Does Stigma Toward Anal Sexuality Impede HIV Prevention Among Men Who Have Sex With Men in the United States? A Structural Equation Modeling Assessment

Abstract

Introduction:

Men who have sex with men (MSM) are suboptimally engaged in efficacious HIV interventions, due in part to stigma.

Aim:

We sought to validate the Anal Health Stigma Model, developed based on theory and prior qualitative data, by testing the magnitude of associations between measures of anal sex stigma and engagement in HIV prevention practices, while adjusting for covariates.

Methods:

We conducted a cross-sectional online survey of 1,263 cisgender MSM living in the United States and analyzed data with structural equation modeling. We tested a direct path from Anal Sex Stigma to Engagement in HIV Prevention alongside 2 indirect paths, 1 through Anal Sex Concerns and another through Comfort Discussing Anal Sexuality with Health Workers. The model adjusted for Social Support, Everyday Discrimination, and Sociodemographics.

Main Outcome Measure:

Engagement in HIV Prevention comprised an ad hoc measure of (i) lifetime exposure to a behavioral intervention, (ii) current adherence to biomedical intervention, and (iii) consistent use of a prevention strategy during recent penile-anal intercourse.

Results:

In the final model, anal sex stigma was associated with less engagement (β = −0.22, P < .001), mediated by participants' comfort talking about anal sex practices with health workers (β = −0.52; β = 0.44; both P < .001), adjusting for covariates (R² = 67%; χ²/df = 2.98, root mean square error of approximation = 0.040, comparative fit index = 0.99 and Tucker-Lewis index = 0.99). Sex-related concerns partially mediated the association between stigma and comfort (β = 0.55; β = 0.14, both P < .001). Modification indices also supported total effects of social support on increased comfort discussing anal sex (β = 0.35, P < .001) and, to a lesser degree, on decreased sex-related concerns (β = −0.10; P < .001).

Clinical Implications:

Higher stigma toward anal sexuality is associated with less engagement in HIV prevention, largely due to discomfort discussing anal sex practices with health workers.

Strength & Limitations:

Adjustment for mediation in a cross-sectional design cannot establish temporal causality. Self-report is vulnerable to social desirability and recall bias. Online samples may not represent cisgender MSM in general. However, findings place HIV- and health-related behaviors within a social and relational context and may suggest points for intervention in health-care settings.

Conclusion:

Providers' willingness to engage in discussion about anal sexuality, for example, by responding to questions related to sexual well-being, may function as social support and thereby bolster comfort and improve engagement in HIV prevention.

INTRODUCTION

The pandemic involving the human immunodeficiency virus (HIV) has disproportionately burdened gay, bisexual, and other men who have sex with men (MSM).¹ In the United States, there is an increasing incidence of HIV among specific subpopulations each year, despite some stabilization.^2-4 This pattern occurs alongside the advancement of behavioral and biomedical interventions^5-7 that, in combination, could avert a significant number of new infections.⁸ However, despite promising trends,² engagement of MSM in HIV prevention simply has not occurred at the necessary pace to curb the epidemic across the United States⁴—a particularly urgent problem to prioritize, given the federal goal to end the epidemic within the next 10 years.⁹ Few HIV-negative or HIV-status unknown MSM report participation in behavioral interventions,^10-12 and condom use has decreased in recent years. ¹³ Use of biomedical interventions such as pre-exposure and postexposure prophylaxis (PrEP and PEP) also remains low.^12,14 5 years after the Food and Drug Administration approval, 35% of MSM reported PrEP use,¹⁵ but retention in PrEP-specific health care continued to be suboptimal.¹⁶ Among MSM living with HIV, surveillance data suggest a similar retention problem across the HIV care continuum,^17-19 compromising the potential for community-level viral suppression.²⁰

Leveraging behavioral and biomedical interventions to end the epidemic⁹ will require identifying mechanisms that limit involvement of those MSM least likely to engage in HIV prevention. Stigma, the social and structural labeling of differences that empowers the stereotyping, separation, loss of status, and discrimination of those labeled,²¹ is a major barrier in the prevention and control of HIV globally across all populations.^22,23 For MSM, social and structural barriers related to sex and sexuality (ie, sexual stigmas)^24,25 are potential candidates for intervention to improve engagement, as they intersect across stigmas among MSM, including devaluation by racial group, ethnicity, socioeconomic status, HIV status, and other labels.^1,26 Left unaddressed, sexual stigma particularly impedes MSM engagement in HIV prevention and treatment.^27-31 Fortunately, sexual stigma is also amenable to modification.^32,33

An important and rarely studied aspect of sexual stigma is stigma toward anal sexuality.³⁴ Anal sex is the primary route of HIV transmission among MSM, but the influence of any associated stigma on MSM engagement in HIV prevention practices is fairly uncharted. Limited evidence, largely qualitative, suggests that anal sex may function as a label for social devaluation among some MSM^35-37 and that experiences related to stigma, including the absence of information about anal sexuality, likely influence both men's HIV-relevant decision-making during sex and their health-seeking behavior in health-care settings.^38,39 To date, however, studies have yet to quantify potential associations between exposure to anal sex stigma and MSM engagement in HIV prevention practices.

