| |
Sociodemographic, ecological, and spatiotemporal factors associated with HIV drug resistance in Florida: a retrospective analysis…implications for 'End AIDS'
|
| |
| |
Download the PDF here
This study analyzed the prevalence, sociodemographic, ecological, and spatial-temporal determinants of HIVDR among PWH in Florida during 2012-2017. The results indicate HIVDR prevalence is higher in Florida compared to current NA estimates (9) and may be increasing for PI and INSTI. Compared to previously published estimates for NA from 2007-2016, HIVDR prevalence was higher in Florida for NRTI (15⋅7% vs 19⋅2%), NNRTI (23⋅4% vs. 29⋅7%), and MDR (11⋅5% vs. 13⋅2%) (9). Estimates were comparable for PI (7⋅2% vs 6⋅6%); however, we observed a positive association between increasing genotype year and prevalence of PI resistance. We likewise observed a positive association between increasing genotype year and prevalence of resistance to newer INSTI therapies over the study period, though this result may be linked to the increased frequency of integrase testing in more recent years.
Moreover, from a public health perspective, HIVDR presents substantial programmatic burden, contributing to roughly 9% of new infections by 2030 if current trends persist (7). HIVDR surveillance, prevention and response are therefore critical for achieving elimination of HIV as a public health threat by 2030 (2,8).
The increased burden of HIVDR in Florida observed in this study may reflect the overall US HIV epidemic. Florida, like other southern states, has a disproportionately high burden of HIV infection compared to other regions.
Multivariable analyses revealed HIVDR was higher among PWH aged ≥46 years (vs. ≤25 years), individuals who acquired HIV through MCT (vs. heterosexual contact), Black (vs. White) individuals, and males (vs. females). These findings are similar to those from a previous study reporting comparable racial and sex differences with the odds non-viral suppression in Florida (25). Due to exposure to ART in utero or through breast milk after birth, individuals who acquire HIV through MCT often have high rates of pretreatment drug resistance (26), which may explain why this transmission group had the highest odds of all HIVDR outcomes studied. In contrast, individuals who acquired HIV through IDU had lower odds of HIVDR in the current study.
The increased burden of HIVDR in Florida observed in this study may reflect the overall US HIV epidemic. Florida, like other southern states, has a disproportionately high burden of HIV infection compared to other regions. Nearly half of all HIV diagnoses in the US occur in the south, despite accounting for only one-third of the US population (30). Factors thought to be driving the HIV epidemic in the south include poverty, income inequality, cultural issues (e.g. homophobia, transphobia, and racism), and higher rates of comorbidities (e.g. obesity, diabetes, and cancer) (30). While this analysis could not account for factors such as socio-economic status or comorbiditiesat the individual-level, we did assess some of these factors at the county-level and found HIVDR was significantly associated with socioeconomic status, income level, unemployment, and mental health. This suggests there are socioeconomic and mental health factors contributing to acquired HIVDR, and future studies should investigate this relationship further at the individual-level.
Notably, higher rates of resistance were consistently associated with higher percent unemployed. We also noted associations between NNRTI resistance and county-level rankings. County rankings for health factors, health outcomes, life quality, and social economic factors, as well as percent poor or fair health, percent limited access to healthy foods, percent unemployed, percent diabetic, and physically or mentally unhealthy days were all associated with higher NNRTI resistance. Further, median household income was inversely associated with MDR and NNRTI resistance.
Summary: In this study of 34,447 HIV-1 sequences collected in Florida, we observed high prevalence of drug resistance with significant sociodemographic and geospatial heterogeneity. Resistance was linked to counties with lower socioeconomic status, higher unemployment, and poor mental health.
The US epidemic epicenter is concentrated in the south, which contributed over half of all new HIV diagnoses in 2017 (11). That same year, the southern state Florida had the most new HIV diagnoses in the country (12) and the proportion of PWH virally suppressed (<200 copies/mL) was only 62% (13). The extent to which HIVDR contributes to the burden of HIV in Florida, a contributor to poor treatment outcomes, is not understood.
HIVDR prevalence over the study period was 29⋅7% (95% confidence interval [CI]=29⋅2%-30⋅2%) for NNRTI, 19⋅2% (CI=18⋅8%-19⋅6%) for NRTI, 23⋅5% (CI=22⋅1%-24⋅9%) for TDR, 13⋅2% (CI=12⋅8%-13⋅6 for MDR, and 8⋅2% (CI=7⋅7%-8⋅7%) for INSTI, and 6⋅6% (CI=6⋅3%-6⋅9%) for PI (Figure 1).
