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.
Keywords: transgender women, HIV prevention, research methods, longitudinal cohort
INTRODUCTION
In the United States (USA), transgender (trans) women have a disproportionately high prevalence of HIV infection1,2 and represent a key population in the national Ending the HIV Epidemic strategy.3–5 There is a need to identify modifiable risk factors, ascertain health-promoting resiliencies, and design, deliver, and test culturally-tailored interventions to meet the unique HIV prevention needs of this population. Among trans women, the HIV epidemic and HIV-related vulnerabilities are underscored by myriad social and health concerns,6 including socioeconomic marginalization,7,8 healthcare access barriers,8,9 gender affirmation needs,10,11 mental health conditions,12–14 gender-based violence,15,16 stigma and discrimination,17,18 sexual partnerships and social network characteristics,19 high pre-exposure prophylaxis (PrEP) indication and low PrEP uptake,20,21 and high rates of sexually transmitted infections (STIs)22–24 which may increase HIV acquisition efficiency. Importantly, these are also factors which impact research participation for trans women, including HIV cohort studies.25
Many contemporary HIV cohorts are either in-person facility-based (site-based)26–28 or entirely online (digital) with remote participation.29,30 Exclusively digital cohorts are thought to address many barriers related to transportation, time, and other inconveniences, span geographic distances, reach people “where they are”, and reduce research infrastructure costs associated with site-based cohorts. The push toward online digital research has been further accelerated due to environmental circumstances (e.g., COVID-19 pandemic, anti-transgender legislation contexts). However, exclusively digital cohorts have inherent assumptions including that technology access is high and consistent in the study population, that individuals in online environments trust researchers, and participants accessed exclusively online are representative or generalizable to the same population as those in site-based HIV prevention research and services. It is important to investigate and consider methodological limitations and strengths of site-based and digital-only cohorts that enroll trans women (and, broadly, other populations affected by the HIV epidemic), such as selection bias and limited generalizability, given these may impact the evidence-base upon which to address the HIV epidemic for this high priority population. Despite growing interest in multimodal survey methods,31–34 few studies have examined differences between trans respondents completing surveys in-person or online35,36 and none, to our knowledge, have done so for HIV research more broadly.
This study sought to assess baseline differences in site-based and digital mode participants in a cohort to study HIV incidence among trans women. The LITE Study deployed a hybrid cohort methodology with technology-enhanced site-based and exclusively digital participation modes. This multimodal design allowed us to capture comparable data including on demographic, healthcare access, gender affirmation, and HIV prevention characteristics, examine implementation differences by mode, and observe potential selection biases among trans women in the cohort.
METHODS
Participants and Procedures
The LITE Study was a multisite cohort designed to estimate HIV incidence and vulnerabilities for HIV acquisition among trans women across at least 24 months of follow-up.38–40 The cohort was comprised of 1,312 trans women enrolled between March 22, 2018-August 31, 2020. Qualitative formative research informed the cohort.25,41 A virtual Community Advisory Board composed of national trans leaders consulted on all aspects of the study.42,43
Trans women were recruited to either mode using digital and non-digital convenience sampling methods such as social media, dating apps, geotargeted online advertisements, peer referrals, and community events. Adult trans women were enrolled using a hybrid cohort design with two modes: a site-based technology-enhanced (e.g., electronic data capture via tablets or computers) mode enrolled at clinical and research organizations in 6 eastern and southern USA cities (Atlanta, Baltimore, Boston, Miami, New York City, Washington, DC) (n=734), and a strictly online digital mode from 72 cities with similar demographic characteristics as the site-based cities (n=578). Candidate participants who resided in the 6 cities with the site-based modes could elect to participate in either the site-based mode or the digital mode.
Cohort inclusion criteria (regardless of participation mode) included being >=age 18 years, identifying on the trans feminine spectrum determined at screening using the two-step method (current gender identity as a woman or along the trans feminine spectrum, assigned a male sex at birth),44 speaking or understanding English or Spanish, and having no prior HIV infection diagnosis confirmed by an HIV-negative baseline test. Screening and enrollment protocols and procedures were synced for cohort modes. Study procedures were available in English and Spanish.
Following preliminary screening and electronic consent, trans women completed a baseline socio-behavioral survey and HIV test (self-collected oral fluid specimens for HIV-1 testing and laboratory-verification of HIV status). All participants had access to their study timeline, survey questionnaires, and HIV testing information via a secure hybrid app, which was freely available in Apple and Google app stores, as well as available as a web application for participants without a smartphone or tablet. All study activities were reviewed and approved by the Johns Hopkins Single Institutional Review Board (sIRB; IRB00142429). Additional methodological details have been published elsewhere,38,39 including procedures to ensure non-fraudulent enrollments.
Measures
Demographics, socio-structural vulnerabilities, language and literacy, and recruitment source.
Demographics included region (North, Mid-Atlantic, South) coded from reported city of residence within US Census regions and divisions, age in years (continuous and grouped: 18–24, 25–29, 30–39, 40+), race (Black, Multiracial, Other Race, White), ethnicity (Latina/e/x vs not), gender identity (female/woman, trans female/trans woman, nonbinary/agender/another identity), and sexual orientation (straight/heterosexual, another sexual identity). Socio-structural vulnerabilities included education (<=high school, >=some college), employment (employed full-/part-time, not employed), income in last 30 days (>=$1,000, <=$0–999 consistent with 2018 federal poverty level),45 food insecurity with the US Department of Agriculture indicator, “How often do you run out of food or money to purchase food at the end of the month?”46 (Likert-scale responses collapsed to yes, including sometimes/most of the time/almost always, and no, including never/seldom/rarely), and lifetime and recent (past 3-month) homelessness (yes, no), incarceration (yes, no), and sex work (yes, no). Indexes of lifetime and of recent (past 3-month for site-based, or past 6-month for digital) socio-structural vulnerability were created by summing each of these seven binary indicators. Survey language for all participants (Spanish, English) and literacy level for site-based participants via Rapid Estimate of Adult Literacy in Medicine (low <8 words read correctly, high >=8)47 were assessed. Participants were asked the source from which they were recruited (e.g., peer, healthcare organization or medical provider).
Healthcare access, healthcare utilization, and gender affirmation.
Healthcare access and utilization variables were health insurance (uninsured, public, private), regular healthcare provider (yes, no), where they get healthcare when sick (e.g., community health center, private doctor’s office), perception of provider knowledge about trans health (not at all, somewhat, or very knowledgeable), and barriers to care (e.g., time, transportation). Gender affirmation included age of first trans self-recognition, clinical gender affirmation (hormones, age at hormone initiation, source of hormones, surgery, age at first surgery, silicone fillers), tried to get gender transition-related care (yes, no), unmet need for gender-related care (yes, no), and legal name and gender marker on identification documents (IDs) (none vs some, most, or all IDs and records list the name/gender I prefer).
Mental health, substance use, violence victimization, stigma, and social support.
Mental health measures included the validated Kessler-6 (K-6) with score >=13 considered clinically elevated psychological distress,48 and Primary Care Post Traumatic Stress Disorder (PC-PTSD) screener where scores >=3 indicated clinically significant elevation in PTSD symptomatology.49 Substance use variables were the 3-item Alcohol Use Disorders Identification Test (AUDIT-C) for hazardous drinking (summary score >=4),50 the 10-item Drug Abuse Screen Test (DAST-10) for substance use disorder (summary score >=3),51 and lifetime injection drug use (yes, no). The 8-item Adverse Childhood Experiences Scale (ACES) measured childhood violence (scores summed and categorized as 0, 1–3, or >=4).52 Physical, sexual, and psychological violence in adulthood were captured with adapted World Health Organization measures.53 Stigma was measured using the 9-item anticipated discrimination subscale of the Intersectional Discrimination Index (continuous, range 0–36).54 Social support variables included the 5-item Multidimensional Scale of Perceived Social Support (continuous, range 0–30),55 family acceptance (strongly disagree/disagree, neither agree nor disagree, agree/strongly agree), and number of individuals they know personally who are trans women.
Sexual relationships, HIV acquisition risk behaviors, biomedical HIV prevention, HIV knowledge and perceived risk, and self-reported STI diagnosis.
Variables were current relationship status (single, casually dating, committed relationship, legally married or civil union, other relationship), gender(s) of sexual partners (cisgender man, cisgender woman, transgender woman, transgender man, nonbinary assigned female at birth [AFAB], nonbinary assigned male at birth [AMAB]), number of sexual partners in last 6 months, partner concurrency (more than one partner), HIV acquisition risk behaviors (vaginal sex, anal sex, vaginal and anal condomless sex, shared injection drug use), and where met most recent partner (e.g., online dating apps, public spaces). Biomedical HIV prevention included PrEP awareness (yes, no), PrEP indicated based on trans women-specific criteria20 (yes, no), PrEP uptake ever and current in past 30 days (yes, no), reasons not used PrEP (e.g., side effects, people think you have HIV), PEP awareness (yes, no), and PEP ever taken (yes, no). Participants reported on five binary HIV knowledge questions (0–1, >=2 correct answers), HIV risk perception (no risk, low risk, medium/high risk), and self-reported STI diagnosis in the last 6 months (yes, no).
Statistical Analysis
Primary analyses.