To better understand the potential role of stigma specifically toward anal sex, we developed an Anal Health Stigma Model (Figure 1) based on the theories of concealable stigmatized identities,⁴⁰ sexual stigma,²⁴ fundamental causes of disease,⁴¹ and our own qualitative work.³⁴ Our qualitative work found that MSM harbor anal sex–related concerns and anxieties, often posed as questions about anal physiology and sexual functioning, connected to the absence of sexual education.³⁴ These concerns are likely related to HIV risk and concealment. For example, interest in hygiene, pain reduction, and pleasure motivate the use of douches,^42,43 substances,^44-47 and lubricants,^48,49 which can potentially exacerbate HIV risk,^50-52 but which, similar to sexual behavior in general, MSM rarely discuss with sex partners or in health-care settings.³⁴ Therefore, we hypothesize that stigma toward anal sexuality contributes to elevated sex-related concerns along with compromised comfort discussing anal sex, a precursor to concealment. These mediators, as well as stigma, likely contribute to behavioral health responses among MSM that impede both their health-seeking engagement in sexual risk reduction and in HIV prevention and care services.

Figure 1.

Anal Health Stigma Model positing direct effects of high anal sex stigma linked to less engagement, mediated through elevated anal sex–related concerns, and lessened comfort discussing anal sexuality. A circle denotes a latent factor; a square denotes a manifest (observed) variable per.¹⁰⁴

To study these determinants, we surveyed a national, ethnically and racially diverse sample of cisgender MSM. We conducted structural equation modeling (SEM) to identify whether anal sex stigma does indeed impede HIV prevention practices through our proposed mediators.

METHODS

Following upon a qualitative study,³⁴ we developed a set of scales and an inventory of sex-related questions and concerns, which we subsequently pilot tested and evaluated for scale performance,¹⁰⁵ then used in model testing among 1,263 sexually active cisgender MSM. The ethical review boards at both the University of Washington and New York State Psychiatric Institute approved the research.

Sample

Recruitment relied on snowball and targeted sampling, including paid and donated advertisements on social media and men-seeking-men platforms. We also sent electronic announcements to affinity groups, with at least 1 in each state or territory dedicated to serving people of color. Images for announcements reflected racial and ethnic diversity.

Eligibility criteria required reporting as a cisgender man (using a 2-step method⁵³), with at least 1 male anal sex partner in the past year; living in the United States; speaking, reading, and writing English; and being 18 years of age or older. We defined “anal sex” broadly, including penile, oral, and manual stimulation. Screening, consent, and surveys were hosted through an online platform.⁵⁴ We redirected non-Latino white MSM away from the survey after this group comprised a quota of more than one-third of the targeted sample size, to ensure sufficient racial and ethnic diversity.

Of the 1,936 respondents who responded beyond the information statement, 549 (28%) partially completed the survey, and 1,387 (72%) completed the survey. Among these 1,387 completers, 1,263 (91%) reported penile-anal intercourse with a male partner in the past 3 months, forming our analytic sample. Their median completion time was 27 minutes (interquartile range: 18 minutes).

Procedures

On the 1st survey screen, participants could opt into viewing sexually explicit cartoon images which were featured intermittently to encourage retention and attention.^55,56 We randomized items within measures to lessen response bias from a fixed order. Respondents who completed the survey could voluntarily choose to enter a raffle to win 1 of 12 $50 gift certificates. We collected responses over 10 weeks, between July and September 2017.

To flag potential fraudulent responses, we checked repetition across key variables, including eligibility criteria, age, zip code, start/stop times, and internet protocol and email addresses, when available.⁵⁷ We further examined responses that indicated: a non-U.S. location; a previously excluded respondent who ended participation within the prior 2 hours; straight-line responding; completion times less than one-third of the median survey length; and final survey answers that misaligned with related earlier questions. This process excluded 17 respondents from inclusion in our sample of 1,263, or 1.3%.

Measures

Primary Explanatory Variable

Anal Sex Stigma.

We developed a 17-item measure of stigma toward anal sexuality, the Anal Sex Stigma Scales (ASS-S, α = 0.81), based on earlier qualitative and quantitative studies,^33,52 then validated the measure with the 1,387 respondents of the present study who completed measures, regardless of their recent sexual activity or positional preferences during intercourse.¹⁰⁵ Among the 1,263-person analytic sample, 1 subscale comprised self-stigma (α = 0.72; eg, “I may never let go of the shame I feel about anal sex”), and 2 other subscales comprised a combination of experienced and anticipated stigma in the forms of omission stigma (α = 0.73; eg, “Most guys don't understand how to ease into anal sex”) and provider stigma (α = 0.80; eg, “Health workers would treat me badly if they knew the ways I have anal sex”). Likert response categories ranged from 0 (disagree strongly) to 3 (agree strongly).

