Spatial and temporal analysis of HIV clinical outcomes in Florida reveals counties with persistent racial and ethnic disparities during 2012-2019

The Southern region within the United States (US) is considered an epicenter of the HIV epidemic, with the largest population of people with HIV (PWH) living in Florida. There are over 117,000 Floridians with HIV, with a disproportionate number of these individuals (45%) identifying as Black (27% White, 25% Hispanic/Latinx) [1]. The Ending the HIV Epidemic in the US (EHE) initiative aims to reduce HIV incidence by 90% by 2030 [2]. The EHE lists 57 priority jurisdictions across the US, seven of them are in Florida. A key pillar of the EHE is getting PWH to achieve and sustain viral suppression quickly. This has been made possible by the advent of combination antiretroviral therapy (ART) enables PWH to medically manage their infection and effectively eliminate the risk of transmission [3, 4]. However, in Florida it is estimated that less than two-thirds of PWH have achieved viral suppression [5]. Research is needed to identify actionable factors to improve HIV clinical outcomes in this high incidence and prevalence setting.

Racial/ethnic disparities throughout the HIV care continuum have been well documented in the US [6‐9], with especially striking inequalities in viral suppression rates between White and Black PWH [10]. For example, based on the National HIV Surveillance System, an estimated 36.1% of Black PWH aged 13-29 achieved sustained viral suppression, lower than the 46.7% observed in Hispanic/Latinx PWH and 50.8% in White PWH within the same age group [11]. Data from Florida indicate that it takes longer for Black PWH to achieve viral suppression than White and Hispanic PWH during the first several years after HIV diagnosis [12] . These disparities are partially explained by individual-level factors, but the remaining disparities could be attributed to socio-ecological factors. Factors including local healthcare and transportation infrastructure, the health behaviors of those around them, and the local economic context can influence health behaviors, including those that influence HIV health outcomes [13‐16]. In Florida, one cross-sectional analysis of the state’s enhanced HIV/AIDS reporting system (eHARS) in 2015 revealed that Black and Latinx PWH more likely than White PWH to have unsuppressed HIV viral load (47% among Black PWH, 37% among Latinx PWH, and 32% among White PWH) [13]. After individual and community factors were controlled, the disparity between Latinx and White PWH reduced precipitously to 7%, however, Black PWH was still 55% more likely not to have viral suppression [13]. Immune reconstitution on ART, defined as a restoration of CD4 T cell counts soon after initiation of ART [17], is another clinical outcome important for overall prognosis, yet few studies have investigated the presence of racial/ethnic disparities for this outcome. One study conducted among PWH in Texas reported that the rate of immune recovery was not significantly different by race/ethnicity [18] though more studies are needed to confirm this finding.

Additionally, geographic variation in HIV burden by race has been documented [19]. For example, an analysis using HIV mortality data from Florida and found that disparities between White and Black PWH widened over time before stabilizing [20]. However, that study did not examine spatial variation in HIV mortality by race. Spatial-temporal variations in racial disparities have been found in Florida for non-HIV disease burdens [21‐23]. For instance, greater racial disparities in hypertensive disorders of pregnancy burden have been found to occur in the North Central Florida counties throughout the past decade [23]. However, little is known about the spatial-temporal variation in racial/ethnic disparities in HIV clinical outcomes in Florida. Since much of the HIV response is conducted at the local level, a better understanding of HIV clinical outcomes by race across space and time could guide targeted intervention to improve HIV care outcomes and reduce racial/ethnic disparities.

As Florida seeks to achieve the EHE goals, it is important to assess the progress made in reaching HIV-related health targets and identify populations that are lagging behind and where these lags are occurring within the state. The objectives of this study were to 1) investigate longitudinal trends in clinical outcomes among PWH in Florida between 2012 and 2019, 2) examine the impact of socio-ecological factors on the association between race/ethnicity and HIV clinical outcomes, 3) explore spatial and temporal variations in HIV clinical outcome by race/ethnicity.

Data were available for 42,369 PWH in Florida between 2008 and 2020. Due to data quality concerns with reporting changes and incomplete data during the period prior to 2012 and after 2019, we restricted our subsequent analyses to this time frame. Most of the population was aged 45-54 (26.5%) or 55–64 years (28.5%), male (68.0%), non-Hispanic Black (52.2%), and heterosexual (39.7%) or men who have sex with men (42.2%) (Table 1).

The proportion of PWH with suppressed viral loads in this cohort rose from about 64% in 2012 to 84% in 2019. Similarly, the proportion of PWH who experienced immune reconstitution also rose from about 44% in 2012 to about 63% in 2019 (Fig. 1, Panel A). The increase over time was statistically significant for both outcomes (P<0.001). The levels of CD4 reconstitution by transmission risk group were overlapping throughout the study period in general, however, from 2015 and onward, we observed a lower proportion of individuals with immune reconstitution within the perinatal group compared to the other risk groups (data not shown). Although the levels of CD4 reconstitution improved significantly throughout the study period for all racial/ethnic and ethnic groups, the proportions were consistently higher among White PWH compared to Hispanic and Black PWH (Fig. 1, Panel B). Similarly, we observed significant improvement in viral suppression for all race/ethnicity groups over time, however, the proportions were consistently higher among White and Hispanic PWH compared to Black PWH (Fig. 1, Panel C).

Despite annual improvements in the proportion of Ryan White clients who achieve viral suppression and immune reconstitution, significant racial and ethnic disparities in reaching these milestones remained and with variations by county. These data can be used to better target populations and areas where additional services and support are needed to help achieve the goals set out in the EHE in an equitable manner.

