Patterns of Polysubstance Use in Young Black and Latinx Sexual Minority Men and Transgender Women and Its Association with Sexual Partnership Factors: The PUSH Study

Abstract Background: Adult studies have demonstrated that polysubstance use increases HIV acquisition risk through increased sexual behaviors, however, few studies have examined polysubstance in young Black and Latinx sexual minority men (SMM) and transgender women (TW). Methods: We used cross-sectional data from 466 young Black and Latinx SMM and TW living in four high HIV-burden US cities enrolled in the PUSH Study, a status-neutral randomized control trial to increase HIV prevention and treatment adherence. We examined data for patterns of polysubstance use comparing age differences of use and explored associations between substance use and sexual partnership factors - inconsistent condom use, pressure to have condomless anal sex, and older partner, using bivariate and multivariate analyses. Results: Most participants described prior substance use with alcohol and cannabis being most common (76% each) and 23% described other illicit drug use, including stimulants, cocaine, hallucinogens, sedatives, opioids, and inhalants. Polysubstance use was common with nearly half (47%) of participants reporting alcohol and cannabis use, 20% reporting alcohol, cannabis, and one other illicit drug use, and 19% reporting alcohol or cannabis use plus one other illicit drug use. Polysubstance use was associated with greater adjusted odds of pressure to have condomless anal sex, older partner (>5 years older), and inconsistent condom use. Conclusions: Associations of polysubstance use with sexual practices and sexual partnerships that are known predictors of HIV acquisition or transmission among Black and Latinx SMM and TW underscore the need for combination interventions that include substance use treatment alongside antiretroviral-based and partner-based HIV prevention and treatment interventions. Trial Registration: ClinicalTrials.gov Identifier: NCT03194477


Introduction
Age and race-related disparities exist for sexual minority men (SMM) and transgender women (TW) in the US.In 2020, SMM accounted for 71% of new HIV infections with young Black and Latino SMM aged 13-24 years, which accounted for 54% and 27% of new infections in this age group (Centers for Disease Control and Prevention, May 2022).Incidence rates are thought to be as high as 8.81-10.9per 100 person-years in young Black SMM, 3.15 per 100 person-years in Latinx, and 3.75 per 100 person-years in mixed-race youth (Mustanski et al., 2020;Sullivan et al., 2015).Transgender women accounted for 2% of new HIV diagnoses in 2020, with rates highest among Black and Latinx TW and 28% of new infections occurring among youth aged 13 to 24 (Centers for Disease Control & Prevention, 2022).Among Black and Latinx TW and nonbinary or gender queer youth, incidence rates may be around 2.91 per person-years, slightly lower than some rates of Black SMM (Mustanski et al., 2020).Researchers have called for further explorations of racial disparities, including partner factors associated with HIV incidence, including experiences of condomless anal intercourse, sexual coercion, substance, and polysubstance use, that may predispose young SMM and TW to HIV (Martino et al., 2021;Jalil et al., 2022).

HIV risk and substance use
Stimulant use is strongly associated with sexual HIV acquisition risk among SMM and risk is significantly increased with co-occurring binge drinking and cannabis use (Swartz & McCarty-Caplan, 2018;Torres et al., 2020).Other work has identified methamphetamine use as a key predictor of HIV seroconversion among SMM (Grov et al., 2020;Wirtz et al., 2023).In one study, men who used club drugs defined as ketamine, ecstasy, GHB, cocaine, or methamphetamine had a significantly higher number of missed pre-exposure prophylaxis (PrEP) doses compared to men who did not use club drug; however, this work did not specifically explore the role of coexisting other illicit drug use in younger men (Grov et al., 2020).
Polysubstance use, defined as the use of more than one psychoactive substance of abuse, has been identified as a risk factor in HIV acquisition among adult SMM and TW participants (Ostrow et al., 2009;Swartz & McCarty-Caplan, 2018).Adult studies have demonstrated that polysubstance use increases HIV acquisition risk through increased sexual behaviors, such as group sex (Hirshfield et al., 2015), condomless anal sex (Daskalopoulou et al., 2014), and exchange sex (Bond et al., 2019).However, little work has sought to understand the role of polysubstance use in sexual partnerships and whether the combination of polysubstance use might increase young SMM and TW's risk for HIV.In a prior scoping review, young SMM were found to be more likely to engage in polysubstance use than older SMM in three of five studies that examined polysubstance use and sexual risk (Jeffries et al., 2018).The two studies which found either no association or less polysubstance use in young SMM used an imprecise measure of younger age (< =39 years) or focused only on Asian/Pacific Islander SMM.Lifetime substance use as high as 90% has been described among young TW, with some studies reporting that over 50% of study participants describe sexual intercourse under the influence of polysubstance use (Brennan et al., 2012;Garofalo et al., 2007;Wilson et al., 2010).