Compared to Black individuals, White individuals tended to have lower odds of HIVDR – particularly for NRTI, NNRTI, INSTI, and MDR – but higher odds of PI resistance (OR=1⋅31; CI=1⋅16-1⋅48).
There were several associations between HIVDR and county-level socioecological factors (Table 2). Higher HIVDR rates were associated with higher socioeconomic factors rank (i.e. lower socioeconomic status), higher percent unemployed, more mentally unhealthy days, and higher percent elderly population (Table 2). Alternatively, lower rates of HIVDR were associated with higher crime rates and lower percent rural population. Univariate model associations of sociodemographic, clinical, and socioecological factors with resistance outcomes are presented in Supplementary Tables 1-3.
Older individuals (≥46 years), had significantly higher resistance compared to individuals ≤25 years in all models except for TDR (Table 2).
The most frequent race/ethnicity was Black for both periods, however, the proportion of Hispanic/Latino individuals increased from 19⋅0% (CI=18⋅3-19⋅7%) in earlier years to 21⋅4% (CI=20⋅8-21⋅9%) in later years. Likewise, the proportion of Latin American-born individuals increased from 3⋅9% (CI=3⋅5-4⋅3%) in earlier years to 5⋅4% (CI=5⋅1-5⋅7%) in later years. MTM was the most frequent transmission category, increasing from 43⋅0% (CI=42⋅0%-43⋅9%) in earlier years to 47⋅4% (CI=46⋅7-48.0%) in more recently.
Notably, higher rates of resistance were consistently associated with higher percent unemployed. We also noted associations between NNRTI resistance and county-level rankings. County rankings for health factors, health outcomes, life quality, and social economic factors, as well as percent poor or fair health, percent limited access to healthy foods, percent unemployed, percent diabetic, and physically or mentally unhealthy days were all associated with higher NNRTI resistance. Further, median household income was inversely associated with MDR and NNRTI resistance.
Overall estimates for North America (NA) suggest prevalence of HIVDR among PWH with molecular sequences is 7⋅2% for protease inhibitors (PI), 15⋅7% for nucleoside reverse transcriptase inhibitors (NRTI), 23⋅4% for non-nucleoside reverse transcriptase inhibitors (NNRTI), 11⋅5% for two-class (or "multi-drug") resistance (MDR), and 13⋅0% for transmitted drug resistance (TDR) for the period 2007-2016 (9).
Moreover, from a public health perspective, HIVDR presents substantial programmatic burden, contributing to roughly 9% of new infections by 2030 if current trends persist (7). HIVDR surveillance, prevention and response are therefore critical for achieving elimination of HIV as a public health threat by 2030 (2,8)
In 2019, seven largely urban Florida counties (Broward, Duval, Hillsborough, Miami-Dade, Orange, Palm Beach, and Pinellas), which include the major cities Jacksonville, Orlando, Tampa, and Miami, were identified as target regions for Ending the HIV Epidemic: A Plan for America (EHE) (14). One of the key strategies of EHE is to rapidly respond to potential outbreaks using cluster detection techniques (e.g. molecular surveillance), made possible through drug resistance testing (15). Over the past decade the Florida Department of Health (FDOH) has been collecting viral isolate sequences on individuals with a recent HIV diagnosis. Collection of these sequences started as part of the CDC funded: Variant, Atypical and Resistant HIV Surveillance (VARHS) program designed to monitor HIVDR starting in 2007 in Florida. Around 60,000 HIV sequences have been collected by FDOH since 2007.
Discussion
This study analyzed the prevalence, sociodemographic, ecological, and spatial-temporal determinants of HIVDR among PWH in Florida during 2012-2017. The results indicate HIVDR prevalence is higher in Florida compared to current NA estimates (9) and may be increasing for PI and INSTI. Compared to previously published estimates for NA from 2007-2016, HIVDR prevalence was higher in Florida for NRTI (15⋅7% vs 19⋅2%), NNRTI (23⋅4% vs. 29⋅7%), and MDR (11⋅5% vs. 13⋅2%) (9). Estimates were comparable for PI (7⋅2% vs 6⋅6%); however, we observed a positive association between increasing genotype year and prevalence of PI resistance. We likewise observed a positive association between increasing genotype year and prevalence of resistance to newer INSTI therapies over the study period, though this result may be linked to the increased frequency of integrase testing in more recent years.
