Latent Profile Patterns of Network-Level Norms and Associations with Individual-Level Sexual Behaviors: The N2 Cohort Study in Chicago

Abstract

Individual-level behavior can be influenced by injunctive and descriptive social network norms surrounding that behavior. There is a need to understand how the influence of social norms within an individual’s social networks may influence individual-level sexual behavior. We aimed to typologize the network-level norms of sexual behaviors within the social networks of Black sexual and gender minoritized groups (SGM) assigned male at birth. Survey data were collected in Chicago, Illinois, USA, between 2018 and 2019 from Black SGM. A total of 371 participants provided individual-level information about sociodemographic characteristics and HIV vulnerability from sex (i.e., condomless sex, group sex, use of alcohol/drugs to enhance sex) and completed an egocentric network inventory assessing perceptions of their social network members’ (alters’) injunctive and descriptive norms surrounding sexual behaviors with increased HIV vulnerability. We used Latent Profile Analysis (LPA) to identify network-level norms based on the proportion of alters’ approval of the participant engaging in condomless sex, group sex, and use of drugs to enhance sex (i.e., injunctive norms) and alters’ engagement in these behaviors (i.e., descriptive norms). We then used binomial regression analyses to examine associations between network-level norm profiles and individual-level HIV vulnerability from sex. The results of our LPA indicated that our sample experienced five distinct latent profiles of network-level norms: (1) low HIV vulnerability network norm, (2) moderately high HIV vulnerability network norm, (3) high HIV vulnerability network norm, (4) condomless sex dominant network norm, and (5) approval of drug use during sex dominant network norm. Condomless anal sex, group sex, and using drugs to enhance sex were positively and significantly associated with higher HIV vulnerability social network norm profiles, relative to low HIV vulnerability norm profiles. To mitigate Black SGM’s HIV vulnerability, future HIV risk reduction strategies can consider using network-level intervention approaches such as opinion leaders, segmentation, induction, or alteration, through an intersectionality framework.

Introduction

Despite the availability of biomedical HIV prevention strategies such as pre-exposure prophylaxis (PrEP), antiretroviral treatment (ART), and Treatment as Prevention (TasP), Black sexual and gender minoritized groups assigned male at birth such as cisgender sexual minoritized men (SMM) and transgender women (TW), continue to be subpopulations at exacerbated vulnerability to HIV in the US (Centers for Disease Control Prevention, 2021; Hess et al., 2017; Matthews et al., 2016). Due largely in part to the benefits of PrEP, ART, and TasP, HIV incidence decreased among White SMM from 2010 to 2019 but did not for Black SMM (Kanny et al., 2019). While 1 in 257 US White heterosexual men will be diagnosed with HIV in their lifetimes and 1 in 11 for white SMM, this rate is 1 in 2 for Black SMM (Hess et al., 2017; Matthews et al., 2016). Alarmingly, a 2021 Centers for Disease Control and Prevention (CDC) surveillance study found that 62% of Black transgender women were living with HIV (Centers for Disease Control Prevention, 2021) and a simulation study by Matthews et al. (2016) found that up to 61% of Black SMM are predicted to be diagnosed with HIV by age 40 years (Matthews et al., 2016).

Structurally, Black SMM and TW experience co-occurring and intersecting discrimination from racism, homophobia, transphobia, biphobia, and HIV-related stigma (Arnold et al., 2014; Cahill et al., 2017; Purdie-Vaughns & Eibach, 2008; Quinn et al., 2019), which can compound to create a new type of discrimination, as best described using the Intersectionality Framework (Bauer, 2014; Bowleg, 2008, 2012, 2013; Callander et al., 2019; Crenshaw, 2017; Duncan et al., 2019, 2020, 2021; English et al., 2020; Feelemyer et al., 2021; Quinn et al., 2019). Social network-based interventions may surmount intersecting structural and systemic oppressions faced by Black sexual minority men and transgender women to mitigate health disparities. Social networks can disseminate information, channel peer/personal influence, and enable attitudinal or behavioral change (Huang et al., 2016; Jaganath et al., 2012; Pagkas-Bather et al., 2020; Rosengren et al., 2016; Rössler, 2017; Wang & Muessig, 2017). This suggests that until health equity in HIV incidence and supranormal levels in the implementation of biomedical strategies are reached, researchers must consider investigating the reduction in individual-level HIV sexual vulnerability behaviors (e.g., condomless sex, group sex, using drugs or alcohol to enhance sex) among Black sexual and gender minoritized groups (SGM), including SMM and TW, using social network approaches which can account for intersectional stigma.

An approach to promote social justice and health equity for Black SMM and TW is to provide SMM and TW with HIV status-neutral care, which confers protection to HIV acquisition (i.e., PrEP and condoms) or prevention of HIV transmission (i.e., ART and condoms) through equitable access pathways for HIV care (Gardner et al., 2011; Myers et al., 2018). However, HIV status-neutral care is not yet widely available and until it is, opportunities to address individual-level sexual behaviors with vulnerability to HIV must be addressed. Previous research surrounding HIV-related health disparities among Black SGM has extensively examined individual-level factors, without prioritizing the multilevel influences of HIV vulnerability and related health disparities (Baral et al., 2013; Bronfenbrenner, 1977), such as those at the network level. For example, the majority of research has been at the individual level and has identified sex-related behaviors associated with HIV vulnerability including sex without using a condom consistently and correctly (De Santis, 2009; Operario et al., 2011), engaging in group sex (Friedman et al., 2008; Prestage et al., 2009), and using drugs or alcohol during sex (Morgan et al., 2016; Washington et al., 2021). There were some differences among these behaviors by race (with Black SMM reporting less risk behaviors relative to White SMM) but due to health disparities and structural oppression, the adverse consequences of sexual vulnerability behaviors disproportionately impact Black SMM (Maulsby et al., 2014; Millett et al., 2007, 2012). This necessitates an examination of health inequities and disparities at higher levels (Halkitis et al., 2013). As Black SGM continue to experience poor HIV outcomes relative to other racial groups due to multilevel influences, HIV vulnerability must account for an individual’s social networks and norms surrounding sexual behaviors. This is a current gap in our understanding of health disparities. Social networks can impact behaviors through various mediating pathways: one of these pathways include through social influence which can be transmitted through networks (Kohler et al., 2001), social network norms (Bandura & Walters, 1977; Marsden, 2006), and an individual’s perceived engagement of their network members in a behavior (Valente et al., 2013).