Polysubstance use trends & HIV risk
The data around race and age are less clear.Some prior work has suggested that polysubstance use is heightened in adolescents and young adults of color (Native American, Hispanic/Latinx, or Black) (Wohl et al., 2007;Martino et al., 2021), while other studies have demonstrated higher rates among White non-Hispanic youth for certain substances, such as methamphetamine (Irwin & Morgenstern, 2005), and have suggested that use is dependent on living within certain geographical areas (Reback et al., 2013).Despite national data showing that youth engage in polysubstance use and use is associated with sexual exploration (Martino et al., 2021), few studies have explored age-specific patterns of polysubstance use among young SMM and TW and whether the severity of use changes during this developmental period.According to the 2021 Monitoring the Future Survey, lifetime substance use among 10 th and 12 th graders in 2021 were respectively: 25.0% and 41.3% for illicit drugs, 34.7% and 54.1% for alcohol, 22.0% and 38.6% for cannabis, 5.2% and 4.9% for amphetamines, 3.5% and 7.1% for hallucinogens (NAHDAP/ICPSR, 2021).Similarly, use in the past year among 10 th and 12 th graders were: 18.7% and 32.0% for illicit drugs, 28.5% and 46.5% for alcohol, 17.3% and 30.5% for cannabis, 2.7% and 2.3% for amphetamines, 2.2% and 4.1% for hallucinogens, respectively.However, data presenting polysubstance use trends in this sample were limited.
Polysubstance use has been tied to sexual violence and condomless anal sex and can occur with and reinforce risk for HIV (Arrington-Sanders et al., 2022;De Boni et al., 2018;Logie et al., 2020).Both Black and Latinx SMM and TW experience age-discordant partnerships (e.g., partners greater than five years older), sexual trauma, or abuse during adolescence, in the context of substance use (Arrington-Sanders et al., 2013;Downing et al., 2020;VanDevanter et al., 2011).These experiences may further predispose youth to potential syndemics, which are the multiple co-occurring conditions (e.g., substance use, condomless sex) in the context of deleterious social and/or physical conditions (e.g., sexual violence, younger age/older partner) that mutually reinforce and increase vulnerability to an adverse condition (e.g., HIV) (Mustanski et al., 2007;Stall et al., 2003).
The main objective of the current study was to characterize patterns of polysubstance use in young Black and Latinx SMM and TW comparing age differences of use; and explore associations between substance use and HIV risk behaviors focusing on three core partnership factors -inconsistent condom use, pressure to have condomless anal sex, and older partner.We hypothesized that more severe substance use would be associated with greater sexual risk.

