Does Racial Concordance Between HIV-positive Patients and Their Physicians Affect the Time to Receipt of Protease Inhibitors?
Received from the UCLA Robert Wood Johnson Clinical Scholars Program (WDK); Division of General Internal Medicine and Health Services Research (MDW, MFS, WEC), Department of Medicine, and Department of Health Services (MFS, WEC), School of Public Health, UCLA, Los Angeles, Calif; and Department of Health Care Policy (BEL), Harvard Medical School, Boston, Mass.
Abstract
BACKGROUND: Compared to whites, African Americans have been found to have greater morbidity and mortality from HIV, partly due to their lower use of effective antiretroviral therapy. Why racial disparities in antiretroviral use exist is not completely understood. We examined whether racial concordance (patients and providers having the same race) affects the time of receipt of protease inhibitors.
METHODS: We analyzed data from a prospective, cohort study of a national probability sample of 1,241 adults receiving HIV care with linked data from 287 providers. We examined the association between patient-provider racial concordance and time from when the Food and Drug Administration approved the first protease inhibitor to the time when patients first received a protease inhibitor.
RESULTS: In our unadjusted model, white patients received protease inhibitors much earlier than African-American patients (median 277 days compared to 439 days; P < .0001). Adjusting for patient characteristics only, African-American patients with white providers received protease inhibitors significantly later than African-American patients with African-American providers (median 461 days vs. 342 days respectively; P < .001) and white patients with white providers (median 461 vs. 353 days respectively; P= .002). In this model, no difference was found between African-American patients with African-American providers and white patients with white providers (342 vs. 353 days respectively; P > .20). Adjusting for patients’ trust in providers, as well as other patient and provider characteristics in subsequent models, did not account for these differences.
CONCLUSION: Patient-provider racial concordance was associated with time to receipt of protease inhibitor therapy for persons with HIV. Racial concordance should be addressed in programs, policies, and future racial and ethnic health disparity research.
Mortality rates from acquired immunodeficiency syndrome (AIDS) is much higher among African Americans than among whites.1,2 This disparity is partly due to the fact that African Americans are less likely than whites to receive HIV medications.3–9 Understanding why disparities in the use of antiretroviral medications exist may lead to improved treatment and outcomes of HIV in African Americans.
Previous studies have shown that income, health insurance, and severity of HIV disease only partly explain racial disparities in antiretroviral use.8,10–13 One unexplored explanation is that aspects of the patient-provider relationship may influence racial disparities in HIV care. Specifically, African-American patients cared for by providers of the same race (race concordant) report greater participation and satisfaction with their provider than those with white providers (race discordant).14–17 This may explain why African Americans in racially concordant relationships are also more likely to use preventive care,17 and less likely to delay seeking care.18 Thus, we hypothesize that race concordance may also explain why African Americans are less likely than whites to receive antiretroviral treatment.
We set out to examine whether African Americans who have providers of the same race are more likely to receive protease inhibitor therapy than those in racially discordant provider relationships. This is an important issue to study in HIV because the timely receipt of antiretroviral medications can have a profound impact on clinical outcomes. We examined data from the HIV Cost and Services Utilization Study (HCSUS), a cohort study of a nationally representative sample of HIV-infected adults receiving medical care and their providers.
METHODS
Study Population
Full details of the HCSUS sampling design are presented elsewhere.19,20 In brief, the reference population was persons at least 18 years old with known HIV infection who made at least one visit, in the context of regular or ongoing care, to a nonmilitary, nonprison medical provider (other than an emergency department) in the contiguous United States, during the period January 5 to February 29, 1996. The HCSUS used a 3-stage sampling design, in which geographical areas, medical providers, and patients were sampled. In the first stage, we sampled 28 metropolitan areas and 25 clusters of rural counties within the United States. In the second stage, we sampled a total of 148 urban and 51 rural providers. In the third stage, we sampled patients from de-identified lists of all eligible patients who visited participating providers during January and February of 1996.
Of the 4,034 subjects who were sampled, 2,864 (71%) individuals completed the baseline survey. All participants gave informed consent and interviews were usually conducted in person. Respondents (2,466; 86.1% of the baseline cohort) completed the first follow-up survey and 2,267 persons (84.5% of baseline) completed the second follow-up survey. The median time from baseline to the second follow-up interview was 15 months, during which time 238 (8%) patients died and 359 (12%) were not located or did not cooperate with follow-up interviews.
