Introduction

Throughout the late 1980s and into the mid-90s, human immunodeficiency virus (HIV) held steady as one of the top 15 leading causes of death in the United States of America (USA), claiming a high of 50,628 American lives in 1995[1–4]. Following the introduction of effective combination antiretroviral therapy (ART) in 1996, HIV diagnoses and mortality began rapidly decreasing and eventually leveled off around 1998. Consequently, HIV infection is no longer a major leading cause of death in the USA [1].

Despite this notable progress, racial disparities are prominent. For example, in the USA, the rates of diagnoses and prevalence of HIV among non-Hispanic Blacks (Blacks) are 9 and 7 times higher, respectively, than those of non-Hispanic Whites (Whites) [5]. Among men, the magnitude of the Black:White disparities in diagnoses and prevalence is 8 and 6, respectively [5]. The differences are larger among women at 20 and 18, respectively [5]. The overall trend has been a decrease in disparities in diagnoses over time, with some populations increasing in diagnoses, namely young black men [6]. Similarly, HIV is the 25th leading cause of death among Whites but is ranked 13th among Blacks [7].

These numbers suggest that the devastating effects of HIV are being disproportionately experienced by Blacks and that racial and gender disparities in HIV mortality, which were present early on, have persisted—and possibly even grown—well into the 2000s [8, 9]. Given the documented declines in HIV diagnoses, as well as the advancements in treatment that have occurred since the mid-90s, it would be instructive to examine mortality rates and disparities therein to gain a better understanding of how these patterns have changed over time and what might underlie these changes [10–14]. Thus, the purpose of this analysis is to examine the race- and gender-specific HIV mortality rates and corresponding rate ratios for Blacks and Whites in the USA for the four 5-year increments from 1990–2009.

Methods

Deaths where the underlying cause was human immunodeficiency virus infection (ICD-9 = 042–044) and human immunodeficiency virus disease (ICD-10 = B20-B24) were included in this analysis. Numerator data for 1990–2009 were abstracted from death files obtained from the National Center for Health Statistics [15]. Population-based denominators for the non-Hispanic White (White) and non-Hispanic Black (Black) populations were obtained from the US Census Bureau for 1990, 2000, and 2010. Population-based denominators for years other than 1990, 2000, and 2010 were estimated using linear interpolation as the intercensal population estimates do not provide the level of detail necessary for this analysis. For each of the two data sources, Black and White classifications were obtained by cross-tabulating two demographic variables: Hispanic ethnicity and racial identity. For the census, these variables are self-reported; for the death files, these variables are reported by the next of kin. Age-adjusted rates per 100,000 population were calculated employing the year 2000 standard US population [16].

Five-year average annual rates were calculated for 1990–1994 (T1), 1995–1999 (T2), 2000–2004 (T3), and 2005–2009 (T4) (2009 is the most recent year for which city-level data were available). Because 1990–1998 mortality data employ the International Classification of Diseases, Ninth Revision (ICD-9) codes while 1999–2009 data employ the ICD, Tenth Revision (ICD-10) codes, we used an age-specific comparability ratio formula designed for transforming age-adjusted rates from the ICD-9 to ICD-10 coding so that the rates were comparable across the four time points [17–19]. Data are provided for the total population, as well as, for males and females.

Statistical Analysis

The disparity in HIV mortality was measured by calculating the Black:White rate ratio (RR) for each population at each time period. An RR was considered statistically significant if the 95 % confidence interval did not contain 1.00. Confidence intervals for the RRs were calculated using a Taylor series expansion technique [20]. All data were analyzed using SAS v 9.2 [21].

Excess deaths due to the racial disparity in HIV mortality were calculated by multiplying the age-specific White mortality rates by the corresponding Black populations in each age category. The sum of these products is the number of Black deaths that would be expected if White death rates were applied to this population. We then subtracted the number of expected deaths from the number of observed deaths to obtain the excess number of deaths for each population group [22].

Results

Figure 1 displays the age-adjusted HIV mortality rates for the Black and White male and female populations in the USA for the intervals 1990–1994 (T1), 1995–1999 (T2), 2000–2004 (T3), and 2005–2009 (T4). At each of the four time points examined, the rates are higher among men than women and among Blacks than Whites. Although the rates decreased for all groups over the study period, the Black male rate remained substantially higher than the rates for all other groups. Additionally, the Black female rate was higher than the White male rate at T2-T4. White females demonstrated the lowest rates at all time points examined.