Although social network behavioral norms have been posited to guide individual-level HIV vulnerability (Bandura, 1994, 2001; Centers for Disease Control & Prevention, 2021; De et al., 2007; Dearing et al., 1994; Latkin et al., 2003, 2010; Sheehan et al., 2019), much of previous research has focused on either the influences of specific network members and sexual partners and less so on the overall influence of the norms of their social networks (Holloway et al., 2015; Tobin & Latkin, 2008). Social networks can be composed of several spheres of influence, including family members, friends, and sexual partners (Miller et al., 2005). Networks can facilitate an environment in which individuals may share norms and behaviors; however, it is unclear how an individual may internalize the diverse and potentially conflicting injunctive and descriptive norms of their social networks. To identify an individual’s network norm, it may be useful to characterize the different typologies of network norms within the social networks of Black SGM. While injunctive norms are behaviors which individuals believe are approved by others within their social networks, descriptive norms are the perceived behaviors that occur within a social network (Baumeister & Vohs, 2007; Cialdini et al., 1990; Kincaid, 2004). Both types of norms have shown to influence individual-level HIV risk behaviors (Bandura, 1994; Latkin et al., 2003, 2010; Sheehan et al., 2019). For example, previous research on descriptive norms has identified that Black SMM who had an “enabler,” or an individual who would not disapprove of the respondent’s HIV sexual vulnerability behavior (i.e., condomless sex, group sex, or drug use during sex), in their social networks were more likely to engage in that behavior, relative to Black SMM who did not have an enabler in their social networks (Schneider et al., 2013). Network characteristics may also be important for consideration: Black SMM with a greater proportion of family members in their social networks were more likely to discourage group sex and drug use during sex to members of their sexual networks (Schneider et al., 2012). This also suggests there may be a threshold effect occurring in networks: individuals may only adopt a norm or behavior once a threshold of their network adopts that norm or behavior (Montgomery & Chung, 1999; Valente, 1996).

These foundational social network norm studies relied on a variable-centered approach as social network sexual norms have yet to be examined synchronously in the context of a person-centered approach such as Latent Profile Analysis (LPA). LPA is a method to identify patterns of social network injunctive and descriptive sexual norms among Black SGM who may be at heightened sexual vulnerability to HIV, using a person-centered approach. Latent class analysis has previously been used in social network research among people from low-income neighborhoods to identify social support patterns within social support networks (Bohnert et al., 2010) and young SMM to identify sexual partner typologies (Janulis et al., 2018). Additionally, LCA aligns with the Intersectionality theory as it allows for overlapping experiences to be captured at the person level and identifies the resulting unique typology. The present study aimed to examine profiles of injunctive sexual norms (i.e., Black SGM network members approve of the individual engaging in condomless sex, group sex, or using drugs or alcohol to enhance sex), and descriptive sexual norms (i.e., Black SGM network members engage in condomless sex, group sex, or using drugs or alcohol to enhance sex) within the social networks of Black SGM. We hypothesize that Black SGM will have distinct network profiles of social network norms surrounding approval and engagement of these sexual behaviors and these norms will be associated with increased individual-level HIV vulnerability behaviors of condomless sex, group sex, and using drugs or alcohol to enhance sex.

Method

Participants

Data were from the baseline wave of the Neighborhoods and Networks (N2) Cohort Study, a prospective longitudinal study of Black SGM (baseline data were collected from January 2018 to December 2019). Black SGM were recruited via peer referral sampling in Chicago, Illinois, United States (US). Trained interviewers obtained participant’s written consent and then, used interviewer-administered computer-assisted assessments to collect information about participants. Assessments were conducted in a private room at the study site and lasted 1–2 hours in length. Additional study information can be found elsewhere (Duncan et al., 2019). Participants were provided $150 for their time and transportation to the study site, and participants were provided $20 for each referral, with a maximum of 6 referrals possible. Participants provided individual-level information surrounding their sociodemographic characteristics, relationship characteristics, HIV sexual vulnerability behaviors, HIV status neutral treatment and care. Participants were also asked to name up to 5 alters whom they considered to be a part of their confidant networks and then, administered an egocentric network inventory which assessed the (1) participant’s perception of each confidant alter’s sociodemographic characteristics, (2) confidant alter’s relationship characteristics with the participant, (3) participant’s perception of the confidant alter’s sexual vulnerability to HIV, and (4) participant’s perception of the confidant alter’s approval of the participant engaging in HIV vulnerability behaviors.

Measures

Participant’s Sociodemographic and Background Characteristics

Participants provided information about their age (in years), gender (cisgender man or transgender woman, non-binary, or other), sexual orientation (gay/homosexual/lesbian, bisexual, straight/heterosexual, or other), relationship status (not single or single), ethnicity (Hispanic/Latine or not Hispanic/Latine), education (high school degree or no high school degree), employment status (employed full-time or not employed full-time), income ($20,000 a year or less, or above $20,000 a year), housing instability in the past year (history of housing instability or did not experience housing instability), number of sexual partners in the past 6 months, self-reported HIV serostatus (HIV negative or living with HIV), and adherence to ART or PrEP.

Confidant Network Member’s Sociodemographic Characteristics

Participants provided information surrounding each alter’s gender (cisgender man, cisgender woman, or another transgender identity) age, ethnicity (Hispanic/Latine or not Hispanic/Latine), and race (Black or not Black).

Alter’s Relationship Characteristics

Participants indicated the relationship type with their alter: family (i.e., immediate, extended, and god parent/sibling), romantic or sexual partner (current or former), chosen family [i.e., play, House Ball (i.e., “a kinship system that is organized to meet the needs of its members for social solidarity and mentoring” (Arnold & Bailey, 2009; Murrill et al., 2008; Phillips et al., 2011), gay sibling, friend, or different relationship (e.g., counselor, minister, therapist). Participants described their emotional closeness to and how often they communicated with the alter.

Alter’s Sexual Vulnerability to HIV

Participants described 3 descriptive and 3 injunctive norms for each confidant alter. Participants indicated descriptive norms such as condomless sex (yes, or no/don’t know), use of drugs or alcohol to enhance sex (yes, or no/don’t know), or engagement in group sex (yes, or no/don’t know). Participants were asked to indicate the alter’s approval (i.e., injunctive norm) of the participant engaging in condomless sex (yes, or no/don’t know), using drugs or alcohol to enhance sex (yes, or no/don’t know), or engaging in group sex (yes, or no/don’t know).

Network Proportions of Sexual Behavior and Approval of Participant Engaging in Behavior

We calculated proportion of networks that responded “yes” to each of the sexual vulnerability behaviors and approval of the participant engaging in sexual vulnerability behaviors.