Participants and procedures
Data for this analysis were drawn from the Providing Unique Support for Health (PUSH) Study of young Black and Latinx SMM and TW living in four cities in the US, (Baltimore MD, Washington DC, Philadelphia PA, and Tampa/St.Petersburg FL).Participants were recruited between August 2017 -June 2021 (Arrington-Sanders & Celentano;Arrington-Sanders et al., 2020).Briefly, the PUSH Study is a mixed-methods study with a qualitative, formative component, baseline survey, and two embedded 18-month randomized clinical trials (RCTs) to improve viral suppression and uptake of pre-exposure prophylaxis in young Black and Latinx SMM and TW at-risk for or living with HIV.Eligibility for the PUSH Study included: (1) 15-24 years old; (2) reside in the four study city sites; (3) male sex at birth; (4) self-identified Black/African American or Hispanic/Latinx race/ethnicity; and (5) oral/anal sex with a man in the prior 12 months.Participants were excluded if they did not consent to the screening HIV test to determine their HIV status or if there was no verified test result on their medical records.This analysis used cross-sectional survey data from the baseline assessment, which was conducted prior to randomization and enrollment into the embedded trials.
The PUSH study protocol was approved by the Institutional Review Boards of all participating institutions including The Johns Hopkins University in Baltimore, MD; Children's National Medical Center and Whitman-Walker Health in Washington DC; Children's Hospital of Philadelphia (CHOP), Philadelphia, PA; and Western IRB, FL.
Participants were recruited using modified respondent driven sampling (RDS) with targeted seed selection, in which seeds were identified through direct and indirect recruitment (Wirtz et al., 2021).Direct seed selection consisted of informational flyers directly given to youth from study staff at a clinical site, community-based organization, or in-person events.Indirect seed selection consisted of informational flyers posted at clinical sites that serve young SMM and TGW, offices of community-based organizations as well as a web-based approach with electronic ads placed on webpages, social and sexual networking sites frequented by young SMM (e.g., Jack' d, Black Gay Chat, Grindr).All eligible and participating youth who were direct and indirect recruits were then eligible to become RDS seeds and refer other participants.Eligible and participating youth were offered electronic or paper coupons to recruit peers.Participants were provided $5-15 depending on IRB restrictions for each eligible and participating peer recruit.Because the purpose of recruitment was for enrollment into the clinical trials, not generation of population-based estimates, we allowed for a large number of seeds and did not aim for long recruitment chains to approximate the underlying networks; thus, we provide sample estimates which would be similar to those estimated through a convenience sample (Arrington-Sanders et al., 2020).
All individuals meeting eligibility and voluntary study consent completed a baseline survey and HIV testing.The baseline survey included questions about demographics (e.g., age, sexual orientation, gender identity, housing status), sexual behavior (e.g., number of partners, type of sex, etc.), substance use, HIV history, engagement in preventive behavior (e.g., condom and PrEP use), disclosure of sexual orientation to provider, and city location of recruitment.Participants were offered a rapid oral HIV test (OraQuick Advance® or INSTI) at baseline unless HIV status was known via medical record.Participants were paid $50 for completion of the baseline survey and the HIV test.

Substance use measures
Substance use was measured using the CRAFFT a validated six-item screening tool designed to detect alcohol and drug use (Knight et al., 1999;2002), and the NIDA Modified (NM)-ASSIST (Humeniuk et al., 2008).Both instruments were used to account for the age range of the sample and to include a measure of substance use severity.The CRAFFT is validated for use in adolescents (Knight et al., 2002) while the NM-ASSIST is validated for use in adolescents and young adults 18 years or older, although it has been used in adolescent populations (Gryczynski et al., 2015).Participants answered the NM-ASSIST first for each substance to assess their history of lifetime use followed by recent (past 3 months) substance use.The substances reported included tobacco, alcohol, cannabis, amphetamine-type stimulants (speed, diet pills, ecstasy, etc.), cocaine, hallucinogens, sedatives, opioids, and inhalants.Drugs used with low frequency including amphetamine-type stimulants (speed, diet pills, ecstasy, etc.), cocaine, hallucinogens, sedatives, opioids, and inhalants were further combined and categorized as other illicit drug use.For 3-month substance use, frequencies (e.g., never, once, or twice, monthly, weekly, daily, or almost daily) were assessed for each substance.For this analysis, we examined lifetime and substance use in the past 3 months.
To evaluate polysubstance use, we created a composite variable using data from the NM-ASSIST with categories based on combinations of substances reported to have been ever used (excluding tobacco): 1) both alcohol and cannabis only; 2) alcohol and cannabis and at least one of the other illicit drugs; 3) alcohol or cannabis with one other illicit drugs; 4) no lifetime substance use.Combined categories were created based on the most frequently used substances reported by participants and reflected the prevalence of use in adolescents and young adults (Goodwin et al., 2022).The CRAFFT score was categorized into − 0 (no dependence), 1-2, and ≥3 based on prior literature in African American adolescents that suggest the use of a combination of cutoffs to demonstrate the severity of substance dependence or problem consumption (Lee et al., 2021).