To select providers for participation in the provider survey, patients in the first follow-up survey were asked to identify the provider that was most important to their HIV care. Of the 2,466 patients, 222 (9%) were unable to identify a provider. Of the 692 providers identified through this method, 551 had a confirmed identity, location, and mailing address. Each of these providers was mailed a self-administered written questionnaire, which was completed in 1998, approximately the same time as completion of the second follow-up patient interview. A total of 412 providers completed the questionnaire with a response rate of 75%. For this study, we excluded patients who did not have an identified provider or whose provider did not complete the survey. We also excluded 5 providers because they were nurses, 5 providers because their patients were dropped from the baseline survey, and 6 providers with missing key data. Finally, we excluded any patients or providers whose racial or ethnic group was not African American or white. Thus, for our analysis, there was a final “patient-provider sample” that included 1,241 (43%) of the baseline HCSUS cohort with linked data from 287 providers.
Outcome Variables
At baseline and follow-up, subjects were asked which antiretrovirals medications they had used, and for protease inhibitors, when they began using the medication (see Appendix available online at http://www.jgim.org). From this information, we constructed a variable representing time of first protease inhibitor use. Time 0 was set as the earliest date that a protease inhibitor could have been prescribed, which was December 6, 1995 (the date when Saquinavir, the first protease inhibitor, was approved by the Food and Drug Administration [FDA]). Some individuals reported using protease inhibitors before this date, perhaps receiving the medication through a clinical trial. For these individuals, time to protease inhibitor use was set to a small positive number (i.e., 0.01). All patients in our cohort were HIV positive and enrolled in the study after the FDA's approval date and were therefore eligible for protease inhibitor therapy if they met prescribing guidelines for protease inhibitors during the course of the study.
Independent Variables
All independent variables were measured at baseline. There were 3 groups of independent variables: patient-provider racial pairs, patient characteristics, and provider characteristics.
Patient-Provider Racial Pairs. To assess race/ethnicity, patients and providers were asked, “Which of these would you say is your main racial or ethnic group?” The response options were non-Hispanic white, non-Hispanic African American, Hispanic, American Indian or Alaskan Native, Asian or Pacific Islander, mixed race, or some other single race. We focused the racial and ethnic comparisons on African Americans and whites because there were too few cases for comparison among the other racial and ethnic groups. We then constructed 4 analytic categories: African-American patients with African-American providers, African-American patients with white providers, white patients with African-American providers, and white patients with white providers.
Patient Characteristics. From the baseline survey, information was collected regarding patient's age, gender, highest educational level completed, annual family income, insurance status (private fee-for-service, private managed care, Medicaid, Medicare, and uninsured), geographic region (Northeast, Midwest, South, and West), type of HIV exposure, and baseline CD4 count (500+, 200–499, 50–199, and less than 50). A previously derived symptom index was included to assess severity of illness, with a higher score reflecting more symptoms reported by the patient (range 0 to 86).21 We assessed heavy alcohol use and drug dependency in the last year using previously validated questions.8,22
Self-reported access to care was measured using a previously developed 6-item scale.23 Patients were asked to rate from strongly agree to strongly disagree statements regarding whether they could get admitted to the hospital without any trouble, could get emergency medical care, had to go without needed medical care because it was too expensive, had easy access to medical specialists, could get medical care whenever needed, and whether places to get medical care were conveniently located.
To assess patient attitudes toward their care, we asked patients to rate the quality of their medical care during their last visit to their physician or clinic, using an excellent, very good, good, fair, and poor response scale.24 We dichotomized individuals between those who rated their care as fair or poor versus good to excellent. Based on a published measure of patient's trust in their provider, we used a scale created from 2 questions. One question asked patients how much they trust their doctor or clinic to provide them with high-quality medical care. The second question asked patients how much they trust their doctor or clinic to make their health care a top concern.25 The 5 response options for the 2 items were completely, mostly, somewhat, a little, and not at all. We dichotomized the trust variable at the median, which because of the skewed distribution of the responses resulted in dichotomizing those who said “completely” to both questions (more trust) versus all others (less trust). Patients were also asked whether they believed that antiretrovirals were worth taking using a strongly agree to strongly disagree response scale. Similar to our method for the trust variable, we dichotomized these individuals into those who agree versus those who disagreed.
