Childhood Exposure to Religions With High Prevalence of Members Who Discourage Homosexuality Is Associated With Adult HIV Risk Behaviors and HIV Infection in Black Men Who Have Sex With Men

Abstract

Exposure to childhood religious affiliations where the majority of members discourage homosexuality may have negative psychological impacts for Black men who have sex with men. This study tested the hypothesis that exposures to these environments during childhood were associated with adulthood human immunodeficiency virus (HIV)/sexually transmitted infection (STI) behavioral risk and HIV infection, because these exposures influenced HIV/STI risk by undermining race/sexual identity congruence and increasing internalized homophobia and interpersonal anxiety. Structural equation modeling as well as logistic and Poisson regressions were performed using baseline data from HIV Prevention Trials Network 061 (N = 1,553). Childhood religion affiliations that were more discouraging of homosexuality were associated with increased likelihood of HIV infection; however, the association was no longer significant after adjusting for age, income, and education. Having a childhood religion affiliation with high prevalence of beliefs discouraging homosexuality was associated with increased numbers of sexual partners (adjusted odds ratio = 4.31; 95% confidence interval [3.76, 4.94], p < .01). The hypothesized path model was largely supported and accounted for 37% of the variance in HIV infection; however, interpersonal anxiety was not associated with HIV/STI risk behaviors. Structural interventions are needed that focus on developing affirming theologies in religious institutions with Black men who have sex with men congregants.

Human immunodeficiency virus (HIV) continues to disproportionately affect Black communities in the United States. Current epidemiologic data indicate that Blacks represented 45% of all new HIV infections, while reflecting approximately 14% of the U.S. population (Centers for Disease Control and Prevention [CDC], 2015). The majority of all new HIV infections among Blacks have been observed in men who have sex with men (MSM). The estimates of MSM representation in new HIV infections among Black males range between 63% (CDC, 2012) and 66% (Prejean et al., 2011), with even higher estimates and greater disparities among MSM living in southeastern states (Lieb et al., 2011). The distribution of new HIV infections among Black MSM is increasing, with those younger than 25 years of age accounting for over half of the new infections (CDC, 2015). The primary mechanism of HIV transmission for Black MSM is condomless anal intercourse; however, likelihood of engagement in HIV risk behaviors is influenced by a variety of psychosocial factors (Egan et al., 2011; E. L. Fields et al., 2012; Jeffries, Marks, Lauby, Murrill, & Millett, 2013; Kelly et al., 2013; Koblin et al., 2013; Millett, Jeffries, et al., 2012; Tieu et al., 2012).

HIV risk for Black MSM must be understood within the context of social and structural factors that shape their vulnerability to HIV infection (Levy et al., 2014; Magnus et al., 2010; Millett, Jeffries, et al., 2012; Millett, Peterson, et al., 2012). In Black communities, the “church” is a revered cultural and spiritual institution that has a rich history of struggle against oppression and marginalization (Jeffries, Dodge, & Sandfort, 2008; Miller, 2005; Woodyard, Peterson, & Stokes, 2000; Quinn & Dickson-Gomez, 2015; Taylor, Chatters, & Levin, 2003). Nonetheless, a number of studies indicate that religious institutions have a variety of attitudes and approaches for dealing with MSM in their congregations—perhaps the most common being that same-sex behavior is an abomination that is prohibited by divine scriptures and therefore cannot be accepted and must be condemned (Foster, Arnold, Rebchook, & Kegeles, 2011; Harris, 2010; Quinn & Dickson-Gomez, 2015; P. A. Wilson, Wittlin, Munoz-Laboy, & Parker, 2011). Religions with punitive, fundamentalist doctrines are associated with increased homophobia (Balkin, Schlosser, & Levitt, 2009; Rosik, Griffith, & Cruz, 2007; Miller, 2007; Ward, 2005). There are at least three separate studies reporting that MSM’s exposure to these types of religious environments have been associated with development of internalized homophobia (Barnes & Meyer, 2012; Pitt, 2009; Wilkerson, Smolenski, Brady, & Rosser, 2012). Moreover, research identifies that internalized homophobia is associated with HIV risk (Jeffries et al., 2013) as well as a number of mental health sequela, including depression (Barnes & Meyer, 2012), and conflicts between religious/sexual identities (Durell, Chiong, & Battle, 2007; Rodriguez, 2010) and racial/sexual identities (Harawa et al., 2008; Malebranche, Gvetadze, Millett, & Sutton, 2012; Williams, Ramamurthi, Manago, & Harawa, 2009).