Proposed Mediators

Anal Sex–Related Concerns.

In our earlier formative work, we validated an index of 45 frequently asked questions about anal sexuality, the Anal Sex Questions Index (ASQx), derived from 35 in-depth interviews with key informants and MSM participants, pilot tested among 218 MSM online, and validated among the present study's larger sample.¹⁰⁵ We selected the 10 least skewed items with the highest variability in our sample of sexually active MSM, which collectively indicated a measure of worry or concern (α = 0.89; eg, “Why does anal sex feel different for me than it used to feel?” “How many other guys have problems with anal sex like the problems I have?”). Respondents rated their interest in hearing an answer to each question, from 0 (not at all) to 3 (very interested).

Comfort Discussing Anal Sex and Sexual Orientation.

1 item measured respondent comfort talking with medical providers about their specific anal sex practices,⁵⁷ with Likert response categories ranging from 0 (not at all) to 4 (extremely comfortable). We also developed a parallel item for comfort discussing attraction to men.

Primary Dependent Variable

Engagement in HIV Prevention Practices.

We operationalized engagement as a latent construct in SEM, based on 3 dichotomous variables (Figure 2): (i) behavioral intervention, (ii) biomedical intervention, and (iii) prevention strategy during recent penile-anal intercourse (standardized Cronbach's α = 0.54).

Figure 2.

Final model showing only significant paths with standardized beta coefficients, adjusting for covariates in dotted lines, *P <.05; **P <.01; ***P <.001. A circle denotes a latent factor; a square denotes a manifest (observed) variable per.¹⁰⁴ Anal Sex Stigma comprises 3 latent subscales (self-stigma, provider stigma, and omission stigma). Engagement comprises 3 observed dichotomous variables (behavioral intervention, biomedical intervention, prevention practice during recent PAI). PAI refers to penile-anal intercourse.

Behavioral Intervention.

Those who reported ever participating in a one-to-one or small group conversation with a health worker about HIV prevention practices, as operationalized by the Centers for Disease Control and Prevention,⁵⁸ were coded as having engaged in a behavioral intervention.

Biomedical Intervention.

We followed conventions for engagement in HIV treatment for people living with HIV,¹⁹ HIV testing recommendations for MSM,⁵⁹ treatment guidelines for sexually transmitted infection,⁶⁰ and protocols for PrEP.⁶¹ To be considered not engaged in biomedical intervention, seropositive respondents needed to report one of the following: most recent medical visit > 6 months ago, > 2 missed antiretroviral therapy (ART) doses in the past week, or detectable viral load. For seronegative respondents, those who reported no HIV test result within the past 2 years or current PrEP use but no HIV test in the past 3 months were coded as not engaged in biomedical intervention. Finally, respondents of any serostatus who suspected an incident sexually transmitted infection in the past 3 months but who did not go to a health-care provider, or who were never tested for HIV or never received an HIV test result, were coded as not engaged in biomedical intervention.

Prevention Strategy During Penile-Anal Intercourse.

We developed a dichotomous variable for prevention strategy based on the protective benefits of condom use,⁶² adherence to antiretroviral medication as either treatment or PrEP,^5,7,61 and harm reduction strategies that MSM use to minimize HIV risk.^63,64 We coded men as engaged in a prevention strategy during recent penile-anal intercourse who met any of the following conditions: (i) consistent (100%) use of condoms; (ii) consistent use of ART or PrEP (ie, undetectable viral load if seropositive, and ≤ 2 missed doses of ART or PrEP in the past week regardless of serostatus); or (iii) condomless intercourse with only one partner in the past year whom the respondent had not recently met, who had no concurrent sexual partnerships in the past 3 months, and who was either seroconcordant or adhering to ART or PrEP protocols.

Covariates

In interviews that informed the development of the ASS-S, respondents described an amalgamation of anal sex stigma and sexual racism,¹⁰⁵ an intersection of multiple targets of devaluation⁶⁵ with relevance to sexual behavior and health-seeking behavior.^66,67 Social support may also play a role as a buffer against stigma and attendant urges to conceal sexuality and as encouragement for health-seeking behavior.^68,69 Our model therefore adjusts for exposure to these constructs and tests for additional sociodemographic confounds of the association between anal sex stigma and HIV prevention.

Social Support.

We adapted the 8-item emotional and informational subscale of the Medical Outcomes Study Social Support Scale⁷⁰ to be specific to anal sexuality, which yielded high reliability in our sample (α = 0.97). Participants endorsed the availability of support (eg, “Someone to share my most private worries and fears with about anal sex” and “Someone who's advice about anal sex I really want”) with Likert response categories ranging from 1 (none of the time) to 5 (all of the time).

