Disparities in HIV-related risk and socio-economic outcomes among trans women in the sex trade and effects of a targeted, anti-sex-trafficking policy

Abstract

Introduction:

Marginalization of sex work presents numerous risks for trans women (TW) engaged in the sex trade, including criminalization, traumatization, and contracting HIV. We identified socio-economic and HIV risk disparities among trans women sex workers and others who do sex work (TWSW/OWSW), and evaluated these disparities for TWSW/OWSW compared to TW not engaged in sex work from pre- and post-implementation of the US 2018 “Allow States and Victims to Fight Online Sex Trafficking Act” and “Stop Enabling Sex Traffickers Act” (FOSTA-SESTA).

Methods:

We analyzed 429 trans women (TW) from the Trans*National cohort study (2016–2019). Generalized estimating equations (GEE) characterized differences in socio-economic and HIV risk outcomes for TWSW/OWSW compared to TW not engaged in sex work over the study period. Adjusted, pre-to-post law changes in these outcomes for TWSW/OWSW versus TW not engaged in sex work were compared using difference-in-differences GEE regression analyses.

Results:

Over 18 months, TWSW/OWSW had higher adjusted odds of being unstably housed, having income from criminalized sources, experiencing transphobic hate crimes, experiencing discrimination from police/courts, being incarcerated, meeting sex partners in the street/public settings, meeting sex partners on Craigslist or other online forums (except dating apps), or engaging in condomless anal intercourse, (p<0.01 for all comparisons); TWSW/OWSW also had a higher mean number of income sources (p=0.03). One difference-in-differences analysis showed additive interaction: the adjusted mean number of income sources reported by TWSW/OWSW compared to those not engaged in sex work decreased from pre- to post-FOSTA-SESTA (from 1.79 to 1.48 for TWSW/OWSW and from 1.52 to 1.47 for TW not engaged in sex work; p=0.01).

Conclusions:

Disparities in socio-economic and HIV-related risk outcomes exist for TWSW/OWSW in San Francisco. There is an urgent need for comprehensive, long-term follow-up data of TW to accurately analyze policy effects, especially given the recent enactment of a number of other policies targeting TW.

Introduction

Compared with the rest of the adult population, trans women (TW) carry a disproportionate burden of HIV. Globally, almost one in five (19.1%) trans women lives with HIV¹. For trans women who do sex work, HIV prevalence is estimated to be even higher at 27.3%². Compared to white trans women, sex work is more prevalent among trans women of color³ who were shown in a national report to have a HIV prevalence of 40.6%⁴. It follows that the intersection of gender identity, race, and sex worker status amplifies HIV risk for trans women who do sex work^2,5, most of whom are trans women of color⁴.

The sex work profession encompasses a “complex HIV risk environment”, and trans women who do sex work experience stigma and violence which are both important drivers of HIV risk⁶. According to the Report of the 2015 U.S. Transgender Survey, participants who did sex work were more likely to experience violence than their counterparts, with almost half of them reporting that they were physically abused or harassed, and over one third being sexually assaulted.³

Sex work can be a source of economic empowerment⁷, while at the same time a marker of marginalization and necessity for trans women. According to a report on the 2015 National Transgender Discrimination Survey, 69.3% of respondents who engaged in sex work experienced some form of workplace discrimination and they were more than twice as likely to live in extreme poverty than their counterparts.⁴ For the many TW who experience employment discrimination, sex work is a means of survival^6,8–10. Moreover, studies have shown that sex workers are placed into “poverty traps”¹¹, where they engage in increasingly “risky” sexual behaviors in order to make day-to-day ends meet, without being able to generate long-term savings^12,13. This traps them in a cycle of poverty, where socio-economic constraints may be both a precursor to and an outcome of sex work engagement. However, to our knowledge, no such socio-economic consequences of sex work have been explored for trans women in the sex trade.