Provider Characteristics. We asked providers about their gender, years of experience as a provider, clinical specialty (general medicine and family practice, infectious disease, and nurse practitioner/physician assistant), whether they defined themselves as an HIV specialist, number of HIV patients in their practice, and patient mix. HIV knowledge was assessed by a previously developed measure of 11 true-false questions about HIV treatment and was calculated based on the percent answered correctly.26 Type of practice setting (private university, public university, public facility, or private setting) was derived from information on the specific location where the physician practiced and from information obtained through the sampling process. Patient mix was derived from a set of questions asking providers to estimate the proportion of their patients who had various HIV exposure risk factors. Providers had a practice primarily of injection drug users, for example, if they indicated that more than 50% of their patients were injection drug users. Patient-mix categories were injection drug users, men having sex with men, heterosexuals, none (less than 50% listed for all categories), and mixed (more than 50% listed for more than one category). We also asked providers whether they believed that many of their patients cannot adhere to protease inhibitor regimens, whether antiretrovirals should be delayed or withheld for nonadherent patients, and whether they preferred not to treat intravenous drug users, using a 5-point response scale from strongly agree to strongly disagree.
Analysis
First, we performed bivariate comparisons of patient and provider characteristics among the 4 patient-physician racial pairs. We used χ2 statistics for comparing proportions, weighted to account for the complex sampling methods.
We used parametric time-to-event models to analyze the effect of racial concordance on time to first protease inhibitor use. We controlled for potential confounders using 4 staged models in which sets of covariates were progressively added at each stage. The first stage was unadjusted and included only the 4 patient-provider race groups as independent variables. In the second stage, we added to the model patient characteristics, which were age, gender, income, highest educational level completed, type of insurance, self-reported access to care, geographic region, HIV exposure, CD4 count, symptom index, and drug dependence or heavy alcohol use in the past year. The third stage model additionally controlled for provider characteristics: gender, years in practice, medical specialty, HIV expertise, HIV knowledge, patient caseload, practice setting, patient mix, belief that patients cannot adhere to antiretrovirals, preference not to treat injection drug users, and belief that antiretroviral should be withheld or delayed for nonadherent patients. The fourth stage additionally controlled for factors related to patients’ attitudes toward their physician and their care. These attitudes were patients’ rating of their medical care, trust in the provider, and the belief that antiretrovirals are worth taking.
For all comparisons, standard errors and statistical tests were adjusted for the complex sampling design and weighting scheme using linearization methods available in Stata (Stata Corporation, College Station, Tex). We used bootstrapping to estimate confidence intervals for predicted median times to first protease inhibitor use.27
RESULTS
Patient Characteristics
Of the 1,241 patients, 61% (803) were white with white providers, 32% (341) were African-American with white providers, 6% (86) were African-American with African-American providers, and less than 1% (11) were white with African-American providers (Table 1). Compared to other patients, African-American patients with African-American providers were most likely to be female (41.0%, P < .0001), have less than a high school education (37%, P < .0001), have annual incomes less than $10,000 (75%, P < .0001), have Medicaid insurance (43%), and live in the southern United States (64%, P < .02). In contrast, white patients with white providers were most likely to have graduated from college (30%), have an annual income of more than $25,000 (45%), have any health insurance (87%), have homosexual intercourse as their primary HIV exposure risk (66%), and live in the western United States. (46%). Self-reported access to care was lowest among African-American patients with white providers (P= .0001). Patients in the 4 different patient-provider race groups were similar with regard to their age distribution, recent drug or heavy alcohol use, lowest reported CD4 count, symptom burden, overall ratings of their medical care, trust in their provider, and the proportion believing that antiretrovirals are worth taking.