Analyses utilized baseline data only. Descriptive statistics (frequency, proportion, median, interquartile range) were obtained for the overall sample and by site-based and digital-only cohort participation. Bivariate tests (χ2 for proportions, Wilcoxon rank-sum for continuous variables) were calculated comparing site-based and digital participation modes. We selected a set of variables for modeling based on a combination of conceptual salience, prior research literature, and χ2 or Wilcoxon rank-sum tests that were statistically significant (p<0.05). Variables selected for modeling were geographic region, age group, race, ethnicity, education, employment, history of sex work, food insecurity, incarceration history, survey language, health insurance, regular healthcare provider, hormones, IDs list preferred name, psychological distress, PTSD symptomatology, psychological violence, anticipated discrimination, number of trans women personally know, cisgender man partner, cisgender woman partner, PrEP indicated, perceived HIV acquisition risk, and healthcare organization/provider as recruitment source. Poisson regression models with robust standard errors estimated crude and adjusted prevalence ratios of participation in site-based (vs. online) mode. Multivariable model 1 did not adjust for recruitment source; model 2 added recruitment source as a potential confounding variable. Analyses were conducted in R.
Sensitivity analysis.
A sensitivity analysis was performed to assess whether the differences observed between site-based and online modes were due to the mode of participation rather than simply inherent differences due to comparing individuals from 6 major cities to individuals from 72 other cities with similar demographic characteristics. Bivariate tests (χ2 for proportions, Wilcoxon rank-sum for continuous variables) were used to compare demographics, socio-structural vulnerabilities, language and literacy, and recruitment source for participants from the 6 cities who were site-based (n=733) and online participants from those same 6 cities (n=130).
RESULTS
Demographics, Socio-Structural Vulnerabilities, Language and Literacy, and Recruitment Source (Table 1 and Supplemental Table 1)
Table 1.
Demographics, Socio-structural Vulnerabilities, Language and Literacy, and Recruitment Source Among Transgender Women in the LITE Cohort: Overall and Stratified for Site-based and Digital Modes.
| Characteristic | Overall, N = 1,3121 | Site-based, N = 7341 | Digital, N = 5781 | p-value2 |
|---|---|---|---|---|
|
| ||||
| Demographics | ||||
| Geographic region | <0.001 | |||
| North | 655 (49.9%) | 382 (52.0%) | 273 (47.2%) | |
| Mid-Atlantic | 301 (22.9%) | 194 (26.4%) | 107 (18.5%) | |
| South | 356 (27.1%) | 158 (21.5%) | 198 (34.3%) | |
| Age in Years, Median (IQR) | 28 (23–35) | 29 (24–37) | 27 (22–34) | <0.001 |
| Age group | <0.001 | |||
| 18–24 | 428 (32.6%) | 214 (29.2%) | 214 (37.0%) | |
| 25–29 | 316 (24.1%) | 185 (25.2%) | 131 (22.7%) | |
| 30–39 | 333 (25.5%) | 178 (24.3%) | 155 (26.8%) | |
| 40+ | 235 (17.9%) | 157 (21.4%) | 78 (13.5%) | |
| Race | <0.001 | |||
| Black | 197 (15.0%) | 171 (23.3%) | 26 (4.5%) | |
| Multiracial | 175 (13.3%) | 110 (15.0%) | 65 (11.2%) | |
| Other | 154 (11.7%) | 119 (16.2%) | 35 (6.1%) | |
| White | 786 (59.9%) | 334 (45.5%) | 452 (78.2%) | |
| Ethnicity | <0.001 | |||
| Not Latina/e/x | 1,066 (82.2%) | 529 (73.3%) | 537 (93.4%) | |
| Latina/e/x | 231 (17.8%) | 193 (26.7%) | 38 (6.6%) | |
| Unknown | 15 | 12 | 3 | |
| Gender identity | <0.001 | |||
| Female or woman | 366 (27.9%) | 246 (33.6%) | 120 (20.8%) | |
| Trans female, trans woman, male-to-female, Trans feminine, woman of trans experience | 843 (64.4%) | 433 (59.1%) | 410 (71.1%) | |
| Nonbinary, agender, another identity | 101 (7.7%) | 54 (7.4%) | 47 (8.1%) | |
| Unknown | 2 | 1 | 1 | |
| Sexual orientation | <0.001 | |||
| Straight/heterosexual | 299 (23.3%) | 243 (34.0%) | 56 (9.8%) | |
| All others | 987 (76.7%) | 471 (66.0%) | 516 (90.2%) | |
| Unknown | 26 | 20 | 6 | |
| Socio-structural vulnerabilities | ||||
| Educational attainment | <0.001 | |||
| High school or less | 362 (27.8%) | 259 (35.6%) | 103 (18.0%) | |
| Some college or higher | 938 (72.2%) | 468 (64.4%) | 470 (82.0%) | |
| Unknown | 12 | 7 | 5 | |
| Employment | <0.001 | |||
| Not employed | 523 (41.1%) | 330 (46.0%) | 193 (34.7%) | |
| Employed (full-time or part-time) | 750 (58.9%) | 387 (54.0%) | 363 (65.3%) | |
| Unknown | 39 | 17 | 22 | |
| Income | <0.001 | |||
| Above the federal poverty level | 698 (62.7%) | 332 (54.0%) | 366 (73.5%) | |
| Below the federal poverty level | 415 (37.3%) | 283 (46.0%) | 132 (26.5%) | |
| Unknown | 199 | 119 | 80 | |
| Ever homeless | 567 (43.8%) | 334 (46.1%) | 233 (40.8%) | 0.058 |
| Unknown | 16 | 9 | 7 | |
| Recent^ homeless | 137 (10.6%) | 83 (11.5%) | 54 (9.5%) | 0.200 |
| Unknown | 21 | 14 | 7 | |
| Food insecure | 518 (39.6%) | 312 (42.7%) | 206 (35.7%) | 0.010 |
| Unknown | 5 | 4 | 1 | |
| Ever incarceration | 202 (15.8%) | 143 (20.1%) | 59 (10.4%) | <0.001 |
| Unknown | 36 | 24 | 12 | |
| Recent incarceration | 23 (1.8%) | 15 (2.1%) | 8 (1.1%) | 0.4 |
| Unknown | 31 | 21 | 10 | |
| Ever sex work | 470 (36.4%) | 326 (45.0%) | 144 (25.4%) | <0.001 |
| Unknown | 21 | 10 | 11 | |
| Recent sex work (past 6 months) | 172 (13.4%) | 126 (17.5%) | 46 (8.2%) | <0.001 |
| Unknown | 30 | 16 | 14 | |
| Index of socio-structural vulnerability (lifetime), Median (IQR) | 2 (1–4) | 3 (1–4) | 2 (1–3) | <0.001 |
| Unknown | 274 | 162 | 112 | |
| Index of socio-structural vulnerability (recent), Median (IQR) | 1 (0–3) | 2 (1–3) | 1 (0–2) | <0.001 |
| Unknown | 281 | 167 | 114 | |
| Language and literacy | ||||
| Survey language | <0.001 | |||
| Spanish | 59 (4.5%) | 57 (7.8%) | 2 (0.3%) | |
| English | 1,253 (95.5%) | 677 (92.2%) | 576 (99.7%) | |
| Literacy screener+ | <0.001 | |||
| High | 1,212 (94.1%) | 634 (89.3%) | 578 (100%) | |
| Low | 76 (5.9%) | 76 (10.7%) | 0 (0%) | |
| Unknown | 24 | 24 | 0 | |
| Recruitment source 3 | ||||
| Friend | 356 (27.6%) | 235 (32.5%) | 121 (21.3%) | <0.001 |
| Healthcare organization or medical provider | 342 (26.5%) | 304 (42.0%) | 38 (6.7%) | <0.001 |
| Community organization | 85 (6.6%) | 71 (9.8%) | 14 (2.5%) | <0.001 |
| Flyer | 77 (6.0%) | 70 (9.7%) | 7 (1.2%) | <0.001 |
| 106 (8.2%) | 52 (7.2%) | 54 (9.5%) | 0.13 | |
| Study website | 101 (7.8%) | 57 (7.9%) | 44 (7.7%) | >0.9 |
| Dating app | 42 (3.2%) | 30 (4.1%) | 12 (2.1%) | 0.042 |
| Other website | 58 (4.5%) | 33 (4.5%) | 25 (4.4%) | 0.9 |
| Email or phone call from study staff | 128 (9.9%) | 123 (16.9%) | 5 (0.9%) | <0.001 |
| Other | 78 (6.0%) | 39 (5.4%) | 39 (6.8%) | 0.3 |
n (%); Median (IQR)
Pearson’s Chi-squared test; Wilcoxon rank sum test
Item was “select that apply” response
Digital participants did not receive the literacy screener; the literacy screener implemented in the site-based cohort was not validated for remote self-administration. We therefore were not able to screen for low literacy among digital-only participants.
Recent is past-3 months for the site-based mode and past 6-months for the digital mode
The overall cohort had a median age of 28 years (interquartile range=23–35), was diverse in geographic location (49.9% North, 22.9% Mid-Atlantic, 27.1% South), and in racial and ethnic identity (15.0% self-identified as Black, 13.3% Multiracial, 11.7% Another Race, 17.8% Latina/e/x). One quarter (27.9%) identified as female/women, and 23.3% were straight/heterosexual. Socio-structural vulnerabilities were high: 27.8% completed high school or less, 41.1% were unemployed, 37.3% lived below the federal poverty level, 43.8% were ever homeless (10.6% recent), 39.6% were food insecure, 15.8% were ever incarcerated (1.8% recent), and 36.4% ever engaged in sex work (13.4% recent). Peer referrals (27.6%) and healthcare organizations (26.5%) were common recruitment sources.
There were several statistically significant differences in demographic and socio-structural vulnerabilities by cohort mode. For example, the proportion identifying as people of color was higher in site-based than digital modes—Black (23.3% vs 4.5%), Multiracial (15.0% vs 11.2%), another race (16.2% vs 6.1%), and Latina/e/x (26.7% vs 6.6%)—whereas the reverse was found for White respondents (45.5% vs 78.2%) (all p<0.05). Further, lifetime and recent socio-structural vulnerability indexes were higher for the site-based vs digital mode (both p<0.001).