Everyday Discrimination.

The 5-item Everyday Discrimination Scale⁷¹ measured perceived frequency of discrimination (α = 0.87; eg, “You are treated with less respect than others”) based on a range of potential stigmatized identities, including racial and ethnic identity, with Likert response categories ranging from 0 (never) to 5 (almost every day).

Analytic Plan

SEM analyses were conducted using Mplus. We followed a common 2-step procedure: 1st, confirmatory factor analysis to assess the fit of the measurement model and then path analysis with model specification.⁷² Given the number of ordinal variables in our model, we chose a weighted least squares multivariate estimator using a diagonal weight matrix, with theta parameterization, with theta parameterization. To assess model goodness of fit, we examined the comparative fit index (CFI), Tucker-Lewis index (TLI), and the root mean square error of approximation (RMSEA), using their widely accepted cutoff scores (CFI and TLI > 0.9; and RMSEA < 0.05).^73,74 We also examined the ratio of the chi-square to degrees of freedom (χ²/df), as a guide toward increasing fit in successive models, with a target between 2 and 3, as long as all other fit indices were favorable.^73,75 To examine sources of substantial model misfit, we requested modification indices > 50 to assess options to increase model fit that might be congruent with theory. To select measures for inclusion as covariates in model testing, we examined bivariate statistics, using Spearman's rho (r_s), point biserial correlations for dichotomous components of engagement, and chi-square and t-test analyses. We used the joint significance test, where both paths have to be significant, to conclude mediation. Literature supports that joint significance is almost as powerful as bootstrapping without the computational demands and potential for alpha inflation.⁷⁶

Mplus conducts full information maximum likelihood estimation by default,⁷⁷ thereby including participants with missing endogenous items. We conducted Little's Missing Completely at Random test,⁷⁸ and the results were significant (χ² = 833.53(662), P = .000), indicating that data were not missing completely at random. However, missing data were minimal, less than 0.3% at the item level across respondents.

RESULTS

Sample and Descriptive Findings

The sample reflected a broad distribution across ages, incomes, racial/ethnic group identification, and geographic distribution (see Table 1). Most reported a frequency of engaging in anal sex between twice a week to monthly. The top quartile reported more than 13 anal sex partners in the past year, and most expressed a preference for versatile sexual positioning, with exclusively insertive anal intercourse least preferred. Most men reported that their last HIV test was seronegative, and more than one-third of these men reported currently using PrEP. The vast majority of respondents living with HIV reported engagement with health care and had achieved viral suppression. Most samples reported past engagement in a behavioral intervention, with slightly more currently engaged in biomedical intervention, and fewer engaged in an HIV prevention strategy during recent penile-anal intercourse.

Table 1.

Characteristics for cisgender MSM reporting recent penile-anal intercourse (N = 1,263)

Variable	n (%)
Age in years (M, SD)	36.1 (11.0)
Geography
Rural	93 (7.4)
Small	145 (11.5)
Medium-sized	279 (22.1)
Large	586 (46.4)
Suburb	160 (2.7)
Region
Northeast	278 (22.0)
Midwest	224 (17.7)
South	520 (41.2)
West	240 (19.0)
Race and ethnicity
Latino, Hispanic, or Spanish ethnicity (of any race)	270 (21.4)
American Indian/Alaska Native (AIAN)	17 (1.3)
Asian or Asian American	78 (6.2)
Black or African American	204 (16.2)
White or Caucasian (non-Latino)	673 (53.3)
Native Hawaiian or Pacific Islander (NHPI)	4 (0.3)
Biracial or multiracial	80 (6.3)
Education
≤ High school/GED	99 (7.8)
Some college	289 (22.9)
2-year college	73 (5.8)
4-year college	415 (32.9)
Masters	266 (21.1)
Doctoral/professional	121 (9.6)
Income
Less than $15,000	159 (12.6)
$15,000–$29,999	217 (17.2)
$30,000–$44,999	208 (16.5)
$45,000–$59,999	175 (13.9)
$60,000–$74,999	132 (10.5)
$75,000–$89,999	92 (7.3)
$90,000 or more	270 (21.4)
Housing
Shelter/dorm/drug treatment/other	21 (1.7)
Someone else's home	157 (12.4)
Rent	717 (56.8)
Own	368 (29.1)
Relationship status*
Single	607 (48.1)
Casually dating several people	91 (7.2)
Boyfriend or girlfriend	146 (11.6)
Partner or lover	207 (16.4)
Legal, civil, committed partnership	202 (16.0)
Open relationship or not sure if open	410 (32.5)
Sexual orientation
Gay	1,022 (80.9)
Bisexual/heterosexual	120 (9.5)
Queer	62 (4.9)
No label	45 (3.6)
Additional/2 spirit	14 (1.1)
Sexual position preference
“Bottoming” (receptive)	391 (31.0)
“Topping” (insertive)	282 (22.3)
“Versatile” (both receptive and insertive)	550 (43.5)
No preference/not sure	40 (3.2)
Outness about sexual attraction to men
Nobody	32 (2.5)
Select friends	161 (12.7)
All friends and select family	213 (16.9)
All friends and all family	94 (7.4)
Almost everyone (friends, family, coworkers, and so on)	763 (60.4)
Never tested/never received HIV result	71 (5.6)
HIV seronegative	999 (79.1)
Tested within the last year	887 (88.8)
Current PrEP use	350 (35.0)
HIV seropositive	183 (14.5)
Diagnosed within past 2 years	23 (12.6)
Linked to care at time of diagnosis	170 (92.9)
Retained in care (most recent visit < 6 mos ago)	175 (95.6)
Prescribed ART	178 (97.3)
Undetectable viral load	167 (91.3)
Health-care engagement
No health insurance or primary care provider (PCP)	90 (7.1)
Out to PCP about anal sex with men (among 1,092 with PCP)	811 (74.3)
Behavioral HIV intervention	942 (74.6)
Thought you had an STI but did not see medical provider	118 (9.3)
Sexual Behavior
13 or more anal sex partners in past year†	325 (25.7)
Receptive penile-anal intercourse in past 3 months	1,019 (80.7)
Insertive penile-anal intercourse in past 3 months	963 (76.2)
HIV-related Sexual Behavior in past 3 months
Any condomless receptive anal intercourse	824 (80.9)
Any condomless insertive anal intercourse	789 (82.0)
Chemoprophylaxis (PrEP/ART)	470 (37.2)
Any HIV prevention strategy during anal intercourse	805 (63.7)

