Introduction

Race is widely-recognized as a primarily social, not biological, construct. Generally in health services and outcomes research, race/ethnicity is measured as respondent self-identification. However, race/ethnicity is also ascribed to individuals by others in social interactions, referred to as socially-assigned race. One's self-identification may or may not be the same as his/her socially-assigned race. Social assignment may be a largely unrecognized determinant of observed racial/ethnic differences in healthcare outcomes.

Differences in healthcare outcomes between patients who self-identify as racial/ethnic minority and those who self-identify as white are widely-recognized [1], [2]. However, the broad demographic classifications currently used (i.e. White, Black/African American, Asian, American Indian/Alaska Native and Native Hawaiian/Pacific Islander and Hispanic), may obscure differences within racial/ethnic groups where there is often variability in phenotypic characteristics. Thus, re-examining and expanding the definitions of racial/ethnic categories is one recommended way to enhance the quality of health services research and delivery [3], and isolate factors associated with racial/ethnic healthcare disparities that may be obscured by the current categorization strategy [1], [3]. A novel study recently found self-identified racial/ethnic minority individuals who reported being perceived in societal interactions as white had higher levels of self-reported health compared with racial/ethnic minority individuals who reported being perceived by society as minority [4]. This intriguing new area of research suggests that race as it is perceived by others or socially-assigned race, in addition to self-identified race/ethnicity, may be associated with key health and healthcare outcomes. Notably, this prior work concluded being socially-assigned as non-Hispanic white conveyed an advantage in health outcomes regardless of individual self-identification. It is unknown whether this previously observed advantage of being socially-assigned as non-Hispanic white extends beyond health status to areas of persistent healthcare inequity such as receipt of preventive health services.

Patients who self-identify as racial/ethnic minorities underutilize recommended preventive health services such as age-appropriate vaccinations [5]–[10] and disease screening [11]–[17]. These racial/ethnic inequities remain despite adjusting for insurance coverage [16], socioeconomic status [11], [12], [15], [16], often raising the question of whether the experience of healthcare discrimination may be an important contributor. Patient-reported healthcare discrimination has already been shown as independently associated with limited healthcare utilization, self-reported quality of care, low adherence to care plans and poor health outcomes [18]–[23].

The Institute of Medicine identified closing the healthcare utilization gap and eliminating any contribution of healthcare provider bias to observed racial/ethnic healthcare inequities as two priorities for reducing racial/ethnic disparities in healthcare [1]. Recognizing race/ethnicity as a complex phenomenon that takes into account both self-identification and social interactions with individuals and institutions [24], we sought to first examine the agreement between self-reported race and self-report of socially-assigned race, and then characterize its association with both reported healthcare discrimination and self-reported receipt of preventive healthcare services. We hypothesized that self-identified racial/ethnic minority respondents who report being socially-assigned as minorities would report higher rates of healthcare discrimination when compared with: 1) racial/ethnic minority respondents who report being socially-assigned as non-Hispanic white and 2) self-identified non-Hispanic white respondents. We further hypothesized that self-identified racial/ethnic minority respondents who report being socially-assigned as non-Hispanic white would have higher rates of self-reported recommended preventive health service utilization compared with self-identified racial/ethnic minority respondents who report being socially-assigned as minorities. In addition, we expected that rates of utilization would be similar between self-identified minorities socially-assigned as non-Hispanic white and self-identified non-Hispanic whites.

Methods

Results

Sample Characteristics

The overall sample included 33,679 respondents (Table 1). The majority (78.1%) of the sample self-identified as non-Hispanic white, 15.5% self-identified as black, 4.4% self-identified as Hispanic, and 1.2% self-identified as multiracial (not presented in Table 1). Other racial/ethnic groups comprised less than 1% of the sample. The majority was married (60.1%) and employed (63.5%). About one-half was female (51.8%) and had annual household incomes greater than $35,000 (50.9%). The vast majority of respondents had completed at least high school (90.0%) and reported having health insurance (85.5%).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Sample demographic characteristics of 2004 BRFSS Reactions to Race respondents.