Sexual partnership measures
Participants were asked to report lifetime and last 3-month sexual behavior (oral, anal, vaginal) including sexual behavior with the last 4 partners (Donenberg et al., 2001).
Participants were asked details about sexual partners, including dating status (not dating, casual, exclusive), age, HIV status, and condom use in the last three months.Inconsistent condom use was based on the question "In the last 3 months, how often did you/your male partner wear a condom?" and was coded as 1 if the response was "Never" or "Half the time" and 0 if the response was "Always".The outcome pressure to have condomless anal sex was based on the question: "To what extent do you feel pressure from other people, including sex partners, to have anal sex without a condom?"It was coded as 0 if the response was "Not at all" and 1 if the response was "A little", "Somewhat", or "Very much".Partner age discrepancy was defined as having a partner in the last three months that was five years or older than the participant.The main exposure factor of interest was polysubstance use.

Statistical analysis
We explored associations between polysubstance use, the independent variable of interest, and HIV risk behaviors focusing on dependent variables: inconsistent condom use, pressure to have condomless anal sex, and partner age discrepancy.Covariates known previously to be associated with sexual risk behaviors and potential confounders included age, race/ethnicity, education, gender identity, sexual identity, HIV status, disclosure of sexual orientation, history of transactional sex, and unstable housing (Arrington-Sanders et al., 2022;Mena et al., 2017).Disclosure of sexual orientation was assessed with the following question, "Have you disclosed [or told] your sexual orientation or sexual preferences to your medical provider?"Transactional sex was a binary response to the question, "Have you ever had sex with another male in exchange for money, a place to stay, or food?"Unstable housing was defined as an affirmative response to the following question, "In the past 12 months, have you been without a regular place to stay?" We first examined patterns of drug use using trend graphs over age and the distribution of specific drugs used within each category of polysubstance used for "ever" and "3-month use".We focused on lifetime use rather than 3-month use patterns to allow us to be powered to investigate variance in effects related to other illicit drug use and not just variance in effects related to cannabis, alcohol, and tobacco, which we would be restricted to if we used past 3-month or past 30-day rates.Descriptive statistics are summarized as simple unweighted sample proportions; we did not use statistical weighting to calculate population estimates (as is common with RDS), given the objective of examining patterns and variables associated with polysubstance use.
Next, we examined the association between lifetime polysubstance use and our defined dependent variables (inconsistent condom use and having an older male partner in the last three months, and ever experiencing pressured sex to engage in sex without a condom), using bivariate, and multivariate analyses.Bivariate analysis was performed using frequency tables and associations were estimated with logistic regression models accounting for clustering by geographic location.Similarly, multivariate logistic models with robust variance accounting for clustering within geographic site were developed for each outcome separately.The initial model included all sexual risk factors and potential confounders associated with the outcome with a pvalue of <0.3, to ensure all measured potential confounders are considered.We chose to set the pvalue at a higher level to explore potential factors associated with outcomes, examine confounders, and avoid masking or blocking associations that could occur with a low pvalue cutoff.All models included the substance use pattern as well as the CRAFFT score and adjusted for clustering by city.The final model for each outcome was developed using stepwise procedures.Model results are shown as odd ratios (OR), and associated pvalue.