Repeated early-life exposure to environments that disaffirm and/or attempt to dismantle essential aspects of one’s personhood, such as culture, spirituality, and sexuality, has been reported to have residual psychosocial impacts in adulthood (Gone, 2007; Ironson et al., 2011; Moore, 2010). There is limited research examining childhood exposure to religions with high prevalence of members who discourage homosexuality as a potential early contributor to the production of longer term HIV/STI risk for Black MSM. This study hypothesized that childhood exposure to religions that discouraged homosexuality would be associated with increased HIV/STI behavioral risk and HIV infection and that the associations would be mediated by internalized homophobia, racial/sexual identity incongruence, and interpersonal anxiety.

Method

Participants and Procedures

The current study used baseline secondary data collected between July 2009 and October 2010 as part of the HIV Prevention Trials Network (HPTN) 061 study (Koblin et al., 2013). HPTN 061 was a multisite study to determine the feasibility and acceptability of a multicomponent HIV prevention intervention for Black MSM (N = 1,553). The study was conducted with a nonprobability sample of Black MSM recruited from Atlanta, Boston, Los Angeles, New York, San Francisco, and Washington, District of Columbia. Men were eligible to enroll if they: self-identified as a man who is Black, African American, Caribbean Black, or multiethnic Black, were at least 18 years old, and reported ≥1 episode of condomless anal intercourse with a man in the past 6 months. The institutional review boards of the clinical research sites in the cities where HPTN 061 was conducted approved the study. Black MSM were recruited directly from the community or as sexual network partners referred by index participants. Further details regarding procedures for HPTN 061 have been published elsewhere (Koblin et al., 2013; Mayer et al., 2014). The U.S. Religious Landscape Survey (USRLS; Pew Forum on Religion and Public Life, 2008) was utilized to establish those religions with the highest prevalence of negative views of homosexuality. Survey data were collected via interviews between May and August 2007. The survey was conducted using a probability sample of U.S. residents aged 18 years and older (N = 35,556). The study focused on understanding religious affiliations, religious practices and beliefs, and their social and political views in the U.S. population. The survey also identified and grouped the religious traditions and/or protestant family of the participants.

Measures

Demographic characteristics included standard measures for age, education, income, HIV serostatus as well as measures of childhood, and current religious affiliation.

Childhood Religious Affiliation—Weighted for Prevalence of Beliefs Discouraging Homosexuality

For this variable, participants were asked to respond to the item “What was your religious denomination affiliation in childhood?” Participants responded to a categorical list of 10 common religious denominations in African American communities. Each of these denominations were weighted based on data from the population-based (N = 35,556) USRLS, which indicated the prevalence of members, within the religion/denominations, that discouraged homosexuality. In the USRLS, a random representative sample of American residents were asked to indicate both their present religious affiliation and the denomination to which they most closely identified—if any (PewPew Forum on Religion and Public Life, 2008). Next, those respondents who reported religious affiliations were presented with the following pairs of statements: “Homosexuality is a way of life that should be accepted by society” or “Homosexuality is a way of life that should be discouraged by society.” They were then asked to indicate which of the two statements came closer to their own views even if neither was exactly right. For each affiliation, percentages were calculated for those who endorsed the statement that homosexuality should be discouraged in society. The prevalence of people discouraging homosexuality (Table 1) ranged from lows of 0.14 (Agnostic) and 0.28 (Episcopal) to highs of 0.76 (Jehovah’s Witness) and 0.79 (Pentecostal). The cases in HPTN 061 were weighted by assigning the “discouraging homosexuality” prevalence value from the religion categories in the USRLS to the corresponding childhood religious affiliations in HPTN 061.

Table 1.

Childhood Religious Affiliations Reported in HIV Prevention Trials Network (HPTN) 061 and Corresponding Prevalence of Beliefs Discouraging Homosexuality From the U.S. Religious Landscape Survey.

HPTN 061 (N = 1, 553)		U.S. Religious Landscape Survey (N = 35,556)
Religious affiliations listed in HPTN 061	MSM reporting childhood participation in religion (n)a	Prevalence of beliefs discouraging homosexuality in society
Agnostic	2	0.14
Baptist	596	0.58
Catholic	109	0.30
Episcopal	14	0.28
Jehovah’s Witness	42	0.76
Lutheran	6	0.37
Methodist	39	0.38
Muslim	20	0.61
Pentecostal	91	0.79
Presbyterian	9	0.39

Note. MSM = men who have sex with men.

^aOnly includes study participants who listed one religious affiliation. Data are not included for study participants who did not list a childhood religious affiliation or who listed more than one childhood religious affiliation.

