Can this be done by at 8 pm 4/22

AFRICAN-AMERICAN PREFERENCE FOR SAME-RACE HEALTHCARE PROVIDERS: THE ROLE
OF HEALTHCARE DISCRIMINATION

Objective: To determine the extent to which
African Americans prefer same-race clinicians

and the extent to which: 1) knowledge of

historical mistreatment; 2) perceptions of

current racial inequities in medical treatment;

and 3) personal experiences of discrimination

are associated with preference for same-race

healthcare providers among African Ameri-

cans.

Design: Statistical analysis of a nationally

representative telephone survey designed by

the Henry J. Kaiser Family Foundation and

conducted by Princeton Survey Research

Associates (PSRA). Bivariate significance is

determined by using chi-square tests of

association. Multinominal logistic regression

models adjust for age, gender, income, edu-

cation, and self-reported health status.

Results: Approximately one in five African

Americans states a preference for a same-race

healthcare provider. Neither knowledge of

historical mistreatment nor perceptions of

current racial inequities in medical treatment

are related to preferred race of healthcare

providers. In contrast, personal experiences of

discrimination in health care are associated

with a preference for same-race healthcare

providers.

Conclusions: The results suggest that while
knowledge of unfair treatment historically and

perceptions of current racial inequity do not

affect preferences, personal experiences of

unfair treatment may have a significant effect

on African-American patients’ preferences re-

garding health care. Findings suggest that

rather than focusing on how historical mis-

treatment and current inequities in medical

treatment affect individual patients, research

should focus on individual experiences. (Ethn

Dis. 2005;15:740–747)