Results
A total of 466 participants contributed to this analysis.The study sample is described in Table 1.Participants came from all four cities.The mean age of participants was 21.3 years (SD = 2.45) with 10% being under the age of 18, most participants identified as African American/Black (n = 265, 57%), with 15% (n = 69) identifying as Latinx.Most participants self-identified as gay (n = 277, 59%) or bisexual (n = 115, 25%), completed high school or higher education (78%), had a cell phone (92%), and had health insurance (87%).Fewer participants (14%, n = 67) identified as transgender or gender diverse.Twenty-four percent of participants described unstable housing in the last year.Participants had an average of 23.0 total lifetime partners, with 50% reporting 10 or more lifetime partners.
The frequency of reported substance use is shown in Table 2.The minority (n = 67, 14%) described no prior substance use.Nearly forty percent (38%) described lifetime tobacco use, 76% (n = 354) alcohol use, 76% (n = 353) cannabis use, and 23% other illicit drug use with amphetamines reported by 10%.Polysubstance use pattern is also shown in Figures 1-3 with grouping defined by the most common combinations of substances.The frequency of using specific substances for each of the three groups defined by the combinations of reported use is reported for lifetime ever use (Figure 1) and the last three months of use (Figure 2), with similar patterns emerging in both timeframesnearly half (47%) of participants reporting alcohol and cannabis use, 20% of participants reporting alcohol, cannabis, and one other illicit drug use, and 19% reporting alcohol or cannabis use plus at least one other illicit drug use.Most participants who reported lifetime alcohol and cannabis use described using alcohol (98%) or cannabis (94%) in the last 3 months.Among participants that reported alcohol, cannabis, and other substance use, the most common other substance used was cocaine (35%) followed by hallucinogens (30%).Age patterns emerged in the data (Figure 3), with younger adolescents (15-19 years old) describing more lifetime cannabis use than alcohol use, though with a notable inflection at age 21, in which young adult participants (> =20 years) described more lifetime and past 3-month alcohol than cannabis.Over the age groups, amphetamine, cocaine, and sedative use increased in the higher age groups.Similar patterns were seen with reported use in overall and last 3 months of use (Figure 3).The bi-variate association of lifetime substance use with sexual risk behaviors is shown in Table 2. Participants who described alcohol and cannabis had 1.99 greater odds of inconsistent condom use, 2.67 greater odds of pressure to engage in condomless anal sex, and 2.10 greater odds of having a partner 5 years older in the last 3 months than non-substance-using peers.This effect was magnified in participants who described alcohol, cannabis, and other substance use-where participants had over 3 times greater odds of inconsistent condom use and 2.9 odds of pressure to engage in anal sex without a condom and have a partner 5 years older.A higher CRAFFT score >2 was significantly associated with all outcomes with increased odds of 2 or higher.Looking at specific substances, patterns emerged with alcohol and cannabis each having an association with inconsistent condom use, pressure to engage in anal sex without a condom, and having a recent older partner, which is consistent with the associations shown for the group that reported using both.Amphetamine, cocaine, and hallucinogen use were associated with pressure to engage in anal sex without a condom, whereas cocaine use was also associated with having a recent older partner, and hallucinogen use was associated with inconsistent condom use.
The association of substance use with the outcomes adjusting for participants' characteristics and potential confounders is shown in Table 3. History of polysubstance use with alcohol and cannabis was associated with 1.73 greater adjusted odds of pressure to have condomless anal sex and 1.69 greater adjusted odds of having an older partner.Reporting alcohol, cannabis, and other substances was associated with 1.58 greater adjusted odds of inconsistent condom use.Alcohol or cannabis use plus other substance use was not associated with outcomes in the final adjusted model.A higher score on CRAFFT was also associated with pressure for condomless anal sex in the adjusted model.Independent relationships were seen for Latinx ethnicity and disclosure of one's sexual orientation with inconsistent condom use and having an older partner and higher number of sexual partners associated with all outcomes.