ART = antiretroviral therapy; GED = General Education Diploma; HIV = human immunodeficiency virus; M = mean; PrEP = pre-exposure prophylaxis; STI = sexually transmitted infection.

^*N = 1,253 due to missingness.

^†“How many people have you had anal sex within the past year? (Anal sex here includes any sexual contact with the ass, like touching, licking or penetration.)”

Overall, participants reported a low level of anal sex stigma (mean [M] = 1.04, SD = 0.50), approximately one-third of the possible 0 to 3 range, with higher levels of omission stigma (M = 1.60, SD = 0.28) than provider stigma (M = 0.94, SD = 0.12) and self-stigma (M = 0.56, SD = 0.21). Participants harbored moderate sex-related concerns (M = 1.72, SD = 0.14), about 57% of the possible 0 to 3 range. Social support specific to anal sexuality was also low to moderate (M = 2.00, SD = 0.13); on average, support was available “a little of the time.” They reported moderate comfort discussing anal sex and attraction to men, with much higher variability than for the other measures (M = 2.10, SD = 1.39 and M = 2.73, SD = 1.31, respectively).

Bivariate Analyses

We detected significant differences in how respondents ranked each of the 2 comfort with discussion items (Wilcoxon Signed Ranks Test, Z = −20.5, P < .0001). Overall, nearly half (47.4%) felt more comfortable discussing attraction to men than anal sex practices, 2.8% reported the inverse, and 49.7% reported the same comfort level regardless of the topic. We therefore opted to use the single item related to anal sexuality in our model.

Bivariate analyses with sociodemographic variables indicated that greater age, higher household income, higher education, greater outness about attraction to men, medical coverage, and racial group identification other than black/African-American were each negatively associated with stigma and positively associated with engagement (all P < .05). We therefore constructed a formative latent factor for sociodemographics, comprising these 6 variables, to adjust for potential confounding while accounting for intercorrelation and measurement error. As a linear composite of these risk factors, higher values indicate better outcomes.

Measurement Model

Our measurement model comprised 6 latent factors, including our primary variables and our 3 covariates, and converged normally with acceptable fit: χ²(1,109) = 3324.33, P < .0001, χ²/df = 3.00, RMSEA = 0.040 (95% CI: 0.038–0.041), probability of RMSEA < 0.05 = 1.00, CFI = 0.99, TLI = 0.99. Standardized factor loadings (λ) on engagement were all significant (P < .001): behavioral intervention (λ = 0.58), biomedical intervention (λ = 0.58), and prevention strategy during recent penile-anal intercourse (λ = 0.43). Table 2 details means, standard deviations, correlations, and covariances among study variables.

Table 2.