In addition to being entrenched within a complicated socio-economic environment, the sex work profession resides within a deep history of large-scale criminalization. Repressive sex work laws and policing – from “unlawful arrest and detention” to “physical and sexual violence by law enforcement” to “lack of recourse to justice” – are consistently shown to increase sex workers’ risk from HIV^2,14. Almost nine out of ten transgender people who engaged in sex work experienced violence from police³. Upstream precursors to HIV risk, including exposure to violence and trauma, and engagement in condomless sex, are also linked with sex work criminalization^15–17. Mathematical modelers found that decriminalization of sex work would have the greatest effect on HIV transmission, averting approximately 33–46% of new HIV infections in the next 10 years¹⁸. These findings underscore that health disparities research among sex workers must pivot from focusing on the profession itself to the structural-level marginalization process for sex workers. The implication is that the process of marginalization, rather than the profession of sex work, is the modifiable risk factor. One step in that direction is to examine how repressive policies exacerbate adverse health outcomes for sex workers.

Two such policies, the “Stop Enabling Sex Traffickers Act” (SESTA) and “Allow States and Victims to Fight Online Sex Trafficking Act” (FOSTA), were introduced in Congress with the intention of curbing sex trafficking by making it illegal to advertise, participate in, and/or financially benefit from sex trafficking. Several major advertising sites frequented by sex workers to vet their clients were shut down immediately following the passage of FOSTA-SESTA into law on April 11, 2018. Sex workers’ rights and national advocacy organizations posited that such “end demand” approaches, which conflate sex trafficking with sex work, would effectively push consensual sex work from online forums to the streets, making sex work less safe by eliminating the ability to screen potentially violent clients and by lowering sex workers’ abilities to negotiate condom use⁴. Similar campaigns in Chicago subjected Black trans women to increased arrest and associated fines¹⁹. With an already disproportionate burden of socio-economic scarcity, stigma, violence, and HIV compared to the rest of the population^1,20–21, trans women who do sex work are especially vulnerable to the adverse effects of FOSTA-SESTA. Policy analyses are needed to determine the impact of the passing of this law on trans women who do sex work to determine if prevention and care efforts are needed to address their specific HIV needs related to sex work.

In this study, we conducted the first-ever assessment of longitudinal disparities in socio-economic and upstream HIV-related risk outcomes among trans women in the sex trade. We then evaluated these risks for TW in the sex trade compared to those who are not from pre- and post-implementation of FOSTA-SESTA. We hypothesized that, despite the enactment of FOSTA-SESTA, engagement in the sex trade would be stable over time, and that any changes in outcomes from pre- to post-law would not be due to changes in the prevalence of sex work.

Methods

We analyzed data from 429 adult TW who were HIV-negative at baseline and participated in the 18-month Trans*National cohort study in San Francisco from 2016 to 2019. Interviewer-administered assessments captured data on socio-demographic characteristics and risk behaviors at baseline, 6, 12, and 18 months of follow-up.

The “exposure” of interest was whether participants reported engaging in sex work (i.e., reported having sexual partners in the last 6 months with whom they exchanged sex for pay, drugs, services, or other goods and/or reported income from sex work in the last month). This exposure was captured at all 4 visits and could vary over time. Since participants who reported income from sex work or exchange partners may not refer to themselves as sex workers, or may have engaged in other aspects of the sex trade (such as web camming or exotic dancing), we will refer to the exposure group as trans women sex workers and others who do sex work (TWSW/OWSW).

Outcomes of interest included both upstream and downstream drivers of HIV, to which we will refer as “HIV-related risk behaviors”. First, we focused on five socio-economic effects of sex work, namely: (1) whether participants’ current living situations were stable (e.g., owning or renting a house or apartment, living in a single room occupancy unit) or unstable (e.g., couch surfing, experiencing homelessness or living in a shelter, living in a residential treatment program, or living in transitional/supportive housing); (2) whether or not participants reported income in the last month from non-criminalized, part-time/full-time employment (excludes income from sex work); (3) whether or not participants reported income in the last month from non-criminalized, supplemental sources (excludes income from sex work or non-criminalized employment, and includes income from general assistance, food stamps, social security, disability, unemployment, alimony/child support, student loans, or income from friends/family/partners); (4) whether or not participants reported income in the last month from criminalized sources (excludes income from sex work or non-criminalized sources, and includes income from drug dealing/selling or stealing/scamming); and (5) the combined number of income sources other than sex work that participants reported in the last month (sums “yes” response to items asking whether participants’ income came from full- or part-time jobs, general assistance, food stamps, social security, disability, unemployment, alimony/child support, student loans, friends/family/partners, drug dealing/selling, alimony/child support, scamming/stealing, or sources other than sex work). All of these socio-economic outcomes were collected at each study visit.