Fig. 1
figure 1

Age-adjusted HIV mortality rates by race and gender: USA, 1990–2009

Full size image

Table 1 presents age-adjusted HIV mortality rates for the Black and White total, male, and female populations in the USA for T1-T4, and the corresponding RRs and confidence intervals. For the total, male, and female populations, we observed a statistically significant increase in the Black:White RR between T1 and T4. The total population RR in the USA was 4.33 at T1, and it grew steadily to 11.38 by T4. Within the male population, the RR was 3.91 at T1, and it more than doubled by T4 to 9.66. For females, the RR was 14.00 at T1, and it grew by about 1.5 times to 21.22 by T4. Bold-faced rate ratios indicate statistical significance. At all time points examined, the RRs for the total, male, and female populations were statistically significant. Taken together, these observations demonstrate that the Black:White HIV mortality disparity in the USA worsened over this 20-year period.

Table 1 Five-year average annual HIV mortality rates and rate ratios for non-Hispanic Black (Black) and non-Hispanic White (White) total population, men, and women in the USA, 1990–2009
Full size table

A comparison of the male and female populations reveals that, while the rates of HIV mortality are higher among men than women for both races at each time point, the racial disparities are significantly higher within the female than the male population (Table 1). For example, the T1 female RR (14.00) is about 3.5 times larger than the T1 male RR (3.91). By T4, the difference between the male (9.66) and female (21.22) RRs had narrowed but was still over twice as large among the female population. This is largely attributable to the substantially lower rates observed among White women.

Table 2 displays the within race T1 to T4 percent decrease in HIV mortality rates for Blacks and Whites. For the total, male, and female populations, we observed that the decrease in mortality rates from T1 to T4 was greater among Whites than Blacks. For example, while the White rate decreased by 83 % between T1 and T4, the Black rate decreased by just 56 % over the same period. For the male population, the White rate decreased by 85 %, while the Black rate decreased by 64 %. Among females, Whites saw a 53 % decrease, while Blacks experienced only a 29 % decrease.

Table 2 Percent decrease in T1 (1990–1994) to T4 (2005–2009) Non-Hispanic Black (Black) and Non-Hispanic White (White) HIV mortality rates for the USA
Full size table

We also calculated the number of excess Black deaths due to the racial disparity in HIV mortality (Table 2). Excess deaths are a function of the size of the population of each area and the size of the disparity. The annual number of Black excess deaths in the USA at T4 was 5603. Among men and women in the USA, there were 3580 and 2050 annual excess Black deaths, respectively.

Discussion

Despite decreasing mortality rates for both Blacks and Whites in the USA, the mortality disparity between Blacks and Whites increased significantly between 1990–1994 (T1) and 2005–2009 (T4) for the total, male, and female populations, with the largest disparity occurring among women. Among the total population, the disparity was 11.38, indicating that Blacks were more than 11 times more likely to die from HIV compared to Whites at T4. Among men and women, Blacks were 9.66 and 21.22 times more likely to die from HIV at T4, respectively (Table 1). The Black:White HIV mortality disparity was 2.6 times higher at T4 compared to T1, 2.5 times higher among men, and 1.5 times higher among women. The rate at which mortality decreased over time varied by both race and sex, with White rates decreasing at a faster pace than Black rates regardless of gender, and Black women showing the slowest decline (Table 2). Notably, at every time point except T1, the death rates for Black women were higher than those for White men (Table 1). Finally, the Black:White HIV mortality disparity resulted in 5603 excess Black deaths annually in the USA. These data demonstrate an indisputable White advantage with respect to HIV mortality, which has been occurring since at least 1990 [23, 9, 24, 25, 13].

The HIV Care Continuum (treatment cascade) outlines the steps a person living with HIV should go through in order to achieve viral suppression and thus reduce or eliminate the risk of death from HIV. The steps include the following: diagnosis of HIV infection, linkage to care, retention in care, receipt of antiretroviral therapy, and achievement of viral suppression [26]. The population engaged at each step has been documented, and areas for improvement have been identified. However, racial differences in engagement at various stages in the Care Continuum, which are less systematically documented, likely contribute to the observed disparities in HIV mortality. Furthermore, access to care is imperative to one’s ability to follow the Care Continuum, and any differences in access to care will thus impact individual success in achieving viral suppression.

The first step in being diagnosed with HIV is being tested for HIV. However, there have been notable differences in testing among Blacks versus Whites. One reason for this may be that early prevention and testing messaging focused mainly on White risk groups (e.g., White gay men) as opposed to risky behavior (e.g., unprotected sex) and the initial classification of AIDS as gay-related immunodeficiency disease (GRID) resulted in a stigma related to HIV/AIDS and an exclusion of Blacks, women, and heterosexuals from initial messaging [27]. This highly targeted messaging may have resulted in a lack of perceived risk among certain segments of the population, impacting their motivation to prevent the transmission of or be tested for HIV [27]. Indeed, since 2007, more black men have been infected with HIV compared to white men; however, an often overlooked population has been women of color, and most of whom report a risk category of heterosexual contact with fewer reporting injection drug use [28, 12, 5]. Thus, historical differences in perceived risk and testing behaviors may have contributed to observed racial disparities in HIV mortality.