A sensitivity analysis restricted to participants from the 6 cities where site-based participants were enrolled (n=863) found similar results (Supplemental Table 1). For example, a significantly higher proportion of site-based participants (n=733) vs participants who chose exclusively digital participation from those same 6 cities (n=130) were: people of color (Black 23.3% vs 4.6%; Latina/e/x 26.7% vs 10.9%), had lower educational attainment (high school or less 35.6% vs 17.8%), were unemployed (46.0% vs 35.8%), had an income below the federal poverty level (46.0% vs 27.5%), had ever been homeless (46.1% vs 35.2%), were food insecure (42.7% vs 30.2%, had a lifetime history of incarceration (20.1% vs 8.9%), reported lifetime (45.0% vs 23.8%) and past 6-month (17.5% vs 8.1%) sex work, had higher index of lifetime (median: 3 vs 1) and recent (median: 2 vs 1) socioeconomic vulnerability, completed the survey in Spanish (7.8% vs 0.8%), and were recruited from a friend (32.5% vs 19.0%) or healthcare organization (42.0% vs 24.6%) (all p<0.05).
Healthcare Access, Healthcare Utilization, and Gender Affirmation (Table 2)
Table 2.
Healthcare Access, Healthcare Utilization, and Gender Affirmation Among Transgender Women in the LITE Cohort: Overall and Stratified by Site-based and Digital Modes.
| Characteristic | Overall, N = 1,3121 | Site-based, N = 7341 | Digital, N = 5781 | p-value2 |
|---|---|---|---|---|
|
| ||||
| Health insurance | <0.001 | |||
| Uninsured | 125 (10.2%) | 71 (10.4%) | 54 (9.8%) | |
| Public insurance | 500 (40.6%) | 353 (51.8%) | 147 (26.8%) | |
| Private insurance | 606 (49.2%) | 258 (37.8%) | 348 (63.4%) | |
| Unknown | 81 | 52 | 29 | |
| Regular healthcare provider | 919 (70.9%) | 557 (77.0%) | 362 (63.1%) | <0.001 |
| Unknown | 15 | 11 | 4 | |
| Primary location for healthcare services | <0.001 | |||
| Health department clinic | 168 (13.0%) | 113 (15.7%) | 55 (9.7%) | |
| Community health center | 365 (28.3%) | 271 (37.7%) | 94 (16.5%) | |
| Private doctor’s office | 372 (28.9%) | 137 (19.1%) | 235 (41.3%) | |
| Student health center | 39 (3.0%) | 10 (1.4%) | 29 (5.1%) | |
| Hospital emergency room | 131 (10.2%) | 103 (14.3%) | 28 (4.9%) | |
| Mobile health unit | 1 (0.1%) | 0 (0.0%) | 1 (0.2%) | |
| Chain of patient care and urgent care centers | 87 (6.8%) | 32 (4.5%) | 55 (9.7%) | |
| Some other place | 33 (2.6%) | 21 (2.9%) | 12 (2.1%) | |
| Don’t have regular source of healthcare | 60 (4.7%) | 16 (2.2%) | 44 (7.7%) | |
| There is no place | 32 (2.5%) | 16 (2.2%) | 16 (2.8%) | |
| Unknown | 24 | 15 | 9 | |
| Provider knowledge of trans health | <0.001 | |||
| Not knowledgeable | 162 (13.9%) | 45 (6.8%) | 117 (23.0%) | |
| Somewhat knowledgeable | 324 (27.7%) | 151 (22.9%) | 173 (34.1%) | |
| Very knowledgeable | 682 (58.4%) | 464 (70.3%) | 218 (42.9%) | |
| Unknown | 144 | 74 | 70 | |
| Barriers to healthcare 3 | ||||
| Time | 668 (51.5%) | 333 (45.9%) | 335 (58.5%) | <0.001 |
| Transportation | 520 (39.9%) | 299 (41.0%) | 221 (38.4%) | 0.30 |
| Safety | 270 (20.8%) | 151 (20.8%) | 119 (20.7%) | >0.90 |
| Childcare | 30 (2.3%) | 8 (1.1%) | 22 (3.8%) | 0.001 |
| Cost | 728 (55.8%) | 360 (49.3%) | 368 (64.1%) | <0.001 |
| No health insurance | 332 (25.6%) | 169 (23.3%) | 163 (28.4%) | 0.035 |
| Inconvenient hours | 420 (32.3%) | 191 (26.4%) | 229 (39.8%) | <0.001 |
| Mistreatment | 343 (26.6%) | 161 (22.2%) | 182 (32.2%) | <0.001 |
| Bad experience in past | 531 (41.1%) | 277 (38.3%) | 254 (44.6%) | 0.021 |
| You feel like healthcare providers are not comfortable caring for transgender patient | 518 (40.9%) | 236 (33.1%) | 282 (51.0%) | <0.001 |
| Other | 75 (5.9%) | 43 (6.0%) | 32 (5.8%) | 0.80 |
| Gender affirmation | ||||
| Age at first trans recognition, Median (IQR) | 13 (6–18) | 11 (5–17) | 14 (8–20) | <0.001 |
| Unknown | 3 | 1 | 2 | |
| Hormones | 1,085 (83.5%) | 646 (88.7%) | 439 (76.9%) | <0.001 |
| Unknown | 13 | 6 | 7 | |
| Age first accessed hormones, Median (IQR) | 23 (19–29) | 22 (19–28) | 24 (20–30) | 0.003 |
| Unknown | 290 | 139 | 151 | |
| Any gender-affirming surgery | 621 (47.3%) | 394 (53.7%) | 227 (39.3%) | <0.001 |
| Age first accessed surgery, Median (IQR) | 25 (21–32) | 25 (21–31) | 27 (22–34) | 0.015 |
| Unknown | 690 | 339 | 351 | |
| Silicone fillers | 101 (7.8%) | 89 (12.2%) | 12 (2.1%) | <0.001 |
| Unknown | 12 | 7 | 5 | |
| Tried to get gender transition-related care | 1,082 (83.4%) | 623 (85.8%) | 459 (80.4%) | 0.009 |
| Unknown | 15 | 8 | 7 | |
| Unmet need for gender-related care | 110 (10.3%) | 59 (9.6%) | 51 (11.3%) | 0.40 |
| Unknown | 244 | 119 | 125 | |
| Source where obtained hormones | ||||
| Parties | 10 (1.0%) | 6 (1.0%) | 4 (0.9%) | >0.90 |
| Unknown | 281 | 130 | 151 | |
| Prescription | 980 (95.3%) | 572 (95.2%) | 408 (95.6%) | 0.80 |
| Unknown | 284 | 133 | 151 | |
| Friend, lover, or family member | 74 (7.2%) | 47 (7.8%) | 27 (6.3%) | 0.40 |
| Unknown | 283 | 132 | 151 | |
| Street | 24 (2.3%) | 21 (3.5%) | 3 (0.7%) | 0.004 |
| Unknown | 283 | 131 | 152 | |
| Other | 1 (0.1%) | 0 (0.0%) | 1 (0.2%) | 0.40 |
| Unknown | 283 | 132 | 151 | |
| Internet | 74 (7.2%) | 27 (4.5%) | 47 (11.0%) | <0.001 |
| Unknown | 283 | 131 | 152 | |
| Legal gender affirmation (legal name and gender marker) | ||||
| IDs and records list preferred name | <0.001 | |||
| None of my IDs and records list the name I prefer | 563 (44.1%) | 265 (37.3%) | 298 (52.6%) | |
| Some, most, or all IDs and records list the name I prefer | 715 (55.9%) | 446 (62.7%) | 269 (47.4%) | |
| Unknown | 34 | 23 | 11 | |
| IDs and records list preferred gender | <0.001 | |||
| None of my IDs and records list the gender I prefer | 693 (54.0%) | 317 (44.4%) | 376 (66.0%) | |
| Some, most, or all IDs and records list gender I prefer | 591 (46.0%) | 397 (55.6%) | 194 (34.0%) | |
| Unknown | 28 | 20 | 8 | |
n (%)
Pearson’s Chi-squared test; Wilcoxon rank sum test
Item was “select that apply” response
Overall, 10.2% of the cohort did not have health insurance, 40.6% had public insurance, and 49.2% were privately insured. In the site-based mode, 10.4% were uninsured, 51.8% publicly insured, and 37.8% privately insured, compared to the digital mode where the insurance distribution was 9.8%, 26.8%, and 63.4%, respectively (p<0.01). The majority (70.9%) reported a regular healthcare provider, higher for site-based vs digital participants (77.0% vs 63.1%; p<0.001). Some barriers to care were lower in site-based than digital modes, such as cost (49.3% vs 64.1%), time (45.9% vs 58.5%), access to health insurance (23.3% vs 28.4%), inconvenient hours (26.4% vs 39.8%), mistreatment (22.2% vs 32.2%), and providers not being comfortable caring for trans patients (33.1% vs 51.0%) (all p<0.05).
A higher proportion of site-based than digital mode participants reported medical gender affirmation including hormones (88.7% vs 76.9%), surgery (53.7% vs 39.3%), and silicone fillers (12.2% vs 2.1%) (all p<0.05). The prevalence of street-obtained hormones was low overall, but higher for site-based than digital modes (3.5% vs 0.7%), while Internet-obtained hormone prevalence was higher (4.5% vs 11.0%) (all p<0.05). Site-based vs digital participants were younger at age of self-recognizing being trans (median age=11 vs 14 years), accessing hormones (median age=22 vs 24), and obtaining surgery (median age=25 vs 27) (all p<0.05). A higher proportion of site-based than digital mode participants attempted to get transition-related care (85.8% vs 80.4%; p=0.009), but no significant differences were observed in unmet need for gender-related care by mode. Trans women in the site-based mode had higher prevalence of legal affirmation than digital mode respondents for name (62.7% vs 47.4%) and gender (55.6% vs 34.0%) (both p<0.05).
