Sexual Behaviors Associated with HIV Transmission Among Transgender and Gender Diverse Young Adults: The Intersectional Role of Racism and Transphobia

Abstract

HIV prevalence and engagement in sexual behaviors associated with HIV transmission are high among transgender people of color. Per intersectionality, this disproportionate burden may be related to both interpersonal and structural racism and transphobia. The goal of this study was to estimate the association between interpersonal and structural discrimination and sexual behaviors among transgender and gender diverse (TGD) U.S. young adults. We used logit models with robust standard errors to estimate the individual and combined association between interpersonal and structural racism and transphobia and sexual behaviors in a national online sample of TGD young adults of color (TYAOC) aged 18–30 years (N = 228). Racism was measured at the interpersonal and structural level using the Everyday Discrimination Scale and State Racism Index, respectively. Transphobia was measured at the interpersonal and structural level using the Gender Minority Stress Scale and the Gender Identity Tally, respectively. We found that interpersonal racism was associated with transactional sex, and interpersonal transphobia was associated with alcohol/drug consumption prior to sex and transactional sex among TYAOC. We also found evidence of a strong joint association of interpersonal and structural racism and transphobia with alcohol/drug consumption prior to sex (OR 3.85, 95% CI 2.12, 7.01) and transactional sex (OR 3.54, 95% CI 0.99, 12.59) among TYAOC. Racism and transphobia have a compounding impact on sexual behaviors among TYAOC. Targeted interventions that reduce discrimination at both the interpersonal and structural level may help reduce the HIV burden in this marginalized population.

Introduction

HIV prevalence is disproportionately high among transgender individuals in the U.S [1, 2]. A recent metaanalysis estimated that the prevalence of self‑reported HIV infection among transgender individuals in the U.S. was 16.1% overall, including 21.0% among transgender women and 1.2% among transgender men [2], compared to 0.4% in the population overall [1]. The same meta‑analysis also documented severe ethnoracial inequities, reporting that HIV prevalence among Black and Hispanic transgender women was 44.2% and 25.8%, respectively [2]. HIV prevalence and engagement in sexual behaviors associated with HIV transmission among transgender individuals is shaped by societal factors at the individual, interpersonal, institutional, community, and structural levels [3]. In particular, interpersonal (i.e., person‑to‑person) and structural (e.g., laws, policies, regulations, norms, governance practices) discrimination limit access to social, economic, and healthcare resources and in turn contribute to high rates of HIV in this marginalized population [3].

A few studies have evaluated the relationship between discrimination and sexual behaviors among transgender individuals. These studies have been limited by focusing on a narrow set of sexual behaviors and/or are limited to transbinary gender identities (i.e., transgender women and transgender men). Most of these studies have centered transgender women and demonstrated the association between interpersonal transphobia and unprotected receptive anal intercourse [4, 5] or unprotected sex with a primary male partner [6]. By focusing exclusively on receptive anal intercourse and placing constraints on the gender identity of partners, these studies implicitly applied inappropriate assumptions about the anatomy and sexual practices of transgender individuals that do not capture the full diversity of transgender experiences. Additionally, by restricting exposure measurement to interpersonal measures of transphobia, prior studies neglect the impact of other axes of discrimination, such as racism [7]. These additional axes may have distinct or compounding effects, so accounting for them is necessary for a complete assessment of the relationship of discrimination and sexual behaviors. Two studies have integrated interpersonal racism and transphobia in the context of sexual behaviors [8] and HIV prevalence [9]. However, even these studies are incomplete because they excluded measures of structural discrimination.

The goal of this study is to extend the current literature by using an intersectional approach to analyze the relationship between racism and transphobia and sexual behaviors associated with HIV transmission. Intersectionality is a theoretical framework and analytical tool pioneered by Kimberlé Crenshaw and Patricia Hill Collins that centers multiply marginalized individuals and emphasizes how co‑constituted and interlocking power structures of oppression and privilege, such as transphobia and racism, can synergize with one another and cause unique social and health effects among individuals subject to multiple oppressive forces [10–12]. Core to intersectionality is that these power structures manifest on multiple levels, including individual, interpersonal, and structural levels. Previous conceptual work has enumerated each of these levels in the transgender [13] and racial [14] health inequities literature. Unfortunately, an appreciation of these overlapping systems of oppression, particularly on the structural level, has been lacking in the empirical study of HIV health inequities among the transgender population. In fact, in their recent review, Bowleg et al. identified structural racism as a fundamental “social‑structural” determinant of HIV inequities that must be addressed to end the HIV epidemic. The authors explicate how many of the remaining barriers to HIV prevention and treatment are manifestations of structural racism, including carceral and police violence and discrimination, neighborhood disadvantage, and limited access to social services among others. This rationale can also be applied to other forms of structural discrimination, including structural transphobia, which similarly underlies social and healthcare‑related inequities, such as the exclusion of transgender healthcare needs from research [15]. Such exclusion may function as a barriers to HIV treatment and prevention, and demonstrates how structural discrimination is a negative health determinant for transgender persons.