Early detection of HIV infection coupled with timely treatment are of great importance for reducing mortality [29]. Routine testing improves the likelihood of identifying HIV at this early stage, reducing transmission and decreasing the risk for progression to AIDS and/or death [29, 10, 11, 30, 12]. Unfortunately, it was not until 2006 that the Centers for Disease Control and Prevention (CDC) released recommendations for routine screening regardless of risk in medical settings with an HIV prevalence greater than 0.1 % [31]. These issues as well as the fact that some groups have historically been less able to access health care regularly have likely contributed to differences in who has been screened, generally, and who has done so routinely, specifically [32]. The CDC estimates that approximately 14 % of persons in the USA are unaware of their infection and, as of 2009, a majority of these undiagnosed infections were among people of color, predominantly young and male, and are often identified at late stages of the disease when they are identified [1, 10, 33–38]. Given that treatment is contingent upon knowledge of infection, it is not hard to see how these differences would contribute directly to disparities in mortality.

For those persons who are diagnosed with HIV, the next step in the Care Continuum is linkage to care. However, there are documented disparities in the percent linked to, as well as engaged or retained in care, favoring Whites [39]. These differences will undoubtedly impact the receipt of prescribed antiretroviral therapy, the next stage in the Care Continuum.

Treatment is arguably the most important avenue through which HIV mortality can be reduced. Combination antiretroviral medications (ART) have allowed infected persons to live healthy lives by improving immune function and enabling the body to defend itself against opportunistic infections or an AIDS diagnosis [10, 40, 41]. Sharp declines in mortality associated with the introduction of ART demonstrate its effectiveness in treating HIV and preventing death [10, 11]. However, the present analysis, along with previous research, has shown that the declines in mortality have not been shared equally by all race groups in the USA.

Inherent in the notion of a continuum is the idea that differences present at any given stage will directly impact differences in the stages that follow. In order to be prescribed ART, one must be linked to care and in order to achieve viral suppression, one must be adherent to ART. However, racial differences are observed at each of these stages with a higher percentage of Whites being prescribed ART (among those retained in care) and achieving viral suppression (among those prescribed ART) [39]. Some portion of the racial difference in the percent prescribed ART may be accounted for by guidelines at the time which recommended ART based on disease severity [42]. And some portion of the racial difference in the percent achieving viral suppression may be accounted for by differences in either response or adherence to ART [9, 13, 23, 40, 43–50].

However, the crucial element to achieving viral suppression is gaining access to care, without which one cannot be diagnosed with HIV nor obtain ART. Thus, we argue that the racial disparity in HIV mortality is the result of differential access to health care. Despite the availability of effective treatment for HIV, the mortality rates declined much more slowly for Blacks compared to Whites (Table 2), suggesting differences in access to and/or timing of diagnostic testing and treatment favoring Whites.

This explanation aligns closely with the “Fundamental Cause Perspective”, which posits that as tests become more sophisticated at detecting early stages of diseases, and treatments become more effective at treating disease, those with resources and power are most likely to benefit and those of lower socioeconomic status (SES) and minorities benefit least [9, 51, 52].

Exploring this in the context of HIV mortality disparities, we infer that the disparities documented in the continuum measures in diagnosis, linkage to care, prescription to ART, and viral suppression among Blacks have contributed substantially to the widening disparities. Blacks in the USA have routinely faced barriers to receiving health care, and even when seeking care, perceived risk may impact the likelihood of being tested [32, 53]. Additionally, factors such as poor reimbursement for HIV tests, competing health needs, and possible provider biases or comfort level in providing positive results to certain patients have been cited as potential barriers to HIV testing by clinicians [54–56]. Thus, even when connected to health care, certain populations are less likely to be offered testing [57, 58].

The notion that Blacks experience unequal access to treatment is supported by studies demonstrating disparities in ART receipt by race (favoring Whites) and by insurance status (favoring the privately insured) [59–65]. Some studies have suggested that disease severity qualifiers or income restrictions for public insurance plans contribute to delays in care or mortality from HIV [11, 64]. These studies reveal that Blacks are more likely to experience delays in HIV care and less likely to receive ART at earlier stages, both indicators of disparities in HIV care quality, again, favoring Whites.