Mental Health, Substance Use, Violence Victimization, Stigma, and Social Support (Table 3)
Table 3.
Mental Health, Substance Use, Violence Victimization, and Social Support Among Transgender Women in the LITE Cohort: Overall and Stratified by Site-based and Digital Modes.
| Characteristic | Overall, N = 1,3121 | Site-based, N = 7341 | Digital, N = 5781 | p-value2 |
|---|---|---|---|---|
|
| ||||
| Mental health | ||||
| Psychological distress (Kessler-6 summary score >= 13) | <0.001 | |||
| Yes | 778 (60.2%) | 485 (67.5%) | 293 (51.0%) | |
| No | 514 (39.8%) | 233 (32.5%) | 281 (49.0%) | |
| Unknown | 20 | 16 | 4 | |
| PTSD symptomatology (PC-PTSD summary score >= 3) | 0.002 | |||
| Yes | 670 (52.5%) | 404 (56.3%) | 266 (47.6%) | |
| No | 606 (47.5%) | 313 (43.7%) | 293 (52.4%) | |
| Unknown | 36 | 17 | 19 | |
| Substance use | ||||
| Hazardous drinking (AUDIT-C summary score >= 4) | >0.90 | |||
| Yes | 858 (67.0%) | 475 (67.0%) | 383 (67.0%) | |
| No | 423 (33.0%) | 234 (33.0%) | 189 (33.0%) | |
| Unknown | 31 | 25 | 6 | |
| Drug abuse (DAST-10 summary score >= 3) | 0.50 | |||
| Yes | 931 (71.8%) | 515 (71.0%) | 416 (72.7%) | |
| No | 366 (28.2%) | 210 (29.0%) | 156 (27.3%) | |
| Unknown | 15 | 9 | 6 | |
| Lifetime history of injection drug use | 46 (3.5%) | 26 (3.6%) | 20 (3.5%) | >0.90 |
| Unknown | 15 | 9 | 6 | |
| Violence victimization | ||||
| Childhood violence, ACES | 0.40 | |||
| Score of 0 | 110 (8.6%) | 62 (8.7%) | 48 (8.4%) | |
| Score of 1–3 | 593 (46.2%) | 318 (44.5%) | 275 (48.3%) | |
| Score of >= 4 | 580 (45.2%) | 334 (46.8%) | 246 (43.2%) | |
| Unknown | 29 | 20 | 9 | |
| Physical violence, lifetime | 843 (65.2%) | 457 (63.2%) | 386 (67.7%) | 0.091 |
| Unknown | 19 | 11 | 8 | |
| Sexual violence, lifetime | 557 (43.3%) | 319 (44.2%) | 238 (42.0%) | 0.40 |
| Unknown | 25 | 13 | 12 | |
| Psychological violence, lifetime | 1,101 (85.2%) | 594 (81.9%) | 507 (89.3%) | <0.001 |
| Unknown | 19 | 9 | 10 | |
| Stigma | ||||
| Anticipated discrimination, Median (IQR) | 22.00 (16.00, 27.00) | 21.00 (13.00, 27.00) | 24.00 (19.00, 28.00) | <0.001 |
| Unknown | 50 | 32 | 18 | |
| Social Support | ||||
| Multidimensional Perceived Social Support | 11 (7–16) | 12 (7–16) | 11 (7–16) | 0.30 |
| Unknown | 39 | 23 | 16 | |
| My family is accepting and supportive of my gender identity | <0.001 | |||
| Strongly disagree + disagree | 394 (30.9%) | 203 (28.4%) | 191 (34.1%) | |
| Neither agree nor disagree | 287 (22.5%) | 142 (19.9%) | 145 (25.9%) | |
| Agree + strongly agree | 594 (46.6%) | 370 (51.7%) | 224 (40.0%) | |
| Unknown | 37 | 19 | 18 | |
| How many different people do you know personally who are trans women? Median (IQR) | 5 (1–20) | 10 (3–40) | 3 (0–10) | <0.001 |
| Unknown | 17 | 2 | 15 | |
n (%)
Pearson’s Chi-squared test
The prevalence of mental health, substance use, and violence victimization were high in the whole cohort. However, site-based vs digital mode participants had higher prevalence of psychological distress (67.5% vs 51.0%; p<0.001) and PTSD symptoms (56.3% vs 47.6%; p=0.002), and lower prevalence of psychological violence (81.9% vs 89.3%; p<0.001). There were no statistically significant differences in hazardous alcohol use, drug abuse, injection drug use, ACES, physical violence, or sexual violence across participation modes.
Anticipated discrimination scores were lower for trans women in the site-based than digital mode (median=21 vs 24; p<0.001). There were no statistically significant differences in social support. However, there was significant heterogeneity in family acceptance of gender identity for site-based vs digital modes with 51.7% vs 40.0%, respectively, reporting they Agree/Strongly Agree that their family is accepting and supportive of their gender identity (p<0.001). Site-based participants reported knowing a greater number of trans women personally than those in the digital mode (median=10 vs 3; p<0.001).
Sexual Relationships, HIV Acquisition Risk Behaviors, Biomedical HIV Prevention, HIV Knowledge, Perceived HIV Risk, and STIs (Table 4)
Table 4.
Relationships and Sexual Partners, HIV Acquisition Risk Behaviors, Biomedical HIV Prevention, HIV Knowledge and Perceived HIV Risk, and Self-Reported STI Diagnosis Among Transgender Women in the LITE Cohort: Overall and Stratified by Site-based and Digital Modes.
| Characteristic | Overall, N = 1,3121 | Site-based, N = 7341 | Digital, N = 5781 | p-value2 |
|---|---|---|---|---|
|
| ||||
| Current relationship status | <0.001 | |||
| Single, not in a relationship | 542 (41.7%) | 328 (45.1%) | 214 (37.3%) | |
| Casually dating | 211 (16.2%) | 154 (21.2%) | 57 (9.9%) | |
| In a committed relationship | 350 (26.9%) | 158 (21.7%) | 192 (33.5%) | |
| Legally married or in a civil union | 120 (9.2%) | 45 (6.2%) | 75 (13.1%) | |
| Other relationship type | 78 (6.0%) | 43 (5.9%) | 35 (6.1%) | |
| Unknown | 11 | 6 | 5 | |
| Gender of sexual partner(s) | ||||
| Man (non-transgender) | 615 (64.2%) | 452 (78.1%) | 163 (43.0%) | <0.001 |
| Unknown | 354 | 155 | 199 | |
| Woman (non-transgender) | 329 (34.3%) | 142 (24.4%) | 187 (49.3%) | <0.001 |
| Unknown | 352 | 153 | 199 | |
| Transgender woman/Male-to-Female (MTF) | 230 (23.9%) | 111 (19.1%) | 119 (31.4%) | <0.001 |
| Unknown | 351 | 152 | 199 | |
| Transgender man/Female-to-Male (FTM) | 101 (10.5%) | 52 (8.9%) | 49 (13.0%) | 0.044 |
| Unknown | 354 | 152 | 202 | |
| Genderqueer/gender non-conforming (female at birth) | 142 (14.8%) | 72 (12.4%) | 70 (18.4%) | 0.010 |
| Unknown | 352 | 154 | 198 | |
| Genderqueer/gender non-conforming (male at birth) | 118 (12.3%) | 66 (11.4%) | 52 (13.7%) | 0.3 |
| Unknown | 352 | 154 | 198 | |
| Number of sexual partners, last 6 months | ||||
| # of sex partners in last 6 months, Median (IQR) | 1 (0–3) | 1 (0–4) | 1 (0–2) | <0.001 |
| Unknown | 6 | 2 | 4 | |
| Concurrent sexual partners | ||||
| More than one sexual partner concurrently | 458 (35.6%) | 282 (39.1%) | 176 (31.3%) | 0.004 |
| Unknown | 27 | 12 | 15 | |
| HIV risk behaviors | ||||
| Any vaginal sex | 68 (5.2%) | 35 (4.8%) | 33 (5.7%) | 0.4 |
| Unknown | 8 | 4 | 4 | |
| Condomless vaginal sex | 40 (3.1%) | 20 (2.7%) | 20 (3.5%) | 0.4 |
| Unknown | 10 | 4 | 6 | |
| Any anal sex | 583 (44.7%) | 390 (53.4%) | 193 (33.6%) | <0.001 |
| Unknown | 8 | 4 | 4 | |
| Condomless anal sex | 339 (26.1%) | 220 (30.4%) | 119 (20.7%) | <0.001 |
| Unknown | 15 | 11 | 4 | |
| History of needle sharing for drug injection | 25 (1.9%) | 12 (1.7%) | 13 (2.3%) | 0.4 |
| Unknown | 17 | 10 | 7 | |
| Where did you meet most recent regular partner? | <0.001 | |||
| Online dating apps (like Tinder, Grindr) | 219 (32.8%) | 123 (33.5%) | 96 (32.0%) | |
| Public spaces (park, public bathroom, street) | 56 (8.4%) | 41 (11.2%) | 15 (5.0%) | |
| Work or school | 129 (19.3%) | 52 (14.2%) | 77 (25.7%) | |
| Bar or club | 30 (4.5%) | 23 (6.3%) | 7 (2.3%) | |
| Party | 29 (4.3%) | 17 (4.6%) | 12 (4.0%) | |
| Hotel | 7 (1.0%) | 5 (1.4%) | 2 (0.7%) | |
| Other | 96 (14.4%) | 58 (15.8%) | 38 (12.7%) | |
| Other online (like Facebook, Backpage, Craigslist) | 101 (15.1%) | 48 (13.1%) | 53 (17.7%) | |
| Unknown | 645 | 367 | 278 | |
| Where did you meet your most recent sex work client? | 0.084 | |||
| Online dating apps (like Tinder, Grindr) | 59 (40.1%) | 43 (40.6%) | 16 (39.0%) | |
| Public spaces (park, public bathroom, street) | 20 (13.6%) | 18 (17.0%) | 2 (4.9%) | |
| Work or school | 4 (2.7%) | 1 (0.9%) | 3 (7.3%) | |
| Bar or club | 11 (7.5%) | 9 (8.5%) | 2 (4.9%) | |
| Party | 2 (1.4%) | 2 (1.9%) | 0 (0.0%) | |
| Hotel | 3 (2.0%) | 3 (2.8%) | 0 (0.0%) | |
| Other | 8 (5.4%) | 6 (5.7%) | 2 (4.9%) | |
| Other online (like Facebook, Backpage, Craigslist) | 40 (27.2%) | 24 (22.6%) | 16 (39.0%) | |
| Unknown | 1,165 | 628 | 537 | |
| PrEP awareness | ||||
| Have you ever heard about PrEP? | 997 (76.6%) | 612 (84.2%) | 385 (67.0%) | <0.001 |
| Unknown | 10 | 7 | 3 | |
| PrEP indication | ||||
| Indicated for PrEP based on trans women criteria | 619 (47.2%) | 398 (54.2%) | 221 (38.2%) | <0.001 |