Additionally, we analyzed outcomes related to violence, discrimination, and incarceration. These included: (1) experiences of transphobic hate crimes in the last 6 months (dichotomized as “yes” or “no” and measured at 6-, 12-, and 18-month surveys); (2) experiences of discrimination from the police or in the courts in the last 6 months (dichotomized as “yes” or “no” and measured at 6-, 12-, and 18-month surveys); and (3) incarceration in the last 6 months (dichotomized as “yes” or “no” and measured at all 4 study visits).

Finally, we assessed sexual health outcomes that we hypothesized would differentially affect TWSW/OWSW and be exacerbated by FOSTA-SESTA. We included the following outcomes: (1) whether or not participants met sexual partners in the streets or other public settings in the last 6 months; (2) whether or not participants met sexual partners via dating apps in the last 6 months; (3) whether or not participants met sexual partners on Craigslist in the last 6 months; (4) whether or not participants met sexual partners on some other online forum in the last 6 months; and (5) whether or not participants engaged in any condomless anal intercourse in the last 6 months. All aforementioned sexual health outcomes were collected at each study visit for up to five sexual partners. Participants who responded “yes” to engaging in the sexual health behavior of interest with at least one of up to five sexual partners were coded as engaging in that behavior.

We adjusted for baseline social transition and race/ethnicity since these factors were identified a priori as the minimal sufficient adjustment set based on a directed acyclic graph.

Social transition was operationalized as the degree to which participants felt that they fit into the gender binary of “female” and was categorized as “a lot or somewhat”, “a little”, or “not at all”. We included it for adjustment because we hypothesized that involvement in the sex trade could be source of gender affirmation for trans women. At the same time, we hypothesized that sex trade client demand may shape gender transition. Specifically, TWSW/OWSW who desire gender affirming surgeries (e.g., penectomies, vaginoplasties, and so on) may delay these procedures to appease clients who threaten to discontinue payment for services (or worse, bodily harm) should TWSW/OWSW undergo these procedures. We also hypothesized that TW who were not perceived as conforming to societal gender binary norms were more likely to experience violence and discrimination, both of which put them at greater risk for HIV.

Race/ethnicity was categorized as Black and non-Hispanic/Latinx, Hispanic/Latinx, white and non-Hispanic/Latinx, or “other”/multiple and non-Hispanic/Latinx race/ethnicity. We included race/ethnicity for adjustment given that the prevalence of sex work and HIV is higher for trans women of color.

In terms of statistical analyses, we first evaluated whether the prevalence of sex work engagement changed over 18 months of follow-up for TW. Baseline socio-demographic characteristics, experiences of violence, discrimination, and criminalization, and sexual health behaviors were described for TWSW/OWSW and trans women not engaged in sex work. Bivariable Poisson binomial regression models were specified to directly calculate the prevalence of baseline sex work for each of the aforementioned factors compared to its reference group. We then ran generalized estimating equations (GEE) regression models to test whether these factors were different for TWSW/OWSW compared to those not engaged in sex work over the 18-month study period, after adjustment for social transition and race/ethnicity. Finally, we ran difference-in-differences analyses using GEE to examine whether FOSTA-SESTA moderated adjusted associations between sex work and the aforementioned factors from pre- to post-implementation. Adjusted estimates from statistically significant difference-in-differences analyses were converted to marginal probabilities in order to plot changes in outcome probabilities from pre- to post-law for TWSW/OWSW compared to those not engaged in sex work.