Discussion
In this analysis, we sought to characterize patterns of polysubstance use in young Black and Latinx SMM and TW comparing age differences of use; and explore associations between substance use and HIV risk behaviors focusing on three core partnership factors -inconsistent condom use, pressure to have condomless anal sex, and partner age disparity.Participants in this sample described very high substance use overall and in the last 3 months.Polysubstance use was independently associated with inconsistent condom use, pressure to engage in anal sex without a condom, and engaging in sex with an older partner.We hypothesized that more severe substance use would be associated with greater sexual risk; in fact, participants who described severe or problem substance use, as marked by a higher CRAFFT score, reported greater odds of pressure to engage in condomless anal sex.Polysubstance use was very common with 20% of participants reporting alcohol, cannabis, and at least one other illicit drug use, and 19% reporting alcohol or cannabis use plus one other illicit drug use.This rate is slightly lower than adult analyses that suggest that non-medical prescription substance use with corresponding alcohol use may be as high as 27% lifetime use (Votaw et al., 2019).However, patterns of polysubstance use increased with older age with participants describing more illicit drug use with amphetamine, cocaine, and sedatives in the higher age groups in addition to alcohol and cannabis use.Polysubstance use was very common among participants who described both alcohol and cannabis use, with common substances including cocaine (35%) followed by hallucinogens (30%).These relationships were independently seen among youth who reported a Latinx ethnicity, had disclosed their sexual orientation to their provider and reported a higher number of sexual partners.The findings of this work suggest that substance use in young Black and Latinx SMM and TW may not be limited to one substance, but multiple substances, which may predispose youth who use substances to different inherent risks for HIV acquisition than non-using peers.
Other studies have demonstrated that the co-occurrence of high rates of condomless anal sex and substance use can predispose young SMM and TW to HIV (Brennan et al., 2012;Halkitis et al., 2015).Experiences with polysubstance use and characteristics of sexual partners, for example, older partnerships and pressure to engage in sex without a condom during substance use provide evidence that these youth may be particularly vulnerable to social contexts that predispose them to HIV (De Boni et al., 2018).Polysubstance use, combined with partner-specific factors like coerced sex and partners of discrepant age, may provide more opportunities for inconsistent condom use for youth.Youth who have a higher number of sex partners may inherently be particularly vulnerable.In this sample, participants had an average of 23.0 total lifetime partners, with 50% reporting 10 or more lifetime partners.This is similar to earlier work on youth living with HIV and high substance use (Jennings et al., 2009).Higher number of sex partners may inherently increase one's risk for more opportunities of inconsistent condom use, coerced sex, and partners of discrepant ages.The results of this analysis also suggest that Latinx youth and youth with a higher number of sex partners and more severe substance use may be particularly vulnerable to risk.More research is needed to understand this relationship and whether polysubstance use in young SMM and TW is intermittent occurring predominantly only within sexual partnerships, chronic, or as a means of coping.
A major goal of the current study was to examine associations between lifetime polysubstance use categories and HIV risk behaviors among young Black and Latinx SMM and TW.Other aspects of substance use behaviors related to frequency and temporal patterns of use and co-use (e.g., simultaneous use vs. concurrent use), use motives, and substance-related problems may also contribute variance to HIV outcomes in this population.Regarding frequency, a recent study by Morgan and colleagues examined cannabis use patterns in young Black SMM and found that three quarters of respondents used cannabis, with intermittent use (41%) being more common than heavy use (32%), and heavy use being associated with lower awareness of HIV status (Morgan et al., 2016).These findings parallel other studies in youth samples showing that chronic/heavy substance use, particularly with cannabis and alcohol, can impair cognitive functioning (Scott et al., 2018) and short-term memory (Dougherty et al., 2013).In contrast, intermittent substance use may impact different mechanisms of decision-making within partnerships and contribute to condomless anal sex and impaired communication around condoms (Sander et al., 2013;Wray et al., 2019).These factors (substance frequency, cognitive functioning, and substance-related problems) and their associations with decision making and HIV risk behaviors warrant further investigation in young Black and Latino SMM and TW who co-use cannabis and alcohol (Yurasek et al., 2017).
Temporal aspects of substance use, and social and contextual factors related to when and with whom different substances are used/co-used and how substances are used in relation to sex may also be important in this population.Combined with partner-specific factors, such as older partners and sexual coercion, younger SMM and TW may be particularly vulnerable to substance use at time of sex.Tailored interventions that address the different ways youth may be using/co-using multiple substances are needed to address the nuance of substance use within and outside of relationships.Such approaches would also require providers to address assumptions about the context of substance use in Black and  Latinx SMM and TW and factors that contribute to sexual pressure or coercion within relationships (Brooks et al., 2021), including existing marginalized social identities (race, gender, sexuality, class, etc.) and having to navigate and experience larger macro systems of oppression (racism, heteronormativity, homophobia, etc.) that intersect to impact an individual's health, relationships, and quality of life (Crenshaw, 1989).Stimulant use, particularly cocaine and amphetamine use before or during sex has been associated with increased risk for HIV acquisition in SMM and TW (Nerlander et al., 2018).Cocaine and amphetamine use has been reported predominantly during Chemsex for the pharmacological effects during sex (e.g., orgasm, enhanced pleasure).In this analysis, we found that 35% of polysubstance users also reported cocaine use.This rate is higher than other reports of stimulant use in young SMM and TW (De Boni et al., 2018;Mimiaga et al., 2019) and may have significant implications for the participants in this sample.In bivariate analysis, we observed a direct relationship between participants who reported amphetamine, cocaine, and hallucinogen use and experiencing pressure to have condomless anal sex.Recent work by Xavier Hall suggests that polysubstance use that is inclusive of methamphetamine and cocaine use may contribute to poly-victimization among young SMM and TW (Xavier Hall et al., 2022).Partner-related coercive sex or pressure to engage in sex without a condom is a form of intimate partner violence experienced within relationships.
With prior work suggesting that younger SMM and TW may be at more risk for cocaine and amphetamine use before or during sex than older peers (De Boni et al., 2018;Jalil et al., 2022), interventions will need to be multi-factorial and address not only individual substance use but also incorporate aspects victimization, violence and pressure that may occur within the sexual partnerships of young SMM and TW (Peitzmeier et al., 2021).Continued research is needed to disaggregate components of HIV risk for young Black and Latinx youth-for example, how aspects of condomless sex, older partners, and sexual violence may be interrelated to substance use.Intersectional screening tools are needed to alert providers to vulnerability, and training to be able to address the social context that youth experience which may include aspects of racism, homophobia, and heteronormativity that may increase substance use and impact power dynamics and vulnerability within sexual relationships.Disclosure of sexual orientation in youth who also report polysubstance use and inconsistent condom use provides a key opportunity for providers to link youth to intersectional care that addresses both sexual risk and substance use.