Key Words: African Americans, Discrimina-
tion, Professional Patient Relations

Jennifer Malat, PhD; Michelle van Ryn, PhD, MPH

INTRODUCTION

While only 4% of physicians are

Black, <20% of African Americans report having a same-race physician, 1 which makes African-American patients much more likely than other-race patients to receive health care from African-American physicians. 2–4 Re- searchers have assumed that patient preferences influence the race of pa- tients’ healthcare providers, but the nature of preferences and what factors inform them have not been clarified. 5 Explicating this relationship can provide insight into the sometimes problematic relationship between healthcare workers and African-American patients and improve efforts to ensure that appropri- ate health care is provided to African Americans. This paper examines pre- ferred healthcare provider race among African-American adults and assesses the extent to which perceptions of racial discrimination are associated with these preferences. BACKGROUND Research on how patients choose physicians suggests that patients usually do not undertake a systematic review of physicians. 6,7 Nonetheless, interperson- al expectations or belief systems appear to influence preferences for healthcare provider characteristics. Research in this area generally examines women’s pref- erence for female clinicians and finds, for example, that women tend to prefer a female gynecologist because of factors like religious beliefs and interpersonal comfort. 8,9 Healthcare provider race has not been adequately explored as a social factor influencing patients’ choice of provider. Extant research is based on local samples or asks about factors that influenced selection of one’s regular physician. 2,10 The limitation of the latter approach is that if preference for a Black physician exceeds the supply, some respondents are not able to express their preference. Research is needed that overcomes these limitations in the assessment of preferred provider race. The poor relationship historically between the African-American commu- nity and the medical and public health communities may lead to a preference among African Americans for same-race healthcare providers. For example, when medicine sought professional status in the United States, African Americans’ unjust legal and social standing, along with White physicians’ prejudice, led to medical experimentation and abuse. 11 The oft-cited Tuskegee Syphilis Study, which occurred in the middle of the 20th century, is one incident in a long history of mistreatment. 12 Some authors From the Department of , University of Cincinnati, Cincinnati, Ohio (JM); Department of Family Medicine and Community Health and Division of Epide- miology, University of Minnesota, Minne- apolis, Minnesota (MVR). Address correspondence and reprint requests to Jennifer Malat; Department of , University of Cincinnati; PO Box 210378; Cincinnati, OH 45221; 513-556- 4709; 513-556-0057 (fax); jennifer.malat@ uc.edu This paper examines preferred healthcare provider race among African-American adults and assesses the extent to which perceptions of racial discrimination are associated with these preferences. 740 Ethnicity & Disease, Volume 15, Autumn 2005 have reported that mistrust of medical and public health workers among African Americans is one consequence of this history. 5 However, while re- searchers and practitioners have specu- lated that this history affects African Americans’ preference for the race of their healthcare providers, no empiric research has assessed this re- lationship. While medical and public health practices have changed in the past several decades, inequities in medical care persist. African Americans are less likely than Whites to receive appropri- ate medical care, from basic treatment to high technology services, for a num- ber of health problems. 13 For example, studies of the racial disparity in treat- ment of heart disease and stroke generally report that Whites are more likely to undergo invasive medical procedures. 14–17 While researchers have persuasively documented this gap, little is known about its effect on the thinking of African-American patients. LaVeist and colleagues 18 have shown that per- ceptions of unequal treatment in health care are related to lower patient satis- faction, which suggests that racial dis- parities in care influence African-Amer- icans’ attitudes toward care. In addition to knowledge of group- level inequities, individuals may have personal experiences of racial discrimi- nation in medical encounters. Percep- tions of personal unfair treatment have been linked to patient sociodemo- graphic characteristics, 19 and some evi- dence from a small sample shows that these perceptions can affect patient behavior. 20 In sum, both knowledge of unequal treatment of African Americans and personal experiences of discrimination may increase Afri- can-American patients’ likelihood of preferring a same-race healthcare pro- vider. The aim of this paper is to de- termine the extent to which African Americans prefer same-race healthcare providers and to assess how perceptions of racial discrimination in health care affect racial preferences. Specifically, the analysis will evaluate the extent to which: 1) knowledge of historical mis- treatment; 2) perceptions of current racial inequities in medical treatment; and 3) personal experiences of discrim- ination predict preference for same- race healthcare providers. Understand- ing how perceptions of racial discrimi- nation affect preferences can provide insight into decision making by African Americans and help guide medical out- reach to African Americans, who con- tinue to suffer poor health and reduced access to care at a higher rate than Whites. 