The choice of GEE versus a mixed effects approach was motivated by the former method’s robustness to model misspecification.²² GEE was also the method of choice in a similar, recently published study.²³ Moreover, there was a sufficiently large number of clustering units (i.e., more than 50 individuals with repeated measures) in this dataset, making it suitable for GEE.

All analyses were conducted in Stata 14 software (StataCorp LP, College Station, TX, USA). As needed, results were visualized in R console²⁴.

Results

Sex work engagement was stable over time (Figure 1). Bivariable generalized estimating equations confirmed that the odds of sex work engagement did not change over the 18-month study period (odds ratio, OR = 0.95, 95% CI = 0.88 – 1.02, p = 0.16).

Fig. 1.

Prevalence of sex work among trans women in San Francisco, by study visit, Trans*National, 2016–2019.

From Table 1, we observed that baseline engagement in sex work was more likely among Black trans women (adjusted prevalence ratio, aPR = 2.75, 95% confidence interval, 95%CI = 1.35 – 5.60, p < 0.01), Hispanic/Latinx trans women (aPR = 3.02, 95%CI = 1.81 – 5.04, p < 0.01), and trans women of “other” or multiple races/ethnicities (aPR = 1.93, 95%CI = 1.06 – 3.51, p = 0.03) compared to white trans women. Those with unstable housing (aPR = 2.28, 95%CI = 1.51 – 3.46, p < 0.01) and those with income from criminalized sources other than sex work (aPR = 2.59, 95%CI = 1.72 – 3.89, p < 0.01) were also more likely to engage in sex work at baseline compared to their counterparts. Having a higher number of income sources at baseline was also associated with sex work (aPR = 1.30, 95%CI = 1.09 – 1.54, p < 0.01).

Table 1.

Baseline socio-demographic characteristics and health outcomes for trans women sex workers and other who do sex work (TWSW/OWSW) compared to those not engaged in sex work, Trans*National, 2016 – 2019 (n = 429)

	TWSW/OWSW		Not engaged in sex work		Bivariable baseline comparison
	N	%a	N	%a	PRb	95% CIc	p-value
Total	90	(20.98)d	338	(78.79)d		--
Socio-demographics
Race/ethnicity
White, non-Hispanic/Latinx	17	(18.89)	144	(42.60)	Reference
Black, non-Hispanic/Latinx	9	(10.00)	22	(6.51)	2.75	(1.35 – 5.60)	<0.01
Hispanic/Latinx	44	(48.89)	94	(27.81)	3.02	(1.81 – 5.04)	<0.01
Other/Multiple, non-Hispanic/Latinxe	20	(22.22)	78	(23.08)	1.93	(1.06 – 3.51)	0.03
Social transition: self-reported passing with gender binary
A lot / somewhat	71	(78.89)	222	(65.68)	Reference
A little	12	(13.33)	64	(18.93)	0.65	(0.37 – 1.14)	0.13
Not at all	6	(6.67)	48	(14.20)	0.46	(0.21 – 1.00)	0.05
Current living situation
Stablef	41	(45.56)	245	(72.49)	Reference
Unstableg	49	(54.44)	93	(27.51)	2.41	(1.67 – 3.46)	<0.01
Income source in the last month: non-criminalized employmenth
No	55	(61.11)	179	(52.96)	Reference
Yes	35	(38.89)	159	(47.04)	0.77	(0.53 – 1.12)	0.17
Income source in the last month: non-criminalized supplemental incomei
No	22	(24.44)	96	(28.40)	Reference
Yes	68	(75.56)	242	(71.60)	1.18	(0.76 – 1.81)	0.46
Income source in the last month: criminalized sourcesj
No	74	(82.22)	321	(94.97)	Reference
Yes	16	(17.78)	17	(5.03)	2.59	(1.72 – 3.89)	<0.01
Number of income sources in the last monthk, mean (standard deviation)	1.86	(1.27)	1.53	(0.81)	1.30	(1.09 – 1.54)	<0.01
Experiences of violence, discrimination, and criminalization
Victim of transphobic hate crimel, last 6 months
No	67	(83.75)	257	(91.79)	Reference
Yes	13	(16.25)	22	(7.86)	1.80	(1.11 – 2.91)	0.02
Experienced discrimination from the police or in courtsl, last 6 months
No	59	(73.75)	269	(96.07)	Reference
Yes	15	(18.75)	25	(8.93)	2.08	(1.31 – 3.31)	<0.01
Incarcerated, last 6 months
No	84	(93.33)	328	(97.04)	Reference
Yes	6	(6.67)	9	(2.66)	1.96	(1.02 – 3.76)	0.04
Sexual health behaviors
Met any sexual partners in last 6 months in street or other public setting
No	44	(48.89)	301	(89.05)	Reference
Yes	46	(51.11)	37	(10.95)	4.35	(3.10 – 6.09)	<0.01
Met any sexual partners in last 6 months via dating app
No	78	(86.67)	304	(89.94)	Reference
Yes	12	(13.33)	34	(10.06)	1.28	(0.76 – 2.16)	0.36
Met any sexual partners on Craigslist in the last 6 months
No	76	(84.44)	329	(97.34)	Reference
Yes	14	(15.56)	9	(2.66)	3.24	(2.21 – 4.77)	<0.01
Met any sexual partners in last 6 months on some other online forum
No	67	(74.44)	309	(91.42)	Reference
Yes	23	(25.56)	29	(8.58)	2.48	(1.71 – 3.61)	<0.01
Any condomless anal intercourse, last 6 months
No	33	(36.67)	240	(71.01)	Reference
Yes	54	(60.00)	94	(27.81)	3.02	(2.05 – 4.43)	<0.01