Multivariable analyses revealed HIVDR was higher among PWH aged ≥46 years (vs. ≤25 years), individuals who acquired HIV through MCT (vs. heterosexual contact), Black (vs. White) individuals, and males (vs. females). These findings are similar to those from a previous study reporting comparable racial and sex differences with the odds non-viral suppression in Florida (25). Due to exposure to ART in utero or through breast milk after birth, individuals who acquire HIV through MCT often have high rates of pretreatment drug resistance (26), which may explain why this transmission group had the highest odds of all HIVDR outcomes studied. In contrast, individuals who acquired HIV through IDU had lower odds of HIVDR in the current study. This funding was initially puzzling since rates of homelessness and unemployment (linked to decreased care linkage/retention), are high among people who inject drugs (PWID) (27); yet, another study reported lower incidence of drug resistance mutations among PWID as a result of the test-and-treat initiative (28). Further analysis of this finding is warranted.
Geographic analyses revealed considerable heterogeneity in HIVDR prevalence by Florida county. Interestingly, high prevalence of resistance was not detected in counties with the highest proportion of PWH or resistance testing. The seven Florida counties listed in the EHE, which together represent 73% of total PWH in the state, did not have significantly higher rates of HIVDR in the current study. The EHE targets largely urban Florida counties in which fewer barriers to care, such as lack of specialty health care providers, exist (29). Given that HIVDR contributes to poorer overall treatment outcomes, continued monitoring of HIVDR in rural regions is needed. Analysis of HIV-1 subtypes in Florida revealed most sequences were covered by the B subtype, which is typical in the US, but the increasing presence of recombinants warrants further investigation of the contribution of imported HIV transmissions in Florida.
The increased burden of HIVDR in Florida observed in this study may reflect the overall US HIV epidemic. Florida, like other southern states, has a disproportionately high burden of HIV infection compared to other regions. Nearly half of all HIV diagnoses in the US occur in the south, despite accounting for only one-third of the US population (30). Factors thought to be driving the HIV epidemic in the south include poverty, income inequality, cultural issues (e.g. homophobia, transphobia, and racism), and higher rates of comorbidities (e.g. obesity, diabetes, and cancer) (30). While this analysis could not account for factors such as socio-economic status or comorbiditiesat the individual-level, we did assess some of these factors at the county-level and found HIVDR was significantly associated with socioeconomic status, income level, unemployment, and mental health. This suggests there are socioeconomic and mental health factors contributing to acquired HIVDR, and future studies should investigate this relationship further at the individual-level.
This analysis had many strengths. To our knowledge, it was the largest and most comprehensive study of transmitted and acquired HIVDR in a state to date. Previous smaller-scale studies have reported the prevalence and correlates of TDR in other US regions (31–33); however, fewer studies exist to describe trends in acquired HIVDR at the state-level, and no studies have assessed associations with spatial-temporal or socioecological factors. Our multivariable models combined individual-level sociodemographic and clinical factors with county-level health indicators to account for socioecological factors contributing to HIVDR patterns. The study provides important epidemiological information on the geographic regions and subpopulations with the greatest burden of HIVDR in a region with disproportionately high incidence of HIV. Moreover, these results justify the need for clinicians to order genotype tests to ensure ART regimen compatibility and for continued molecular surveillance for public health Florida.