| Patient Characteristic | White Providers | African-American Providers | P value * | ||
|---|---|---|---|---|---|
| White Patients | African- American Patients | White Patients | African- American Patients | ||
| N (weighted %) | 803 (61.0) | 341 (31.7) | 11 (0.7) | 86 (6.5) | |
| Age, y | .32 | ||||
| <35 | 29.2 | 35.0 | 31.3 | 36.9 | |
| 35 to 45 | 43.9 | 45.0 | 41.0 | 36.4 | |
| >45 | 26.9 | 20.0 | 27.7 | 26.7 | |
| Female | 9.9 | 36.5 | 14.2 | 41.0 | <.0001 |
| Highest level of education | <.0001 | ||||
| Some high school | 12.2 | 24.6 | 20.4 | 37.2 | |
| High school diploma | 25.3 | 30.0 | 26.5 | 32.0 | |
| Some college | 32.3 | 35.2 | 20.6 | 26.7 | |
| College graduate | 30.2 | 10.2 | 32.5 | 4.2 | |
| Annual family income | <.0001 | ||||
| <$5,000 | 10.2 | 31.3 | 14.9 | 21.4 | |
| $5,000 to $10,000 | 21.8 | 26.7 | 8.3 | 53.5 | |
| $10,000 to $25,000 | 22.9 | 29.1 | 63.8 | 17.7 | |
| >$25,000 | 45.1 | 12.8 | 13.1 | 7.5 | |
| Insurance | <.0001 | ||||
| Uninsured | 13.4 | 22.2 | 43.3 | 33.2 | |
| Medicaid | 16.1 | 40.8 | 25.1 | 43.0 | |
| Managed care | 26.6 | 13.6 | 22.6 | 3.6 | |
| Fee-for-service | 22.9 | 5.1 | 0.0 | 3.5 | |
| Medicare | 21.0 | 18.3 | 8.9 | 16.8 | |
| Self-reported access to care | .37 | ||||
| Low | 37.7 | 46.9 | 34.6 | 42.6 | |
| Medium | 28.2 | 24.8 | 28.6 | 22.6 | |
| High | 34.1 | 28.3 | 36.9 | 34.8 | |
| HIV exposure | <.0001 | ||||
| Injection drug use | 20.7 | 32.5 | 45.0 | 26.4 | |
| Men who have sex with men | 66.3 | 29.9 | 41.6 | 17.9 | |
| Heterosexual intercourse | 7.3 | 25.7 | 9.5 | 33.3 | |
| Other | 5.8 | 11.9 | 3.9 | 22.4 | |
| Drug or heavy alcohol use in last 12 months | 16.8 | 20.0 | 0.0 | 13.4 | .40 |
| Lowest reported CD4 count (cells/ml) | .13 | ||||
| >500 | 8.1 | 10.5 | 4.8 | 12.9 | |
| 200 to 499 | 34.4 | 41.3 | 30.5 | 48.4 | |
| 50 to 199 | 31.7 | 28.6 | 42.2 | 23.3 | |
| 0 to 49 | 25.8 | 19.6 | 22.6 | 15.4 | |
| Symptom burden index | .03 | ||||
| Low | 33.8 | 42.1 | 14.1 | 39.5 | |
| Medium | 31.8 | 30.3 | 46.2 | 38.7 | |
| High | 34.4 | 27.6 | 39.8 | 21.8 | |
| Region | .0008 | ||||
| Northeast | 20.6 | 39.3 | 29.6 | 32.9 | |
| Midwest | 13.1 | 18.2 | 9.5 | 2.9 | |
| South | 20.0 | 27.9 | 42.7 | 63.7 | |
| West | 46.3 | 14.7 | 18.3 | 0.5 | |
| Overall rating of medical care | .39 | ||||
| Poor | 1.7 | 1.7 | 0.0 | 2.3 | |
| Fair | 4.0 | 9.8 | 0.0 | 13.9 | |
| Good | 11.7 | 17.0 | 30.7 | 29.7 | |
| Very good | 27.9 | 23.7 | 3.9 | 22.7 | |
| Excellent | 54.8 | 47.9 | 65.4 | 31.6 | |
| More trust in provider | 51.5 | 51.8 | 25.7 | 57.0 | .39 |
| Agree that antiretrovirals are worth taking | 79.7 | 80.5 | 70.8 | 82.9 | .88 |
- * Statistical comparison adjusted for complex sampling methods.