Notes:

^aPercentages column-calculated (i.e., the denominator is the total number of participants in the column of interest), unless otherwise specified.

^bBivariable ratio comparing the baseline prevalence of sex work by socio-demographic characteristics, experiences of violence, discrimination, and criminalization, and sexual health behaviors to their respective reference groups.

^cConfidence interval.

^dPercentage calculated out of total baseline participants (n = 429).

^e Other/multiple race/ethnicity includes: Asian, non-Hispanic/Latinx (n = 12); American Indian/Alaska Native, non-Hispanic/Latinx (n = 4); Native Hawaiian or Pacific Islander, non-Hispanic/Latinx (n = 8); “Other” race/ethnicity, non-Hispanic/Latinx (n = 9); and more than one race, non-Hispanic/Latinx (n = 65)

^fStable housing includes: owning or renting a house/apartment, or living in a single room occupancy unit.

^gUnstable housing includes: couch surfing, homelessness, living in a shelter, living in a residential treatment program, or living in transitional/supportive housing.

^hIncludes income from non-criminalized employment sources, such as a full- or part-time job. Excludes income from sex work or other criminalized sources; excludes non-criminalized supplemental income.

ⁱIncludes income from supplemental sources, such as general assistance, food stamps, social security, disability, unemployment, alimony/child support, student loans, or income from friends/family/partners. Excludes income from sex work or other criminalized sources; excludes income from part- or full-time, non-criminalized employment.

^jIncludes income from criminalized sources other than sex work, drug dealing/selling or scamming/stealing. Excludes income from part- or full-time, non-criminalized employment; excludes non-criminalized supplemental income.

^kSum of income sources (excluding sex work). Includes income from full- or part-time jobs, general assistance, food stamps, social security, disability, unemployment, alimony/child support, student loans, income from friends/family/partners, drug dealing/selling, alimony/child support, scamming/stealing, or other source other than sex work.

^lNot collected at baseline, therefore data taken from each participant’s 6-month survey (n = 80 sex workers and n = 280 participants not engaged in sex work).