Internalized Homophobia

A seven-item, 5-point Likert-type scale adapted from Herek and Glunt’s (1995) Internalized Homophobia Scale was used to create the latent factor “internalized homophobia.” The scale included items such as “In the past 90 days, I have tried to stop being attracted to men” and “As a Black man, I try to act more masculine to hide my sexuality.” Responses ranged from 1 (disagree strongly) to 5 (agree strongly). For regression analyses, the sum of scores were categorized as having low (≤10), medium (from 11 to 19), or high (≥20) internalized homophobia. This measure was previously validated and demonstrated strong internal consistency reliability (α = .91). The Internalized Homophobia Scale also demonstrated strong internal consistency reliability (α = .91) in the present study.

Racial and Sexual Identity Incongruence

Five items from the racial/sexual identity integration scale were used to create the latent factor “racial/sexual identity incongruence” in the structural equation model (SEM). This measure assessed the degree to which there was incongruence between one’s racial identity and one’s identity as an MSM. Participants responded using a 5-point Likert-type scale from 1 (strongly disagree) to 5 (strongly agree) to indicate how true each of five statements were for them. The five statements included (1) “Both my race and my sexuality are important to who I am as a man,” (2) “I cannot imagine a loving sexual relationship between two Black men,” (3) “Black men who are attracted to other men can play an important role in Black families,” (4) “A Black man who is attracted to other men can still be a strong man,” and (5) “Black men who are attracted to other men only make it more difficult for Black people in general.” Items 1, 3, and 4 were reverse coded. The mean score of these five items was also calculated for use in a regression. Higher scores indicate greater incongruence between one’s racial and sexual identities. This scale demonstrated good internal consistency reliability in the current study (α = .69).

Interpersonal Anxiety

The latent construct interpersonal anxiety was assessed with four items selected from the Center for Epidemiologic Studies depression scale (CES-D; Radloff, 1977). Confirmatory factor analyses of CES-D items from studies with African Americans identified factors on interpersonal anxiety (Callahan & Wolinsky, 1994; Makambi, Williams, Taylor, Rosenberg, & Adams-Campbell, 2009). Items in the interpersonal anxiety factors included “I felt lonely” and “I was bothered by things that don’t usually bother me.” Participants selected one of five response values that reflected the frequency with which they experienced the symptoms described. The values ranged from 0 (less than 1 day per week [rarely/none of the time]) to 4 (5 to 7 days per week [most of the time]). In regression analyses, the sum score of these four items was calculated. Higher scores indicate higher interpersonal anxiety. This measure demonstrated strong internal consistency reliability in the current study (α = .78).

HIV/STI Behavioral Risk

Three observed variables were used to represent the latent factor HIV/STI behavioral risk. “Condomless insertive anal intercourse” (CIAI) and “condomless receptive anal intercourse” (CRAI) were categorical items measuring whether the participant had any condomless episode of insertive or receptive anal intercourse with men. The variables were derived from several questions related to reported frequency of insertive and receptive anal intercourse and condom use in the previous 6 months with various partner types. Participants were also asked how many of their sexual partners over the past 6 months were male, female, or transgender. In regression analyses, the sum for all gender categories was used as the measure of “total number of sex partners.”

HIV Infection

This variable was assessed using a rapid HIV antibody test conducted after participants received HIV/STI risk reduction counseling. Reactive rapid HIV tests were confirmed by Western blot. Retrospective quality assurance testing was performed at the HPTN Network Laboratory to confirm the HIV infection status of all study participants at enrollment (1 = HIV seropositive, 0 = HIV seronegative). For participants with low or undetectable HIV ribonucleic acid who did not report a prior HIV diagnosis, enrollment samples were tested for the presence of antiretroviral drugs after the end of the study; men whose samples contained antiretroviral drugs indicative of antiretroviral therapy were considered to have had a prior HIV diagnosis at baseline and coded as HIV seropositive. These data on HIV infections are baseline data and do not represent HIV incidence in the sample, but reflect prevalence of HIV.

Analysis

Descriptive analyses and regression models were conducted using SAS 9.3. Frequencies were used to summarize categorical variables. Continuous variables were summarized using means and standard deviations. Differences in associations between HIV/STI behavioral risk variables, HIV infection, and the model predictor variables were assessed using logistic regression for binary outcomes and Poisson regression for number of sexual partners. Interval ratios were calculated for “number of sexual partners” and adjusted odds ratios for all other variables, adjusting for participant age, income, and education. Each regression included data on participants with complete data on the variables for that model. Therefore, the sample sizes for each regression varied somewhat depending on the particular outcome.

