Effect of Race on Insurance Coverage and Health Service Use for HIV-Infected Gay Men

Objective: To determine whether race is associated with health insurance coverage and health service use among gay and bisexual men in the Baltimore center of the Multicenter AIDS Cohort Study.

Methods: Data from eight semiannual study visits between 1991 and 1996 were used. Descriptive, stratified, and logistic regression analyses were conducted to determine whether race is associated with insurance coverage, medical, or dental service use, after controlling for socioeconomic variables.

Results: No difference was found between blacks' and whites' likelihood of having health insurance, private insurance, using inpatient, emergency department services, or antiretroviral medications. Whites were more likely to use outpatient services, particularly if CD4 cell counts were high, and were more likely to use dental services, although blacks were more likely to have dental insurance.

Conclusions: Further research must be conducted to examine cultural, social, and psychological factors that help explain why white gay men use more outpatient and dental services, when other service use is unrelated to race. Investigators should be precise when using race as a variable in health services and epidemiologic research, emphasizing when racial differences truly exist versus when the variable race is a surrogate for another factor.

Race frequently is used as an analytic predictor in epidemiologic and health services research ^(1,2). Many studies have found race to be associated with differences in clinical outcome, access to care, or insurance coverage. These studies, however, often present results as if race per se is responsible for the reported differences, and little discussion is provided as to whether a different factor actually is responsible for the inferred differences among races.

For example, a publication examined the relation between race and health service use among persons infected with HIV-1 ⁽³⁾. The authors reported that whites had more than six more outpatient visits per year than people of color. Controlling for risk factors for HIV, and looking only at persons with CD4 counts <50 cells/mm³, they found a 20% greater rate of hospital admissions for people of color. However, in their report, analyses were not adjusted for other variables that might be relevant to health service use, such as health insurance status or socioeconomic background. Thus, one is left to conclude that the authors believe that race itself is responsible for these differences.

The Study to Help the AIDS Research Effort (SHARE) is the Baltimore center of the Multicenter AIDS Cohort Study (MACS), a study of the natural history of HIV infection among gay and bisexual men. Compared with other samples of HIV-infected men, SHARE is a relatively homogenous cohort in terms of socioeconomic background and risk exposure. Most SHARE participants are highly educated, highly insured, and earn medium to high incomes. Given this socioeconomic homogeneity, we hypothesized that racial differences in health insurance coverage and health service use reported by other researchers would be undetectable in the SHARE cohort. If this were true, it would suggest that factors other than race explain whether people have health insurance and whether they use medical and dental services.

METHODS

Population

The MACS is a prospective study of the natural history of HIV-1 infection in homosexual and bisexual men begun in 1984. The organization and rationale of the MACS have been described previously ⁽⁴⁾. MACS data are collected at four centers throughout the country: Baltimore/Washington, DC, Chicago, Pittsburgh, and Los Angeles. This analysis includes participants from only the Baltimore Center, known as SHARE. Pittsburgh and Los Angeles sites had insufficient numbers of African-American participants for valid analyses, and the Chicago sample was recruited predominantly through attendance at a health clinic, which was thought to confound analyses related to health service use. The Baltimore site was recruited through community outreach. Of SHARE participants, 29% were seropositive for HIV-1 at the time of enrollment.

Data Collection and Definition of Variables

SHARE visits are scheduled every 6 months. At each study visit, laboratory work, a physical examination, self-administered questionnaires, and a structured interview are completed. In total, 856 men attended at least one study visit during the period from April 1991 to March 1996. Data for this analysis were taken from the most recent visit that each man attended during that 5-year period. Of these men, 32% attended a study visit in the last year, and 81% had attended a study visit in the last 2 years.

Serostatus

HIV-1 serostatus was determined by a repeatedly positive enzymelinked immunosorbent assay (ELISA) and confirmed with Western blot testing. AIDS diagnoses, as defined by the Centers for Disease Control and Prevention's 1987 criteria ⁽⁵⁾, were ascertained initially by self-report and confirmed by review of medical records. Because a significant association was found between race and serostatus in this cohort, and because serostatus is correlated with health service use, analysis was restricted to men who were HIV-infected.

Clinical Symptoms

Participants were asked in the interview whether they had experienced any of the following symptoms during the previous 6 months: persistent (≥2 weeks) thrush, diarrhea, fatigue, fever, or unexplained weight loss of ≥10 pounds. Men who reported having had two or more symptoms were considered symptomatic for the purpose of this analysis.

CD4 Count

CD4 count was measured at each semiannual visit ⁽⁶⁾. For this analysis, the CD4 count was grouped into three categories: >500 cells/mm³, 201 to 500 cells/mm³, and <200 cells/mm³.