Table 1 also shows that participants who reported experiencing transphobic hate crimes (aPR = 1.80, 95%CI = 1.11 – 2.91, p = 0.02), discrimination from police or in courts (aPR = 2.08, 95%CI = 1.31 – 3.31, p < 0.01), and incarceration (aPR = 1.96, 95%CI = 1.02 – 3.76, p = 0.04) were more likely to be engaged in sex work at baseline compared to their counterparts. Sex work was significantly associated with a number of sexual health behaviors at baseline, including meeting recent sexual partners in the streets/public settings (aPR = 4.35, 95%CI = 3.10 – 6.09, p < 0.01), meeting sexual partners on Craigslist (aPR = 3.24, 95%CI = 2.21 – 4.77, p < 0.01), meeting sexual partners on other online forums (aPR = 2.48, 95%CI = 1.71 – 3.61, p < 0.01), or engaging in condomless anal intercourse (aPR = 3.02, 95%CI = 2.05 – 4.43, p < 0.01).

Results from multivariable GEE models (adjusting for race/ethnicity and social transition) of the relationship between time-varying sex work engagement and outcomes of interest are displayed in Figure 2. Over 18 months, TWSW/OWSW had higher adjusted odds of being unstably housed (adjusted odds ratio, aOR = 1.84, 95% CI = 1.39 – 2.45, p < 0.01) and having income in the last month from criminalized sources (aOR = 5.86, 95% CI = 3.37 – 10.19, p < 0.01) compared to those not engaged in sex work. TWSW/OWSW also had higher mean income sources compared to their counterparts over 18 months (mean income sources = 1.23, 95% CI = 1.06 – 1.42, p < 0.01). TWSW/OWSW had greater odds of experiencing transphobic hate crimes (aOR = 1.10, 95% CI = 1.03 – 1.17, p < 0.01), discrimination from police and in the courts (aOR = 1.12, 95% CI = 1.04 – 1.22, p < 0.01), and incarceration (aOR = 3.13, 95% CI = 1.70 – 5.79, p < 0.01) compared to those not engaged in sex work over the study period. TWSW/OWSW compared to their counterparts had higher adjusted odds of meeting sexual partners in the streets/public settings (aOR = 5.20, 95% CI = 3.53 – 7.66, p-value < 0.01), on Craigslist (aOR = 3.07, 95% CI = 1.53 – 6.16, p < 0.01), or on some other online forum (aOR = 2.70, 95% CI = 1.64 – 4.42, p < 0.01) over 18 months of follow-up. Finally, engagement in condomless anal intercourse was higher for TWSW/OWSW compared to those not engaged in sex work over the study period (aOR = 2.83, 95% CI = 2.01 – 3.97, p < 0.01).

Fig. 2.

Adjusted odds of select outcomes over the 18-month study period for trans women sex workers and others who do sex work (TWSW/OWSW) compared to those not engaged in sex work, Trans*National, 2016–2019.

One difference-in-differences analysis showed additive interaction (Figure 3). From pre- to post-FOSTA-SESTA, the adjusted mean number of income sources (excluding income from sex work) reported by TWSW/OWSW compared to those not engaged in sex work decreased (from 1.79 to 1.48 for TWSW/OWSW and from 1.52 to 1.47 for TW not engaged in sex work; p=0.011).

Fig. 3.

Predicted number of income sources for trans women sex workers and others engaged in sex work (TWSW/OWSW) compared to those not engaged in sex work, pre-versus post-FOSTA-SESTA, Trans*National, 2016–2019.

Discussion

Numerous income and HIV-related disparities existed for TWSW/OWSW compared to those who did not do sex work in San Francisco over a period of 18 months. TWSW/OWSW had higher adjusted odds of being unstably housed, reporting income from criminalized sources other than sex work, and experiencing victimization, discrimination, and incarceration. They also had greater adjusted odds of meeting partners in public or online settings, and engaging in condomless anal intercourse. Difference-in-difference analyses did not indicate whether FOSTA-SESTA exacerbated these socio-economic and HIV-related risk outcomes.