Individual-Level HIV Vulnerability Behaviors

Participants indicated if they engaged in condomless sex in the past 6 months, if they had ever had group sex (sex between three or more people) in the past 6 months, and if they used drugs or alcohol to enhance sex in the past 6 months.

Analysis

In line with intersectionality theory, we used LPA to identify latent profiles (Wilson & Urick, 2022). Previous studies have also used LPA to construct network norm profiles surrounding HIV sexual vulnerability, based on an individual’s sexual partner network, but have not assessed the impact of social network-level norms (Dangerfield et al., 2018). We used the tidyLPA package on the R environment to examine latent profiles until the best fitting profile emerged. We used the following indices to identify the optimal profile solution: Bayes information criterion (BIC), Akaike’s information criterion (AIC), Entropy, Bootstrap likelihood ratio test (BLRT), Prob. Min/Prob. Max, and the unique sample size of each profile. Lower values of BIC and AIC indicate a more parsimonious fit (Akaike, 1974; Akaike, 1980; Schwarz, 1978). Entropy values closer to 1 indicate the model’s ability to discriminate between profiles (Celeux & Soromenho, 1996). The BLRT assesses if model fit is superior relative to the model with one profile less (Lo et al., 2001; McLachlan & Peel, 2004). We also considered the Prob. Min (minimum of the diagonal of the average latent profile probabilities for most likely profile membership, by assigned profile) and Prob. Max (maximum of the diagonal of the average latent profile probabilities for most likely profile membership, by assigned profile) as these indices signify greater classification as values increase (Jung & Wickrama, 2008). Finally, we considered the sample size of each profile (Min n), as models with profiles that are less than or equal to 25 members could indicate a spurious finding (Jung & Wickrama, 2008).

After identifying the most parsimonious latent profile solution, we used bivariate associations to examine the relationship between network profile membership and other measures. We used binomial regression models to assess the associations between the network-level LPA profiles, and five individual-level HIV sexual vulnerability outcomes (condomless anal sex in the past 6 months, condomless insertive anal sex in the past 6 months, condomless receptive anal sex in the past 6 months, group sex in the past 6 months, and the use of drugs or alcohol to enhance sex in the past 6 months).

Results

Analytic Sample

Study staff recruited and enrolled 412 Black SGM into the parent study. Of participants, 377 reported sex in the past 6 months and named up to 5 confidant network alters (n = 1036 alters). There were missing data for 2 participants’ confidant networks, so information from 375 participants and 1016 social network alters was included in the LPA. There were missing data for 2 participants’ sexual orientation and 2 participants’ laboratory-confirmed HIV serostatus. We used listwise deletion to identify the final sample size (n = 371).

Latent Profile Analysis for Identifying Injunctive and Descriptive Sex Behavior Norms

We assessed several fit criteria in the selection of the latent profile model of the most parsimonious fit (Pastor et al., 2007). Based on BIC and AIC values which were lowest in the 5-profile model solution, we identified the 5-profile model solution as the best fit to the data. In accordance with the BLRT values, the 5-profile model solution was identified as superior to the 4-profile solution (BLRT = 230.09; BLRT p-value < 0.01). Although the BLRT suggested that the 6-profile solution was a significantly better fit to the data relative to the 5-profile solution, the minimum proportion of participants who were grouped into the smallest size group was 0, indicating that this model contains a group which is a spurious finding or may not have converged. Additional information about latent profile model fit indices can be found in Supplemental Table 1. Figure 1 visualizes the mean probability of being included in each social network norm profile. Social network sexual norms were categorized as: (1) Low HIV vulnerability network norm (low injunctive and descriptive norm of sexual behaviors), (2) Moderately high HIV vulnerability network norm (high injunctive and descriptive norms of group sex and using drugs to enhance sex, and moderately high injunctive and descriptive norms of condomless sex), (3) High HIV vulnerability network norm (high injunctive and descriptive norms of condomless sex and drug use to enhance sex, and moderately high injunctive and descriptive norms of group sex), (4) Approval of drugs to enhance sex dominant network norm (high injunctive norm of participant engaging in using drugs or alcohol to enhance sex but low injunctive and descriptive norms of all other behaviors), and (5) Condomless sex dominant network norm (high descriptive norm of condomless sex; low injunctive and descriptive norm of other behaviors).

Fig. 1.

Five profile latent class distribution of HIV network perceptions among Black sexual and gender minoritized groups

Comparison of Participant Background Information, by Network Norm Profilel

Participants were a mean age of 26 years old (± 4 years). The majority of participants identified as a cisgender man (87%), gay/homosexual/lesbian (61%), not Hispanic/Latine (93%), and single (62%). The majority of participants reported an income less than $20,000 a year (66%), had more than a high school diploma (49%), did not experience housing instability in the past year (69%), were HIV negative (67%), and were not adherent (i.e., missed a dose in the past 30 days) to PrEP or ART (90%). Approximately half of participants reported full-time employment. Participants reported a mean of 3.7 (± 3.8) sexual partners and 17% reported selling or exchanging sex for money, drugs, food and/or shelter in the past 6 months. Approximately half of participants were employed in the past week. The majority of participants reported condomless anal sex in the past 6 months (84%), and 95% engaged in insertive anal sex of which 52% reported condomless insertive anal sex, and 96% engaged in receptive anal sex of which 46% reported condomless receptive anal sex (Tables 1 and 2). Slightly less than half of participants engaged in group sex (45%), and the majority of participants reported the use of alcohol or drugs during sex to enhance the experience (58%). Additional participant background information, by network norm profile, can be found in Table 1.

Table 1.