This analysis also had limitations. Importantly, we lacked data on prescription ART information which prevented the ability to consider the impact of prescribing practices on HIVDR patterns. This may explain the significant annual increase in resistance to the newest group of therapies (INSTI) observed in this study, since we were unable to account for expected increased rates of INSTI prescriptions in more recent years. Another limitation of our analysis was the method of MDR determination. Although we did not assess specific mutations, presumably, the majority of NRTI resistance was M184V, and the majority of NNRTI resistance was K103N and related efavirenz (EFV) resistance mutations. Since many people with resistance to EFV also have resistance to M184V (due to the use of combination drugs like Truvada), this explains the similarly high prevalence of resistance to these two classes, and the apparent high prevalence of MDR. Thus, MDR estimates may be artificially inflated due to resistance to EFV+3TC/FTC. Because cross-resistance to specific drugs within ART classes can be common and pooled estimates contain less measurement error than resistance to single drugs, we preferred to run the analysis by ART class. Future studies should examine single drugs as well as commonly prescribed combination drugs in drug resistance analyses. Additionally, although coverage of genotype testing was near or above 50% throughout the study period according to the FDOH, selection bias likely occurred since our study population only included diagnosed PWH who received a genotype test, representing less than half of PWH in the state. These findings do not necessarily represent the burden of resistance among non-virally suppressed PWH in Florida. Another potential limitation is that our modeling approach did not account for individuals contributing more than one sequence; however, the impact was likely inconsequential given the mean number of sequences available per person was 1.2. Further, results of the socioecological analysis of health indices at the county-level should not be interpreted at the individual-level. This approach was selected to improve model fitness and provide a source of socioeconomic data that would have otherwise been omitted. Deeper analysis of the individual-level sociodemographic/behavioral factors contributing to HIVDR patterns in the community (e.g. medication adherence) is needed.
Conclusion
This was the most comprehensive analysis of HIVDR in Florida to date. It covered all 67 Florida counties, encompassed several consecutive years of genotype sampling, and analyzed associations with numerous epidemiological, spatial-temporal and socioecological factors to provide a complete depiction of HIVDR in a region with a disproportionately high burden of HIV. Our findings indicate prevalence of HIVDR in Florida is higher than published North American estimates, with considerable heterogeneity by geographic region. These results warrant further surveillance of HIV molecular epidemiology in Florida in support of EHE.
Sociodemographic, ecological, and spatiotemporal factors associated with HIV drug resistance in Florida: a retrospective analysis
09 July 2020 Journal of Infectious Diseases
Abstract
Background
Persons living with HIV (PWH) with resistance to antiretroviral therapy (ART) are vulnerable to adverse HIV-related health outcomes and can contribute to transmission of HIV drug resistance (HIVDR) when non-virally suppressed. The degree to which HIVDR contributes to disease burden in Florida –the US state with the highest HIV incidence– is unknown.
Methods
We explored sociodemographic, ecological, and spatial-temporal associations of HIVDR. HIV-1 sequences (n=34,447) collected during 2012-2017 were obtained from the Florida Department of Health. HIVDR was categorized by ART class: nucleoside reverse transcriptase inhibitors (NRTI), non-NRTI (NNRTI), protease inhibitors (PI), and integrase inhibitors (INSTI). Multi-drug resistance (MDR) and transmitted-drug resistance (TDR) were also evaluated. Multivariable fixed-effects logistic regression models were fitted to associate individual and county-level sociodemographic and ecological health indicators with HIVDR.
Results
HIVDR prevalence was 19.2% (NRTI), 29.7% (NNRTI), 6.6% (PI), 23.5% (TDR), 13.2% (MDR), and 8.2% (INSTI) with significant variation by Florida county.
Individuals who were older, Black, or acquired HIV through mother-to-child transmission had significantly higher odds of HIVDR. HIVDR was linked to counties with lower socioeconomic status, higher unemployment, and poor mental health.
Conclusions
Our findings indicate HIVDR prevalence is higher in Florida than aggregate North American estimates with significant geographic and socioecological heterogeneity.
In the United States, aggregate estimates of HIVDR prevalence appear stable over time when compared to other regions (3,9); however, HIV epidemic dynamics differ dramatically across regions within the US (10). Overall estimates for North America (NA) suggest prevalence of HIVDR among PWH with molecular sequences is 7⋅2% for protease inhibitors (PI), 15⋅7% for nucleoside reverse transcriptase inhibitors (NRTI), 23⋅4% for non-nucleoside reverse transcriptase inhibitors (NNRTI), 11⋅5% for two-class (or "multi-drug") resistance (MDR), and 13⋅0% for transmitted drug resistance (TDR) for the period 2007-2016 (9). HIVDR prevalence at the regional-level is not well understood, however. The US epidemic epicenter is concentrated in the south, which contributed over half of all new HIV diagnoses in 2017 (11). That same year, the southern state Florida had the most new HIV diagnoses in the country (12) and the proportion of PWH virally suppressed (<200 copies/mL) was only 62% (13). The extent to which HIVDR contributes to the burden of HIV in Florida, a contributor to poor treatment outcomes, is not understood.
|
|
| |
| |
|
|
|