| PrEP uptake | ||||
| Have you ever taken PrEP? | 248 (19.1%) | 209 (28.8%) | 39 (6.8%) | <0.001 |
| Unknown | 11 | 8 | 3 | |
| Are you currently taking PrEP (in the last 30 days)? | 147 (11.3%) | 123 (17.0%) | 24 (4.2%) | <0.001 |
| Unknown | 14 | 11 | 3 | |
| Reasons not used PrEP | ||||
| Side effects | 93 (37.8%) | 78 (37.7%) | 15 (38.5%) | >0.9 |
| People think you have HIV | 50 (20.3%) | 46 (22.2%) | 4 (10.3%) | 0.088 |
| Interaction with hormones | 23 (9.3%) | 19 (9.1%) | 4 (10.5%) | 0.8 |
| Don’t like taking a pill every day | 92 (37.7%) | 78 (37.9%) | 14 (36.8%) | >0.9 |
| People think you have a lot of different sex partners | 99 (40.4%) | 80 (38.8%) | 19 (48.7%) | 0.2 |
| Your sex partner(s) don’t want to use condoms because you’re on PrEP | 74 (30.6%) | 61 (29.9%) | 13 (34.2%) | 0.6 |
| You feel like you’re not at risk for HIV | 79 (32.8%) | 64 (31.5%) | 15 (39.5%) | 0.3 |
| Don’t like the clinical visits and testing that are required to stay on PrEP | 62 (25.6%) | 51 (25.0%) | 11 (28.9%) | 0.6 |
| Other experiences | 22 (9.1%) | 20 (9.8%) | 2 (5.3%) | 0.4 |
| PEP | ||||
| Heard of PEP | 749 (57.6%) | 469 (64.5%) | 280 (48.8%) | <0.001 |
| Unknown | 11 | 7 | 4 | |
| Ever taken PEP | 135 (10.4%) | 117 (16.1%) | 18 (3.1%) | <0.001 |
| Unknown | 13 | 9 | 4 | |
| HIV knowledge | <0.001 | |||
| 0–1 correct answers | 281 (21.5%) | 65 (8.9%) | 216 (37.5%) | |
| 2–5 correct answers | 1,028 (78.5%) | 668 (91.1%) | 360 (62.5%) | |
| Unknown | 3 | 1 | 2 | |
| Perceived HIV risk | ||||
| How high do you think your risk for HIV infection is? | <0.001 | |||
| No risk | 233 (23.8%) | 126 (20.3%) | 107 (29.9%) | |
| Low | 466 (47.6%) | 298 (48.1%) | 168 (46.9%) | |
| Medium/high risk | 279 (28.5%) | 196 (31.6%) | 83 (23.2%) | |
| Unknown | 334 | 114 | 220 | |
| STI diagnosis (self-reported) | ||||
| Have you tested positive for a STI in the last 6 months? | 45 (3.5%) | 35 (4.9%) | 10 (1.7%) | 0.002 |
| Unknown | 21 | 16 | 5 | |
n (%); Median (IQR)
Pearson’s Chi-squared test; Wilcoxon rank sum test; Fisher’s exact test
Item was “select that apply” response
Relationship status and gender of sexual partners differed by mode of participation. The proportion of participants with a cisgender man sexual partner was higher for site-based vs digital modes (78.1% vs 43.0%; p<0.001) but was lower for a cisgender woman sexual partner (24.4% vs 49.3%), transgender woman (19.1% vs 31.4%), transgender man (8.9% vs 13.0%), and nonbinary AFAB (12.4% vs 18.4%) partners (all p<0.05). Site-based compared to digital mode participants reported a higher number of sexual partners and higher prevalence of concurrent sex partners, any anal sex, and condomless anal sex (all p<0.05). No significant differences were found for vaginal sex, condomless vaginal sex, or needle-sharing. Site-based participants had higher levels of perceived HIV risk than digital mode participants (no HIV risk, low risk, medium/high risk: 20.3%, 48.1%, 31.6% site-based; 29.9%, 46.9%, 23.2% digital; p<0.001) and higher self-reported STI prevalence (4.9% vs 1.7%; p=0.002).
Site-based participants had higher awareness of and engagement with biomedical HIV prevention than digital mode participants, including PrEP awareness (84.2% vs 67.0%), lifetime PrEP uptake (28.8% vs 6.8%), PrEP in past 30 days (17.0% vs 4.2%), PEP awareness (64.5% vs 48.8%), and lifetime PEP uptake (16.1% vs 3.1%) (all p<0.05). There were no statistically significant differences in reasons why PrEP was not used across cohort modes.
Modeling Site-based Versus Digital Mode Participation
In a multivariable Model 1, unadjusted for recruitment source, variables independently associated with site-based vs digital mode participation were: residing in the South compared to the North (aPR=0.71; 95% CI=0.55–0.91); identifying as Black (aPR=1.49; 95% CI=1.14–1.96), Multiracial (aPR=1.30; 95% CI=1.00–1.69), or another race (aPR=1.30; 95% CI=1.00–1.71) compared to White; reporting use of gender-affirming hormones (aPR=1.34; 95% CI=1.01–1.79); having a cisgender man sexual partner (aPR=1.32; 95% CI=1.05–1.66); and reporting low compared to no HIV acquisition risk (aPR=1.39; 95% CI=1.12–1.72). No other variables in this model achieved statistical significance. Healthcare organization or medical provider recruitment source was added in Model 2 and was significant (aPR=1.76; 95% CI=1.46–2.12); residing in the South, gender-affirming hormones, and identifying as Multiracial or another race no longer reached statistical significance. Identifying as Black, having a cisgender man sexual partner, and perceived HIV acquisition risk remained associated with in-person vs online participation.
DISCUSSION
This study compared site-based to digital-only participants in the LITE cohort and found that site-based and digital modes engaged trans women with different baseline demographic characteristics, life experiences, and HIV acquisition risks. A higher proportion of site-based participants resided in the North, identified as a person of color, and reported greater HIV acquisition risk than digital mode participants. Before conducting the adjusted model, we also observed that site-based participants were more likely to have experienced socio-structural vulnerabilities, had a regular healthcare provider, accessed medical gender affirmation, endorsed poorer mental health, had experiences with biomedical HIV prevention, and indicated larger social network size. Findings shed light on the potential selection bias that would occur for studies that utilize an exclusively site-based or digital cohort mode for research with trans women. Non-coverage bias is also a potential issue, wherein some individuals may not be reached at all if only one mode is used. Mixed-mode approaches are increasingly used in national survey efforts (e.g., National Survey on Drug Use and Health56) to minimize response and measurement error in subpopulations. Findings suggest that hybrid studies combining site-based and digital modes may achieve a more diverse and potentially representative sample of trans women than either site-based or digital cohorts alone.
For the cohort, site-based data collection occurred at clinics selected with express missions to serve trans and economically marginalized communities (e.g., Medicaid patients), which may explain the high engagement in healthcare and gender-affirming care. Indeed, when adjusting for recruitment source by a healthcare organization or medical provider, the association of hormones with mode of participation was no longer statistically significant. Thus, findings also highlight the importance of appropriate selection of sites for HIV research. Site-based vs digital-only participants were more likely to report referral from a health facility provider, but there was no difference in whether people were recruited online between site-based vs digital-only modes. This may suggest that participants recruited online are more willing to either participate in-person at sites or digitally, whereas those recruited at facilities may be less likely to participate in a digital-only mode (potentially explained by non-coverage bias, trust and relationships with providers, convenience of accessing services, and/or lower technology access). Additional research is needed on experiences of participation across modes.
There were several characteristics with no observed differences among trans women by mode of cohort participation including self-reported childhood adversity, physical and sexual violence, hazardous drinking, drug misuse, and condomless vaginal sex. These seem to be ubiquitous experiences among trans women, non-differentiable by mode of participation. Given that some characteristics and health outcomes evidenced differences by mode and others did not, careful consideration is warranted concerning the research questions, exposures, and outcomes of interest when designing studies and in selecting site-based, digital-only, or hybrid cohort modes.
