were viewed as contributing to health behaviors that resulted in poor health outcomes.17 Physicians related exposure to racism with distrust of the healthcare system17 and both were used to explain patient behaviors. Nevertheless, a survey study23 of family medicine physicians found no association between the extent to which physicians believed environmental factors contributed to racial differences in health outcomes and the value of race in patient care.

Theme 3: Race Used in Providing Medical Counseling and Determining Care

Physicians used race in patient counseling17,18,19,20,22 and to assess disease risk17,18,22 and management.17,18,19,20,21,22 The diseases physicians described as having racialized risk were hypertension,17,22 diabetes,17,18,22 renal disease,22 prostate cancer,17 Tay-Sachs,17 and cystic fibrosis.17

Only one quantitative study provided insight into the extent to which physicians used race to manage disease. In this study, 24 physicians scored an average of 70 on a 100-point Likert scale (0=not at all; 100=very much) in response to the question “How comfortable are you taking race into account when treating patients with congestive heart failure?” While two survey studies6,25,26,27,28 evaluated the general use of race in disease management, neither reported specific clinical practices in which clinicians engaged nor described criteria to determine high or low level of race-based practice.

Theme 4: Justifications for Race-Based Medical Care

Physicians’ justifications for race-based clinical care were as follows: support in medical literature;17,18,20,21 personal clinical experience;17,20 desire to address racial disparities;21 allows for personalized care;19,20,21 and belief that race-based care encouraged patient compliance.20 A few studies17,19,22 provided examples of physicians speaking of biological differences between racial groups as uncontested medical facts without specifically attributing this knowledge to medical literature. One physician19 stated, “We clearly know that biologically there are different chemicals and different functional processes going on in African-Americans” to explain why race was important.

Quantitative studies provided support for these findings. In a survey study on personalized medicine,29 physicians had a median score of 7 on a 10-point Likert scale (1=strongly disagree; 10=strongly agree) in response to the statement, “I believe the delivery of personalized medicine requires consideration of a patient’s racial background.” In the BiDil study,24 34% (17/50) of physicians believed that BiDil reduced health disparities in cardiovascular disease.

Theme 5: Concerns About Race-Based Medical Care

Physicians articulated several concerns regarding race-based medicine. They believed race failed to categorize patients adequately;17,19,20,21 normalized the poor health of racial minorities;17,20 caused patient distrust;19,21,22 created physician discomfort;19,21,22 and legitimized the concept of biological race.20 However, even when arguing against race-based clinical care, some physicians used the language of racial heritability.20,21 For example, some physicians argued that race was a poorly defined construct due to genetic mixing between racial groups. No quantitative studies evaluated concerns regarding race-based care.

Quantitative Findings: Physician Factors Associated with Race-Based Care

Studies found the following demographic characteristics to be positively associated with the support of (or engagement in) race-based clinical care: older age;6,25,28 female (v male) gender;23 racial/ethnic minority (v non-Hispanic White) identity;25 foreign-born (v US-born).6

Training characteristics positively associated with race-based practice were lack of fellowship training (v fellowship-trained);26 general internal medicine (v other specialty);6 osteopathic (v allopathic) medical education;6 and international (v US) medical training.26

Practice characteristics positively associated with support of (or engagement in) race-based practice were greater23 and fewer25 years in clinical practice; percent time seeing patients;25 percent racially/ethnically minoritized patients;6,25,26 innovative practice style (v non-innovative style);23 use of direct questioning (v perception) to collect patient race;25 practice location in an area with poverty rates greater than 10% (v ≤ 10%);23 urban (v non-urban) practice location;23 rural (v urban) practice location.28

The following attitudes and beliefs were positively associated with support of race-based clinical care: greater comfort collecting race and ethnicity;25 belief in race as a biological trait;6,26 belief that race is clinically relevant;6 belief that genetics has relevance in clinical practice;6 anxiety due to uncertainty;26 and colorblind racial beliefs.28 Notably, belief in genetic difference as the source of racial health disparities was not associated with belief in the importance of race in clinical decision-making.23

DISCUSSION

We found that physicians believed race had biological components and used race to contextualize patients’ health conditions and guide disease management. Some physicians perceived race-based care as evidence-based, personalized care providing superior outcomes for patients. Others distrusted race as a measure that could appropriately capture genetic differences between individuals and questioned the validity of race-based study findings. In quantitative studies, age was the only characteristic consistently shown to be positively associated with the use of race in clinical care.

We found evidence that clinicians used race-specific study findings to justify their engagement in race-based clinical care and their belief in biological race. No study has evaluated whether physicians’ perceptions of race are shaped by medical literature or how their racial beliefs evolve through medical training and practice. We posit that the way race is employed in medical research,30 education, 31 and training may encourage physicians to perceive race as a biological trait. Furthermore, several studies in our review found a positive relationship between age, practice duration, percent time spent seeing patients, and race-based care. While these associations must be interpreted cautiously given the small number of underlying studies, they collectively suggest that, in addition to generational differences in perceptions of race, increased immersion in the medical field may affect how physicians perceive race, possibly through increased exposure to medical literature providing evidence of racial difference. It is important to note that the relationship between practice duration and race-based practice was negative in the Bonham et al. analysis, emphasizing the need for additional studies in this area.