Loadings, means, standard deviations, reliability, correlations, and covariances among all variables (N = 1,263)

	Variable	λ†	M (SD)	α (95% CI)	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16
	Anal Sex Stigma Scales		1.04 (0.50)	0.81 (0.80–0.83)
1	Omission	0.80	1.60 (0.28)	0.73 (0.70–0.75)	–	0.61	0.65	0.33	−0.29	−0.87	0.36	−0.08	−0.10	−0.14	−0.14	−0.13	−0.25	−0.17	−0.16	−0.10
2	Self-stigma	0.75	0.56 (0.21)	0.72 (0.69–0.74)	0.53***	–	0.66	0.55	−0.59	−1.26	0.60	−0.23	−0.24	−0.17	−0.30	−0.30	−0.46	−0.34	−0.43	−0.20
3	Provider-stigma	0.49	.94 (0.12)	0.80 (0.78–0.81)	0.50***	0.32***	–	0.31	−0.53	−0.63	0.70	−0.02	0.20	−0.02	−0.10	−0.01	−0.04	−0.11	−0.25	−0.23
4	Anal sex concerns		1.72 (0.14)	0.89 (0.88–0.90)	0.41***	0.41***	0.21***	–	−0.14	−0.37	0.29	−0.06	−0.19	−0.11	−0.17	−0.15	−0.20	−0.11	−0.13	−0.04
5	Comfort discussing anal sex		2.10 (1.40)		−0.34***	−0.44***	−0.35***	−0.15***	–	1.09	−0.22	0.10	0.03	0.06	0.12	0.08	0.38	0.35	0.37	0.23
6	Social support		2.00 (0.13)	0.97 (0.97–0.98)	−0.36***	−0.32***	−0.14***	−0.13***	0.38***	–	−0.29	0.03	0.07	0.24	0.30	0.33	0.73	0.73	0.53	−0.50
7	Multiple stigmas		1.61 (1.10)	0.87 (0.86–0.88)	0.31***	0.33***	0.34***	0.22***	−0.16***	−0.07*	–	−0.42	−0.01	−0.21	−0.36	−0.30	0.01	−0.04	−0.26	−0.13
	Sociodemographics
8	Age	0.51			−0.09**	−0.17***	−0.01	−0.06	0.10**	0.01	−0.31***	–	0.07	0.24	0.53	0.30	0.08	0.19	0.18	0.28
9	Race	0.34			−0.12**	−0.17**	0.13**	−0.20***	0.03	0.02	−0.01	0.07	–	0.25	0.21	0.05	0.41	0.13	0.15	0.03
10	Education	0.52			−0.17***	−0.13***	−0.02	−0.11***	0.06*	0.08**	−0.16***	0.24***	0.25***	–	0.48	0.44	0.21	0.20	0.16	0.19
11	Income	0.74			−0.17***	−0.22***	−0.07*	−0.18***	0.12***	0.11***	−0.27***	0.53***	0.21***	0.48***	–	0.41	0.24	0.18	0.25	0.20
12	Medical coverage	0.57			−0.15**	−0.22**	−0.01	−0.15**	0.08	0.11*	−0.22***	0.30***	0.05	0.41***	0.44***	–	0.11	0.14	0.10	0.28
13	Outness	0.60			−0.30***	−0.34***	−0.03	−0.21***	0.38**	0.25***	0.01	0.08*	0.41***	0.21***	0.24***	0.11	–	0.35	0.25	0.27
	Engagement
14	Behavioral intervention	0.58			−0.20***	−0.25***	−0.07	−0.12**	0.35***	0.26***	−0.03	0.19***	0.13**	0.20***	0.18***	0.14*	0.35***	–	0.33	0.24
15	Biomedical intervention	0.58			−0.19***	−0.32***	−0.16***	−0.14**	0.37***	0.18***	−0.19***	0.18***	0.15**	0.16***	0.25***	0.10	0.25***	0.33***	–	0.27
16	Prevention during PAI†	0.43			−0.11**	−0.15***	−0.15***	−0.05	0.23***	−0.17***	−0.09*	0.18***	0.03	0.19***	0.20***	0.28***	0.27***	0.24***	0.27***	–

Covariances are in bold; correlations are below the diagonal

^*P < .05

^**P < .01

^***P < .001 (all 2-tailed).

PAI = penile-anal intercourse.

^†Factor loadings are all significant (P < .001)

Anal Health Stigma Model

The structural paths hypothesized in our conceptual model (Figure 1), adjusting for covariates, provided good fit for the data: χ²(1,159) = 3575.66, P < .0001, χ²/df = 3.09; RMSEA = 0.041 (95% CI: 0.039–0.042), probability of RMSEA < 0.05 = 1.00, CFI = 0.99, TLI = 0.99. However, modification indices (ranging from 57 to 195) suggested additional and theoretically congruent associations. Their inclusion in the final model (Figure 2, showing only significant paths) improved fit (χ²(3) = 48.18, P < .0001) and indicated good fit: χ²(1,156) = 3449.72, P < .0001, χ²/df = 2.98; RMSEA = 0.040 (95% CI: 0.038–0.041), probability of RMSEA < 0.05 = 1.00, CFI = 0.99, TLI = 0.99. This final model accounted for 67% of the variance in engagement, 39% in anal sex stigma, and 27% in anal sex concerns (all P < .001).