Provider Characteristics
We compared the provider characteristics of those caring for the different patient-provider racial groups (Table 2). Among African-American patients with African-American providers, only 6% were cared for by infectious disease specialists compared to 42% of whites with white providers, 48% of African-American patients with white providers, and 18% of white patients with African-American providers. African-American patients with African-American providers were most likely to be cared for by providers who are HIV experts but not infectious disease specialists (general medicine specialists; 78%). African-American patients with white providers were least likely to be cared for by providers who are general medicine nonspecialists (neither general medicine HIV experts nor infectious disease specialists). Compared to other patients, white patients with white providers were most likely to have a provider whose predominant patient population are homosexual men (P < .0001), and have sought care in a private office (P= .002). Approximately 38% of African-American patients with African-American providers received care in a public health care facility as compared to 22% of white patients with African-American providers, 4% of African-American patients with white providers, and 1% of white patients with white providers.
| Provider Characteristic | White Providers | African-American Providers | P value * | ||
|---|---|---|---|---|---|
| White Patients | African- American Patients | White Patients | African- American Patients | ||
| Female | 12.9 | 34.5 | 50.1 | 31.4 | .02 |
| More than 10 years in practice | 89.8 | 85.9 | 64.2 | 78.4 | .4 |
| Specialty/HIV expert | .001 | ||||
| Infectious disease | 42.3 | 47.9 | 18.3 | 6.0 | |
| GIM expert | 43.8 | 33.0 | 52.2 | 78.2 | |
| GIM nonexpert | 12.0 | 7.1 | 29.6 | 14.2 | |
| NP or PAs | 1.8 | 12.0 | 0.0 | 1.6 | |
| HIV knowledge (>80% correct) | 56.6 | 67.5 | 26.5 | 48.1 | .21 |
| Predominant patient population | <.0001 | ||||
| Injection drug users | 14.5 | 23.4 | 5.5 | 16.6 | |
| Homosexual men | 77.7 | 48.0 | 75.3 | 14.8 | |
| Heterosexual persons | 1.3 | 20.1 | 19.2 | 55.3 | |
| Mix | 3.3 | 2.3 | 0.0 | 4.2 | |
| None | 3.2 | 6.2 | 0.0 | 9.1 | |
| Site type | .002 | ||||
| Private, university-based | 33.0 | 42.7 | 33.5 | 18.6 | |
| Public, university-based | 10.5 | 24.5 | 45.0 | 25.9 | |
| Public | 1.4 | 3.7 | 21.5 | 38.6 | |
| Private office | 38.3 | 12.0 | 0.0 | 1.3 | |
| Other | 16.8 | 17.1 | 0.0 | 15.6 | |
- * Statistical comparison adjusted for complex sampling methods.
- GIM, general internist; NP, nurse practitioner; PA, physician assistant.
Receipt of Protease Inhibitors
First, we examined time to first protease inhibitor use by patient race. In bivariate analysis, white patients received protease inhibitors a median of 277 days (95% confidence interval [CI], 256 to 301) after the FDA approved the first protease inhibitor, compared to 439 days for African Americans (95% CI, 393 to 494; P < .0001). After controlling for patient demographics, CD4 count, symptom burden, insurance status, self-reported access to care, type of HIV exposure risk, and heavy alcohol or illicit drug use, African Americans still had later use of protease inhibitors than whites (440 vs. 348 days; P= .005).
We then examined time to first protease inhibitor use for each of the patient-provider race groups. Without adjustment for patient or provider characteristics (Table 3, Stage 1 model), white patients with white providers received protease inhibitors 278 days (95% CI, 258 to 304) after the first protease inhibitor was approved by the FDA. Compared to this patient-provider group, African-American patients with white and African-American providers received protease inhibitors later (443 days, P < .001 and 419 days, P= .003, respectively).
| White Providers | African-American Providers | |||
|---|---|---|---|---|
| White Patients | African-American Patients | White Patients | African-American Patients | |
| Stage 1: unadjusted | 278 (258 to 304) | 443 (392 to 503)‡ | 206 (99 to 458) | 419 (337 to 551)† |
| Stage 2: adjusted for patient characteristics | 353 (317 to 386) | 461 (404 to 520)† | 251 (160 to 422) | 342 (269 to 429) |
| Stage 3: adjusted for patient and provider characteristics | 377 (335 to 413) | 460 (393 to 516)* | 227 (125 to 375) | 285 (219 to 369)* |
| Stage 4: adjusted for patient, provider and attitude characteristics | 383 (340 to 419) | 467 (397 to 522)* | 223 (122 to 351)* | 288 (222 to 368)* |
- * P < .05 for comparison to white patients with white providers.