To address the dearth of research on transgender HIV inequities that investigate the role of multiple levels of discrimination, we apply intersectionality and estimate the association of interpersonal and structural racism and transphobia with sexual behaviors associated with HIV transmission among an ethnoracially diverse sample of transgender and gender diverse (TGD) U.S. young adults of color. Specifically, we estimate these associations for measures of each type of discrimination on the interpersonal and structural level, separately (i.e., interpersonal racism separate from structural racism) and jointly (i.e., interpersonal racism along with structural racism), as well as the association of racism and transphobia across both levels with sexual behaviors (joint association of both types of discrimination on both levels).

Methods

Study Participants

Our analyses used data from the Body Image, Sexual Health and Relationships Project (B*SHARP), a national cross‑sectional online survey of transgender and genderexpansive U.S. young adults conducted from February to July of 2019 (N = 716). The inclusion criteria for the survey were: ages 18–30 years, United States residency, and transgender, non‑binary, or another gender identity different from gender assigned at birth, such as genderqueer or agender.

B*SHARP employed multiple recruitment strategies including outreach to community‑based organizations, college student groups, LGBTQ health centers and social media groups that serve transgender and gender‑expansive youth, as well as paid, targeted social media advertisements. Potential participants were provided a link to an online screening tool to determine eligibility. After the initial screen, potential participants were provided an additional link to an informed consent form and those who consented were able to complete the anonymous survey online. This study was reviewed and approved by the Boston Children’s Hospital Institutional Review Board.

The survey was administered online via REDCap [16], and responses were validated using a three‑stage protocol that employed automatic and manual procedures to exclude fraudulent or repeat survey responses. In total, 4604 initial screening forms were completed; of these, 1677 survey responses were received, inclusive of repeat and non‑human entries. After validation, the final sample included 716 (43%) of survey responses. Individuals who completed the survey were eligible to receive a $10 gift card or to donate their incentive to one of two community‑based transgender health and advocacy organizations. This study uses information on respondent demographics, experiences of discrimination, and sexual behaviors ascertained from the B*SHARP survey. In the B*SHARP study, questions related to sexual behaviors were only presented to respondents who reported having penetrative sex, defined as vaginal/front hole or anal sex using the genitals of both partners, at least once in the past year, further reducing the size of the cohort to 388 individuals.

Outcomes

In this study we investigated the following sexual behaviors during the year prior to completing the survey, (1) history of consuming alcohol or drugs prior to sex (2) history of inconsistent condom usage during penetrative sex and (3) history of transactional sex. Each outcome was operationalized as a dichotomous variable for any history of the sexual behaviors over the past 12 months. See Table 1 below for details of the survey item corresponding to each outcome and how responses were dichotomized.

Table 1.

Outcomes and corresponding survey items

Outcome	Survey item	Model response variable
Alcohol or drug consumption prior to sex	In the past 12 months, about how often did you drink alcohol or use drugs before you had vaginal/front hole or anal sex? Response options: (1) Never, (2) Sometimes, (3) Always, (4) Prefer not to answer *	Dichotomized: Yes = (2) Sometimes or (3) Always No = (1) Never
Inconsistent condom usage	In the past 12 months, how often did you or your partner(s) use a condom or barrier when you had this type of sex [vaginal/ front hole or anal sex using your and your partner's genitals (not including sex using a prosthetic or sex toy)]? Response options: (1) Never, (2) Less than half the time, (3) Not always, but more than half of the time, (4) Always	Dichotomized: Inconsistent = (1) Never, (2) Less than half of the time, (3) Not always, but more than half of the time Consistent = (4) Always
Transactional sex	In the past 12 months, have you had sex with anyone in exchange for things you needed (like money, drugs, food, shelter, etc.)? Response options: (1) Yes or (2) No	Dichotomized as original response options

Measures of Interpersonal Discrimination

Interpersonal Racism

Several measures of discrimination were embedded in the B*SHARP survey, including the Everyday Discrimination Scale (EDS) by David Williams [17]. This scale is composed of nine items that capture chronic, routine instances of unfair treatment that occur in daily life [17]. The items were summed across a 6‑point Likert scale based on frequency (0 = Never, 1 = Less than least once a year, 2 = A few times a year, 3 = A few times a month, 4 = at least once a week, 5 = Almost everyday) to generate a final score that measures the severity of discrimination experienced [18]. The scale was originally applied to study differences in Black‑White health disparities [17] and is most often used in the context of racial health inequities [19–21]. However, the instrument includes a follow‑up question, “What do you think is the main reason for these experiences,” that allows respondents to attribute their discrimination to other social identities. We used the EDS score for those who select “race or ethnicity” or “nationality” as an attribution for their discrimination as our measure of interpersonal racism. See Supplemental Appendix 1 for the full EDS as included in the B*SHARP survey.

Interpersonal Transphobia

In addition to the EDS, B*SHARP also included the Gender Minority Stress and Resilience (GMSR) scale [22]. The GMSR is based on extending the Meyer minority stress model for lesbian, gay, and bisexual (LGB) individuals to include transgender and gender nonconforming individuals [23]. The minority stress model proposes external and internal sources of stress specific to LGB individuals as potential drivers of poor mental health [23]. In their extension, Testa et al. argue that gender minority individuals are subject to external, or distal, sources of stress, such as gender‑related discrimination, rejection, victimization, and non‑affirmation of their gender identity, which can have direct health effects as well as indirect health effects, mediated by internal stressors such as internalized transphobia, negative expectations for future events and concealment [22]. For the purpose of this study, we used the four distal stressor subscales, which include a 5‑item discrimination scale and 6‑item subscales for rejection, victimization, and non‑affirmation subscales, and summed them to generate an overall gender minority stress score as a measure of interpersonal transphobia. See Supplemental Appendix 2 for details of the GMSR items included in the B*SHARP survey and used in this study.