Reliance on medical literature may also explain why physicians who scored higher on anxiety due to uncertainty and practice innovation were more likely to use race in clinical care and found race clinically relevant, respectively, than less anxious and innovative physicians. Of note, in a German study evaluating how general practitioners managed anxiety due to uncertainty, the focus group assessing the study measure suggested guideline use as a way to cope.32 However, guideline use was not included in the final measure due to a lack of correlation with the first two themes elicited in the scale. To our knowledge, no study has evaluated the relationship between anxiety due to uncertainty and guideline use.

Irrespective of the motivation for guideline use, engagement with guidelines may still be moderated by trust in the underlying medical research used in its construction. In medical research, race is not exclusively a marker of biological difference. Researchers often do not state the relevance of race to the research question30 nor explain whether race represents biological or environmental factors. Support of race-based clinical practice necessitates approval of how race is measured, utilized, and interpreted in research from which the practice is derived. Indeed, we found several instances where physicians disapproved of race-based practice due to concern about the validity and interpretability of race-specific outcomes. As such, conversations regarding the relevance of race in clinical care may be conversations about the appropriateness of race as a variable in medical research.

While there are biological consequences of being a minoritized race,33 there are no inherent biological differences between individuals belonging to different racial groups.1 The belief that race is inherently biological lends itself to the opinion that race is a quantifiable and reliable measure. Nevertheless, we found that physicians also associated race with patients’ values regarding health. Therefore, it is also possible that the belief in race as a social factor representing shared cultural values is positively associated with the race-based care, albeit in ways that likely differ from practice patterns associated with the belief in race as primarily biological. To our knowledge, no study has evaluated the relationship between the belief in race as a cultural trait and the use of race in clinical care.

Strengths and Limitations

There were several strengths of this study. First, this is the first systematic review to evaluate how physicians conceptualize race and physician factors associated with race-based practice. Second, a mixed-methods approach allowed for greater depth in study findings. Finally, the use of Google Scholar to conduct citation searching on all studies included from the primary database search increased the comprehensiveness of the search and allowed for the inclusion of relevant anthropological studies that may have otherwise been missed.

There were several limitations of this review. First, most studies were conducted at least 5 years ago and, given the recent movements away from race-based care,34,35,36 physicians’ racial beliefs may have evolved since these studies were conducted. Second, while qualitative data is rich and robust, data from qualitative studies were limited by the original study authors’ interpretations and selection of exemplary quotes. Ideas not explored by study authors or excluded quotes may have affected our ability to derive a comprehensive set of themes. However, while there may be additional explanations regarding how and why physicians use race in care, the absence of these explanations does not invalidate the themes elucidated thus far. Third, quantitative studies were cross-sectional, so only associations could be explored. Fourth, there were a small number of quantitative studies, with findings often explored in only one analysis and occasionally contradictory. Fifth, survey studies had low response rates and, therefore, findings may not fully reflect beliefs or practices of US physicians. However, conversations regarding race can be contentious and provider unwillingness to engage in these conversations should not deter researchers from reporting their findings. Finally, survey studies also likely suffered from selection and response bias. However, there is no evidence to suggest that younger physicians are more likely to underreport their use of race-based care than older physicians or that younger physicians and older physicians with high and low, respectively, use of race in clinical care were less likely to participate in the study. Therefore, the positive association between age and race-based practice can be considered valid.

CONCLUSION

Physicians had varying perceptions of race, but most believed race had biological components. Moreover, age was associated with race-based clinical practice, which could result from repeated exposure to racialized care in medical literature in addition to generational changes in how physicians perceive race. No studies have evaluated the association between reliance on medical guidelines and using race to guide clinical care. Furthermore, nothing is known regarding how physicians’ racial beliefs evolve during training and practice. Additional research on the evolution of physicians’ racial beliefs, and the role of medical literature in shaping these beliefs, is needed.

Given the extent to which race is embedded in medical infrastructure, moving towards a unified view of race as a sociopolitical construct that does not adequately capture human genetic diversity34,35,37 requires system-wide efforts challenging how race is studied, taught, and utilized in medical research, education, and care.38 To this point, several journals39,40,41 have developed guidelines regarding how authors can describe race in their research, which may have the beneficial effect of reorienting clinicians from viewing race as a proxy for biological or cultural difference and encourage consideration of the structural factors42 that contribute to racial differences in health. Greater scrutiny31,43 of how medical trainees are taught race is also needed. Medical students should be taught the history of race in medicine and the association between race and other social determinants of health, and discouraged from using race as a genetic proxy for disease risk. We should continue removing race from medical equations44 and guidelines, where appropriate, and reorient our understanding of race to account for the impact of racism on health when needed. Reframing race requires both inclusion of language relating genetic ancestry45 to disease risk and considering the ways in which structural factors9,42,46 correlate with race and, thus, disease. Finally, more work is needed to explore the factors for which race serves as proxy and the mechanisms through which racism47 affects biology and, thus, health.