21 METHODS Data The data come from a subsample of the 1999 telephone survey, ‘‘Americans’ Perceptions of Racial Disparities in Health Care.’’ Detail on the methods can be found in Lillie-Blanton et al. 22 Briefly, the survey included a nationally representative sample of 3886 adults living in households with telephones in the continental United States. A dispro- portionate stratified sample of random- digit telephone numbers was used to oversample African-American and Latino respondents. The analysis in this paper was limited to the non- Hispanic Black sample (n51,189). Seventy-two percent of the residential numbers in the sample were contacted by an interviewer; of these, 69% answered screener questions, 93% of those screened were found eligible for the interview, and 98% of eligible respondents completed the interview. Therefore, the final response rate was 49%. 23 Measures The dependent variable, preferred provider race, was assessed by the survey question, ‘‘If you had to choose, would you prefer to be treated by a doctor or nurse of your own race or ethnic group, or not?’’ Knowledge of historical mistreat- ment is indicated by knowledge of the Tuskegee Syphilis Study. In the survey, respondents were asked whether they had heard of the Tuskegee Syphilis Study. Those who responded ‘‘yes,’’ were asked which of three options described the Tuskegee Syphilis Study: 1) a much-criticized government study of syphilis treatment involving African- American men (correct); 2) the African- American airmen who fought in World War II; or 3) a study of heart disease among African-American men. The response choice order was randomized in the administration of the survey. For the present analysis, these two variables were used to create a new variable. Report of having heard of the study and identifying the correct description of the study were labeled as ‘‘correctly identi- fied’’ on the new variable. Responding that one had not heard of the study, or failing to correctly identify it, were labeled ‘‘did not know/incorrectly iden- tified.’’ Nearly 72% of African Amer- icans who had heard of the study correctly identified it. Perceptions of current inequities in the delivery of health care were assessed with two survey items. Respondents were asked, ‘‘how often do you think a person’s race or ethnic background affects whether they can get routine medical care when they need it’’ and ‘‘specialized treatments or surgery when they need it.’’ Response categories were very often, somewhat often, not too often, and never. Perceptions a of personal experiences of racial discrimination in health care a Qualifying discrimination reports as ‘‘per- ceptions’’ may be less than ideal because it implies doubt about the veracity of respon- dents’ reports. Nonetheless, because the data are not the result of researcher observation, respondent reports of discrim- ination are often referred to as ‘‘percep- tions’’ of discrimination (eg, references 19,20,41) PREFERENCE FOR SAME-RACE PROVIDERS - Malat and van Ryn Ethnicity & Disease, Volume 15, Autumn 2005 741 were measured with two items. The first asked whether during the past few years respondents had been treated unfairly because of their racial or ethnic background. The second item asked the same about the respondents’ family. Response categories were yes and no. Several control variables were in- cluded in the multivariate models. The survey collected respondents’ age in years. Based on findings from research on cohort differences in racial atti- tudes, 24 age was grouped into three categories: 18–44 years, 45–54 years, and $55 years. Education data were collected by asking respondents the highest grade or degree completed. The responses were recorded in eight categories, which were recoded into four categories: less than high school, high school diploma or equivalent, some college, and a college degree or more. The respondents’ household income was assessed by a pair of questions. The first asked whether the respondents’ income was more or less than $25,000. Based on this response, respondents were asked to place their income in a more precise income category. More than 15% of the African Americans in the sample were missing data on this control variable. Based on responses to these variables, a new variable with five categories was created: ,$20,000, $20,000–$35,000, $35,001–$50,000, .$50,000, and missing. Last, health status was included in the analysis with the standard self-reported health item, ‘‘In general, how would you describe your own health? Is it excellent, good, only fair or poor?’’ Self-reported health is a good indicator of overall health status. 25 Analytic Plan Except when noted, weights were applied to the data presented here. Weights took into account region of residence, gender, age, race, and educa- tion as well as known nonresponse biases in telephone interview surveys. The demographic weighting parameters were developed from an analysis of the March 1998 Current Population Survey. The weights were derived by using an iterative technique that simultaneously balances the distribu- tions of all weighting parameters. 