Abstract
Background: Transgender women (TW) are highly burdened by HIV. There is increasing interest in digital (i.e., through internet-based interfaces) HIV research; yet few studies have assessed potential biases of digital compared to site-based data collection. This study examined differences in characteristics between TW participating via site-based versus digital-only modes in an HIV incidence cohort. Methods: Between March 2018-Aug 2020, a multisite cohort of 1,312 adult TW in the eastern and southern USA was enrolled in site-based and exclusively digital modes. We evaluated differences in baseline demographics, socio-structural vulnerabilities, healthcare access, gender affirmation, mental health, stigma, social support, and HIV acquisition risk comparing site-based vs digital modes using chi square tests and Poisson regression modeling with robust standard errors. Results: The overall median age was 28 (interquartile range=23-35) years and over half identified as people of color (15% Black, 13% Multiracial, 12% Another Race, 18% Latina/e/x). A higher proportion of site-based (vs. digital mode) participants resided in the Northeast, were younger, identified as people of color, experienced socio-structural vulnerabilities, had a regular healthcare provider, received medical gender affirmation, endorsed mental health symptoms and stigma, reported HIV acquisition risk but also greater experience with biomedical HIV prevention (pre-exposure and post-exposure prophylaxis), and had larger social networks (all p<0.05). Conclusion: Site-based and digital approaches enrolled TW with different demographics, life experiences, and HIV acquisition risks. A hybrid cohort model may achieve a more diverse and potentially representative sample of TW than either site-based or online cohorts alone for HIV research.