Background information of participants stratified by latent profile, n = 375

	Low HIV vulnerability network norm (N = 140)	Approval of drugs to enhance sex network norm (N = 41)	Moderately high HIV vulnerability network norm (N = 81)	High HIV vulnerability network norm (N = 34)	Condomless sex network norm (N = 79)	Overall (N = 375)	Test statistic
Age							F = 2.17
Mean (SD)	25.6 (4.04)	26.2 (4.00)	26.8 (3.93)	25.2 (3.78)	25.1 (4.02)	25.8 (4.01)
Median [Min, Max]	25.0 [18.0, 34.0]	26.0 [19.0, 36.0]	27.0 [19.0, 34.0]	25.0 [19.0, 33.0]	24.0 [17.0, 34.0]	25.0 [17.0, 36.0]
Gender							χ2 = 4.41
Cis man	118 (84.3%)	33 (80.5%)	72 (88.9%)	31 (91.2%)	72 (91.1%)	326 (86.9%)
Transgender or non-binary	22 (15.7%)	8 (19.5%)	9 (11.1%)	3 (8.8%)	7 (8.9%)	49 (13.1%)
Sexual orientation†							χ2 = 9.42
Gay, homosexual, or Lesbian	88 (63.8%)	22 (53.7%)	46 (56.8%)	24 (70.6%)	45 (57.0%)	225 (60.3%)
Bisexual	32 (23.2%)	13 (31.7%)	25 (30.9%)	4 (11.8%)	22 (27.8%)	96 (25.7%)
Different sexual orientation††	8 (5.8%)	3 (7.3%)	7 (8.6%)	2 (5.9%)	7 (8.9%)	27 (7.2%)
Straight or heterosexual	10 (7.2%)	3 (7.3%)	3 (3.7%)	4 (11.8%)	5 (6.3%)	25 (6.7%)
Hispanic or Latine*
No	134 (97.1%)	39 (95.1%)	72 (88.9%)	33 (97.1%)	70 (88.6%)	348 (93.3%)	χ2 = 9.48
Yes	4 (2.9%)	2 (4.9%)	9 (11.1%)	1 (2.9%)	9 (11.4%)	25 (6.7%)
Relationship status							χ2 = 4.54
Not single	54 (38.6%)	16 (39.0%)	23 (28.4%)	13 (38.2%)	35 (44.3%)	141 (37.6%)
Single	86 (61.4%)	25 (61.0%)	58 (71.6%)	21 (61.8%)	44 (55.7%)	234 (62.4%)
Education							χ2 = 14.49
More than high school	21 (51.2%)	44 (54.3%)	57 (40.7%)	15 (44.1%)	45 (57.0%)	182 (48.5%)
High school diploma or equivalent (for example, GED)	13 (31.7%)	33 (40.7%)	59 (42.8%)	15 (44.1%)	21 (26.6%)	141 (37.8%)
No high school diploma	7 (17.1%)	4 (4.9%)	23 (16.7%)	4 (11.8%)	13 (16.5%)	51 (13.7%)
Employment status							χ2 = 3.63
No	66 (47.1%)	16 (39.0%)	44 (54.3%)	17 (50.0%)	43 (54.4%)	186 (49.6%)
Yes	74 (52.9%)	25 (61.0%)	37 (45.7%)	17 (50.0%)	36 (45.6%)	189 (50.4%)
Income							χ2 = 7.01
Above $20,000 a year	38 (27.1%)	17 (41.5%)	35 (43.2%)	11 (32.4%)	28 (35.4%)	129 (34.4%)
$20,000 or less a year	102 (72.9%)	24 (58.5%)	46 (56.8%)	23 (67.6%)	51 (64.6%)	246 (65.6%)
History of unstable housing							χ2 = 5.00
No	90 (64.3%)	31 (75.6%)	54 (66.7%)	27 (79.4%)	58 (73.4%)	260 (69.3%)
Yes	50 (35.7%)	10 (24.4%)	27 (33.3%)	7 (20.6%)	21 (26.6%)	115 (30.7%)
Proportion of network that participant has had sex with							F = 2.24
Mean (SD)	0.473 (0.218)	0.450 (0.233)	0.527 (0.169)	0.561 (0.192)	0.513 (0.234)	0.499 (0.213)
Median [Min, Max]	0.500 [0,1.00]	0.500 [0, 0.833]	0.500 [0.200, 1.00]	0.500 [0.167, 1.00]	0.500 [0, 1.00]	0.500 [0, 1.00]
Number of sex partners in the past 6 months							F = 4.20
Mean (SD)	3.16 (3.73)	3.07 (3.40)	5.07 (4.02)	2.85 (2.23)	3.92 (4.21)	3.70 (3.83)
Median [Min, Max]	2.00 [0, 29.0]	2.00 [0, 16.0]	4.00 [0, 18.0]	2.50 [0, 9.00]	3.00 [0, 18.0]	2.00 [0, 29.0]
Exchanged sex for money, drugs, food, or shelter							χ2 = 8.93
No	121 (86.4%)	31 (75.6%)	60 (74.1%)	28 (82.4%)	70 (88.6%)	310 (82.7%)
Yes	19 (13.6%)	10 (24.4%)	21 (25.9%)	6 (17.6%)	9 (11.4%)	65 (17.3%)
HIV status
HIV Negative	98 (70.0%)	23 (56.1%)	51 (63.0%)	20 (58.8%)	59 (74.7%)	251 (66.9%)	χ2 = 6.50
Living with HIV	42 (30.0%)	18 (43.9%)	30 (37.0%)	14 (41.2%(	20 (25.3%)	124 (33.1%)
Adherent to PrEP or ART							χ2 = 2.82
No	123 (87.9%)	35 (85.4%)	75 (92.6%)	31 (91.2%)	73 (92.4%)	337 (89.9%)
Yes	17 (12.1%)	6 (14.6%)	6 (7.4%)	3 (8.8%)	6 (7.6%)	38 (10.1%)
Engaged in insertive anal sex in past 6 months**							χ2 = 2.19
Engaged in insertive anal sex	133 (95.0%)	39 (95.1%)	79 (97.5%)	31 (91.2%)	75 (94.9%)	357 (95.2%)
Engaged in condomless insertive anal sex	60 (42.9%)	18 (43.9%)	54 (66.7%)	20 (58.8%)	42 (53.2%)	194 (51.7%)	χ2 = 13.24
Engaged in receptive anal sex in past 6 months*
Engaged in receptive anal sex	134 (95.7%)	39 (95.1%)	79 (97.5%)	34 (100%)	74 (93.7%)	360 (96.0%)	χ2 = 3.14
Engaged in condomless receptive anal sex	55 (39.3%)	19 (46.3%)	37 (45.7%)	24 (70.6%)	37 (46.8%)	172 (45.9%)	χ2 = 9.71
Engaged in both insertive and receptive (versatile) anal sex in past 6 months*
Engaged in versatile anal sex	38 (92.7%)	78 (96.3%)	130 (94.2%)	31 (91.2%)	73 (92.4%)	350 (93.8%)	χ2 = 1.67
Engaged in condomless versatile anal sex	19 (46.3%)	37 (45.7%)	54 (39.1%)	24 (70.6%)	37 (46.8%)	171 (45.8%)	χ2 = 10.93
Had condomless sex in past 6 months*							χ2 = 11.03
No	34 (24.3%)	7 (17.1%)	8 (9.9%)	3 (8.8%)	10 (12.7%)	62 (16.5%)
Yes	106 (75.7%)	34 (82.9%)	73 (90.1%)	31 (91.2%)	69 (87.3%)	313 (83.5%)
Had group sex in past 6 months***							χ2 = 28.36
No	97 (69.3%)	26 (63.4%)	29 (35.8%)	16 (47.1%)	36 (45.6%)	204 (54.4%)
Yes	43 (30.7%)	15 (36.6%)	52 (64.2%)	18 (52.9%)	43 (54.4%)	171 (45.6%)
Used drugs to enhance sex***							χ2 = 25.84
No	79 (56.4%)	15 (36.6%)	20 (24.7%)	9 (26.5%)	34 (43.0%)	157 (41.9%)
Yes	61 (43.6%)	26 (63.4%)	61 (75.3%)	25 (73.5%)	45 (57.0%)	218 (58.1%)