23 In addition, all of the parameter estimates presented in this paper were estimated by using the statistical pack- age, Stata version 7.0. 26 Stata can adjust standard errors to reflect complex (rather than simple random) survey designs. In this analysis, the five strata used in the sample selection are ac- counted for in the calculation of the standard errors. Weights were applied to the models by using Stata’s ‘‘svy’’ commands. Significance values for bivariate associations were determined by using chi-square tests of association. Multi- variate analyses use multinominal logis- tic regression models, which simulta- neously estimate binary comparisons among the categories of the dependent variable. The explanatory variables were added in conceptually meaningful blocks to a baseline model that included only the control variables. As will be seen, many of the explanatory variables do not achieve significance. Entering the variables in conceptual blocks allows one to easily observe the effect of each conceptual block. In analysis not pre- sented, all variables were included in a single model and produced similar results. The relative risk ratios compar- ing those who prefer a same-race healthcare provider and those who have no preference, and those who prefer a different-race healthcare provider and those who have no preference are presented. Significance values were de- termined with a Wald test for the coefficients or block of coefficients. Significance values are not calculated for a single comparison on the de- pendent variable (eg, between prefer same race and no preference only), but for the complete multinomial model and all possible comparisons on the dependent variable. Further, significance values are based on the block of variables entered simultane- ously. RESULTS Table 1 presents the distribution of the variables for African Americans. Approximately 20% of African Amer- icans stated a preference for a same-race provider, while two thirds responded that they had no preference. Forty-two percent of African Americans correctly identified the Tuskegee Syphilis Study. More than 60% of African Americans feel that race affects routine and spe- cialized medical treatment either very often or somewhat often. Finally, re- ports of unfair treatment of family members are slightly more common than reports of personal unfair treat- ment (18.6% vs 14%). The distribution of the latter four variables was presented previously by Lillie-Blanton et al 22 in their analysis of these data. Table 1 also presents the distribution of the control variables. Table 2 presents the distribution of preferred healthcare provider race by the explanatory and control variables. The association between correctly iden- tifying the Tuskegee Syphilis Study and preferred healthcare provider race is not statistically significant (P..05). Similarly, the perceived diffi- culties of obtaining routine and special- ty medical treatment are not signifi- cantly associated with pr eferr ed healthcare provider race (P..05). In contrast to these results, personal and familial experiences of discrimination in health care are significantly associated with preference for same-race healthcare providers. African Americans who re- port racial unfair treatment in health care of themselves or a family member are more likely to prefer a same-race healthcare provider (36.6% vs 18.1%, P,.01, and 35.3% vs 17.5%, P,.01, respectively). PREFERENCE FOR SAME-RACE PROVIDERS - Malat and van Ryn 742 Ethnicity & Disease, Volume 15, Autumn 2005 Overall, demographic and health variables have disparate relationships with preferred race of healthcare pro- vider. Education and self-reported health status are not significantly related to preferred provider race. Gender is marginally significant, with African-American men being more likely than African-American women to prefer a same-race healthcare pro- vider (P5.09). Income also has a mar- ginally significant relationship to the dependent variable; those with the highest incomes are most likely to state a preference for a same-race healthcare provider (P5.05). Age is significantly associated with preference (P,.01). African Americans older than 55 are the least likely to state a prefer- ence for a same-race healthcare pro- vider; the middle age group (45– 54 years of age) is most likely to state a preference for same-race healthcare providers. The multivariate analyses replicate nearly all of the bivariate findings. Neither knowledge of the Tuskegee Syphilis Study nor perceptions that the provision of health care is inequal are significantly related to preferred health- care provider race for African Americans (see Table 3, models 1 and 2). In contrast, familial and personal experi- ences of unfair treatment in health care are significant predictors of preferred healthcare provider race (P,.05). Ad- justing for the other variables in the model, African Americans who report having been treated unfairly because of race in the medical setting are 1.9 times more likely to prefer a same-race healthcare provider versus stating no preference and 1.84 times more likely