- † P < .01 for comparison to white patients with white providers.
- ‡ P < .001 for comparison to white patients with white providers.
- Patient covariates are age, gender, education, annual family income, type of health insurance, self-reported access to care, primary HIV risk factor, recent heavy alcohol or drug use, CD4 count, symptom burden index, and geographic region. Provider covariates are gender, sexual orientation, years in practice, specialty/expertise, HIV knowledge, practice setting, belief that patients cannot adhere to antiretrovirals, preference not to treat injection drug users, and belief that antiretroviral therapy should be withheld or delayed for nonadherent patients. Attitude covariates are trust in provider, rating of medical care, and belief that antiretrovirals are worth taking.
Adjusting for all patient covariates, we found no difference in time to first protease inhibitor use between African-American patients with African-American providers and white patients with white providers (342 vs. 353 days; P > .20; Table 3 Stage 2 model). However, African-American patients with white providers received protease inhibitors later than white patients with white providers (461 vs. 353 days; P= .002). African-American patients with white providers still had significantly later use of protease inhibitors compared to other patients after additionally adjusting for provider characteristics, which were specialty training, HIV expertise, HIV knowledge, years in practice, gender, sexual preference, patient care mix, type of practice, preference not to treat injection drug users, and the belief that most of their patients can adequately adhere to their medication regimen (Table 3, Stage 3 model), as well as after adjusting for patient's ratings of their medical care, trust in their provider, and belief that antiretrovirals are worth taking (Table 3, Stage 4 model). These results were similar in a sensitivity analysis excluding nurse practitioners/physician assistants and their patients.
DISCUSSION
Racial disparities in a variety of HIV treatments and services have been widely documented.11,28 Disparities in the delivery of potent antiretrovirals is particularly important, because these treatments have a profound impact on AIDS mortality, which is disproportionately greater among African Americans.29 Previous studies have not been able to fully explain why racial disparities in antiretroviral medication use exists.8,10–13 However, these studies have primarily focused on patient characteristics, such as income and insurance status, as explanatory factors and have not examined the impact of racial concordance. Others studies have found that racial concordance between patient and physician influences satisfaction and use,14,15,17,30 but have not found differences in treatment for health conditions.31 In the present study, we found that African-American patients with white providers received protease inhibitors later than white patients with white providers and African-American patients with African-American providers. Although white patients with African-American providers appeared to have the shortest time to protease inhibitor use, this group represents such a small fraction of the patients (<0.1%) that we recommend great caution should be exercised in reaching any conclusions about them.
A number of possible explanations for our findings should be considered. Differences in trust and satisfaction may influence a patient's willingness to accept new therapies. We controlled for trust in their medical provider using a 2-item measure and patient ratings of their medical care. These factors did not appear to explain our findings, however. Still, there may be other characteristics of the patient-provider relationship, including aspects of trust not accounted for in our 2-item scale, that might explain the study results. For example, communication may be better in racially concordant doctor-patient relationships, which may improve patients’ acceptance of and readiness to take antiretroviral treatment. Recent studies found that African-American patients in racially concordant relationships reported more participation in decision making, as well as longer doctor visits and greater satisfaction, than those in racially discordant relationships.32
Our findings might also result from differences in physician prescribing behavior, which may be unintentional or possibly represent overt racial discrimination. Prior research has demonstrated that physicians are susceptible to having prior biases and stereotypes in assessing African-American intelligence, likelihood of substance abuse, and ability to adhere to regimens, which ultimately may affect their clinical decisions about African Americans.33,34 For example, most physicians delay treatment for patients whom they believe will be nonadherent.35 African-American patients in a white physician's panel may be more likely to be perceived as nonadherent. Ultimately, physicians might then be more likely to delay treatment for African Americans than for whites. We asked providers whether they considered patient adherence in their decision to prescribe protease inhibitors. This factor, however, did not explain our findings.