Over a third of TWSW/OWSW reported additional income from non-criminalized part- or full-time employment and a majority of them received non-criminalized supplemental income, suggesting that many TWSW/OWSW did not solely rely on income from the sex trade. While TWSW/OWSW reported a higher number of income sources (in addition to sex work) over time compared to those not engaged in sex work, the difference-in-difference regression model for this outcome suggested that FOSTA-SESTA may have reduced TWSW/OWSW’s income sources to where they reported the same number of income sources post-law as those not engaged in sex work. This finding adds nuance to findings from other research. In a study of male sex workers, researchers found that most sex workers reported livable wages, and that income from sources other than sex work was important¹². The present analysis demonstrates that income from other sources is important for TWSW/OWSW, but may not be livable. While TWSW/OWSW relied more on income from sources other than sex work, they also were more likely to suffer from other economic constraints such as unstable housing and incarceration. Moreover, TWSW/OWSW compared to those not engaged in sex work were more likely to report income from criminalized sources (e.g., drug dealing/selling, stealing/scamming).

Taken together, these findings provide preliminary evidence that FOSTA-SESTA may reduce income options for TWSW/OWSW and further trap them in a cycle of poverty, where they may be more likely to participate in HIV-related risk behaviors in order to make ends meet. Though street-based sexual partnering and condomless anal intercourse were not shown to increase from pre- to post-law, these behaviors were consistently higher for TWSW/OWSW compared to those not engaged in sex work over the study period.

To provide stronger evidence of the effect of FOSTA-SESTA on income sources, we need to rule out other explanations for the observed finding. One alternative explanation could be that the prevalence of sex work decreased over time (i.e., that participants who engaged in sex work pre-law no longer engaged in sex work post-law). This could have caused TWSW/OWSW and those not engaged in sex work to appear more similar in terms of number of income sources. However, sex work engagement was shown to be stable over time, and, according to Figure 3, the number of income sources appeared to decrease for both TWSW/OWSW and those not engaged in sex work.

Another explanation for the observed results involves unmeasured or residual confounding. Based on the directed acyclic graph motivating the analysis, we are confident that we included all known confounders. However, it is possible that unmeasured confounding could bias the findings.

A third explanation involves misclassification bias. This could be an issue if participants who engaged in sex work under-reported their sex work status. However, given that we operationalized sex work as “income” from sex work, we are confident that this provided more accurate reporting than if we had directly asked them whether they engaged in sex work. Even still, it is possible that participants who were classified as TWSW/OWSW only engaged in sex work on an episodic basis if they were in a bind, or engaged in other behaviors under the umbrella of sex work such as web camming. In that case, HIV risk may have been underestimated. Moreover, different kinds of sex work could have been impacted differently be FOSTA-SESTA, with participants who have in-person interactions being more affected. If this were the case, we would expect that we underestimated the effect of FOSTA-SESTA on number of income sources.

The observed results could also be influenced by selection bias. This could be an issue if our study overrepresented TWSW/OWSW who are more likely to experience poverty. In our analytic sample, most TWSW/OWSW lived in extreme poverty at baseline (76/90 = 84.44%). Most TW not engaged in sex work also experienced extreme poverty (247/338=73.98%). Thus, results may not be generalizable outside of contexts where extreme poverty is common. Another source of selection bias may occur by design of the cohort study. The present analysis also includes participants who were HIV-negative at baseline, possibly excluding those who would have been in a riskier context and therefore attenuating observed effects.

A final explanation for the results would be reverse causation (i.e., that participants’ number of income sources causes their engagement in sex work). However, given that the analyses were longitudinal, we are confident that temporality between exposures (sex work and law timing) and outcome (number of income sources) was upheld.

With alternative explanations seeming less likely, we are confident that the observed results are valid and represent the effect of FOSTA-SESTA on income sources for TWSW/OWSW compared to those not engaged in sex work.

Though results from difference-in-differences analyses of other outcomes were not statistically significant, these analyses were likely underpowered or reflect other underlying mechanisms at work. Our intent was to also run sub-group analyses by race/ethnicity, but we encountered precision issues and or models did not converge. Had we been able to run interactions by race/ethnicity, it is likely that these results would show null findings even if the underlying associations between sex work and HIV risk were present because of the small sample size. However, given that most of those who did sex work in the sample were trans women of color, we suspect that the estimates observed in the present analysis are largely driven by the risk context for trans women of color, who experience intersecting layers of marginalization on the basis of gender, sex worker status, and race.