https://doi.org/10.1371/journal.pone.0064522.t001

Respondents were categorized into 3 groups, defined by self-identified race/socially-assigned race (minority or non-Hispanic white): M/M (n = 6,837, 19.0%), M/W (n = 929, 3.4%), and W/W (n = 25,913, 77.6%). There was high agreement between self-identified and socially-assigned race for white (98.5%) and black (95.6%) respondents. However, 26% of Hispanic respondents reported being socially-assigned as white, 4.7% as black and 7.4% as other. The M/W group was comprised almost exclusively of participants who self-identified as Hispanic (98.5%). The M/M group differed significantly from the M/W group on a number of sociodemographic variables (Table 1). Compared with the M/M group, the M/W group was more likely to be married (55.6% vs. 41.5%, p<0.001), more likely to have completed high school (85.0% vs. 79.4%, p = 0.004), had higher annual household incomes (p<0.001) and was more likely to have health insurance (80.1% vs. 74.3%, p = 0.01). Both groups were less likely to have health insurance compared with the W/W group (88.5% among W/W respondents, p-values <0.001). Compared with either self-identified minority group, the W/W group was significantly older (51.0±16.8 years) and more likely to be married (64.8%), to have completed high school (92.8%) and to have a higher annual household income (Table 1) (p-values <0.01). The W/W group was more likely to be employed (64.1%, p = 0.005) compared to the M/M group only, and there was no significant difference in employment rates between the M/W group and either comparison group.

Adjusted analysis of association between socially-assigned race and healthcare outcomes

Although both W/W and M/W groups had higher odds of having a personal physician compared with the M/M group in initial analyses, this relationship was significant for only the W/W group (AOR = 1.15, 95% CI: 1.02, 1.29, Table 2). M/W respondents were more likely to receive both influenza (AOR = 1.83, 95% CI: 1.16, 2.87) and pneumococcal (AOR = 1.43, 95% CI: 0.93, 2.20) vaccinations than the M/M group, although the latter relationship was in part explained by sociodemographic variables (Table 2). Similarly, the W/W group was significantly more likely to receive either vaccination than the M/M group (influenza: AOR = 1.81, 95% CI: 1.48, 2.21; pneumococcal: AOR = 2.20, 95% CI: 1.78, 2.71). There were no statistically significant differences between the M/M and M/W groups for uptake of any of the appropriate cancer screening tests. In contrast, after adjustment, W/W women were less likely to have had appropriate breast cancer screening (AOR = 0.87, 95% CI: 0.77, 0.98) and cervical cancer screening (AOR = 0.67, 95% CI: 0.54, 0.84) compared with M/M women (Table 2). Unadjusted prostate and colorectal cancer screening odds ratios were higher for W/W respondents compared with M/M respondents; however, these relationships were attenuated in the adjusted analysis (Table 2). The M/W respondents had lower odds of healthcare discrimination compared with M/M respondents (AOR = 0.61, 95% CI: 0.39, 0.95, Table 2), after adjustment for potential confounders. W/W respondents also had lower odds of healthcare discrimination than M/M respondents (AOR = 0.27, 95% CI: 0.22, 0.33) and M/W respondents.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Logistic regression analysis of the association between racial/ethnic assignment group and health-related outcomes (Minority/Minority as reference group).

https://doi.org/10.1371/journal.pone.0064522.t002

Discussion

We found that U.S. adults in this study who self-identify as racial/ethnic minorities, but report being socially-assigned as non-Hispanic white, reported better healthcare outcomes compared with self-identified racial/ethnic minorities who are socially-assigned as racial/ethnic minorities. This minority/white (M/W) group was also significantly less likely to report healthcare discrimination compared with the minority/minority (M/M) group. However, the M/W group was significantly more likely to report healthcare discrimination than the white/white (W/W) group in our sample (results not shown). M/M respondents were less likely to have a personal physician or a medical home, and had lower rates of annual influenza and pneumococcal vaccinations compared with both the M/W and W/W groups. The M/W and W/W groups had similar rates of influenza immunization, and the M/W group had lower rates of pneumococcal immunization than the W/W group. These key differences in healthcare outcomes persisted after we adjusted for potential explanatory factors including health insurance, marital status, education, employment, and income.

Previously published comparisons with whites have repeatedly demonstrated lower rates of both influenza and pneumococcal vaccinations among African Americans and Hispanic Americans [5]–[10], populations disproportionately burdened with bacterial pneumonia [32] and associated, preventable hospitalizations [33]. Yet, our finding that M/W group immunization and cancer screening rates were most frequently similar to those of W/W respondents, suggests M/W represents a unique subset of racial/ethnic minority patients. These findings are consistent with earlier research demonstrating that self-identified racial/ethnic minorities who are socially-assigned as white had better self-reported overall health status compared to those who are socially-assigned as non-white [4]. Contrary to our initial hypothesis, there were no differences in prostate or colorectal cancer screening rates between groups, and, after adjustment, M/M women were more likely to receive breast and cervical cancer screening compared with W/W women. Prior studies have found no clear association between perceived discrimination and cancer screening [18], [34], [35], suggesting that although racial/ethnic minority patients experience discrimination in healthcare settings, they may participate in cancer screening programs that employ culturally-appropriate outreach strategies. In turn, immunization programs may seek to adopt some of these methods.

