Objective Male-to-female transgender women (transwomen) have a disproportionate burden of HIV. We sought to estimate HIV treatment cascade indicators among transwomen in San Francisco.
Methods We conducted a respondent driven sampling (RDS) study of 314 transwomen from August to December 2010. The study tested participants for HIV and collected self-reported data on linkage and access to care, viral load and antiretroviral treatment (ART). We derived population-based estimates and 95% CIs of cascade indicators using sampling weights based on established RDS methods. We conducted RDS-weighted logistic regression analyses to evaluate correlates of being on ART and being virologically suppressed (viral load ≤200 copies/mL).
Results The RDS-weighted population-based estimate of HIV prevalence was 39% (95% CI 32% to 48%) among transwomen tested for HIV. Among HIV-positive transwomen, 77% (95% CI 70% to 93%) reported being linked to care within 3 months of diagnosis and 87% (95% CI 76% to 98%) accessed care in the past 6 months. In addition, 65% (95% CI 54% to 75%) were on ART, and less than half (44%; 95% CI 21% to 58%) were virologically suppressed. Housing instability was associated with lower odds of being on ART and being virologically suppressed.
Conclusions We observed a high prevalence of HIV in our population-based estimates of transwomen in San Francisco, coupled with modest ART use and low virological suppression rates, indicating high potential for forward transmission. Poor HIV treatment outcomes were consistently associated with housing instability. These data suggest that multi-level efforts, including efforts to address housing insecurity, are urgently needed to ameliorate disparities in HIV clinical outcomes among transwomen and reduce secondary HIV transmission to their partners.
- HIV Testing
- HIV Clinical Care
- Epidemiology (General)
Statistics from Altmetric.com
Male-to-female transgender women (transwomen) have a disproportionate burden of HIV worldwide and have a 34.21-fold greater odds of being HIV infected compared with the general adult population.1 In San Francisco—where 35.5% of transwomen are estimated to have HIV2—transwomen have a significantly higher mean population viral load compared with other demographic groups, suggesting that a large proportion of transwomen are not receiving optimal care.3 Despite the high prevalence of HIV among transwomen, little is known about their HIV treatment outcomes. The aims of this study were to estimate population-level HIV treatment cascade indicators4—steps along the continuum of HIV care from diagnosis of infection to linkage to care and HIV virological suppression—and to evaluate correlates of HIV virological suppression among transwomen in San Francisco.
We conducted a secondary data analysis of HIV-positive transwomen from a respondent driven sampling (RDS) study in San Francisco conducted from August to December 2010. The methodology of this study has previously been described.5 In brief, the study recruited 11 ethnically/racially diverse “seeds” –initial participants enrolled to generate subsequent network referrals for the study –at least 18 years of age who identified as transwomen. Seeds and enrolled participants received coupons (3–5 each) to recruit through word of mouth other transwomen into the study from their respective networks. The coupon return rate was 36.8%. Among the 318 transwomen who were screened for eligibility, 95.3% were deemed eligible and agreed to participate. This study reached sample stability in key variables and satisfied other ideal RDS criteria.6 ,7 For example, the study had many long recruitment chains (mean recruitment wave=6; range=1–15) and generally moderate homophily with respect to race/ethnicity (range 0.15–0.63). The study reached equilibrium by the seventh wave of recruitment.5
All participants, regardless of self-reported HIV serostatus, were offered an HIV test and the study collected self-reported data on linkage and access to care, most recent HIV viral load, current antiretroviral treatment (ART) use and age, race/ethnicity, injection drug use and insurance status. We derived population-based estimates and 95% CIs of HIV cascade indicators using sampling weights adjusted for homophily and probability of being recruited into the study using established RDS methods via RDS Analysis Tool 6.0 (RDSAT). The cascade indicators we assessed were: (1) previous awareness of HIV infection, (2) linkage to care within 3 months of diagnosis, (3) access to care in the prior 6 months, (4) current ART use and (5) HIV virological suppression (viral load of ≤200 copies/mL). Using STATA V.12.0 (College Station, Texas, USA), we fitted a RDS-weighted multivariable logistic regression model to evaluate the association among virological suppression, current ART use and social/demographic/behavioural factors (eg, age, education, race/ethnicity, hormone use, relationship status, mental health conditions and sexual behaviours), while adjusting for individual sampling weights for outcomes (eg, current ART use and HIV virological suppression), generated from RDSAT.8–10 For model-building, we used the algorithm suggested by Hosmer and Lemeshow in which predictors that were statistically significant in the bivariate level using a p value cut-off point of 0.25 were included in the larger multivariable model.11 The final multivariable model was arrived at using a stepwise backward procedure with hypothesised confounders (age, race–ethnicity, insurance) retained in the models. Likelihood ratio tests were used to confirm that nested-models fit the data as well as larger models. All study procedures received approval from the Committee on Human Research at the University of California San Francisco (Institutional Review Board study # 13–11212).
The characteristics of the HIV-positive participants in the study are summarised in table 1. The RDS-weighted population-based estimate of HIV prevalence was 39.5% (95% CI 31.8% to 47.8%) among transwomen tested for HIV. Among transwomen who tested positive for HIV, 95% (95% CI 88% to 99%) were previously aware of their HIV-positive status, 77% (95% CI 70% to 93%) reported being linked to primary care within 3 months of their HIV diagnosis and 87% (95% CI 76% to 98%) had accessed care in the past 6 months. In addition, 65% (95% CI 54% to 75%) were currently on ART, and less than half (44%; 95% CI 21% to 58%) reported being virologically suppressed (viral load ≤200 copies/mL) (see table 2).