Other than sample size limitations, another possible reason that we did not observe pre- to post-law differences in other outcomes could be that laws targeting sex workers also target trans women of color who are profiled as sex workers. Edelman posits that designation of physical spaces as “prostitution-free zones” fuels trans mobilization and perpetuates their marginalization.²⁵ While FOSTA-SESTA places no direct attack on physical space, it legalizes the policing of internet spaces once used to safely vet sex work clients, and pushes sex work to the streets which may, as a consequence, increase policing of sex workers and trans people profiled as sex workers. Thus, the comparison groups in this analysis – designated as TWSW/OWSW and those not engaged in sex work – may in fact be subject to contamination effects. That is, policies (such as FOSTA-SESTA) that target sex workers may not only affect TWSW/OWSW, but all trans women regardless of sex worker status.

It could also be that the number income sources is a more sensitive indicator of FOSTA-SESTA’s impact, both because we defined it continuously and because it could be a downstream effect. For example, increases in incarceration, which was experienced by TWSW/OWSW more over the study period, could have limited income opportunities from pre-to-post law. Laws such as FOSTA-SESTA also put sex workers at increased risk of being outed and consequently subjected to increased harm and reduced access to social and health services¹⁵. In a descriptive post-hoc analysis, we found that the percent of TWSW/OWSW reporting income from non-criminalized employment or income from general assistance and food stamps had both decreased. Had FOSTA-SESTA had its intended effect of reducing sex work, we would have expected non-criminalized income sources to increase for TWSW/OWSW, which was not the case here.

With these observations in mind, there are several directions that future research and interventions can take. This study highlights the urgent need for comprehensive, long-term follow-up data of trans women and TWSW/OWSW to accurately analyze policy effects, especially given the recent enactment of a number of other policies targeting trans women. Larger samples will also allow for subgroup analysis, such as assessing whether policy effects are larger for trans women of color.

Similar difference-in-differences approaches may help characterize the exacerbation of health disparities for TWSW/OWSW due to the COVID-19 pandemic. Just as the most marginalized communities are hit the hardest by HIV, so too do they bear the brunt of COVID-19.²⁶ Unemployment, incarceration, poverty, food insecurity, health care barriers, and stigma place trans women, especially Black and Latinx trans women and those in the sex trade, at disproportionate risk of COVID-19 and related adverse outcomes.^27–28 To date, a handful of letters to scientific journal editors have called for supporting community-led interventions that expand TWSW/OWSW’s access to emergency housing, hardship funds, personal protective equipment, COVID-19 testing, and health services without financial or legal repercussions.^26–28 As these immediate calls to action are addressed, it is imperative to collect data for analyzing policy and intervention impacts that will inform future public health responses.

Conclusions

Echoing the report on “Transgender Experiences in the Sex Trade”⁴ and other studies on sex workers and those involved in the sex trade^15,18, results from the present analysis add evidence that structural decriminalization efforts and economic empowerment interventions could eliminate health disparities for TWSW/OWSW and the rest of the sex work population. Full sex work decriminalization has been implemented in jurisdictions outside of the U.S, such as New Zealand. Decriminalization efforts in New Zealand removed laws penalizing engagement in the sex trade and instituted new laws to protect sex workers from violence and promote occupational health and safety standards²⁹. These efforts have been shown to reduce discrimination against sex workers, improve sex workers’ abilities to vet clients and negotiate condom use, and also promote emotional well being, workplace safety, and access to health and social services^29–34.

Though decriminalization efforts should be tailored differently for each country depending on the legislation in place, similar institutional-level actions in the U.S. could yield invaluable benefits for TWSW/OWSW and the rest of the sex work community. National LGBT organizations are taking concrete steps toward identifying and replacing laws that criminalize sex work, suggesting that “the push for the decriminalization of sex work in the United States is working to gain traction”⁴.

• Trans women in the sex trade experience socio-economic and health disparities.

• Engagement in the sex trade was stable over 18 months of follow-up.

• An anti-sex trafficking policy may exacerbate income inequity for sex workers.