^*p < 0.05

^**p < 0.01

^***p < 0.001

^†Missing information from 2 participants

^††Different sexual orientation included write in responses (i.e., “Transgender,” “Male,” “Female”), “Confused,” “Curious,” “Girl,” “Nothing,” “Open to all,” “Pansexual,” “Queer,” and “Trade.”

Table 2.

Confidant network member information, n = 1016

	Overall (N = 1016)
Gender
Cisgender man	532 (52.4%)
Cisgender woman	408 (40.2%)
Transgender or non-binary	73(7.2%)
Missing	3 (0.3%)
Age
Mean (SD)	31.3 (11.4)
Median [Min, Max]	28.0 [15.0, 93.0]
Missing	14 (1.4%)
Hispanic or Latine	58 (5.7%)
Black	948 (93.3%)
Relationship status
In a committed relationship with one other person	337 (33.2%)
Married	90 (8.9%)
Single	498 (49.0%)
Different relationship status*	91 (9.%)
Relationship to participant
Best Friend	295 (29.0%)
Family	263 (25.9%)
Friend	162 (15.9%)
Current or Former Romantic/Sex partner/spouse	127 (12.5%)
Chosen family	55 (5.4%)
Different relationship**	21 (2.1%)
Closeness
Not close	16 (1.6%)
Somewhat close	169 (16.6%)
Very close	831 (81.8%)
Communication frequency
Several times a day	247 (24.3%)
Every day	361 (35.5%)
Several times a week	258 (25.4%)
Once a week	68 (6.7%)
Once every two weeks or less frequently	82 (8.1%)
Has condomless sex	460 (45.3%)
Approves of participant having condomless sex	191 (18.8%)
Engages in group sex	180 (17.7%)
Approves of participant engaging in group sex	247 (24.3%)
Uses alcohol or drugs to enhance sex	243 (23.9%)
Approves of participant using alcohol or drugs to enhance sex	360 (35.4%)

^*Includes don’t know, widowed, complicated, in a relationship with more than one person

^**Includes colleague, counselor, minister, therapist, among other relationship types

Confidant network characteristics—Participants reported a mean of 2.7 confidant network alters (± 1.25 alters) for a total of 1,016 alters. Alters were perceived as being mean age 31.3 years (± 11.4 years), and participants perceived alters to be cisgender men (52%), cis women (40%), or of transgender experience (7%). Alters were perceived as majority Black (93%), non-Hispanic/Latine (94%), unmarried (49%), and either a friend, family, or chosen family member of the participant (85%). Of alters, 13% were current or former sexual partners of participants. Participants reported being very close to alters (82%) and communicated with alters frequently, with 92% of participants reporting communicating with alters at least once a week, suggesting high potential for social influence. The majority of alters did not engage in the HIV vulnerability behaviors (condomless sex = 45%, group sex = 18%, use of drugs to enhance sex = 24%) or approve of participants engaging in these behaviors (condomless sex = 25%, group sex = 35%, use of alcohol or drugs to enhance sex = 24%). Additional information can be found in Table 2.

Binomial Regression Model Examining Condomless Anal Sex

Relative to participants who reported being in a relationship, participants who reported being single had lower likelihood of reporting condomless sex in the past 6 months (OR = 0.49, 95% CI 0.25–0.91, p = 0.027). Participants who reported having a higher number of sex partners in the past 6 months (OR = 1.27, 95% CI 1.09–1.53, p = 0.005) had higher likelihood of reporting condomless sex in the past 6 months. There was no statistical significance between latent profile and condomless sex. Additional information can be found in Table 3: Model 1a.

Table 3.

Results of the binomial regression models examining condom use

Predictors	Model 1a: Condomless anal sex			Model 1b: Condomless insertive anal sex			Model 1c: Condomless receptive anal sex
	Odds Ratios	CI	p	Odds Ratios	CI	p	Odds Ratios	CI	p
(Intercept)	1.05	0.11–10.07	0.968	1.04	0.16–6.73	0.963	0.71	0.11–4.80	0.729
Age	1.02	0.94–1.10	0.629	0.98	0.92–1.04	0.477	0.96	0.90–1.02	0.218
Transgender	1.16	0.41–3.59	0.785	0.35	0.15–0.79	0.013	3.83	1.69–9.22	0.002
Sexual Orientation
Ref=Gay
Bisexual	0.73	0.36–1.51	0.386	1.55	0.88–2.75	0.128	0.24	0.13–0.44	< 0.001
Different sexual orientation	2.45	0.57–17.87	0.288	0.99	0.40–2.50	0.978	0.59	0.24–1.44	0.241
Straight or heterosexual	0.89	0.25–3.50	0.860	1.20	0.41–3.58	0.734	0.25	0.08–0.74	0.014
Not in a relationship	0.49	0.25–0.91	0.027	0.87	0.54–1.40	0.567	1.03	0.64–1.68	0.894
No high school education	1.58	0.66–4.22	0.330	1.67	0.82–3.46	0.162	0.99	0.48–2.03	0.981
Not employed	0.84	0.43–1.61	0.595	1.04	0.63–1.74	0.868	1.18	0.71–1.98	0.528
Income is below $20,000	0.85	0.39–1.79	0.669	0.48	0.27–0.83	0.009	0.99	0.57–1.73	0.975
Housing instability	0.70	0.35–1.42	0.314	1.57	0.91–2.74	0.110	0.89	0.51–1.54	0.667
Living with HIV	1.39	0.74–2.64	0.308	1.06	0.65–1.73	0.813	1.63	0.99–2.70	0.058
Using PrEP or ARTs	1.39	0.67–3.05	0.391	0.88	0.50–1.54	0.644	1.01	0.56–1.79	0.984
Proportion of network that participant has had sex with	2.93	0.58–15.70	0.201	4.40	1.27–15.92	0.021	4.32	1.24–15.61	0.023
Number of sexual partners in past 6 months	1.27	1.09–1.53	0.005	1.00	0.94–1.07	0.971	1.06	0.99–1.14	0.096
Exchanged/sold sex for money, drugs, food, or shelter	1.11	0.43–3.27	0.844	1.11	0.57–2.22	0.758	0.98	0.49–1.95	0.960
Network norm profile
Ref = Low HIV vulnerability
Moderately high HIV vulnerability	2.36	1.00–6.11	0.061	2.40	1.28–4.57	0.007	1.31	0.69–2.47	0.406
High vulnerability	3.02	0.94–13.60	0.095	2.01	0.86–4.86	0.110	3.26	1.40–8.04	0.008
Approval of drugs to enhance sex dominant	1.58	0.62–4.47	0.362	1.05	0.48–2.29	0.908	1.53	0.68–3.44	0.304
Condomless sex dominant	2.10	0.96–4.91	0.072	1.28	0.69–2.37	0.433	1.56	0.84–2.93	0.161
Observations	371			353			356
R2 Tjur	0.127			0.122			0.157

