HIV incidence and mortality in transgender women in the eastern and southern USA: a multisite cohort study

Background: Epidemiological monitoring of HIV among transgender women is minimal despite prioritisation of this group in the US National HIV/AIDS Strategy (2022–2025). We aimed to estimate HIV incidence in a multisite cohort of transgender women in the eastern and southern USA. Participant deaths were identified during follow-up; thus, we felt it was an ethical imperative to report mortality alongside HIV incidence. Methods: In this study, we established a multisite cohort across two modes: a site-based, technology-enhanced mode in six cities (Atlanta, Baltimore, Boston, Miami, New York City, and Washington, DC) and an exclusively digital mode that spanned 72 eastern and southern US cities that matched the six site-based cities based on population size and demographics. Trans feminine adults (≥18 years) who were not living with HIV were eligible and followed up for at least 24 months. Participants completed surveys and oral fluid HIV testing with clinical confirmation. We ascertained deaths through community and clinical sources. We estimated HIV incidence and mortality using the number of HIV seroconversions and deaths, respectively, divided by person-years accumulated from enrolment. Logistic regression models were used to identify predictors of HIV seroconversion (primary outcome) or death. Findings: Between March 22, 2018, and Aug 31, 2020, we enrolled 1312 participants with 734 (56%) in site-based and 578 (44%) in digital modes. At the 24-month assessment, 633 (59%) of 1076 eligible participants consented to extending participation. 1084 (83%) of 1312 participants were retained at this analysis based on the study definition of loss to follow-up. As of May 25, 2022, the cohort participants had contributed 2730 accumulated person-years to the analytical dataset. Overall HIV incidence was 5·5 (95% CI 2·7–8·3) per 1000 person-years and incidence was higher among Black participants and those living in the south. Nine participants died during the study. The overall mortality rate was 3·3 (95% CI 1·5–6·3) per 1000 person-years, and the rate was higher among Latinx participants. Identical predictors of HIV seroconversion and death included residence in southern cities, sexual partnerships with cisgender men, and use of stimulants. Participation in the digital cohort and seeking care for gender transition were inversely associated with both outcomes. Interpretation: As HIV research and interventions are increasingly delivered online, differences by mode highlight the need for continued community and location-based efforts to reach the most marginalised transgender women. Our findings underscore community calls for interventions that address social and structural contexts that affect survival and other health concerns alongside HIV prevention. Funding: National Institutes of Health. Translation: For the Spanish translation of the abstract see Supplementary Materials section.