Race

Participants were asked one question about racial background on the self-administered questionnaire during their first study visit. The question asked them to classify themselves as white, non-Hispanic; white, Hispanic; black, non-Hispanic; black, Hispanic; American Indian or Alaskan Native; Asian or Pacific Islander; or other ⁽⁷⁾. Respondents who reported being white, non-Hispanic or black, non-Hispanic were included in this analysis.

Health Service Use

Information concerning health service use was obtained from the interview. Participants were asked whether, during the last 6 months, they had had an outpatient medical visit, and, if so, how many times; whether they had been hospitalized; whether they had used antiretroviral therapy (ARV); whether they had been to an emergency department, and whether they had been to a dentist. Number of outpatient visits was collapsed into three categories: 0, 1, or ≥2 visits. All other health service visits were classified as "yes/no."

Insurance Coverage

Information concerning health insurance coverage was obtained from the interview. Participants were asked in a series of "yes/no" questions whether they had any of the following types of coverage: health maintenance organization (HMO), private insurance through a group, individual private insurance, Medicaid, Medicare, Veterans' Affairs (VA), CHAMPUS/CHAMPVA, or other. For the purpose of this analysis, if a participant said that he had any of the first three types of insurance, he was classified as having private insurance. If he did not have private insurance, but he did have Medicaid, Medicare, VA, or CHAMPUS/CHAMPVA, he was classified as having public insurance. Participants who responded "other" were reclassified where possible. Participants who answered "no" to all types of coverage were classified as having no health insurance. For certain analyses, a variable was created called "any insurance," combining those with private or public insurance and those who answered "other." Participants were asked whether they had dental insurance. This variable was classified as "yes/no."

Income

In the interview, participants were asked whether their annual income (in U.S. dollars) was <$10,000; $10,000 to $19,999; $20,000 to $29,999; $30,000 to $39,999; $40,000 to $49,999; or ≥$50,000. For this analysis, annual income was collapsed into four categories: <$10,000; $10,000 to $29,999; $30,000 to $49,999; and ≥$50,000.

Education

Participants were asked in the self-administered questionnaire during their first study visit about the highest level of education they completed and were not asked again. For this analysis, education was collapsed into three categories: up to but no more than high school diploma, some college or college degree, some graduate training or graduate degree.

Employment

At each study visit, participants are asked on the self-administered questionnaire whether their employment status was full time, part time, retired, student, unemployed, or other. For this analysis, employment was categorized as full time or not full time.

Age

At the first study visit, participants were asked their dates of birth. In regression analyses, age was defined in terms of years (plus or minus) different from 40.

Statistical Analysis

Descriptive statistics for all variables of interest were calculated. Bivariate analyses were conducted to compare white with black participants on all other variables of interest. These analyses were stratified by CD4 counts where appropriate. Finally, multiple logistic regression models were developed to determine whether race predicted health insurance coverage or health service use, when adjusting for education, income, employment, age, insurance status, serostatus, clinical symptoms, and CD4 count.

RESULTS

Description of Sample

Included in this sample were 307 HIV-infected men (Table 1). Of these men, 38% had AIDS at the time of their study visit; 78% of the men were white; 51% had some college and 38% had some graduate education; 27% earned >$50,000/year; 84% had private insurance; 59% were employed full time, and the mean age of participants was 44.1 years.

Table 1

Bivariate Analyses

Table 1 shows the distributions by race of other descriptive variables. Unadjusted for other variables, white participants are older, have more formal education, higher incomes, are less likely to be employed full time, equally likely to have private insurance, more likely to have public insurance, and more likely to have AIDS than black participants. Both black and white participants in this study are more highly educated, better paid, and more likely to be insured than Americans on average ⁽⁸⁾.

Table 2 shows health service use by race. Unadjusted rates of outpatient services differed between black and white participants, with 19% of black men having not seen a doctor in the last 6 months, compared with only 7% of white men. Unadjusted rates of inpatient, emergency department services and ARV use were not different for black and white participants. Unadjusted rates of dental services showed that white men were significantly more likely than black men to have seen a dentist in the last 6 months (p = .013).

Table 2

To examine further the use of outpatient services by blacks and whites, outpatient service use was stratified by the CD4 level of the participant. Stratified analysis reveals that a statistically significant difference in whether blacks and whites had visited a doctor existed only among those men with CD4 counts >500 cells/mm³ (p < .01). Among men with high CD4 counts, 89% of whites had had an outpatient visit in the last 6 months compared with only 56% of the blacks. No difference in likelihood of having had an outpatient visit was found among men with CD4 counts <500 cells/mm³.

Multivariate Analyses

Table 3 shows results of multivariate analyses to model statistics related to those respondents with health and dental insurance. Because clinical symptoms, AIDS status, and CD4 count were so highly correlated, these and other models are shown using only CD4 counts. In addition, age was deleted from all models due to non-significance. Controlling for CD4 count, education, income, and employment, black participants are just as likely as white participants to have health insurance generally and to have private rather than public insurance. Only income and full time employment predicted having private and/or any health insurance. Black men were more than twice as likely to have dental insurance than white men.