In the multivariable logistic regression model (see table 3), currently being on ART was significantly positively associated with older age (adjusted OR (aOR)=1.10; 95% CI 1.03 to 1.19) and inversely associated with marginal housing/homeless status (aOR=0.03; 95% CI 0.01 to 0.52). Moreover, being virologically suppressed was significantly associated with current ART use and housing status. The odds of virological suppression were 54.28-fold (95% CI 4.21 to 700.1) greater among transwomen currently taking ART compared with those who are not. In addition, the odds of virological suppression were 0.05-fold (95% CI 0.01 to 0.51) lower among homeless or marginally-housed transwomen compared with transwomen with stable housing (table 4).
We observed significant gaps in service delivery along the HIV care continuum for transwomen in San Francisco. We estimated that ART coverage among transwomen was modest (65%), relative to the overall San Francisco ART coverage (83%–89% according to HIV surveillance data).12 This RDS population estimate of current ART use is comparable with San Francisco surveillance data for transwomen, and also supports the surveillance registry's finding that transwomen have the lowest ART coverage (between 6% and 7% lower) of all gender groups and risk categories in this city.12 This finding differs from a recently published cohort study using data from HIV clinics across the USA which did not find significant differences in HIV care and treatment outcomes between transgender and non-transgender persons living with HIV.13 Regional differences may explain the difference between our findings and that study. Though our data suggest poor treatment outcomes for transwomen in San Francisco, these outcomes are still better relative to the general population of persons living with HIV in the USA overall.4
As expected, the relatively low coverage of ART use among transwomen in San Francisco explains the low levels of virological suppression we observed. We estimated that less than half of transwomen are virologically suppressed in San Francisco and virological suppression was significantly associated with housing stability, independent of current ART use. These findings are consistent with San Francisco's surveillance data, which have shown that HIV-positive transgender individuals have higher aggregate HIV community viral load and higher mortality rates than other HIV-positive populations.3 ,12 The low rate of virological suppression for this population has important public health implications as it may potentiate HIV transmission between HIV-positive transwomen and their partners.14 However, our finding that virological suppression was much higher for those currently on ART demonstrates the promise of these therapies for keeping transwomen healthy and the potential for stopping transmission between transwomen and their partners with better access to ART for the population. With the high coverage of HIV care and treatment services in San Francisco, the question remains as to why transwomen have low access to care and treatment. These findings point to housing as an important factor. The association between housing and poor HIV-related outcomes has been empirically documented among other populations,15 but this is the first study, to our knowledge, that has observed a significant, negative association between unstable housing and virological suppression for transwomen. Housing assistance has been shown to be a cost-effective and efficacious intervention in improving health outcomes among homeless and unstably-housed individuals living with HIV; efforts to target these services for homeless transwomen are urgently needed.16 ,17
This study has several limitations. First, the treatment outcomes we collected were based on self-report, which may be subject to recall and social desirability bias. Moreover, participants who are aware of their viral load may have more positive treatment outcomes; medical record data or viral load testing may help increase validity of data and should be used when feasible. Additionally, although we used RDS sampling weights to derive population-based estimates for our outcomes of interest with established methods, it is possible that these estimates may still not be representative of all transwomen in San Francisco. That said, the similarities among our estimates, the San Francisco HIV consensus estimates and the HIV surveillance data provide us with additional confidence in these findings.2 ,12 Finally, because coverage of HIV care and ART in San Francisco may be higher than other geographic areas, our findings may not be generalisable to other populations.
Relative to other persons living with HIV in San Francisco, transwomen fare worse along the continuum of HIV care with only a small proportion having achieved virological suppression. Multi-level efforts, especially those addressing structural-level factors like housing, are urgently needed to ameliorate disparities in HIV clinical outcomes among transwomen and reduce secondary HIV transmission to their partners. In addition, comprehensive efforts to meet evidence-based HIV treatment guidelines from the Department of Health and Human Services—which recommends ART treatment among all HIV-positive individuals, regardless of CD4 count—may be particularly needed for transwomen, given the lower coverage of ART in this population.18 Given the lack of data among transgender populations, it is also imperative that HIV surveillance and cohort studies capture data on sex assigned at birth along with current gender identity to better identify transgender individuals living with HIV and monitor their HIV treatment outcomes.19 Such structural interventions are needed to create equitable access to HIV treatment and full benefit of HIV care to reduce HIV-related morbidity and mortality in this underserved population.
High prevalence of HIV among transwomen in this respondent driven sampling (RDS) study was coupled with modest antiretroviral treatment (ART) use and low virological suppression rates.
Poor HIV treatment outcomes were consistently associated with housing instability in this study.
Multi-level efforts, including efforts to address housing insecurity, are urgently needed to ameliorate disparities in HIV clinical outcomes among transwomen and reduce secondary HIV transmission to their partners.
Handling editor Jackie A Cassell
Contributors G-MS conceptualised the analysis, wrote the manuscript, conducted the data analysis and interpretation. ECW assisted in the writing of the manuscript and interpretation of data. JR was responsible for the coordination of the study, assisted in the writing of the manuscript and interpretation of data. OM assisted in the conduct of the study and with writing of the manuscript and interpretation of data. TP and HFR were responsible for the design of the study, assisted in the writing of the manuscript and interpretation of data. All authors contributed to and approved the final draft of the manuscript.
Funding This study was funded by the San Francisco Department of Public Health. The funder had no role in the conduct of the study or the preparation of this manuscript.
Competing interests None.
Ethics approval University of California San Francisco.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.