Digital vulnerabilities disproportionately impact historically marginalized populations,57–60 have resulted in selection bias in HIV research,61 and underscore the need for enhanced methodologies for digital cohorts in HIV research. Our study population most closely reflects the USA demographic distribution in terms of race and ethnicity when both modes are used, but less so with only one mode. Additionally, our finding that trans women identifying as Black, Multiracial, and Another Race were over-represented in the site-based but under-represented in the digital mode is particularly salient, and consistent with prior research suggesting that a digital divide persists among racial groups.35 Given stark racial inequities in the HIV epidemic for Black and Brown trans women,2 findings suggest that digital-only cohorts may not reach those women hardest-hit in the USA. Concerted efforts to reach trans women of color will be necessary if researchers rely on digital cohorts. Researchers may want to enroll hybrid cohorts, like this study, if a diverse sample of trans women is to be reached. Further, trans women in the site-based mode had high socio-structural vulnerabilities. There is a risk for exclusively digital research to reproduce or potentially worsen HIV and related health inequities, particularly for socioeconomically marginalized trans women. Thus, it is important to take care so as not to conduct research that reproduces or sustains the digital divide and/or exacerbates disparities in digital access by social determinants of health.
Study limitations include not having measured consistency of technology access for participants at baseline, and sampling in the eastern and southern USA which may limit generalizability to the West or Midwest. Self-reported STIs are described because only site-based participants received STI testing; digital mode participants were not due to cost. This was a study of HIV incidence and therefore restricted to trans women without HIV; replication of findings is needed for HIV care cohorts. Despite these limitations, our study has strengths including utilizing a hybrid design with site-based technology-enhanced and exclusively digital modes which allowed us to capture comparable data, examine implementation differences in modes, and observe potential selection biases. Further, the sensitivity analysis comparing site-based and online-only participants restricted to 6 cities where site-based participants resided, strengthens the inference that observed differences were due to mode of participation, rather than to inherent differences due to comparing individuals from 6 major cities to those from 72 other cities with similar demographic characteristics. Findings from this study may apply to other HIV key population groups and warrant additional research.
Effective and acceptable research methods that support recruitment and retention for trans women are vital to HIV research participation. Findings suggest that mode of cohort participation has implications for sampling trans women populations in HIV research. Our findings may be generalizable to other digital HIV research, given the association between community-level poverty rates–which likely impact consistent access to technology and internet–and HIV diagnoses.62 Utilizing only a site-based technology-enhanced cohort or a strictly remote digital cohort will have both strengths and limitations for HIV research. A hybrid cohort methodology may circumvent some of these but may also introduce other challenges such as funding for multi-modal infrastructures. Identifying potential sources of bias and ways these may impact research findings represents an important future direction for observational and interventional studies with trans women.
Supplementary Material
Table 5.
Modeling Site-based vs Digital Mode Participation Among Transgender Women in the LITE Cohort.
| Unadjusted | Adjusted Model 1 | Adjusted Model 2 | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Characteristic | PR1 | 95% CI1 | p-value | PR1 | 95% CI1 | p-value | PR1 | 95% CI1 | p-value |
| Age group | |||||||||
| 18–24 | 1.00 | — | 1.00 | — | 1.00 | — | |||
| 25–29 | 1.17 | 0.96, 1.43 | 0.12 | 1.00 | 0.79, 1.26 | >0.9 | 1.01 | 0.80, 1.28 | >0.9 |
| 30–39 | 1.07 | 0.88, 1.30 | 0.5 | 0.88 | 0.69, 1.13 | 0.3 | 0.90 | 0.70, 1.15 | 0.4 |
| 40+ | 1.34 | 1.09, 1.64 | 0.006 | 1.14 | 0.86, 1.51 | 0.4 | 1.14 | 0.86, 1.51 | 0.4 |
| Race | |||||||||
| White | 1.00 | — | 1.00 | — | 1.00 | — | |||
| Black | 2.04 | 1.70, 2.46 | <0.001 | 1.49 | 1.14, 1.96 | 0.003 | 1.34 | 1.02, 1.76 | 0.038 |
| Multiracial | 1.48 | 1.19, 1.83 | <0.001 | 1.30 | 1.00, 1.69 | 0.054 | 1.21 | 0.93, 1.59 | 0.2 |
| Other | 1.82 | 1.48, 2.24 | <0.001 | 1.30 | 1.00, 1.71 | 0.054 | 1.30 | 0.99, 1.70 | 0.060 |
| Ethnicity | |||||||||
| Not Latinx | 1.00 | — | 1.00 | — | 1.00 | — | |||
| Latinx | 1.68 | 1.43, 1.99 | <0.001 | 1.26 | 0.98, 1.63 | 0.070 | 1.22 | 0.94, 1.58 | 0.13 |
| Geographic region | |||||||||
| North | 1.00 | — | 1.00 | — | 1.00 | — | |||
| Mid-Atlantic | 1.11 | 0.93, 1.31 | 0.3 | 1.01 | 0.82, 1.24 | >0.9 | 1.10 | 0.89, 1.35 | 0.4 |
| South | 0.76 | 0.63, 0.92 | 0.004 | 0.71 | 0.55, 0.91 | 0.006 | 0.84 | 0.65, 1.08 | 0.2 |
| Educational attainment | |||||||||
| High school or less | 1.00 | — | 1.00 | — | 1.00 | — | |||
| Some college or higher | 0.70 | 0.60, 0.81 | <0.001 | 0.93 | 0.75, 1.14 | 0.5 | 0.92 | 0.75, 1.14 | 0.5 |
| Employment | |||||||||
| Not employed | 1.00 | — | 1.00 | — | 1.00 | — | |||
| Employed (full-time or part-time) | 0.82 | 0.71, 0.95 | 0.007 | 0.95 | 0.79, 1.15 | 0.6 | 0.96 | 0.80, 1.16 | 0.7 |
| Health insurance | |||||||||
| Private insurance | 1.00 | — | 1.00 | — | 1.00 | — | |||
| Uninsured | 1.33 | 1.03, 1.73 | 0.031 | 1.27 | 0.93, 1.74 | 0.13 | 1.23 | 0.90, 1.69 | 0.2 |
| Public insurance | 1.66 | 1.41, 1.95 | <0.001 | 1.13 | 0.90, 1.40 | 0.3 | 1.12 | 0.90, 1.39 | 0.3 |
| Regular healthcare provider | |||||||||
| No | 1.00 | — | 1.00 | — | 1.00 | — | |||
| Yes | 1.38 | 1.16, 1.64 | <0.001 | 1.15 | 0.92, 1.44 | 0.2 | 1.04 | 0.83, 1.31 | 0.7 |
| Ever sex work | |||||||||
| No | 1.00 | — | 1.00 | — | 1.00 | — | |||
| Yes | 1.43 | 1.24, 1.66 | <0.001 | 1.09 | 0.89, 1.33 | 0.4 | 1.08 | 0.89, 1.32 | 0.4 |
| Food insecure | |||||||||
| No | 1.00 | — | 1.00 | — | 1.00 | — | |||
| Yes | 1.14 | 0.98, 1.32 | 0.086 | 0.91 | 0.75, 1.10 | 0.3 | 0.93 | 0.77, 1.12 | 0.4 |
| Ever incarceration | |||||||||
| No | 1.00 | — | 1.00 | — | 1.00 | — | |||
| Yes | 1.34 | 1.12, 1.61 | 0.002 | 1.09 | 0.87, 1.37 | 0.5 | 1.08 | 0.86, 1.36 | 0.5 |
| Survey language | |||||||||
| English | 1.00 | — | 1.00 | — | 1.00 | — | |||
| Spanish | 1.79 | 1.36, 2.34 | <0.001 | 1.17 | 0.79, 1.75 | 0.4 | 1.27 | 0.85, 1.89 | 0.2 |
| Legal gender affirmation (IDs list preferred name) | |||||||||
| None of my IDs and records list the name I prefer | 1.00 | — | 1.00 | — | 1.00 | — | |||
| Some, most, or all IDs and records list the name I prefer | 1.33 | 1.14, 1.54 | <0.001 | 1.05 | 0.87, 1.26 | 0.6 | 1.03 | 0.86, 1.23 | 0.8 |
| Hormones for medical gender affirmation | |||||||||
| No | 1.00 | — | 1.00 | — | 1.00 | — | |||
| Yes | 1.55 | 1.23, 1.96 | <0.001 | 1.34 | 1.01, 1.79 | 0.044 | 1.25 | 0.94, 1.67 | 0.13 |
| Psychological distress (Kessler-6 summary score >= 13) | |||||||||
| Score < 13 | 1.00 | — | 1.00 | — | 1.00 | — | |||
| Score >= 13 | 0.73 | 0.62, 0.85 | <0.001 | 0.87 | 0.71, 1.07 | 0.2 | 0.84 | 0.69, 1.03 | 0.10 |
| PTSD symptomatology (PC-PTSD summary score >= 3) | |||||||||
| Score < 3 | 1.00 | — | 1.00 | — | 1.00 | — | |||
| Score >= 3 | 0.86 | 0.74, 0.99 | 0.040 | 0.95 | 0.78, 1.15 | 0.6 | 1.01 | 0.83, 1.22 | >0.9 |
| Psychological violence | |||||||||
| No | 1.00 | — | 1.00 | — | 1.00 | — | |||
| Yes | 0.79 | 0.65, 0.96 | 0.015 | 0.97 | 0.76, 1.24 | 0.8 | 0.98 | 0.77, 1.26 | 0.9 |
| Anticipated discrimination | 0.98 | 0.97, 0.99 | <0.001 | 1.00 | 0.99, 1.01 | >0.9 | 1.00 | 0.99, 1.01 | 0.7 |
| How many different people do you know personally who are trans women? | 1.00 | 1.00, 1.00 | 0.015 | 1.00 | 1.00, 1.00 | 0.5 | 1.00 | 1.00, 1.00 | 0.3 |
| Man (non-transgender) sexual partner | |||||||||
| No | 1.00 | — | 1.00 | — | 1.00 | — | |||
| Yes | 1.83 | 1.57, 2.12 | <0.001 | 1.32 | 1.05, 1.66 | 0.016 | 1.27 | 1.01, 1.60 | 0.042 |
| Woman (non-transgender) sexual partner | |||||||||
| No | 1.00 | — | 1.00 | — | 1.00 | — | |||
| Yes | 0.71 | 0.59, 0.86 | <0.001 | 0.87 | 0.69, 1.09 | 0.2 | 0.87 | 0.69, 1.09 | 0.2 |
| Indicated for PrEP based on suggested criteria for trans women | |||||||||
| No | 1.00 | — | 1.00 | — | 1.00 | — | |||
| Yes | 1.33 | 1.15, 1.53 | <0.001 | 0.89 | 0.72, 1.09 | 0.3 | 0.90 | 0.73, 1.10 | 0.3 |
| How high do you think your risk for HIV infection is? | |||||||||
| No risk | 1.00 | — | 1.00 | — | 1.00 | — | |||
| Low | 1.57 | 1.32, 1.88 | <0.001 | 1.39 | 1.12, 1.72 | 0.003 | 1.42 | 1.14, 1.75 | 0.001 |
| Medium/high risk | 1.73 | 1.42, 2.10 | <0.001 | 1.28 | 0.98, 1.67 | 0.073 | 1.30 | 1.00, 1.70 | 0.053 |
| Healthcare organization or medical provider recruitment source | |||||||||
| No | 1.00 | — | — | — | 1.00 | — | |||
| Yes | 2.01 | 1.73, 2.33 | <0.001 | — | — | 1.76 | 1.46, 2.12 | <0.001 | |
PR = Prevalence Ratio, CI = Confidence Interval
Note: Variables selected for modeling were: Geographic region, age group, race, ethnicity, educational attainment, employment, ever sex work, food insecure, ever incarceration, survey language, health insurance, regular healthcare provider, hormones, IDs list preferred name, psychological distress, PTSD symptomatology, psychological violence, anticipated discrimination, number of trans women personally know, cisgender man sexual partner, cisgender woman sexual partner, PrEP indicated, and perceived HIV acquisition risk, and healthcare organization or medical provider recruitment source. Model 1 was fit without recruitment source. Model 2 included recruitment source.