Structural equation modeling was performed using Mplus Version 7.0 statistical software (Hair, Black, Babin, Anderson, & Tatham, 2006; Muthén & Muthén, 2010) to examine associations between childhood religious affiliation (weighted for prevalence of members discouraging homosexuality), HIV/STI behavioral risk, and HIV infection while simultaneously examining the mediating roles of internalized homophobia, race/sexual identity incongruence, and interpersonal anxiety. This approach first specifies (1) a measurement model assessing how well selected observed indicators represent latent factors and (2) a structural model that identifies direct and indirect effects, corresponding standard errors, and assessing the overall model fit. The measurement model was assessed using confirmatory factor analysis to identify associations between latent constructs and the observed indicator variables. The latent construct factor loadings are presented in Table 2. Multiple indices of overall model fit were evaluated for both the measurement and structural models and the chi-square (χ²) statistic was used to assess differences between the hypothesized and the observed model. A low, nonsignificant chi-square is desirable as it indicates a low plausibility that there are significant differences between the hypothesized and observed models. Root mean square error of approximation of less than .08 is considered a reasonable approximate fit and low likelihood of model misspecification (Beckstead, 2002; Browne & Cudeck, 1993). The weighted root mean square residual (WRMR) was also used as an indicator of the overall fit of the model. A WRMR value of less than 1 is an indicator of good fit between the hypothesized and the observed structural models. The χ² was used along with other fit indices to make the final determination of overall model fit (H. W. Wilson & Widom, 2011). Standardized linear regression coefficients were reported for continuous factors in factor loadings and on predictor variables in path estimates. Standardized probit regression coefficients were reported for binary outcomes. Full information maximum likelihood estimation was used to handle missing data.

Table 2.

Variable Descriptive Statistics and Factor Loadings in Confirmatory Factor Analysis.

Variable	Range	N	% Yes	M (SD)	Factor loading
Had religious affiliation as a child	0-1	1,524	76.05	NA	NA
Was a member of a church or religious/spiritual institution	0-1	1,526	44.30	NA	NA
Prevalence of childhood religious affiliation discouraging homosexuality	0.14-0.79	1,553		0.52 (0.12)	NA
HIV/STI behavioral risk
CRAI in past 6 months	0-1	1,549	69.27		0.20*
CIAI in past 6 months	0-1	1,549	45.97		0.13*
Number of sex partners	0-3,007	1,549		13.51 (94.86)	0.08*
HIV infection	1-2	1,553	22.41		NA
Racial/sexual identity incongruence	1-5	1,490		2.20 (0.98)
Both my race and my sexuality are important to who I am as a man.a	1-5	1,520		2.25 (1.67)	0.56*
I cannot imagine a loving sexual relationship between two Black men.	1-5	1,510		2.14 (1.54)	0.11*
Black men who are attracted to other men can play an important role in Black families.a	1-5	1,510		2.74 (1.70)	0.65*
A Black man who is attracted to other men can still be a strong man.a	1-5	1,509		1.91 (1.48)	0.89*
Black men who are attracted to other men only make it more difficult for Black people in general.	1-5	1,510		1.98 (1.41)	0.21*
Internalized homophobia	1-5	1,467		2.23 (1.00)
In the past 90 days, I have stopped trying to be attracted to men.	1-5	1,505		2.01 (1.20)	0.70*
If someone offered to the chance to be completely heterosexual, I would accept the chance.	1-5	1,504		2.41 (1.33)	0.79*
I wish I weren’t attracted to men.	1-5	1,494		2.36 (1.28)	0.87*
I feel like being attracted to men is a shortcoming for me.	1-5	1,485		2.24 (1.22)	0.85*
I would like to get professional help change my orientation so that I am no longer attracted to men.	1-5	1,499		1.95 (1.13)	0.78*
I feel bad about being attracted to men because my community looks down on men who are attracted to other men.	1-5	1,500		2.20 (1.28)	0.77*
As a Black man, I try to act more masculine to hide my sexuality.	1-5	1,495		2.49 (1.38)	0.57*
Interpersonal anxiety	1-4	1,497		1.71 (0.74)
I felt that people disliked me	1-4	1,479		1.61 (0.90)	0.62*
I felt fearful	1-4	1,477		1.66 (0.94)	0.62*
I felt lonely	1-4	1,472		1.94 (1.06)	0.70*
People were unfriendly	1-4	1,467		1.63 (0.90)	0.81*

Note. STI = sexually transmitted infection; CIAI = condomless insertive anal intercourse; CRAI = condomless receptive anal intercourse; NA = not applicable.