As seen in Table 3, the direct effect of stigma on engagement was not statistically significant, but total effects indicate that higher stigma was linked to less engagement (β = −0.22, P < .001). This association was mediated (see Figure 2) by comfort discussing anal sexuality, with higher stigma linked to less comfort (β = −0.52) and more comfort linked to more engagement (β = 0.44; both P < .001).

Table 3.

Standardized β from structural equation modeling to test direct, indirect and total effects of Anal Sex Stigmas on Engagement in HIV Prevention among cisgender MSM who reported anal sex in the preceding year (N = 1,263)

		Explanatory variables			Covariates adjusted for confounding
	Dependent variables	Anal sex stigmas	Anal sex concerns	Comfort discussing anal sex with health workers	Social support specific to anal sex	Everyday discrimination	Sociodemographics
Direct effects	Anal sex stigmas	–			−0.35*	0.34*	−0.24†
	Anal sex concerns	0.55*	–		0.09†
	Comfort discussing	−0.52*	0.15*	–	0.19*
	Engagement	−0.02	−0.01	0.44*	0.11‡	0.09	0.56*
Indirect effects	Anal sex stigmas	–
	Anal sex concerns		–		−0.19*		−0.13*
	Comfort discussing	0.08†		–	0.17*		0.10*
	Engagement	−0.23*	0.06†		0.16*	−0.08†	0.05‡
Total effects	Anal sex stigmas	–			−0.35*	0.34*	−0.24*
	Anal sex concerns	0.55*	–		−0.10†		−0.13*
	Comfort discussing	−0.44*	0.15*	–	0.35*		0.10*
	Engagement	−0.22*	0.05	0.44*	0.28*	0.02	0.61*

R² in Engagement = 0.67%; RMSEA = 0.040 (95% CI: 0.038–0.041, probability of RMSEA < 0.05 = 1.00); CFI = 0.99; TLI = 0.99. HIV = human immunodeficiency virus.

^*P < .001

^†P < .01

^‡P < .05

Model testing did not support mediation from stigma to sex-related concerns (β = 0.55, P < .001) and then to engagement (β = −0.01, P = .90), nor a total effect of these concerns on engagement (β = 0.06, P = .35). However, the final model supported a mediation effect of stigma on comfort through elevated concerns: higher stigma was linked to elevated concerns (β = 0.55, P < .001), and elevated concerns were weakly but significantly associated with greater comfort (β = 0.15, P < .01).

Notably, higher informational and emotional social support specific to anal sexuality was linked to lower levels of stigma (β = −0.35, P < .001), greater comfort (β = 0.35, P < .001) and lower concerns (β = −0.10, P < .01), in addition to greater engagement (β = 0.28, P < .001).

DISCUSSION

In our sample, MSM who endorsed less anal sex stigma also reported important HIV-related protective behaviors, including greater comfort talking with health workers about their specific anal sex practices and greater engagement in HIV services and sexual risk reduction. The overall association between stigma and engagement was low, but these other relationships had medium to strong effect sizes, even with adjustment for social support, other forms of discrimination, and confounding sociodemographic factors (ie, age, outness, racial group, education, income and medical coverage). We found no evidence of mediation by elevated interest in anal sex–related questions, our measure of concern, despite a strong negative association with stigma. However, the model also indicated that social support specific to anal sex was significantly and strongly associated with less stigma and more comfort, and with greater engagement.

The finding that discomfort with discussion mediates the relation between stigma and health is consistent with our formative qualitative work³⁴ and literature on sexual stigma.⁷⁹ MSM may have strong reasons to conceal their participation in anal sex in particular. For example, heterosexuals report less willingness to attend a dinner event with a fictionalized gay man who has been characterized as “versatile” as compared with when his involvement in anal sex is disclosed but not specified positionally.³⁵ As we heard in earlier qualitative interviews during our formative research, this may contribute to men's aversion to discussing specific anal sex practices, whether with friends, family, and sex partners, or in health-care settings.³⁴ We detected this aversion to discussing anal sexuality in our current sample, with significantly greater comfort discussing attraction to men than specific anal sex practices. This suggests that discussing anal sex functions somewhat differently than discussing sexual orientation and is a reminder that the more HIV-specific topic of anal sexuality may be more difficult to broach and target for intervention.