to prefer a same-race healthcare provider versus a different-race healthcare pro- vider (see Table 3). Similarly, report of family experiences of unfair racial treatment in health care significantly increases the probability of preferring a Black healthcare provider over having no preference (relative risk ratio51.59, P,.05). The findings Table 1. Distribution of variables Weighted % Unweighted N Dependent variable Preferred provider race Own race 20.7 255 Other 12.6 157 No preference 66.7 768 Explanatory variables Knowledge of Past Unfair Treatment Knowledge of Tuskegee Syphilis Study Correctly identified 42.1 573 Did not know / Incorrectly identified 57.9 616 Knowledge of Current Health Care Inequalities Race affects receipt of routine care Very often 24.4 316 Somewhat often 37.3 434 Not too often 28.9 315 Never 7.4 88 Race affects receipt of specialized treatment Very often 26.7 337 Somewhat often 37.5 414 Not too often 26.3 287 Never 9.6 112 Experiences of Personal Unfair Treatment Respondent treated unfairly due to race Yes 14.0 167 No 86.0 1009 Family member treated unfairly due to race Yes 18.6 241 No 81.4 901 Control variables Gender Male 44.6 489 Female 55.4 700 Age 18–44 59.5 734 45–54 15.7 196 55 or older 24.8 237 Education Less than high school 21.6 176 High school 40.0 443 Some college 24.2 330 College or more 14.2 234 Income Under $20,000 30.4 342 $20,000–$35,000 24.1 285 $35,000–$50,000 16.5 200 Over $50,000 12.6 181 Missing 16.4 181 Self-rated health Excellent 25.3 315 Good 47.7 567 Fair 20.4 249 Poor 6.7 52 Note: Except race/ethnicity variable, table includes only non-Hispanic Black respondents. PREFERENCE FOR SAME-RACE PROVIDERS - Malat and van Ryn Ethnicity & Disease, Volume 15, Autumn 2005 743 are inconsistent in that personal experience of unfair treatment increases the probability of preferring another- race doctor over no preference, while familial experience decreases the prob- ability. Tests for the significance of the control variables show that age is the only consistently significant variable in the multivariate models. Those age $55 are less likely than their younger counterparts to state a preference for a Black healthcare provider over a dif- ferent-race provider or having no pref- erence. Gender is significant in one model (P,.05, model 2); men are more likely to state a preference for same-race providers. DISCUSSION One finding of this analysis is that approximately one in five African Americans reports a preference for a same-race healthcare provider. We are aware of no other study assessing preference for same-race providers in a national survey of African Americans. Extant local studies tend to find no or low stated preference for Black provid- ers 27,28 ; however, this finding is not consistent. 29 The present study finds a preference, though limited, for same- race providers. These results may be because the proportion of African Americans preferring a same-race pro- vider is indeed low. However, other explanations are possible as well. For example, only 4.4% of physicians and 8.8% of nurses are Black, 30 and some respondents may not state a preference for an unavailable provider, feeling that it is useless. Also, respondents may be reluctant to state a preference to an unknown survey interviewer. In partic- ular, perceived race of the interviewer can affect responses to questions about racial topics. 31,32 This dataset contains no information about the race of the Table 2. Preferred race of healthcare provider by explanatory and control variables Same race Different race No preference Explanatory variables Knowledge of Past Unfair Treatment Knowledge of Tuskegee Syphilis Study Correctly identified 23.6 13.3 63.2 Did not know / Incorrectly identified 18.7 12.1 69.3 Knowledge of Current Health Care Inequalities Race affects receipt of routine care Very often 25.8 12.4 61.8 Somewhat often 19.9 14.7 65.5 Not too often 20.8 11.0 68.1 Never 11.1 10.8 78.1 Race affects receipt of specialized treatment Very often 20.5 10.7 68.8 Somewhat often 23.6 12.1 64.3 Not too often 21.4 13.9 64.7 Never 12.6 10.9 76.5 Personal Experiences of Unfair Treatment Respondent treated unfairly due to race* Yes 36.6 13.8 49.6 No 18.1 12.5 69.4 Family member treated unfairly due to race* Yes 35.3 11.3 53.5 No 17.5 13.0 69.6 Control variables Gender Male 24.3 10.2 65.6 Female 17.9 14.5 67.6 Age3 18–44 22.2 11.7 66.1 45–54 31.0 15.4 53.6 55 or older 9.7 13.3 77.1 Education Less than high school 17.3 13.1 69.7 High school 19.3 10.3 70.4 Some college 24.1 14.4 61.5 College or more 24.1 14.8 61.1 Income Under $20,000 19.4 13.6 67.0 $20,000–$35,000 12.9 13.8 73.3 $35,000–$50,000 27.7 10.2 62.1 Over $50,000 32.4 13.5 54.2 Missing 18.4 10.5 71.1 Self-rated health Excellent 23.3 14.7 62.1 Good 20.1 10.9 69.0 Fair 18.1 13.8 68.2 Poor 25.5 13.6 60.8 * P,.01 for chi-square test, 3 p,.05 for chi-square test Note: Includes only non-Hispanic Black respondents. . . .one in five African Americans reports a preference for a same-race healthcare provider. PREFERENCE FOR SAME-RACE PROVIDERS - Malat and van Ryn 744 Ethnicity & Disease, Volume 15, Autumn 2005 interviewer or the respondent’s percep- tion of the interviewer’s race. Previous researchers have speculated that knowledge of past mistreatment of African Americans might influence patients’ behaviors and preferences. 