Bold font indicates significant p-values

Binomial Regression Model Examining Condomless Insertive Anal Sex Among Participants Who Reported Insertive Anal Sex

Relative to participants who reported identifying as a cisgender man, participants who identified as a transgender woman or non-binary and were assigned male at birth (OR = 0.35, 95% CI 0.15–0.79, p = 0.013) and participants who reported an income below $20,000 (OR = 0.48, 95% CI 0.27–0.83, p = 0.009) had lower likelihood of reporting condomless insertive sex in the past 6 months. Participants who reported having sex with a larger proportion of their network (OR = 4.40; 95% CI 1.27–15.92; p = 0.021), and being in the moderately high HIV vulnerability network norm profile (OR = 2.40; 95% CI 1.28–4.57; p = 0.007) relative to the low HIV vulnerability network norm profile had positive and significant associations of engaging in condomless insertive anal sex in the past 6 months. Additional information can be found in Table 3: Model 1b.

Binomial Regression Model Examining Condomless Receptive Anal Sex Among Participants Who Reported Receptive Anal Sex

Relative to participants who reported identifying as a cisgender man, participants who identify as a transgender woman or non-binary and assigned male at birth (OR = 3.83, 95% CI 1.69–9.22, p = 0.002), participants who had sex with a larger portion of their social network (OR = 4.32, 95% CI 1.24–15.61, p = 0.023), and being in the high HIV vulnerability network norm profile (OR = 3.26; 95% CI 1.40–8.04, p = 0.008) relative to the low HIV vulnerability network norm profile had positive and significant associations of engaging in condomless receptive anal sex in the past 6 months. Participants who identified as bisexual (OR = 0.24, 95% CI 0.13–0.44, p < 0.001), or straight/heterosexual (OR = 0.25, 95% CI 0.08–0.74, p = 0.014), relative to gay, had negative and significant associations of engaging in condomless receptive anal sex in the past 6 months. Additional information can be found in Table 3: Model 1c.

Binomial Regression Model Examining Group Sex

Participants who reported not being in a relationship (OR = 1.81; 95% CI 1.07–3.08; p = 0.028), having sex with a larger proportion of their network (OR = 9.96; 95% CI 2.65–39.73; p = 0.001), having a higher number of sexual partners in the past 6 months (OR = 1.23; 95% CI 1.11–1.37; p < 0.001), and relative to the low HIV vulnerability network norm profile being in either the moderately high HIV vulnerability network norm profile (OR = 2.99; 95% CI 1.50–6.02; p = 0.002), high vulnerability network norm profile (OR = 3.13; 95% CI 1.29–7.72; p = 0.012), and the condomless sex network norm profile (OR = 3.08; 95% CI 1.58–6.10; p = 0.001), had positive and significant associations of engaging in group sex in the past 6 months. It may be of interest to note that although not significant at p < 0.05, exchanging or selling sex for money, drugs, food or shelter (OR = 2.10; 95% CI 1.00–4.51; p = 0.053) was found to be trending toward significance for association with group sex in the past 6 months. Additional information can be found in Table 4: Model 2.

Table 4.

Results of the binomial regression models examining group sex and using drugs to enhance sex

Predictors	Model 2: Group sex			Model 3: Used alcohol or drugs to enhance sex
	Odds Ratios	CI	p	Odds Ratios	CI	p
(Intercept)	0.01	0.00–0.10	< 0.001	0.30	0.05–1.84	0.196
Age	1.05	0.98–1.12	0.180	0.99	0.93–1.06	0.840
Transgender	1.63	0.64–4.20	0.308	0.76	0.33–1.72	0.504
Sexual Orientation	1.46	0.79–2.71	0.233	1.50	0.85–2.68	0.163
Ref = Gay
Bisexual	1.16	0.44–3.03	0.761	1.13	0.46–2.82	0.792
Different sexual orientation	0.78	0.23–2.57	0.690	1.61	0.56–4.80	0.382
Straight or heterosexual	1.81	1.07–3.08	0.028	1.06	0.66–1.69	0.823
Not in a relationship	1.82	1.08–3.09	0.026	1.07	0.67–1.71	0.776
No high school education	0.78	0.36–1.69	0.540	0.98	0.50–1.96	0.964
Not employed	0.73	0.42–1.28	0.271	1.22	0.74–2.03	0.431
Income is below $20,000	1.15	0.63–2.10	0.656	1.25	0.72–2.17	0.427
Housing instability	1.14	0.62–2.08	0.677	0.75	0.43–1.30	0.307
Living with HIV	0.85	0.49–1.45	0.543	0.94	0.57–1.53	0.794
Using PrEP or ARTs	0.89	0.48–1.67	0.727	0.83	0.47–1.44	0.501
Proportion of network that participant has had sex with	9.96	2.65–39.73	0.001	3.37	1.05–11.28	0.044
Number of sexual partners in past 6 months	1.23	1.11–1.37	< 0.001	1.04	0.97–1.13	0.279
Exchanged/sold sex for money, drugs, food, or shelter	2.10	1.00–4.51	0.053	2.98	1.41–6.75	0.006
Network norm profile
Ref = Low HIV vulnerability
Moderately high HIV vulnerability	2.99	1.50–6.02	0.002	3.47	1.83–6.76	< 0.001
High vulnerability	3.13	1.29–7.72	0.012	3.39	1.45–8.48	0.006
Approval of drugs to enhance sex dominant	1.51	0.61–3.60	0.361	2.38	1.10–5.31	0.030
Condomless sex dominant	3.08	1.58–6.10	0.001	1.68	0.93–3.06	0.088
Observations	371			371
R2 Tjur	0.295			0.149