Rubin et al., studying the effect of ART on mortality from HIV/AIDS, found that once ART became available, disparities by race and SES began to develop, eventually displaying the same widening trend reported in the present study [9]. These findings provide additional support for the Fundamental Cause Perspective and align with reports on other diseases, including certain screen-detectable cancers [51, 52].

Limitations

Our analysis included deaths for which HIV was the underlying cause of death, but these could potentially be undercounted in cases where the individual completing the death certificate was unaware of the decedent’s HIV status. Our analysis did not include deaths for which HIV was any contributing cause of death. A recent report documenting changes in racial disparities in mortality (from any cause) among known HIV positives using the CDC National HIV Surveillance Database identified higher population-based HIV mortality rates among Blacks and Whites, yet noted similar levels of disparities that were demonstrated in our study [66].

Furthermore, as those infected with HIV live longer, they are more likely to die of a cause other than HIV, meaning that HIV may have become less likely over time to be the underlying cause of death [67, 68]. Additionally, there is some documentation suggesting a small amount of underreporting of HIV as a cause of death in minority populations which could impact the counts of deaths due to HIV among Blacks [69, 70].

Two factors prevented us from including data from the 1980s in our analysis. First, the ICD-9 codes for HIV were not introduced until 1987. Prior to this year, HIV deaths were coded as “human T cell lymphotropic virus-III/lymphadenopathy associated virus (HTLV-III/LAV) infection” [3]. Second, prior to the 1989 revision to the US Standard Certificate of Death, Hispanic origin was not included on the death certificate. Thus, one could not accurately count the deaths of non-Hispanic Blacks and non-Hispanic Whites prior to 1989. As a result, we were unable to look at the full history of HIV mortality disparities beginning in the early 1980s. However, the National Vital Statistics Report published in 1987 shows higher HIV mortality rates among Blacks than Whites, though we cannot determine the effect of including Hispanics in these rates and earlier reports suggest disparities as early as 1985 [3, 71].

Social determinants, which contribute to the health of individuals with HIV, and include mental health conditions, violence, gender inequalities, drug use, SES, HIV knowledge, discrimination, racism and stigma, are not included in death certificate data [72–74]. Clinical indicators, such as ART use and CD-4 count or viral load are also not available from the death certificate. We were therefore unable to include these types of data in our analysis.

Implications/Future Research

Access to care is imperative for early diagnosis and treatment of HIV. However, patient and health care system barriers are often cited as reasons for poor clinical outcomes among those with HIV [26, 72, 73, 75–77]. Incorporating navigation, such as removing barriers to receiving care, providing social support and appointment coordination, into HIV testing programs is effective in both linking patients to HIV care and engaging patients in care [78–82]. Thus, incorporating navigation could both reduce mortality and likely prevent further transmission of HIV, although these effects have not been reported.

Recent initiatives have been developed with the overarching goal of reducing racial disparities in HIV outcomes. These programs, which aim to identify infected persons and link them to care, should be expanded. For example, the CDC funds programs for expanded and routine HIV testing and linkage to care [83]. Additionally, under the Health Resources and Services Administration’s (HRSA) Ryan White program, the AIDS Drug Assistance Program (ADAP) provides life-saving ART to those who cannot afford it, saving lives and averting transmissions and outreach services for persons who have fallen out of care, and Medical Case Management to increase treatment adherence [84]. Continued evaluation of these programs and collaboration between CDC and HRSA will be key to determining their effectiveness and impact on HIV surveillance outcomes.

More research is needed to understand why HIV mortality rates remain so much higher among Black women compared to both White men and women. Literature on this topic is scant, though one study reported that even after controlling for education (a proxy for SES), Black women still had higher HIV mortality rates than White men starting in 1999.[23] As reported in the present study, Black women have experienced the slowest decline in HIV mortality and understanding why this is the case should help us determine how to address the high rates among this group.

Additionally, research is needed to understand the relationship between access to HIV treatment and subsequent mortality, again by race/ethnicity. Changes in clinical recommendations on when to begin ART should be examined as a possible contributor to disparities in HIV mortality as these may have differentially impacted the timing of ART uptake across race/ethnic groups.

Conclusion

Our results have implications for the HIV epidemic in the USA. The HIV mortality rates have declined for both Blacks and Whites in the USA over the last 20 years. However, the pace of the decline has been much steeper for Whites than Blacks, suggesting that existing disparities could be ameliorated if access to care were equal among these groups. Coupled with expanded HIV testing, the addition of patient navigators might be particularly effective in linking to and retaining in care those populations whose mortality rates have been slowest to decline.