5,33 The present data do not support this notion. Knowledge of the Tuskegee Syphilis Study is not associated with preferred healthcare provider race. One possible explanation for this result is that the Tuskegee Syphilis Study is a historical event, which can be used to understand current events, but none- theless lies in the past. Dwelling on past unfair treatment would make navigating daily life difficult for most people. Another potential explanation for the difference between publicly expressed sentiment and the present finding is that, when communicating with public health officials, African Americans may more readily discuss historic abuses than personal experiences. Publicly describ- ing personal experiences of unfair treatment may be difficult for a variety of reasons, such as uncertainty about the cause of the bad experience, a desire for privacy, or fear of being discounted or labeled overly sensitive. Pointing to well-documented cases of poor historic treatment may provide a way to more safely protest personal mis- treatment. We also found that perceptions of present inequities in medical care are Table 3. Models predicting preferred healthcare provider race Model 1 Model 2 Model 3 Own Race Other Race Own Race Other Race Own Race Other Race Risk Ratio Risk Ratio Risk Ratio Risk Ratio Risk Ratio Risk Ratio Explanatory variables Personal unfair treatment 1.89* 1.84* Family member treated unfairly 1.59* 0.88* Race affects receipt of specialized treatment (very often excluded) Somewhat often 1.97 1.13 Not too often 1.70 1.77 Never 1.31 1.44 Race affects receipt of routine care (very often excluded) Somewhat often 0.52 1.16 Not too often 0.62 0.58 Never 0.33 0.67 Know of Tuskegee Syphilis Study 1.06 1.20 Control variables Male 1.33 0.68 1.28* 0.58* 1.26 0.70 Age (under 44 excluded) 45–54 1.593 1.583 1.463 1.353 1.66* 1.64* 55 and more 0.393 1.013 0.333 0.833 0.47* 1.08* Income (under $20,000 excluded) $20,000–$35,000 0.57 0.88 0.57 0.84 0.54 0.96 $35,000–$50,000 1.35 0.69 1.48 0.77 1.39 0.72 Over $50,000 1.64 1.05 1.75 1.14 1.72 1.13 Missing 0.88 0.73 0.99 0.58 0.86 0.77 Education (high school excluded) Less than high school 1.05 1.25 1.12 1.18 1.00 1.09 Some college 1.24 1.54 1.23 1.74 1.16 1.61 College or more 1.04 1.30 1.01 1.47 0.87 1.38 Self Rated Health (excellent excluded) Good 0.86 0.68 0.98 0.80 0.84 0.65 Fair 0.97 0.84 1.13 1.07 0.95 0.89 Poor 1.62 0.90 1.89 1.27 1.48 0.73 (N) (1150) (1099) (1099) * p,.05 for Wald test of coefficients/block of coefficients. 3 p,.01 for Wald test of coefficients/block of coefficients. Note 1: Includes only non-Hispanic black respondents. Note 2: ’No Preference’ omitted category of dependent variable. PREFERENCE FOR SAME-RACE PROVIDERS - Malat and van Ryn Ethnicity & Disease, Volume 15, Autumn 2005 745 not related to preferred race of health- care provider. One explanation is that some African Americans may believe that a healthcare provider’s race is unrelated to the likelihood of African Americans receiving fair medical treat- ment. Racial inequities in medical care may be attributed, for instance, to patient characteristics like insurance coverage. Or inequities may be attrib- uted to other healthcare provider char- acteristics such as class status. Indeed, LaVeist and colleagues 19 found that class-based interpersonal discrimination is the most frequently reported form of discrimination among African Ameri- cans. In these circumstances, racially concordant care would not ensure better treatment. In this analysis, the only form of unfair medical treatment that affected preferences was personal experiences of discrimination. Reports of unfair treat- ment were associated with greater likeli- hood of preferring a same-race provider versus no preference. These results support other research that shows perceptions of being treated unfairly in health care can affect patients’ behav- ior. 20 The present analysis reinforces existing studies by testing the effect in a population-based national sample. However, the present analysis also produces another result: personal unfair treatment is associated with greater likelihood of preferring an other-race provider over no preference. Because this relative risk ratio is only making a comparison between preference for other-race providers and no preference, it does not diminish the effect described for preference for same-race providers. Taking these effects together, experienc- ing unfair treatment appears to increase the likelihood of developing a prefer- ence. Together these results give insight into how African Americans choose physicians; personal experiences of un- fair racial treatment in health care may be influential in decision making. That personal experiences are significant may not be a particularly remarkable finding. Yet, that they are the only significant finding suggests that more attention should be given to individuals’ experi- ences with discrimination along with a focus on group-level unfair treatment. Some effects of control variables are worth considering. The oldest group was least likely to prefer a same-race healthcare provider, while the middle group was most likely. These differences may be influenced by cohort experiences in relation to the civil rights movement. Some researchers have suggested that experiences during young adulthood influence individuals’ …