Bold font indicates significant p-values

Binomial Regression Model Examining the Use of Alcohol or Drugs to Enhance Sex

Participants who reported having sex with a larger proportion of their network (OR = 3.37; 95% CI 1.05–11.28; p = 0.044), exchanging or selling sex for money, drugs, food or shelter (OR = 2.98; 95% CI 1.41–6.75; p = 0.006), and relative to the low HIV vulnerability network norm profile, being in either the moderately high HIV vulnerability network norm profile (OR = 3.47; 95% CI 1.83–6.76; p < 0.001), and high vulnerability network norm profile (OR = 3.39; 95% CI 1.45–8.48; p = 0.006), and approval of drugs to enhance sex group (OR = 2.38; 95% CI 1.08–5.12; p = 0.033), had positive and significant associations with using drugs or alcohol to enhance sex. Additional information can be found in Table 4: Model 3.

Discussion

Our study sought to identify if injunctive and descriptive social network norms would be typologized as latent profiles among Black SGM assigned male at birth, and our findings support our hypothesis. Our study found that Black SGM experienced one of five gradients of HIV risk and vulnerability within their social network sexual norm profiles: low HIV vulnerability norm, moderately high HIV vulnerability norm, high HIV vulnerability norm, approval of drug use dominant norm, and condomless sex dominant norm. Moderately high and high HIV vulnerability norm profiles were associated with condomless sex, group sex, and using drugs or alcohol to enhance sex in the past 6 months. Our findings suggest that social network norms can either exacerbate or mitigate the hazardous effects of intersectional discrimination on HIV-related outcomes.

Importantly, we found that social network norms surrounding engagement and approval of HIV sexual vulnerability behaviors were associated with individual-level sexual vulnerability to HIV. For example, Black SGM who were grouped into the higher HIV vulnerability social network norm profiles (i.e., moderately high and high) were more likely to report engaging in condomless insertive or receptive anal sex, group sex, and the use of drugs/alcohol to enhance sex in the past 6 months, relative to the low HIV vulnerability profile. As our analysis consisted of cross-sectional data, our findings can be interpreted in one of two ways. First, our findings may suggest that participants who perceive their networks to be higher in HIV vulnerability norm profiles could be engaging in higher HIV vulnerability behaviors because they believe the behavior is common. Second, our findings may suggest that participants are embedded in networks in which the norm is the acceptance of higher sexual vulnerability behaviors, and then, themselves engage in higher sexual vulnerability behaviors due to this peer influence. As our study is cross-sectional, perception of behaviors and peer influence cannot be disentangled statistically. However, participants were provided the option to indicate that they do not know if their social network alters engaged in a specific behavior or approved of the participant’s engagement in a specific behavior; thus, our study likely captures the latter option, peer influence (as they participants could indicate not knowing of descriptive norms). A future study could include a longitudinal study which collects data from sociocentric networks (i.e., entire closed networks) to assess peer influence using alter-provided data, instead of perceived data at the ego level.

Our study supports previous literature which suggests that there may be some threshold which must be reached for social network norms to be adopted by an individual (Montgomery & Chung, 1999; Valente, 1996). For example, Black SGM in the moderately high, high, and approval of drugs to enhance sex dominant HIV vulnerability network norm profiles had networks in which over 60% of their alters approved of the participant using drugs or alcohol to enhance sex—these individuals were more likely to report using drugs or alcohol to enhance sex in the past 6 months, relative to the low HIV vulnerability profile. Until health equity or supranormal efforts to increase PrEP and ART accessibility are reached in the USA, interventions to reduce HIV sexual vulnerability for Black SGM may consider network-level interventions as they may surmount structural oppression to reach health equity. Currently, HIV prevention providers and strategies are not equitably accessible to Black SGM, as suggested by research outlining racial gaps in biomedical prevention strategy uptake (Centers for Disease Control Prevention, 2021; Hess et al., 2017; Kanny et al., 2019; Matthews et al., 2016). HIV status-neutral care can also overcome intersectional discrimination faced by Black SMM and TW (Gardner et al., 2011; Myers et al., 2018). Until health equity is reached, social network interventions are needed. For example, SMM of Color were more likely to discuss PrEP and encourage PrEP use with friends they felt more emotionally close to them and interacted with them more frequently (Shrader et al., 2021b). Thus, if Black SGM do not have access to HIV prevention providers or strategies due to structural issues including intersectional discrimination, their social networks may fill this gap. In designing interventions to reduce sexual behaviors related to HIV vulnerability, it is important to consider how network norms can be changed in addition to structural barriers. The Diffusion of Innovation Theory posits that behaviors can spread throughout a network—Valente (1996) emphasized the importance of personal networks in the adoption of behaviors, with an individual’s network position in a network more important than the community’s behavioral norms (Valente & Rogers, 1995). The Diffusion of Innovations theory has also posited that mere exposure to a behavior or the absence of a behavior may be adequate enough for individuals to adopt this new behavior (Rogers & Kincaid, 1981), whereas more recent research suggests that homophily (i.e., similarities between individuals) may be critical in behavior adoption (McPherson et al., 2001). Thus, interventions should consider the utility of staff who are homophilous with the priority population on race, ethnicity, sexual orientation, and gender identity, among other characteristics.

There are generally four types of network-level interventions which can be used to elicit individual-level behavioral change: identification of network opinion leaders, segmentation (prioritizing specific groups of people through an intervention), induction (activating novel interactions between people within a network), and alteration (changing the network) (Pagkas-Bather et al., 2020, 2022; Valente, 2012). To date, there has been much research surrounding the positive effects of network-level interventions which use opinion leaders and social-networking applications to increase HIV self-testing (Huang et al., 2016; Lightfoot et al., 2018; Rosengren et al., 2016), opinion leaders and social media to increase HIV prevention information (Jaganath et al., 2012), and online social network applications to increase HIV prevention information (Tanner et al., 2018; Young, 2012; Young et al., 2014). PrEP Chicago, a social network intervention implemented among cisgender Black sexual minoritized men in Chicago, used a peer change agent approach (i.e., opinion leader workshop with telephonic booster sessions) to leverage existing social networks as pathways for HIV prevention strategy (i.e., PrEP) information diffusion, behavioral influence, social support, and empowerment (Schneider et al., 2021; Young et al., 2017). PrEP Chicago was successful in diffusing PrEP to Black SGM with high HIV susceptibility (Schneider et al., 2021). Interventions to reduce HIV vulnerability from sex can include enrolling sociocentric groups of friends to receive an intervention together (i.e., segmentation) (Kanamori et al., 2017, 2019), prioritizing behavior change in individuals who are central within a network (i.e., induction), and tailoring interventions to each network with the understanding that each individual views and responds to their network influence differently (segmentation and induction) (Dyal, 2015). Other interventions which have shown promise include the alteration or manipulation of networks; however, research such as this is sparse (Schneider et al., 2021). Future network-level interventions can consider altering networks to introduce individuals who do not engage in HIV vulnerability behaviors, removing these individuals who do engage or approve of HIV vulnerability behaviors from a network, or creating ties between network members (Kanamori et al., 2019).