Measures of Structural Discrimination

Structural Racism

We used the State Racism Index (SRI) to measure state level structural racism [24]. The SRI was designed to measure structural racism and has been used in a prior study to demonstrate how structural racism can drive racial inequities in fatal police shootings [24]. The SRI quantifies social inequities that are driven by structural racism and is based on measures of residential segregation and Black‑White disparities in the rate of incarceration, educational attainment, proportion living below the poverty level, median household income, household ownership, and employment at the state level [24]. We linked each individual in the study to an SRI based on their state of residence as a measure of the structural racism they experience.

Structural Transphobia

We used the Movement Advancement Project Gender Identity (MAP‑GI) Tally to measure state‑level transphobia [25]. MAP is an independent, non‑profit organization focused on LGBTQ policy‑related research and advocacy. As part of their “State Policy Tallies” initative, MAP tracks almost 40 LGBTQ‑related state and local laws and policies and assigns a point value with positive points indicating protective or beneficial to LGBTQ populations, and negative point values indicating harmful or detrimental to LGBTQ populations. Fractional points are assigned based on partial implementation of a law or where local laws provide partial protection that does not apply throughout the state. MAP provides three scores, a Sexual Orientation Tally, a Gender Identity (GI) Tally, and an Overall Tally, to disaggregate laws and policies that may differentially impact LGBQ and transgender populations. Therefore, the MAP‑GI tally can be viewed as a summary score of how transprotective or transphobic the laws and policies of that state are, and therefore a proxy of structural transphobia. We reverse‑coded and shifted the score by the minimum such that higher positive values indicated more structural transphobia and the new minimum value was zero for interpretability. We linked each individual in the study to a MAP‑GI based on their state of residence as a measure of the structural transphobia they experience. The scores used in this analysis are current as of March 16, 2021.

Statistical Analyses

We used standard frequency tabulation and summary statistics to report the distribution of respondent sociodemographic characteristics. For the outcomes, we present tables of counts and proportions cross‑stratified by ethnoracial identity and gender, as the primary intersecting identities of interest in this study, reporting proportions for each intersectional position (i.e., Black transgender individuals, or Latinx non‑binary, genderqueer, or gender non‑conforming individuals). To assess the internal validity of the measures of interpersonal discrimination used in this study, we reported Cronbach’s ^α for the GSMR subscales that are summed to create the overall measure of interpersonal transphobia, as well as the Everyday Discrimination Scale used to measure interpersonal racism. We also present the mean and standard deviation of interpersonal racism and transphobia by intersectional position.

The exposures used in this study are interpersonal and structural and include individual and state‑level measures of discrimination. To account for clustering between respondents from the same state, we fitted logistic regression models with cluster‑robust standard errors [26].

To capture the relationship between the different levels of discrimination and sexual behavior outcomes we estimate the following association measures, all from adjusted regression models for each outcome that contain our measures of interpersonal racism, interpersonal transphobia, structural racism, and structural transphobia. The first set of association measures are the single‑level, single‑axis discrimination measures that represent the relationship between each outcome and each of the four types of discrimination, separately. For the second set, because interpersonal racism and interpersonal transphobia are not experienced apart from structural racism and structural transphobia, respectively, we estimate for the combined association of both levels of racism and transphobia (i.e., estimates for the combined association of interpersonal and structural racism with each sexual behavior outcome, as well as estimates for the combined association of interpersonal and structural transphobia with each sexual behavior outcome). We term these estimates “multilevel” racism or transphobia associations as they account for interpersonal and structural discrimination on their respective axes. For the third and final set, because individuals who are transgender and experience ethnoracial minoritization do not experience transphobia and racism in isolation, we estimate the joint association of all four measures of discrimination, racism and transphobia on both the interpersonal and structural levels. We term these “intersectional” associations because, consistent with the theoretical framework, they incorporate multiple axes of discrimination (racism and transphobia) on multiple levels (interpersonal and structural). All regression models are restricted to transgender young adults of color (TYAOC); individuals who self‑reported as Asian, Black or African‑American, Latinx or Hispanic, or were grouped into the Other/multiethnoracial group in the B*SHARP study. This restriction is because we are primarily interested in estimating the association between discrimination and sexual behaviors among individuals exposed to both racism and transphobia, and while Non‑Hispanic White respondents in the B*SHARP study are all transgender and therefore experience transphobia, they are not generally subject to systemic racism. For all models we present odds ratios and 95% confidence intervals for one standard deviation increase in the discrimination measures corresponding to the association measure of interest (i.e., for multilevel racism, the association of one standard deviation increase of interpersonal and structural racism with a given sexual behavior). Confidence intervals that do not span 1 are statistically significant at the type I error rate of ^α= 0.05. Cluster‑robust standard errors were calculated using the sandwich package in R [27].