Avoiding the subject of anal sexuality in health-care may therefore be common among MSM, as protection against the deleterious effects of stigma on health and social wellbeing. In our sample, of those with a primary medical provider, 75% reported being “out” about anal sex with men to their provider, which is higher than the 68% reported in the 2008 National HIV Behavioral Surveillance System⁸⁰ and somewhat consistent with other studies documenting outness about sexual orientation. We know that difficulty disclosing MSM status is associated with both poor health-care access⁸² and less receipt of MSM-specific recommended screenings,⁸³ similar to our study. This pattern suggests that many MSM may be underidentified for continued intervention and that their urges to conceal, albeit protective to some extent,^84,85 also compromise their access to health care and the biomedical and behavioral interventions that could curb the HIV epidemic.^79,80 Our findings add that concealment among MSM is not purely about sexual orientation or general sexual attraction but includes specific reticence to discuss anal sex, the behavior most associated with their disproportionate burden of HIV.

The effects of social support in our sample are particularly compelling in light of the need for novel interventions that encourage disclosure and promote engagement. Our social support measure is an adaptation of the emotional and informational support subscale of the Medical Outcomes Study Social Support Scale, specifying anal sex within each item. Although our measure of interest in answers to questions about anal sex was not significantly associated with poor engagement, responding to MSM-specific anal sex interests, whether in health care or other relationships, might function as emotional and informational support. Our model testing suggests that this, in turn, could increase comfort with discussion about specific anal sex practices. One novel opportunity may be to respond to specific questions MSM harbor about anal sex, as a possible path to improve engagement in HIV services and sexual prevention practices. In pilot testing in a high-stigma context, this sort of sex-positive approach to mitigating stigma, focused on client questions about anal pleasure and health, has demonstrated high acceptability, feasibility, and appropriateness among health workers.⁸⁶

While we found evidence of a negative association of anal sex stigma with engagement in HIV prevention, it is important to note that intersecting forms of devaluation operate on health-related behavior.^87,88 Attention to the intersection of anal sex stigma with other stigmas may inform ideas about how to inoculate the effects of multiple stigmas, particularly given the presence of effects in our study even after adjusting for confounding by other forms of everyday discrimination. One potential option might be to intervene across the diversity of people who engage in anal sex, rather than commit further to stereotypes that equate anal sex with male homosexuality.³⁴ Indeed, the Anal Health Stigma Model is broadly applicable to additional populations that engage in anal sex.^89-94 The extent to which stigma among these populations too is associated with poor engagement would be important to quantify.⁹⁵

Our study has several limitations. We should be cautious to advocate for specific interventions based on a single, cross-sectional observational survey. We found evidence of statistical mediation, but testing true causal pathways requires more advanced designs, such as a longitudinal cohort study. Without testing for temporal precedent, stigma could be the cause of lowered social support, for example, rather than the reverse path posited in our model. However, the ASS-S show strong relations with variables that a priori match theory and literature; our findings, although limited, may function as a further test of the scales' validity and relevance to HIV-related behavior. Self-report also limits measurement in our study,⁹⁶ as our measures are prone to social desirability and recall bias, although the online collection of data along with questions about current and recent experiences may lessen the distorting effects of these forms of bias. 72% of respondents completed our survey, an indication of the potential for selection bias. This may be a function of greater reliance on recruitment through men-seeking-men online venues, but this relatively high completion rate⁹⁷ points to sustained interest while responding to our study, reflected in end-of-survey comments. Our findings also may not be representative of cisgender MSM in general, as participation was limited to those who use the Internet, although almost all adults⁹⁶ and most MSM are engaged online,⁹⁸ in particular to find sex partners.⁹⁹ In addition, our sample also skewed toward greater financial resources. We adjusted for income in our model, but representative sampling could reveal different associations. In addition, although we took precautions to flag and exclude potentially fraudulent and careless responders, we cannot truly detect and accurately excise these kinds of responses from analyses.

Even with these limitations, our findings have important implications for our collective response to the HIV epidemic. The project contributes to behavioral science by testing a broadly applicable model based on how anal sex stigma impedes the 2 main approaches in preventing HIV: engagement in health care and in safer sex practices, the combination critical to curbing and eventually ending the epidemic.¹⁰⁰ Our model also acknowledges that HIV-relevant behaviors occur within a social and relational context^8,101,102 and research informed by social processes holds the potential to reveal health factors at levels further upstream from individual behavior that may not yet be well understood but that, consistent with theory on fundamental causes of disease,¹⁰³ could suggest additional points for intervention on a broader set of health outcomes than just HIV.

CONCLUSIONS

Anal intercourse is both a socially devalued behavior and, under specific circumstances, the most proximate risk factor for HIV among MSM. Our study now quantifies the potential effects of this devaluation. Reluctance to discuss anal sexuality with health workers appears both to derive from anal sex–specific stigma and to impede engagement in prevention services and sexual risk reduction. We did not find evidence of a direct or mediating effect of anal sex concerns on engagement. However, responding to specific questions about anal sex that interest MSM, such as those related to sexual pleasure and well-being, may function as social support and thereby provide avenues to improve comfort discussing anal sexuality and thereby MSM engagement in HIV prevention.