Condomless insertive sex, group sex, and the use of drugs to enhance sex were higher among participants with a higher number of sexual partners and who had sex with a higher proportion of their confidant networks. This may be due to participants having more opportunities to engage in these behaviors with their network members. Alternatively, sexual network members may have a higher influence on sexual behaviors relative to social network members (Kohler et al., 2001; Marsden, 2006; Valente et al., 2013). Dynamics surrounding event-level sexual behaviors remain unclear: future prospective studies can consider the utility of ecological momentary assessments (EMA), such as a sex diary in longitudinal research to further examine this relationship (Wray et al., 2016). EMA could account for specific event-level sexual behaviors, which could vary in HIV vulnerability depending on PrEP/ART use at that time and disclosure of status or PrEP use (Algarin et al., 2022; Shrader et al., 2021a). Additionally, future network-level interventions can also focus on dyadic interventions, as HIV transmission from primary partners accounts for up to 67% of new HIV diagnoses among men who have sex with men (Hess et al., 2017; Sullivan et al., 2009), often due to a low perceived risk of HIV acquisition from primary partners (Stephenson et al., 2015). Other studies have successfully leveraged the relationship between dyads of same sex male couples to promote engagement in HIV care (Stephenson et al., 2017, 2021) and couples HIV counseling and testing and adherence to safer sexual agreements (Stephenson et al., 2022). Our sample of Black SGM’s status-neutral biomedical prevention strategy use (i.e., PrEP or ART) was not associated with condomless sex, group sex, or the use of drugs or alcohol to enhance sex. This finding contradicts the highly stigmatizing concept of “risk compensation” in which individuals are theorized to discontinue condoms or increase the number of concurrent sexual partners when they begin PrEP or ART use (Pasipanodya et al., 2020) and supports those studies which find that PrEP use does not influence the discontinuation of condom use (Algarin et al., 2022; Goodman-Meza et al., 2019). HIV prevention researchers and healthcare providers can continue supporting Black SGM’s PrEP and ART use in a stigma-free manner, having established that PrEP and ART were not found to be associated with condomless sex, group sex, or the use of drugs or alcohol to enhance sex (Rojas Castro et al., 2019).

Although Black SMM and TW are oftentimes clustered into a homogenous group, our findings distinguish important considerations related to gender, sexual identity, income, and sex work when examining HIV-vulnerability related outcomes (Duncan et al., 2020). As aforementioned, HIV prevention interventions must consider addressing inequities in power and privilege to ensure that the comprehensive needs of Black SGM are addressed by meeting Black SGM where they are at by addressing basic needs first, and then addressing the individual’s unique HIV prevention needs. For example, considering the high rate of housing instability in our sample (31%), participants may be engaging in sex work to have their basic needs met. Previous interventions which have successfully done this include strengths-based case management such as ARTAS (Craw et al., 2008; Gardner et al., 2007, 2009) and ASK-PrEP (Ogunbajo et al., 2021; Reback et al., 2019), and client-centered care coordination (C4) models (Nelson et al., 2022; Whitfield et al., 2022). These interventions have a strong peer advocate and service navigation component—supporting the utility of network-based peer interventions. Future qualitative research can aim to explore opportunities to incorporate harm reduction strategies during sex work or the use of drugs to enhance sex.

Limitations

This study had several limitations that are important to note. First, we used convenience and peer referral to recruit participants. Thus, participants may also be present in other participants’ social networks. These sampling procedures can introduce dependencies within the data; however, because this was not a true respondent-driven sampling study, we did not use an estimator to adjust for any potential dependencies. As is typical with social network studies, our findings are not generalizable. Additionally, our data were self-reported by participants and collected by an interview-administered egocentric inventory. For this, our participants may incorrectly perceive the behaviors of others within their networks by overestimating their network members’ behaviors towards their own behaviors. Because the egocentric network inventories may not be as accurate as if other members of the participants’ social networks were assessed (Berkowitz, 2005; Lapinski & Rimal, 2005; Miller & McFarland, 1991), there may have been biases within our data. To address this, we included a “I don’t know” option so that participants are not forced into guessing their network members’ behaviors. In addition, participants may have been subject to recall bias due to the retrospective nature of this study and social desirability bias due to the sensitive nature of the questions. Further, we did not define the different experiences of transgender participants in our study: this may have resulted in less nuanced results and disguised important differences between people with differing gender identities. Finally, despite conducting several analyses, we did not correct for multiple comparisons, which may have increased the family-wise error rate and false discovery rate.

Public Health Implications.

As discrimination from racism, homophobia, and transphobia can intersect and compound to form a new type of discrimination, this may exacerbate racial, ethnic, and sexual minority health disparities, and disrupt HIV status neutral care equity (Duncan et al., 2020). This necessitates behavioral strategies such as the use of condoms, especially when engaging in group sex and drug use during sex, to reduce HIV transmission among Black SGM until PrEP and ART roll-out can be equitably accessed by Black SGM. By using a person-centered approach to construct social network HIV risk profiles, and identifying correlates of engaging in condomless sex, group sex, and the use of drugs/alcohol to enhance sex, our study demonstrated the importance of social network norms and socioecological context in individual-level HIV sexual vulnerability. In addition, as social network norms are an important convention in daily decision making, networks are also an opportunity to intervene. HIV prevention interventions for Black SGM many consider the enrollment of entire social networks, such as groups of friends, or perhaps intervene on specific members within social networks with high social influence or who may be late adopters for maximum impact. Thus, developing community-centered social network HIV prevention interventions, which focus on the reduction in sexual behaviors associated with HIV vulnerability, may be an effective approach to limiting HIV incidence in this community until biomedical prevention strategies are equitably available.