Acknowledgements
The authors would like to express their gratitude to the trans women who took part in this study. This study would not have been possible without their participation. We appreciate the continued involvement and contributions by the Community Advisory Board that supports and guides this study. Research reported in this publication was jointly supported by the National Institute of Allergy and Infectious Diseases, National Institute of Mental Health, and National Institute of Child Health and Human Development of the National Institutes of Health under Award Number UG3/UH3AI133669 (Wirtz/Reisner). Research reported in this publication was also supported by HIV/AIDS, Hepatitis, STD, and TB Administration (HAHSTA), Washington DC Department of Health. The LITE study is also appreciative of support from the CFAR at partner institutions, including JHU (P30AI094189), Emory University (P30AI050409), Harvard University (P30AI060354), DC CFAR (AI117970), and the University of Miami (P30AI073961). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or HAHSTA.
Funding
Research reported in this publication was supported by National Institute of Allergy and Infectious Diseases of the National Institutes of Health (NIH) under award number UH3 AI33669. The content is solely the responsibility of the authors and does not necessarily represent the official reviews of the NIH.
Footnotes
Conflicts of Interest – AW and TP receive separate research funding from ViiV Healthcare to their institution.
REFERENCES
- 1.Becasen JS, Denard CL, Mullins MM, Higa DH, Sipe TA. Estimating the Prevalence of HIV and Sexual Behaviors Among the US Transgender Population: A Systematic Review and Meta-Analysis, 2006–2017. Am J Public Health. 2019;109(1):e1–e8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Centers for Disease Control and Prevention (CDC). HIV Infection, Risk, Prevention, and Testing Behaviors Among Transgender Women—National HIV Behavioral Surveillance, 7 U.S. Cities, 2019–2020. Atlanta, GA: CDC;2021. [Google Scholar]
- 3.Fauci AS, Redfield RR, Sigounas G, Weahkee MD, Giroir BP. Ending the HIV Epidemic: A Plan for the United States. JAMA. 2019;321(9):844–845. [DOI] [PubMed] [Google Scholar]
- 4.Eisinger RW, Fauci AS. Ending the Global HIV Epidemic Begins at the Individual National Level: An Update from the United States. China CDC Wkly. 2020;2(39):764–766. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.The White House. National HIV/AIDS Strategy for the United States 2022–2025. Washington, DC: 2021. [Google Scholar]
- 6.Poteat T, Scheim A, Xavier J, Reisner S, Baral S. Global Epidemiology of HIV Infection and Related Syndemics Affecting Transgender People. J Acquir Immune Defic Syndr. 2016;72 Suppl 3(Suppl 3):S210–219. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Lee K, Trujillo L, Olansky E, et al. Factors Associated with Use of HIV Prevention and Health Care Among Transgender Women - Seven Urban Areas, 2019–2020. MMWR Morb Mortal Wkly Rep. 2022;71(20):673–679. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.James SE, Herman JL, Rankin S, Keisling M, Mottet L, Anafi M. The Report of the 2015 U.S. Transgender Survey. Washington, DC: National Center for Transgender Equality;2016. [Google Scholar]
- 9.Frank J, Restar A, Kuhns L, et al. Unmet Health Care Needs Among Young Transgender Women at Risk for HIV Transmission and Acquisition in Two Urban U.S. Cities: The LifeSkills Study. Transgend Health. 2019;4(1):1–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Safer JD, Coleman E, Feldman J, et al. Barriers to healthcare for transgender individuals. Curr Opin Endocrinol Diabetes Obes. 2016;23(2):168–171. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Sevelius JM. Gender Affirmation: A Framework for Conceptualizing Risk Behavior among Transgender Women of Color. Sex Roles. 2013;68(11–12):675–689. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Storholm ED, Huang W, Ogunbajo A, et al. Gender-Based Violence and Post-traumatic Stress Disorder Symptoms Predict HIV PrEP Uptake and Persistence Failure Among Transgender and Non-binary Persons Participating in a PrEP Demonstration Project in Southern California. AIDS Behav. 2023;27(2):745–759. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Reisner SL, Poteat T, Keatley J, et al. Global health burden and needs of transgender populations: a review. Lancet. 2016;388(10042):412–436. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Rich AJ, Scheim AI, Koehoorn M, Poteat T. Non-HIV chronic disease burden among transgender populations globally: A systematic review and narrative synthesis. Prev Med Rep. 2020;20:101259. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Peitzmeier SM, Malik M, Kattari SK, et al. Intimate Partner Violence in Transgender Populations: Systematic Review and Meta-analysis of Prevalence and Correlates. Am J Public Health. 2020;110(9):e1–e14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Johns MM, Lowry R, Andrzejewski J, et al. Transgender Identity and Experiences of Violence Victimization, Substance Use, Suicide Risk, and Sexual Risk Behaviors Among High School Students - 19 States and Large Urban School Districts, 2017. MMWR Morb Mortal Wkly Rep. 2019;68(3):67–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.White Hughto JM, Reisner SL, Pachankis JE. Transgender stigma and health: A critical review of stigma determinants, mechanisms, and interventions. Soc Sci Med. 2015;147:222–231. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Wilson EC, Chen YH, Arayasirikul S, et al. Differential HIV risk for racial/ethnic minority trans*female youths and socioeconomic disparities in housing, residential stability, and education. Am J Public Health. 2015;105 Suppl 3(Suppl 3):e41–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Ezell JM, Ferreira MJ, Duncan DT, Schneider JA. The Social and Sexual Networks of Black Transgender Women and Black Men Who Have Sex with Men: Results from a Representative Sample. Transgend Health. 2018;3(1):201–209. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Malone J, Reisner SL, Cooney EE, et al. Perceived HIV Acquisition Risk and Low Uptake of PrEP Among a Cohort of Transgender Women With PrEP Indication in the Eastern and Southern United States. J Acquir Immune Defic Syndr. 2021;88(1):10–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Teng F, Sha Y, Fletcher LM, Welsch M, Burns P, Tang W. Barriers to uptake of PrEP across the continuum among transgender women: A global scoping review. Int J STD AIDS. 2023:9564624231152781. [DOI] [PubMed] [Google Scholar]
- 22.Allan-Blitz LT, Menza TW, Cummings V, Gaydos CA, Wilton L, Mayer KH. Differing Correlates of Incident Bacterial Sexually Transmitted Infections Among a Cohort of Black Cisgender Men Who Have Sex With Men and Transgender Women Recruited in 6 US Cities (HIV Prevention Trials Network 061). Sex Transm Dis. 2022;49(7):e79–e84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Van Gerwen OT, Jani A, Long DM, Austin EL, Musgrove K, Muzny CA. Prevalence of Sexually Transmitted Infections and Human Immunodeficiency Virus in Transgender Persons: A Systematic Review. Transgend Health. 2020;5(2):90–103. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Poteat T, White RH, Footer KHA, et al. Characterising HIV and STIs among transgender female sex workers: a longitudinal analysis. Sex Transm Infect. 2021;97(3):226–231. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Reisner SL, Chaudhry A, Cooney E, Garrison-Desany H, Juarez-Chavez E, Wirtz AL. ‘It all dials back to safety’: A qualitative study of social and economic vulnerabilities among transgender women participating in HIV research in the USA. BMJ Open. 2020;10(1):e029852. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Barkan SE, Melnick SL, Preston-Martin S, et al. The Women’s Interagency HIV Study. WIHS Collaborative Study Group. Epidemiology. 1998;9(2):117–125. [PubMed] [Google Scholar]
- 27.Hessol NA, Kalinowski A, Benning L, et al. Mortality among participants in the Multicenter AIDS Cohort Study and the Women’s Interagency HIV Study. Clin Infect Dis. 2007;44(2):287–294. [DOI] [PubMed] [Google Scholar]
- 28.Kaslow RA, Ostrow DG, Detels R, Phair JP, Polk BF, Rinaldo CR Jr. The Multicenter AIDS Cohort Study: rationale, organization, and selected characteristics of the participants. Am J Epidemiol. 1987;126(2):310–318. [DOI] [PubMed] [Google Scholar]