Table 3

Table 4 shows results of multivariate analyses to model inpatient, emergency department, ARV use, and dental visits. As shown in previous MACS analyses of health service ⁽⁹⁾ and medication use ^(10,11), race is not a significant predictor of inpatient, emergency department, or ARV use. Rather, employment and CD4 count are the important predictors. Dental visits do differ by race, however, with white men being almost three times as likely to have had a dental visit in the last 6 months as black men. Table 5 shows outpatient use among all HIV-infected men, and then stratified by CD4. Although a significant difference by race appears in outpatient use overall, with whites three times more likely to have had an outpatient visit than blacks, stratified analyses reveal that this difference is the effect of the men with CD4 counts >500 cells/mm³. Confirming the results of the bivariate stratified analyses, no difference existed in outpatient use between black and white men with CD4 counts <500 cells/mm³. However, among men with CD4 counts >500 cells/mm³, white men are 10 times more likely to have seen a doctor in the last 6 months as black men. After men have had at least one outpatient visit, however, no difference by race is found in whether they have ≥2 visits (data not shown).

Table 4

Table 5

DISCUSSION

It was the hypothesis of this study that, controlling for socioeconomic factors, race would not be relevant to whether HIV-infected gay and bisexual men had health insurance or to their pattern of health service use. Indeed, we found that race was not associated with being insured, nor was it related to whether the insurance held by these men was private or public. What instead predicted having public insurance, not surprisingly, was being unemployed and having little income. Also unrelated to race was whether men had been hospitalized, had been to an emergency department, or had received ARV medications. Men most likely to use these services were not employed, did not have private health insurance, and/or had low CD4 counts.

What was surprising to us, among these well-educated, professional, and well-insured men, was how large the estimated effect due to race was for modeling outpatient health services. Controlling for all socioeconomic and health factors available to us, black men were significantly less likely than white men to go to a doctor, particularly when their CD4 counts were high, and were significantly less likely to go to a dentist. Although outpatient visits were associated with having lower CD4 counts and having higher incomes (as one might expect), even controlling for these factors, HIV-infected white men remained three times more likely to have had an outpatient visit in the 6 months before their study visit than HIV-infected black men. Among those with CD4 counts >500 cells/mm³, white HIV-infected men were 10 times more likely to have seen a doctor in the previous 6 months. Similarly, controlling for other factors, HIV-infected black men were significantly less likely to have seen a dentist in the last 6 months than white men. Whereas 62% of white men had seen a dentist in the last 6 months, only 45% of the black men had. This is particularly remarkable given that the black men were significantly more likely to have dental insurance.

We do not know why HIV-infected white men were so much more likely to have used outpatient and dental services than HIV-infected black men. Limited finances were unlikely to have impeded access to health care for men of either race, so we can only conclude that environmental, social, and/or psychological factors help explain their different use of these services. In terms of outpatient medical visits, we cannot know from these data whether white and black men might have perceived the severity of their illness differently, causing white persons to believe they needed more care than black persons. The data cannot tell us whether white men sought care too often or black men sought care too rarely. Alternatively, it may be that the HIV-infected white men are more integrated into existing gay community networks in which seeking HIV services is a norm. To the extent to which black HIV-infected men try to hide their infection from peers, they may not seek services as readily at early stages of infection.

Black men also may choose to forego what may be thought of as discretionary services because of a lesser trust in biomedical institutions or because they feel less welcome by the health care system. The Tuskegee syphilis study, in which poor, black men with syphilis were enrolled by the Public Health Service, not told they were in research, and denied treatment, has been implicated as a factor in why many African Americans with chronic medical illnesses avoid seeking health services ^(12,13). Further evidence from a published study suggests that African-American people have less trust in biomedical research than whites ⁽¹⁴⁾, and many studies have shown that racism does exist in American medicine in terms of doctors treating their African-American and white patients differently ^(15-18). It also is possible that blacks are less inclined to seek health services, given the shortage of African-American providers. In 1995, just under 5% of all physicians and just under 2% of dentists in the United States were black ⁽¹⁹⁾.

In terms of dental care, other studies have shown that black children are more likely than white children to use emergency dental services and/or to miss days of school due to dental problems, implying that they are less likely to receive routine, preventive services ^(20,21). However, these studies did not control for income and insurance, so it is impossible to tell whether race itself was related to the lack of preventive care, or whether other factors might have been more relevant. Two other studies examined the relation between race and regular dental service use, controlling for socioeconomic factors ^(22,23). The first of these found, as did we, that white persons were more likely than nonwhites to receive regular, preventive dental care. The other study found not only that 36% of nonpoor blacks had seen a dentist in the last year compared with 73% of nonpoor whites but also that blacks had more negative attitudes about dental care. We can only speculate why well-educated, well-insured African-American men were less likely to seek preventive dental care. One possibility is that some of these men, as children, were of a lower socioeconomic group than they are as adults and thus established their dental habits at a time when they were unable to afford regular check-ups.