All models were adjusted for age, immigrant status, and education as potential confounders. Notably, we did not adjust for gender identity or ethnoracial identity, because these social identities are correlated proxies of transphobia and structural racism which are captured by the discrimination measures that are our primary exposures of interest. Therefore, including these additional covariates would lead to overadjustment and would bias our results toward the null [28]. All analyses were conducted in R version 4.0.2.

Results

Sociodemographic Characteristics of Study Cohort

There were 388 B*SHARP participants who had at least one penetrative sex partner in the past year, thereby meeting the inclusion criteria for this study. The cohort was ethnoracially diverse, with 41.2% Non‑Hispanic White respondents, 14.2% Non‑Hispanic Black or African American respondents, 10.8% Latinx or Hispanic respondents, 7.5% Non‑Hispanic Asian respondents, and the remaining responding as another ethnoracial group or multiple ethnoracial groups (Table 2). In sum, there were 228 transgender young adults of color (TYAOC) in the study sample and eligible for inclusion in the regression analyses described below and presented in Table 5. In the overall cohort, most respondents identified as transgender (58.2%), followed by nonbinary, genderqueer, or gender non‑conforming (34.3%) and agender or another gender identity (7.5%). This distribution was somewhat different among TYAOC respondents; the plurality still identified as transgender (48.7), followed by nonbinary, genderqueer, or gender non‑conforming (42.5%), and the remaining TYAOC identifying as agender or another gender identity (8.8%). In the overall cohort and among TYAOC, most respondents identified as bisexual, pansexual, or queer (62.9% and 64.5%, respectively), were ages 18–24 years (59.5% and 62.7%, respectively), US natives (93.0% and 90.4%,respectively), and had at least some college education (77.8% and 77.2%, respectively). In the overall cohort the majority were privately insured (52.8%), however among TYAOC only the plurality were privately insured (43.0%). The overall cohort was also regionally diverse with roughly equal representation from the Northeast, South, and West regions (27.3–29.6%), and somewhat less representation of the Midwest (15.5%). TYAOC were less evenly distributed with the greatest representation in from the West (35.5%) followed by the South (26.8%), the Northeast (22.4%) and the Midwest (15.4%).

Table 2.

Sociodemographic characteristics of B*SHARP participants in study cohort

Sociodemographic	Overall (N = 388)	TYAOC (N = 228)
Gender identity		–
Transgender	226 (58.2)	111 (48.7)
Non-binary, genderqueer, or gender non-conforming	133 (34.3)	97 (42.5)
Agender or another gender identity	29 (7.5)	20 (8.8)
Ethnoracial identity		–
Asian	29 (7.5)	29 (12.7)
Non-Hispanic Black or African American	55 (14.2)	55 (24.1)
Latinx or Hispanic	42 (10.8)	42 (18.4)
Non-Hispanic White	160 (41.2)	0 (0)
Another ethnoracial group or multi-ethnoracial	102 (26.3)	102 (44.7)
Sexual orientation identity		–
Straight	37 (9.5)	26 (11.4)
Bisexual/pansexual/queer	244 (62.9)	147 (64.5)
Lesbian/gay	71 (18.3)	30 (13.2)
Another sexual orientation	36 (9.3)	25 (11)
Age		–
18–24	231 (59.5)	143 (62.7)
25–30	157 (40.5)	85 (37.3)
Immigrant status		–
United States Native	361 (93.0)	206 (90.4)
Immigrant	26 (6.7)	21 (9.2)
Education		–
Less than college	86 (22.2)	52 (22.8)
Some college or more	302 (77.8)	176 (77.2)
Geographic region		–
Northeast	115 (29.6)	51 (22.4)
South	106 (27.3)	61 (26.8)
Midwest	60 (15.5)	35 (15.4)
West	106 (27.3)	81 (35.5)
Health insurance		–
No insurance	62 (16.0)	41 (18.0)
Public health insurance	102 (26.3)	74 (32.5)
Private health insurance	205 (52.8)	98 (43.0)

Table 5.

Association between interpersonal and structural discrimination and sexual behaviors associated with HIV transmission among transgender young adults of color

Type of discrimination	Consuming alcohol or drugs prior to sex (N = 216)	Inconsistent condom usage (N = 216)	Transactional sex (N = 224)
	OR (95% CI)	OR (95% CI)	OR (95% CI)
Interpersonal racism	1.37 (0.86, 2.16)	0.97 (0.69, 1.36)	2.23 (1.49, 3.35)
Structural racism	1.57 (1.11, 2.23)	1.32 (0.89, 1.95)	0.94 (0.52, 1.71)
Multilevel racism	2.15 (1.27, 3.65)	1.28 (0.83, 1.98)	2.11 (1.04, 4.26)
Interpersonal transphobia	1.64 (1.27, 2.12)	0.76 (0.48, 1.23)	1.71 (1.16, 2.51)
Structural transphobia	1.09 (0.76, 1.57)	0.92 (0.57, 1.47)	0.98 (0.57, 1.69)
Multilevel transphobia	1.79 (1.26, 2.55)	0.70 (0.31, 1.55)	1.68 (0.84, 3.36)
Intersectional discrimination	3.85 (2.12, 7.01)	0.89 (0.34, 2.36)	3.54 (0.99, 12.59)

OR odds ratio, CI confidence interval

Confidence intervals are estimated based on cluster robust standard errors

N for each model varies due to missing data

“of Color”—denotes Black or African American, Asian, Latinx or Hispanic, and Other/Multi-ethnoracial B*SHARP Respondents