- 29.Nash D, Stief M, MacCrate C, et al. A Web-Based Study of HIV Prevention in the Era of Pre-Exposure Prophylaxis Among Vulnerable HIV-Negative Gay and Bisexual Men, Transmen, and Transwomen Who Have Sex With Men: Protocol for an Observational Cohort Study. JMIR Res Protoc. 2019;8(9):e13715. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Rendina HJ, Talan AJ, Tavella NF, et al. Leveraging Technology to Blend Large-Scale Epidemiologic Surveillance With Social and Behavioral Science Methods: Successes, Challenges, and Lessons Learned Implementing the UNITE Longitudinal Cohort Study of HIV Risk Factors Among Sexual Minority Men in the United States. Am J Epidemiol. 2021;190(4):681–695. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.de Leeuw ED. To mix or not to mix data collection models in surveys. Journal of Official Statistics. 2005;21(2):233–255. [Google Scholar]
- 32.Tourangeau R, Conrad FG, Couper MP. Measurement Error on the Web and in Other Modes of Data Collection. In: The Science of Web Surveys. New York, NY: Oxford University Press; 2013:129–150. [Google Scholar]
- 33.Hox JJ, De Leeuw ED, Zijlmans EA. Measurement equivalence in mixed mode surveys. Front Psychol. 2015;6:87. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.deLeeuw ED. Mixed-mode: Past, present, and future. Surv Res Methods. 2018;12(2):75–89. [Google Scholar]
- 35.Reisner SL, Conron K, Scout N, Mimiaga MJ, Haneuse S, Austin SB. Comparing In-Person and Online Survey Respondents in the U.S. National Transgender Discrimination Survey: Implications for Transgender Health Research. LGBT Health. 2014;1(2):98–106. [DOI] [PubMed] [Google Scholar]
- 36.Reisner SL, Hughto JMW. Comparing the health of non-binary and binary transgender adults in a statewide non-probability sample. PLoS One. 2019;14(8):e0221583. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.McCambridge J, Witton J, Elbourne DR. Systematic review of the Hawthorne effect: new concepts are needed to study research participation effects. J Clin Epidemiol. 2014;67(3):267–277. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Wirtz AL, Poteat T, Radix A, et al. American Cohort to Study HIV Acquisition Among Transgender Women in High-Risk Areas (The LITE Study): Protocol for a Multisite Prospective Cohort Study in the Eastern and Southern United States. JMIR Res Protoc. 2019;8(10):e14704. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Wirtz AL, Cooney EE, Stevenson M, et al. Digital Epidemiologic Research on Multilevel Risks for HIV Acquisition and Other Health Outcomes Among Transgender Women in Eastern and Southern United States: Protocol for an Online Cohort. JMIR Res Protoc. 2021;10(4):e29152. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Wirtz AL, Humes E, Althoff KN, et al. HIV incidence and mortality in transgender women in the eastern and southern USA: a multisite cohort study. Lancet HIV. 2023;10(5):e308–e319. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Wirtz AL, Cooney EE, Chaudhry A, Reisner SL, American Cohort To Study HIVAATW. Computer-Mediated Communication to Facilitate Synchronous Online Focus Group Discussions: Feasibility Study for Qualitative HIV Research Among Transgender Women Across the United States. J Med Internet Res. 2019;21(3):e12569. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Scheim AI, Appenroth MN, Beckham SW, et al. Transgender HIV research: nothing about us without us. Lancet HIV. 2019;6(9):e566–e567. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Adams N, Pearce R, Veale J, et al. Guidance and Ethical Considerations for Undertaking Transgender Health Research and Institutional Review Boards Adjudicating this Research. Transgend Health. 2017;2(1):165–175. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Tate CC, Ledbetter JN, Youssef CP. A two-question method for assessing gender categories in the social and medical sciences. J Sex Res. 2013;50(8):767–776. [DOI] [PubMed] [Google Scholar]
- 45.Department of Health and Human Services (HHS). 2018. Poverty Guidelines. Federal Register. https://aspe.hhs.gov/topics/poverty-economic-mobility/poverty-guidelines/prior-hhs-poverty-guidelines-federal-register-references/2018-poverty-guidelines. Published 2018. Accessed April 5, 2021. [Google Scholar]
- 46.United States Department of Agriculture (USDA). U.S. Household Food Security Survey Module: Three-Stage Design, With Screeners. Economic Research Service, USDA. https://www.ers.usda.gov/media/8271/hh2012.pdf. Published 2012. Accessed April 5, 2021. [Google Scholar]
- 47.Bass PF 3rd, Wilson JF, Griffith CH. A shortened instrument for literacy screening. J Gen Intern Med. 2003;18(12):1036–1038. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Kessler RC, Andrews G, Colpe LJ, et al. Short screening scales to monitor population prevalences and trends in non-specific psychological distress. Psychol Med. 2002;32(6):959–976. [DOI] [PubMed] [Google Scholar]
- 49.Prins A, Bovin MJ, Smolenski DJ, et al. The Primary Care PTSD Screen for DSM-5 (PC-PTSD-5): Development and Evaluation Within a Veteran Primary Care Sample. J Gen Intern Med. 2016;31(10):1206–1211. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Saunders JB, Aasland OG, Babor TF, de la Fuente JR, Grant M. Development of the Alcohol Use Disorders Identification Test (AUDIT): WHO Collaborative Project on Early Detection of Persons with Harmful Alcohol Consumption--II. Addiction. 1993;88(6):791–804. [DOI] [PubMed] [Google Scholar]
- 51.Skinner HA. The drug abuse screening test. Addict Behav. 1982;7(4):363–371. [DOI] [PubMed] [Google Scholar]
- 52.Felitti VJ, Anda RF, Nordenberg D, et al. Relationship of childhood abuse and household dysfunction to many of the leading causes of death in adults. The Adverse Childhood Experiences (ACE) Study. Am J Prev Med. 1998;14(4):245–258. [DOI] [PubMed] [Google Scholar]
- 53.World Health Organization (WHO). WHO multi-country study on women’s health and domestic violence against women: Initial results on prevalence, health outcomes and women’s responses. https://apps.who.int/iris/handle/10665/43309. Published 2005. Accessed April 4, 2021. [Google Scholar]
- 54.Scheim AI, Bauer GR. The Intersectional Discrimination Index: Development and validation of measures of self-reported enacted and anticipated discrimination for intercategorical analysis. Soc Sci Med. 2019;226:225–235. [DOI] [PubMed] [Google Scholar]
- 55.Zimet GD, Powell SS, Farley GK, Werkman S, Berkoff KA. Psychometric characteristics of the Multidimensional Scale of Perceived Social Support. J Pers Assess. 1990;55(3–4):610–617. [DOI] [PubMed] [Google Scholar]
- 56.Substance Use and Mental Health Services Administration (SAMHSA). National Survey on Drug Use and Health. https://nsduhweb.rti.org/respweb/homepage.cfm. Published 2023. Accessed March 27, 2023. [Google Scholar]
- 57.Friedline T, Chen Z. Digital Redlining and the Fintech Marketplace: Evidence from U.S. Zip Codes. Journal of Consumer Affairs. 2021;55(2):366–388. [Google Scholar]
- 58.Sevelius JM, Gutierrez-Mock L, Zamudio-Haas S, et al. Research with Marginalized Communities: Challenges to Continuity During the COVID-19 Pandemic. AIDS Behav. 2020;24(7):2009–2012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 59.Pew Research Center. Home Broadband Adoption, Computer Ownership Vary by Race, Ethnicity in the U.S. https://www.pewresearch.org/short-reads/2021/07/16/home-broadband-adoption-computer-ownership-vary-by-race-ethnicity-in-the-u-s/. Published July 2021. Accessed July 4, 2024. [Google Scholar]
- 60.Pew Research Center. Digital Divide Persists Even as Americans with Lower Incomes Make Gains in Tech Adoption. https://www.pewresearch.org/short-reads/2021/06/22/digital-divide-persists-even-as-americans-with-lower-incomes-make-gains-in-tech-adoption/. Published June 2021. Accessed July 4, 2024. [Google Scholar]
- 61.Wirtz AL, Logie CH, Mbuagbaw L. Addressing Health Inequities in Digital Clinical Trials: A Review of Challenges and Solutions From the Field of HIV Research. Epidemiol Rev. 2022;44(1):87–109. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 62.Wiewel EW, Bocour A, Kersanke LS, Bodach SD, Xia Q, Braunstein SL. The association between neighborhood poverty and HIV diagnoses among males and females in New York City, 2010–2011. Public Health Rep. 2016;131(2):290–302. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.