Our hypothesis that blacks and whites would have similar insurance coverage and health service use, when adjusting for other variables, was mostly proven correct. And although we are far from the first to suggest that socioeconomic status is related to health ⁽²⁴⁾, that differentials in health status by class are much larger than differences by race ⁽²⁵⁾, or that researchers should include measures of class and social determinants of health ⁽²⁶⁾ in their studies, many research questions still are posed primarily as questions of race, with only secondary or no attention to other factors, and little attention to what a difference by race means.

We assume that most investigators who describe differences in insurance coverage or health service use by race are not concluding that biologic differences exist in these outcomes. Rather, they probably assume that race is a proxy for other factors, such as income or education, that more directly explain an observed difference. Race should not be used in lieu of a true predictor, however. Race and class, although often correlated, do not measure the same thing, and we must understand their separate influences on health and other outcomes. In some situations, race is the true predictor of differences in health status or health service use and is not serving as a proxy. Thomas LaVeist has written that race is a very poor indicator of biology, a somewhat better indicator of ethnicity (culturally determined health and illness behavior) and a very good indicator of degree of exposure to social factors such as racism ⁽²⁷⁾.

Researchers carry more responsibility than we often recognize. The press picks up scientific reports, and the public soon learns of another difference between blacks and whites. Not being careful with our distinctions can result in public health and social harms as we draw conclusions about populations that are incorrect and make medical or social judgments based on false information.

Our findings suggest that the more discretionary health services are less likely to be sought by African-American than white, mostly middle class, gay men. Future qualitative and quantitative research should examine the factors related to health service use among men who financially are able to access services to determine why race so systematically remains relevant to how outpatient medical and dental services are used. And in the mean time, we as researchers should continue to scrutinize how we include race as a study variable ⁽²⁷⁾. This is particularly true in light of findings from this study that remind us that socioeconomic and health status variables, rather than race, are primarily responsible for differences in health service outcomes.

APPENDIX

The Multicenter AIDS Cohort Study (MACS) includes the following: Baltimore: The Johns Hopkins University School of Hygiene and Public Health: Joseph B. Margolick, Principal Investigator; Haroutune Armenian, Homayoon Farzadegan, Nancy Kass, Justin McArthur, Ellen Taylor. Chicago: Howard Brown Health Center and Northwestern University Medical School: John P. Phair, Principal Investigator; Joan S. Chmiel, Bruce Cohen, Maurice O'Gorman, Daina Variakojis, Jerry Wesch, Steven M. Wolinsky. Los Angeles: University of California, U.C.L.A. Schools of Public Health and Medicine: Roger Detels, Principal Investigator; Barbara R. Visscher, Janice P. Dudley, John L. Fahey, Janis V. Giorgi, Oto Martínez-Maza, Eric N. Miller, Hal Morgenstern, Parunag Nishanian, John Oishi, Jeremy Taylor, Harry Vinters. Pittsburgh: University of Pittsburgh Graduate School of Public Health: Charles R. Rinaldo, Principal Investigator; James T. Becker, Phalguni Gupta, Lawrence Kingsley, John Mellors, Sharon Riddler, Anthony Silvestri. Data Coordinating Center: The Johns Hopkins University School of Hygiene and Public Health: Alvaro Muñoz, Principal Investigator; Cheryl Enger, Stephen Gange, Lisa P. Jacobson, Cynthia Kleeberger, Robert Lyles, Steven Piantadosi, Sol Su. National Institutes of Health: National Institute of Allergy and Infectious Diseases: Lewis Schrager, Project Officer; National Cancer Institute: Sandra Melnick.

Acknowledgments: Data in this manuscript were collected by the Multicenter AIDS Cohort Study (MACS) with centers (Principal Investigators) at The Johns Hopkins School of Public Health (Joseph B. Margolick, Alvaro Muñoz); Howard Brown Health Center and Northwestern University Medical School (John Phair); University of California, Los Angeles (Roger Detels, Janis V. Giorgi); and University of Pittsburgh (Charles Rinaldo). The MACS is funded by the National Institute of Allergy and Infectious Diseases, with additional supplemental funding from the National Cancer Institute. UO1-A1-35042, 5-MO1-RR-00722 (GCRC), UO1-A1-35043, UO1-A1-37984; UO1-A1-35039; UO1-A1-35040, UO1-A1-37613; UO1-A1-35041. This work was also supported in part through an interagency agreement with the Agency for Health Care Policy and Research.