Bold indicates statistically significant at the type I error rate of α=0.05

Multilevel racism and multilevel transphobia denote the joint association with the corresponding sexual behavior of one standard deviation increases in in interpersonal and structural racism and transphobia, respectively

Intersectional association measure is the joint association of one standard deviation increases in all four measures of discrimination (interpersonal racism, interpersonal transphobia, structural racism, and structural transphobia)

Prevalence of Sexual Behaviors Associated with HIV Transmission by Race and Gender

Overall, 70.6% of respondents reported consuming alcohol or drugs prior to sex, with substantial variation across intersectional positions (Table 3). Within the transgender group, Black and Latinx respondents reported the highest prevalence of alcohol or drug consumption prior to sex at 84.8% and 75.0%, respectively, while White transgender respondents had the lowest (68.5%). Among nonbinary, genderqueer, and gender non‑conforming respondents, White individuals had the highest prevalence of drug or alcohol consumption prior to sex (80.0%) with the remaining intersectional position prevalence estimates between 65.2 and 73.3%. Among agender and individuals with another gender identity, the prevalence of alcohol or drug consumption prior to sex by intersectional position, was between 50.0 and 58.3%. Inconsistent condom usage varied substantially by intersectional position, but was generally high with a prevalence of 73.5% across the overall cohort. Lastly, prevalence of transactional sex was highest among Black and Latinx respondents across all genders (25.5% and 26.2%, respectively, compared to 3.4% to 10.8% among other ethnoracial groups with shared gender identity), and similarly highest among transgender (32.4% and 33.3%, respectively, compared to 4.3–14.6 among other ethnoracial groups with shared gender identity) and nonbinary, genderqueer, and gender nonconforming respondents (11.8% and 16.7% respectively, compared to 0.0–8.2% among other ethnoracial groups with shared gender identity).

Table 3.

Sexual behaviors associated with HIV transmission by intersectional position

	Transgender	NB/GQ/GNC	Agender/another GI	All genders
Alcohol or drug consumption prior to sex
Asian	9/12 (75.0)	9/13 (69.2)	2/4 (50.0)	20/29 (69.0)
Black	28/33 (84.8)	11/16 (68.8)	2/4 (50.0)	41/53 (77.4)
Latinx or Hispanic	18/24 (75.0)	11/15 (73.3)	–	29/39 (74.4)
Other/multi-ethnoracial	29/41 (70.7)	30/46 (65.2)	7/12 (58.3)	66/99 (66.7)
White	76/111 (68.5)	28/35 (80.0)	4/8 (50.0)	108/154 (70.1)
All ethnoracial groups	160/221 (72.4)	89/125 (71.2)	15/28 (53.6)	264/374 (70.6)
Inconsistent condom usage
Asian	10/12 ( 83.3)	13/13 (100.0)	3/4 ( 75.0)	26/29 (89.7)
Black	23/33 ( 69.7)	15/16 ( 93.8)	4/4 (100.0)	42/53 (79.2)
Latinx or Hispanic	19/24 (79.2)	12/15 (80.0)	–	31/39 (79.5)
Other/multi-ethnoracial	22/41 (53.7)	36/46 (78.3)	10/12 (83.3)	68/99 (68.7)
White	74/110 (67.3)	28/35 (80.0)	5/8 (62.5)	107/153 (69.9)
All ethnoracial groups	148/220 (67.3)	104/125 (83.2)	22/28 (78.6)	274/373 (73.5)
Transactional sex
Asian	1/12 (8.3)	0/13 (0.0)	0/4 (0.0)	1/29 ( 3.4)
Black	11/34 (32.4)	2/17 (11.8)	1/4 (25.0)	14/55 (25.5)
Latinx or Hispanic	8/24 (33.3)	3/18 (16.7)	–	11/42 (26.2)
Other/multi-ethnoracial	6/41 (14.6)	4/49 ( 8.2)	1/12 ( 8.3)	11/102 (10.8)
White	5/115 (4.3)	1/36 (2.8)	0/9 (0.0)	6/160 ( 3.8)
All ethnoracial groups	31/226 (13.7)	10/133 ( 7.5)	2/29 ( 6.9)	43/388 (11.1)

Data in each cell are presented as count/n (%)

count—denotes number of individuals at intersectional position endorsing history of sexual behavior; n— denotes number of individuals at intersectional position with data available; %—denotes percent of individuals among those with data available at intersectional position endorsing history of sexual behavior

NB nonbinary, GQ genderqueer, GNC gender nonconforming, GI gender identity

Distribution of Discrimination Measures by Race and Gender

The discrimination, rejection, and victimization, subscales demonstrated adequate internal reliability, with Cronbach’s ^α values of 0.73, 0.72, and 0.84, respectively (Supplemental Table 1). Notably, these reliability metrics are all higher than in the original GMSR validation study [22]. The non‑affirmation subscale also demonstrated adequate internal reliability, with an ^α of 0.74, though this was lower than the reliability reported in the original study. Lastly, the Everyday Discrimination Scale demonstrated high internal reliability with an ^α of 0.92. To generate the measure of interpersonal transphobia, we summed the scores across the GMSR distal stressor subscales, and for interpersonal racism, we used Everyday Discrimination with a race, ethnicity, or nationality attribution.