The concept of “white advantage” conferred to a sub-group of self-identified racial/ethnic minority individuals should be considered in our sample, and Hispanic Americans are most likely among racial/ethnic minority groups to be socially-assigned as white, thus potentially benefitting from this phenomenon. Among many of the socio-demographic indicators, including having completed high school and household income, the W/W and M/M groups were at disparate ends of the spectrum, while the M/W group consistently had intermediate proportions or mean values. Our findings of relative socioeconomic privilege among the M/W group may reflect the continued legacy of racism and skin colour prejudices that are perpetuated in the United States. For example, there is substantial evidence that among racial/ethnic minorities, darker skin pigmentation is associated with lower educational attainment, occupational status, and income [36], [37], and is adversely associated with health outcomes such as mortality risk [38], and self-reported physical health [39]. Therefore, being a minority who is perceived as non-Hispanic white may confer both socioeconomic and healthcare advantages.

Our findings that M/M respondents were almost twice as likely as M/W respondents to report having experienced healthcare discrimination in the previous year, and that M/W respondents reported healthcare discrimination more frequently than W/W respondents, likely have important implications. Previous reports have associated healthcare discrimination with poor health outcomes [22], [40], [41], and other studies have associated healthcare discrimination with delays in obtaining ordered tests and treatment, not filling prescriptions [21], and low patient satisfaction and adherence [42]. Although both self-identification and social assignment of minority status are associated with reported healthcare discrimination, our findings suggest M/W and M/M respondents represent two distinct groups. Moreover, the M/W status conveys a unique experience within the healthcare setting; rates of reported healthcare discrimination and utilization of vaccinations for the M/W group differs significantly from either the M/M or the W/W groups. Findings from the emerging field of physician implicit bias, suggesting a role for unconscious influences on decision-making, may be particularly relevant in this context [43]–[47].

Our work represents a novel inquiry into how self-identified race and socially-assigned race might influence interactions within healthcare settings. Nevertheless, there are some limitations. Although the BRFSS offered a unique opportunity to conduct this study in a large sample of U.S. adults, there are some challenges inherent to its cross-sectional design and sampling approach. First, we cannot assess causality or directionality; however, we have firmly demonstrated several significant associations which merit further exploration. Second, the study relied on the self-report of preventive health service utilization and self-report of socially-assigned race. Validation studies of self-reported immunization and cancer screening, have demonstrated that these data are generally valid, particularly within one year [48]–[50], and development and validation of self-reported racial/ethnic healthcare discrimination measures remains an area of active research [51], [52]. We also lack data on the demographic or other characteristics of healthcare providers and systems with whom respondents interacted. Still, this work has strength in the capture of divergent healthcare experiences from a broad sample of participants across a spectrum of racial/ethnic identities. Third, we selected the 2004 survey because it is the survey year with the greatest number of states using the optional “Reactions to Race” module. Although the overall response rate was typical for BRFSS, the results may only be applicable to the seven states and District of Columbia which self-selected to obtain this information from their residents. Fourth, all non-white respondents were collapsed together to create the M/W and M/M categories. Because certain minority groups enjoy better overall health status, our dichotomous categorization scheme may have attenuated the impact of socially-assigned race on marginalized racial/ethnic minorities, and it fails to demonstrate all of the inequities existing between racial/ethnic minority groups. Therefore, the differences across groups may be even greater than we observed in our analyses. Finally, our study examined multiple healthcare outcomes, thereby increasing our likelihood of a type I error, or rejecting the null hypothesis. Although there is internal consistency in our results, readers should bear this in mind.

These results suggest that in order to fully understand how race and racism may be mediating health inequities in the United States, it is important to assess not only how patients self-identify their race/ethnicity, but also how they report their socially-assigned race/ethnicity. Future studies are needed to investigate the correlation between self-reported socially-assigned race/ethnicity and race/ethnicity reported by observers and to elucidate the mechanisms by which socially-assigned race leads to biases and inequities in healthcare. Provider bias and patient experience of bias may be barriers to receipt of immunizations and merits further investigation and provider education. Understanding the factors enabling full participation in preventive care by the M/W group will allow future interventions to be designed and targeted in the most efficient ways possible.

Acknowledgments

We would like to extend our appreciation to the CDC's Racism and Health Working Group's Science and Publication Committee for developing and refining the Reactions to Race module, especially Drs. Geraldine Perry and Camara Jones, and also to Tara Rizzo for her assistance revising and formatting the manuscript.