The continued increases in rates of viral suppression found in this study are reflective of trends in the national population of PWH. Steady improvement in viral suppression rates has been reported in the national sample of PWH enrolled in the Ryan White Program, both overall and by race/ethnicity [24]. However, similar to our findings, non-Hispanic White PWH consistently have higher proportions of viral suppression compared to non-Hispanic Black PWH and Hispanic PWH. Samples including PWH who are not enrolled in Ryan White have yielded similar trends in viral suppression and disparities by race/ethnicity in the years after the introduction of integrase inhibitors, although those enrolled in Ryan White tend to have better outcomes than non-enrollees [25]. Individual-level beliefs, behaviors, and conditions can account for a portion of these disparities [26‐28]. For example, medical mistrust is more common among non-Hispanic Black PWH and non-Hispanic White PWH, which contributes to disparities in ART adherence and care engagement, key steps in the care continuum to reaching viral suppression [29‐31]. While the improvements in viral suppression are encouraging, more work is needed to achieve the ambitious targets set by the EHE and to reach these goals equitably.

In this study, racial and ethnic disparities in viral suppression and immune reconstitution remained after accounting for county-level socio-ecological factors. These findings are consistent with other studies in Florida that found disparities in viral suppression remained between White PWH and Black PWH after adjusting for neighborhood factors [13, 32]. A prior study reported several neighborhood-level factors were associated with viral suppression in Miami-Dade County among Hispanic PWH, but not White or Black PWH [32]. Interestingly, they found there were disparities within more socio-economically advantaged neighborhoods and neighborhoods with both low and high levels of racial and language homogeneity, a proxy for residential segregation. Residential segregation leads to different levels of health care access between predominately Black and predominately White communities and contributes to the disparities in several health outcomes including sexually transmitted infection rates, COVID-19 mortality, and cancer [33‐35]. Studies on the impact of segregation on HIV risk outside of Florida have yielded mixed results [36, 37]. Previous studies in Florida studies used ZIP code-level socio-ecological variables, as opposed to county-level variables. Counties are significantly larger geographically and are likely to have a much greater degree of diversity in circumstances [32]. However, results from studies with more granular data are congruous with our findings, supporting the validity of our results.

County-level analyses of disparities between Black and non-Black PWH were found in six counties for immune reconstitution and nine counties for viral suppression. Of the counties where Black PWH were significantly less likely to achieve viral suppression, four of them were identified as priority geographic regions in the EHE (Duval, Pinellas, Hillsborough, and Orange). There were some county characteristics that could be associated with these disparities, such as county-level income inequality by race/ethnicity or degree of residential segregation, but the data were not available for analysis for these CHR variables during this study period. South Florida counties where Black PWH had better clinical outcomes than White and Hispanic PWH were in locations with much higher proportions of Hispanic individuals. Future studies should investigate the determinants of these disparate outcomes to identify the mutable causes that could be targeted to address these disparities.

This study has several limitations. First, our sample of Ryan White enrollees may differ from the overall population of PWH in Florida, since enrollees have to be low income but be able to go through the eligibility verification process every six months. Although efforts have been made to increase the accessibility of Ryan White support, Hispanic PWH are underrepresented in our dataset (24.8% of all PWH but 19.5% of our sample [38]). Second, there may be significant variation within a county in access to care or other socio-ecological variables that influence HIV clinical outcomes. While our results are consistent with other studies that used data from smaller geographic areas, relying on county-level data may lead to biased results due to the ecological fallacy (i.e., relationships present at the population level may not apply at the individual one). For instance, assigning county-level socio-economic status and income indicators to individuals is biased because the within-county distributions of these factors can be highly variable. Individual-level measures of these indicators may better explain some of the disparities identified. Additionally, a limited number of county-level variables were eligible for inclusion in these analyses and some potentially important measures were not included. For example, we were unable to include a direct measure of residential segregation as a variable since this was not added to the CHR until 2016 and was therefore ineligible for inclusion in these analyses. Moreover, our analyses only included the calendar years when the individual had at least one HIV laboratory test. People with long gaps in care engagement may be underrepresented in our analyses. Further, a directed acyclic graph was not constructed to delineate relationships between variables and all included variables were treated as confounders, not mediating or moderating variables, and without a hierarchy to these potential confounders. In the geographic analyses, we classified PWH as Black or non-Black, so we were not able to identify the presence of county-level disparities between other groups (e.g., between Hispanic and White PWH), which may explain the protective association of Black race on the relative risks of non-immune reconstitution and non-viral suppression for some counties. Separate analysis of Hispanic PWH would provide helpful information for developing programs to address disparities in these counties.

The clinical outcomes of PWH in the Ryan White cohort have improved over time in Florida; however, our findings revealed persistent racial/ethnic disparities in both immune reconstitution and viral suppression over the last decade. While the racial/ethnic gap in immune reconstitution appears stable through time, for viral suppression, it may be growing—particularly for Black PWH. Racial/ethnic disparities in achieving viral suppression were especially prominent for six Florida counties, four of which are priority EHE regions. Our findings revealed persistent, spatially heterogeneous, racial/ethnic disparity in the odds of immune reconstitution and viral suppression in Florida. Only a small portion of the effect of race/ethnicity on clinical outcomes could be explained by available socio-ecological factors, suggesting systemic disparity, as well as warranting the search for additional, modifiable factors from individual and societal spheres to improve HIV outcomes among Black and Hispanic PWH. Further analysis of the underlying factors likely driving these findings, including health care inequities, stigma, and distrust in medicine and public health, along with improved data linkage to achieve better granularity and integration with individual-level factors, will be important for ultimately achieving HIV elimination goals in Florida.