We found that the distribution of interpersonal racism and interpersonal transphobia varied by intersectional position. Among transgender respondents, Latinx and Black respondents had the highest mean reported interpersonal racism, 17.18 (SD 13.21) and 18.42 (SD 14.79), followed by respondents in the other/multi‑ethnoracial group (mean 13.59, SD 11.13), and Asian respondents (mean 8.25, SD 11.27), with White respondents having the lowest (mean 0.37, SD 2.68). This trend was similar in the nonbinary, genderqueer, or gender nonconforming and agender or another gender identity groups, as well as the overall cohort (Table 4). Interpersonal transphobia was highest among transgender individuals, compared to individuals in the non‑binary, genderqueer, or gender non‑conforming and agender or another gender identity groups. Within the transgender group, Black and Latinx respondents had the highest mean reported interpersonal transphobia, 17.79 (SD 5.62) and 17.71 (SD 4.71), respectively, followed by Asian (mean 16.00, SD 6.05) and the other/multi‑ethnoracial (mean 15.59, SD 4.06) respondents, with White respondents having the lowest mean reported interpersonal transphobia (mean 14.08, SD 6.09). This trend was similar among nonbinary, genderqueer, and gender nonconforming respondents. Across all genders, Asian respondents (mean 11.93, SD 6.11) had the lowest mean reported interpersonal transphobia, rather than White respondents (mean 13.20, SD 5.82), with Latinx (mean 16.05, SD 5.06) and Black respondents (mean 15.58, SD 5.82) still having the highest mean reported interpersonal transphobia. Estimated structural racism and structural transphobia, by state, along with the number of respondents from each state is presented in Supplemental Table 2. States in the Midwest and Northeast tended to have the highest structural racism, as measured by the SRI, with Wisconsin having the highest 23.50 followed by Alabama and Louisiana each with 22.50 (after reverse-coding and shifting such that all scores are non-negative and higher scores indicate more severe structural transphobia).

Table 4.

Interpersonal discrimination by intersectional position

	Transgender	NB/GQ/GNC	Agender/other GI	All genders
Interpersonal racism
Asian	12: 8.25 (11.27)	13: 10.85 (8.21)	4: 3.50 (2.89)	29: 8.76 (9.26)
Black	34: 17.18 (13.21)	17: 21.59 (8.27)	4: 19.50 (11.68)	55: 18.71 (11.77)
Latinx or Hispanic	24: 18.42 (14.79)	18: 14.11 (13.53)	–	42: 16.57 (14.26)
Other/multi-ethnoracial	41: 13.59 (11.13)	49: 10.90 (12.24)	12: 11.92 (11.07)	102: 12.10 (11.63)
White	115: 0.37 (2.68)	36: 0.94 (4.13)	9: 0.00 (0.00)	160: 0.48 (3.00)
All ethnoracial groups	226: 7.63 (11.81)	133: 10.00 (11.80)	29: 8.10 (10.61)	388: 8.48 (11.75)
Interpersonal transphobia
Asian	12: 16.00 (6.05)	13: 9.69 (4.53)	4: 7.00 (3.37)	29: 11.93 (6.11)
Black	34: 17.79 (5.62)	17: 12.71 (4.10)	4:9.00 (3.46)	55: 15.58 (5.82)
Latinx or Hispanic	24: 17.71 (4.71)	18: 13.83 (4.74)	–	42: 16.05 (5.06)
Other/multi-ethnoracial	41: 15.59 (4.06)	49: 12.84 (4.30)	12: 16.92 (4.81)	102: 14.42 (4.51)
White	115: 14.08 (6.09)	36: 11.42 (4.42)	9: 9.11 (3.82)	160: 13.20 (5.82)
All ethnoracial groups	226: 15.40 (5.72)	133: 12.26 (4.48)	29: 12.03 (5.80)	388: 14.07 (5.55)

Data in each cell are presented as n: M(SD)

n denotes number of individuals at intersectional position; M denotes mean value of discrimination measure for individuals at corresponding intersectional position; SD denotes standard deviation of discrimination measure for individuals at corresponding intersectional positions

NB nonbinary, GQ genderqueer, GNC gender nonconforming, GI gender identity

Racism and Sexual Behaviors Associated with HIV Transmission

We found a significant association between interpersonal racism and engaging in transactional sex, with a one standard deviation increase in interpersonal racism associated with a more than doubling of the odds of transactional sex (OR 2.23, 95% CI 1.49, 3.35, Table 5) among transgender young adults of color (TYAOC). We also found a significant association between structural racism and consuming alcohol or drugs prior to sex (OR 1.57, 95% CI 1.11, 2.23) among TYAOC. Lastly, we found a significant multilevel (interpersonal and structural) association between racism and both consuming alcohol or drugs prior to sex (OR 2.15, 95% CI 1.27, 3.65) and transactional sex (OR 2.11, 95% CI 1.04, 4.26).