We report both statistical and clinical limitations of our study. First, we relied on self-reported dates of antiretroviral medications use. The wrong starting date could lead to either an underestimation or overestimation of the time to first protease inhibitor. Self-report of antiretroviral use is reasonably accurate when compared with pharmacy records,36 but the accuracy of dates of use is not known. There is also little reason to believe that errors in self-report would occur differentially by patient and provider race and result in a biased association. Second, we cannot exclude the possibility of selection bias resulting in some unmeasured differences between patients of the same race who are cared for by African-American versus white providers. African-American patients who have specifically chosen to see African-American providers may be more assertive in seeking medical treatments, may be more receptive to new treatments, or have other characteristics that could improve their access to protease inhibitors. Third, prior research analyzing the impact of patient trust in their provider on patient-provider relationships used more extensive scales to measure trust that the 2-item measure that we used. Thus, we may not have completely controlled for patient's trust in their provider. Finally, unmeasured differences between providers who mainly care for white and African-American patients may also exist. Although we controlled for HIV expertise, physician specialty, patient panel characteristics, and practice location, there may be unmeasured regional or philosophical differences in the standard of care that would affect prescribing patterns. In addition, we controlled for patients’ belief that antiretrovirals are effective and found this variable had little effect on our results. Finally, there were not enough patients of both races cared for by the same physician to conduct within-physician analysis of differences in treatment.
Although the median delay in time translates to greater than 3 months, the clinical implications are not straightforward. Because of side effects37 and the concern about the development of drug resistance, HIV treatment strategy has moved away from early antiretroviral use and toward a “wait and see” approach when initiating treatment.38,39 Thus, the delay in treatment among African-American patients may have actually benefited them. Still, when respondents were enrolled in our study in 1996, the prevailing opinion among HIV experts was that protease inhibitors are crucial in reducing morbidity and mortality and must be included in the initial regimens of patients who could tolerate them.40 During the time of data collection, treatment delay was documented to result in lower CD4 counts and higher mortality rates.41,42 Thus, regardless of the clinical implications, our study indicates that race concordance influences the delivery of state-of-the-art care.
As for the policy implications of our study, one might conclude that segregating patients to providers based upon race may help eliminate health care disparities. We believe, however, that involuntary racial segregation of patients is inappropriate and unethical. Alternatively, we should strive to better understand and improve the relationship between patients and their physician, paying particular attention to the effects of race concordance. Future research should examine how and why race concordance influences satisfaction and care and whether discrimination in physician behavior plays a role. In addition, our study has important policy implications as we consider how to address the paucity of African-American providers available to African-American patients.43 Patients often select providers based upon specific characteristics, including race and ethnicity. African-American patients should have access to a provider of their own race and ethnicity if they so choose. Unfortunately, African-Americans are less than 5% of the total number of physicians, and recent anti-affirmative action judicial and legislative decisions have negatively impacted African-American medical school admissions.44–46
Therefore, policy changes to increase the supply of African-American physicians and meet patient demand are imperative. These changes are not likely to occur very soon. In the meantime, medical educators should expand cultural competence training for physicians, as outlined by the Institute of Medicine Report and the American and National Medical Associations. Increasing the number of African-American physicians and improving the patient-provider relationship are potentially critical to eliminating disparities in care. Only when African Americans achieve equity in health care can racial health disparities be eliminated.
Acknowledgments
Support for this study was provided by a grant to Dr. Wong from the Universitywide AIDS Research program #ID02-LA-031. Dr. Cunningham also received partial support from the UCLA-Drew Project Export, National Institutes of Health, National Center on Minority Health and Health Disparities (P20-MD00148-01), and the UCLA Center for Health Improvement in Minority Elders/Resource Centers for Minority Aging Research, National Institutes of Health, National Institute of Aging (AG-02-004).
Dr. King was partially supported by the UCLA Robert Wood Johnson Clinical Scholars Program and the Minority AIDS Training Program of the Adult AIDS Clinical Trial Group. The HIV Cost and Services Utilization Study was conducted under cooperative agreement U-01HS08578 (M.F. Shapiro, PI; S.A. Bozzette, Co-PI) between RAND and the Agency for Health Research and Quality. Substantial additional funding for this cooperative agreement was provided by the Health Resources and Services Administration, the National Institute of Mental Health, the National Institute on Drug Abuse, and the National Institutes of Health Office of Research on Minority Health through the National Institute of Dental Research. Additional support was provided by The Robert Wood Johnson Foundation, Merck and Company, Inc., Glaxo-Wellcome, Inc., the National Institute on Aging, the Office of the Assistant Secretary for Planning and Evaluation in the U.S. Department of Health and Human Services, and the Mary and Irving Lazar Program in Health Services Research at UCLA.
Appendix