Limitations
There are limitations of this work that should be noted.This data is cross-sectional, limiting our ability to assess temporality or to determine prospectively how substance use changed over time or was modified by specific partnership factors within relationships.The samples are from geographically different, but mostly urban locations, limiting our ability to draw conclusions about youth in other settings.More work is needed to delineate chronic polysubstance use from intermittent polysubstance use, because intermittent use during sexual partnerships may inherently present with different risks for HIV.We did not examine pre-exposure prophylaxis use as a dependent variable because we examined outcomes that would be relevant to both persons who were living with and at risk for HIV.This limited our ability to conclude how substance use impacted protected and unprotected anal sex.The data were drawn from a sample enrolled in two RCTs, thus findings should not be viewed as population prevalence estimates and may have limited generalizability.Additionally, this study focuses only on Black and Latinx SMM and TW, limiting comparisons to other populations.Despite these limitations, this work suggests that in some samples of young Black and Latinx SMM and TW, there are very high levels of polysubstance use which may create a key vulnerability for HIV acquisition within partnerships.

Conclusions
High levels of substance use, polysubstance use, sexual practices and partner characteristics that are known predictors of HIV acquisition were observed in this sample of Black and Latinx sexual and gender minority youth.The relationships between substance use and inconsistent condom use, pressure to have condomless anal sex, and partner age discrepancies likely work through the pharmacologic effects of these substances that are known to impair judgment, decision-making, and negotiation alongside power imbalances within sexual partnerships.These findings highlight highly intervenable relationships that could be addressed through combination interventions that include substance use treatment alongside antiretroviral-based prevention and treatment, partner-based interventions, IPV and safety planning services, and individual support to develop condom negotiation skills to address HIV disparities among Black and Latinx sexual and gender minority youth.

Figure 1 .
Figure 1.Lifetime substance use by groups of polysubstance use.

Figure 2 .
Figure 2. Substance use in last 3 months by groups of polysubstance use.

Table 1 .
Description of study cohort: Participants demographic and socio-economic characteristics.
1 note: HiV status was confirmed by a study HiV test for 88.4% of the participants; 26 of the youth with HiV and 28 of the at-risk participants HiV status was based on self-report.

Table 2 .
the association of substance use with sexual risk outcomes.For this outcome there were 20 missing observations.bi-variate logistic models with robust variance adjusting for clustering at geographic site.pvalue significant at <0.05 have been bolded.

Table 3 .
adjusted associations of substance use patterns with sexual risk outcomes.
*Multivariate logistic regression with robust variance adjusted for clustering by geographic location and continuous age.Results are shown for factors that were significant for at least one outcome.pvalue significant at <0.05 have been bolded.