Transphobia and Sexual Behaviors Associated with HIV Transmission

We found a significant association between interpersonal transphobia and consuming alcohol or drugs prior to sex and engaging in transactional sex among TYAOC in the study sample (Table 5). For a one standard deviation increase in estimated interpersonal transphobia, there was a 64% increase in the estimated odds of consuming alcohol or drugs prior to sex (OR 1.64, 95% CI 1.27, 2.12) and a 71% increase in odds of engaging in transactional sex (OR 1.71, 95% CI 1.16, 2.51). The individual associations between structural transphobia and sexual behaviors were not statistically significant. There was also a statistically significant multilevel association between transphobia and consuming alcohol or drugs prior to sex (OR 1.79, 95% CI 1.26, 2.55) among TYAOC respondents.

Intersectional Association of Racism, Transphobia, and Sexual Behaviors

There was a significant intersectional association (joint association of interpersonal racism, structural racism, interpersonal transphobia, and structural transphobia) between discrimination and consuming alcohol or drugs prior to sex, with one standard deviation increases in the four discrimination measures jointly associated with a nearly quadrupling in the odds of consuming alcohol or drugs prior to sex (OR 3.85, 95% CI 2.12, 7.01, Table 5). In comparison, the intersectional association of discrimination with transactional sex was not statistically significant with a 95% confidence interval lower bound of 0.99, however the estimate is consistent with a 3.5 times increase in the odds of engaging in transactional sex, with an upper bound of 12.59, suggestive of a similarly strong association.

Discussion

HIV remains one of the major health crises facing the transgender community, particularly among ethnoracial minority subpopulations. Black, Indigenous, and Latinx transgender people experience severe adverse health outcomes due to interactions among multiple hierarchies of power and oppression that serve to limit resources and facilitate health inequities relative to their non‑BIPOC and/ or cisgender counterparts [3, 29–31]. Our study focuses on racism and transphobia and demonstrates strong associations between interpersonal racism and transactional sex, and interpersonal transphobia with both consuming alcohol or drugs prior to sex and transactional sex. However, consistent with the principles of intersectionality, separate examination of different levels and forms of discrimination are of limited utility. Interpersonal racism does not impact health separately from structural racism, and for transgender people of color, racism does not operate outside of transphobia. Therefore, when evaluating discrimination at multiple levels, we show that joint exposure to interpersonal and structural forms of racism and transphobia may also contribute to consuming alcohol or drugs prior to sex as well as engaging in transactional sex work. The latter combined estimates provide important context for transgender people of color, as structural and interpersonal discrimination along race and gender are simultaneously experienced by this group.

The differences in self‑reported interpersonal racism and transphobia by intersectional position should also be emphasized. Measured experiences of interpersonal racism and transphobia were, on average, higher for Black, Latinx, and respondents in the other/multi‑ethnoracial group across all genders. While this finding is obvious for racism, the finding for transphobia is also wholly consistent with the theory of intersectionality. Under this framework, systems of oppression are co‑constituted, or comprised of one another. Transphobia is made of racism (and ableism, and other forms of oppression), and so it is reasonable to assume that individuals who experience racism will also experience more severe transphobic discrimination.

We did not find statistically significant associations between structural transphobia and any of the outcomes, and structural racism was only significantly associated with alcohol or drug consumption prior to sex. While it is possible that this indicates there is no relationship between structural transphobia and the sexual behavior outcomes in this study, it is also possible that these null findings are due to data and design limitations. Because structural measures were measured on the group‑level, our study may have been underpowered to detect these associations [32]. Therefore, for the purposes of this study, the primary utility of the structural measures was to estimate multilevel and intersectional associations between discrimination and sexual behaviors and provide interval estimates for these relationships. Future studies with more participants will facilitate more precise estimation of associations between structural discrimination and sexual behaviors. However, the present study still provides an important contribution because it demonstrates an approach to estimating these multilevel and intersectional associations and suggests strong relationships between discrimination and transactional sex and alcohol or drug consumption prior to sex.

Our study is consistent with recent advances in structural approaches to intersectionality in health equity research. In their study of “structural intersectionality”, Homan et al. [33] used aggregate indices similar to the State Racism Index to estimate structural racism and structural sexism and their impact of Black women’s health. Similarly, English et al. used the State Equality Index, an anti‑LGBTQ policy tally from the Human Rights Campaign similar to the MAP‑GI Tally used in our present study, and the State Racism Index, to study the impact of structural oppression along the axes of race and sexuality on suicidality among Black sexual minority men [34]. Our study builds on their approaches by incorporating similar structural measures for transphobia and racism, while also including interpersonal measures to evaluate the intersectional impact of multiple levels of oppression. While such an approach to the study of transgender population health has been theorized for previously [15, 35], to our knowledge this is the first empirical implementation in the transgender HIV health inequities literature.

A few studies have estimated the association between interpersonal discrimination and sexual behaviors among gender minorities. One study conducted by Arayasirikul et al. showed that transphobic discrimination was associated with increased odds of binge drinking among transgender women, and that transwomen of color (TWOC) had increased odds of reporting engagement in condomless receptive anal intercourse compared to White transgender women [5]. The former finding is consistent with our observation of the association between transphobia and drug and alcohol consumption prior to sex, but the latter was not replicated in our study with inconsistent condom usage. Another study, by Sugano et al. [4] also noted an association between interpersonal transphobia and unprotected receptive anal intercourse among TWOC, however, in our study, there was no relationship between inconsistent condom usage and interpersonal transphobia. The discrepancy between our study and the prior literature may be due to study design. The previous studies all limited the outcome to condomless receptive intercourse, whereas our study captures all forms of penetrative sex, including insertive and receptive anal and vaginal/front hole intercourse. While these different forms of penetrative condomless sex are all associated with different levels of HIV transmission risk [36], they are important to capture as part of the full range of sexual activity that transgender individuals can engage in. In a final study, an intersectional analysis of transgender survey respondents in Ontario Canada, Marcellin et al. found that interpersonal transphobic and ethnoracial discrimination had interacting effects on the odds of engagement in high‑risk sex such as receptive or insertive anal or genital intercourse with fluid contact. Our results are consistent with the findings of Marcellin et al., that intersecting racial and transphobic discrimination can drive sexual behaviors. Our study extends their work by incorporating structural measures of discrimination.

It is important to note that mechanisms underlying the relationships between discrimination and transactional sex and alcohol or drug consumption prior to sex may have important distinctions. Both behaviors can be conceptualized as responses to trauma and rejection as a result of discrimination, but transactional sex also serves as a means of economic preservation in the context of severe employment and housing discrimination experienced by transgender individuals [37, 38]. In fact, some transgender individuals report that sex work is the only means of employment they have access to, or that it is easier to engage in sex work than to endure the discrimination faced in other employment settings [39, 40]. Therefore, transactional sex may be viewed as one path through which discrimination, particularly in the form of resource deprivation, is embodied among transgender individuals. This would suggest that providing economic resources may be a potential intervention to reduce the burden of discrimination on this population. Transgender individuals also report benefits to sex work outside of economic gain, including positive experiences related to clientele who affirm their gender in ways that they are deprived of in other contexts [39]. These positive experiences suggest that economic interventions may only reduce, and not eliminate, the need for transgender individuals to engage in sex work and highlight the importance of continued efforts for safety and decriminalization in this industry.

Limitations and Future Directions

This study has several limitations. First, this study was not designed to power analyses of structural measures of discrimination, so the sample size may not be adequate for precise estimation of these effects. Secondly, structural discrimination is multilevel and the policies, practices, and norms that constitute structural racism and transphobia operate on national, state, community, and organizational levels. [14] Therefore, state‑level measures such as those employed in this study can only be used to estimate part of the impact of these ubiquitous societal forces. In the context of this limitation, our findings likely underestimate the impact of structural discrimination on this multiply marginalized population and our results should be viewed as conservative. Future studies should incorporate structural measures for more granular resolution. Thirdly, the scope of the measures of discrimination used in this study likely do not capture the full breadth of interpersonal and structural oppression. For instance, the State Racism Index is derived primarily on measures of Black‑White disparities; however, previous work has demonstrated that for Latina transgender women, a major determinant of engagement in transactional sex work is documentation status [41]. Therefore, policies that affect healthcare and resource access for undocumented transgender individuals residing in the U.S. are likely major structural measures of racial discrimination that are not captured by the State Racism Index. Fourthly, we did not capture PrEP utilization, HIV status, or relationship status of study participants. For participants who are on PrEP or in a monogamous relationship with an HIV‑negative partner or partners living with HIV who are undetectable, inconsistent condom usage is not associated with HIV transmission. Therefore, the relationship between discrimination and inconsistent condom usage as a sexual behavior associated with HIV transmission is not optimally measured by this study design, which may explain why we only observed null findings for the association between inconsistent condom usage and discrimination. Therefore, future work should ascertain additional information on PrEP, HIV, and relationship status of participants, to better understand the influence of discrimination on this outcome. Lastly, this study demonstrates strong associations between transphobia and racism and sexual behaviors, but because of its cross‑sectional nature, causality cannot be inferred. Theory underlying how discrimination impacts health and behavior support the possibility of a causal relationship but longitudinal cohort studies where the exposures and outcomes are ascertained at different time points would be ideal for confirming a true causal relationship.

Despite these limitations, our study is strengthened by its rigorous application of the theory and principles of intersectionality to the study of social‑structural factors and sexual behavior outcomes among TYAOC. Additionally, unlike the previous literature that focused primarily on transbinary experiences, this study includes agender, genderqueer, gender nonconforming and non‑binary individuals, who identify along, or outside of the masculine‑feminine gender identity continuum, making our findings more generalizable to the broader population of transgender young adults of color.

Conclusion

This study, integrated with previous literature, provides evidence for the noxious environment that interpersonal and structural transphobia and racism simultaneously create for transgender and gender diverse individuals, contributing to higher levels of sexual behaviors and exacerbating HIV burden. However, descriptive studies are limited and only serve as an initial step toward achieving health equity. Future work needs to focus on interventions that directly alleviate the burden of multilevel, intersectional discrimination, specifically centering multiply marginalized subgroups within the transgender community. Firstly, on the individual level, these interventions may take the form of direct aid, such as financial or housing resources; an approach that has been effective in the context of other multiply marginalized populations [42]. Secondly, on the interpersonal level these might include mandated transcompetent continuing education and racial bias training for healthcare providers. Lastly, on the structural level these interventions might include policy changes such as decriminalization of sex work and state and federal regulations that facilitate access to gender affirming care among transgender individuals.

Supplementary Material

Supplementary Information The online version contains supplementary material available at https://doi.org/10.1007/s10461-022-03701-w.

Data Availability

The data used in this study is part of ongoing work and can be requested through a data use agreement.