^aReverse coded.

^*p < .05.

Results

Participant Demographics

This analysis included data from 1,553 Black MSM participants in HPTN 061. Variable descriptive statistics are presented in Table 2. Their mean age was 37.7 years (SD = 11.7; range = 18-68). Most participants (59.8%) reported that they earned less than $20,000 annually. The majority (55%) of participants had less than or equal to a high school education. More than two thirds (69%) of the sample reported CRAI within the past 6 months, while slightly less than half (46%) reported CIAI. The mean number of sex partners in the past 6 months was 13.5 (SD = 95.9). The mean and standard deviation for number of sex partners were driven by several extremely large values. The median number of sex partners was 5 and the interquartile range was from 3 to 9. Baseline STI prevalence was high for both chlamydia (6.4%) and gonorrhea (3.5%) infection at one or more anatomical site. Nearly one quarter of the sample had confirmed HIV infection at baseline. In total, 76% of the men in the sample (n = 1,159) reported that they had religious affiliations during their childhood. In general, fewer of those men reported that they had current religious affiliation (n = 676).

Association Between Model Predictors, HIV/STI Behavioral Risk, and HIV Infection

As reported in Table 3, childhood religious affiliation weighted for prevalence of beliefs discouraging homosexuality was significantly associated with increased number of sex partners. In the unadjusted model, childhood religious affiliation was associated with higher likelihood of HIV infection (odds ratio = 3.11; 95% confidence interval [1.07, 9.04]; p < .05); however, the association did not retain significance after adjusting for age, income, and education. Internalized homophobia was associated with increased likelihood of CIAI and number of sex partners and decreased likelihood of CRAI and HIV infection. Racial/sexual identity incongruence was associated with increased likelihood of CRAI and HIV infection but decreased number of sexual partners. Participants who reported anxiety had increased likelihood of engaging in CRAI and having more sexual partners.

Table 3.

Summary of Logistic and Poisson Regression Analyses Predicting HIV Risks and HIV Infection.

Predictor variables	HIV risk and HIV infection variables
	CRAIa (N = 1,549)	CIAI (N = 1,549)	Sex partners (N = 1,549)	HIV (N = 1,499)
	OR [95% CI]	OR [95% CI]	IR [95% CI]	OR [95%CI]
Childhood religious affiliation, weighted for prevalence discouraging homosexuality	1.44 [0.59, 3.54]	1.42 [0.54, 3.72]	4.31 [3.76, 4.94]**	2.68 [0.88, 8.16]
Internalized homophobia	0.74 [0.66, 0.82]**	1.30 [1.15, 1.46]**	1.10 [1.08, 1.12]**	0.72 [0.63, 0.83]**
Racial/sexual identity incongruence	1.38 [1.22, 1.55]**	1.08 [0.96, 1.22]	0.77 [0.76, 0.79]**	1.22 [1.06, 1.41]*
Interpersonal anxiety	1.19 [1.03, 1.37]*	1.16 [0.99, 1.36]	1.27 [1.25, 1.30]**	1.03 [0.86, 1.22]

Note. CIAI = condomless insertive anal intercourse; CRAI = condomless receptive anal intercourse; OR = odds ratio; IR = interval ratio.

^aAdjusted for age, income, and education.

^*p < .05. **p < .01.

Effects of Childhood Religious Affiliation on HIV/STI Behavioral Risk and HIV Infection

Measurement Model

Latent factors were formed for HIV/STI behavioral risk using three indices (CRAI, CIAI, and number of sex partners). For the mediators, latent factors were formed using items from the scales that measured those individual constructs (internalized homophobia, racial/sexual identity incongruence, and interpersonal anxiety). All the latent factors were allowed to freely correlate in a preliminary model using confirmatory factor analysis. Factor loadings on the latent variables were all statistically significant (Table 2). Factor loadings on the latent construct for outcome variable HIV/STI behavioral risk ranged from .08 to .20. Factor loadings on the latent constructs for mediator variables ranged from .57 to .92. This measurement model provided an acceptable fit, χ²(68) = 410.5, p = .05, root mean square error of approximation = 0.07, WRMR = 2.26.

Structural Equation Model

The hypothesized SEM examining the indirect effects of prevalence of childhood religious affiliation discouraging homosexuality on HIV/STI behavioral risk and HIV infection was tested (Figure 1). Correlations were estimated between prevalence of childhood religious affiliation discouraging homosexuality, the two outcome variables (HIV/STI behavioral risk and HIV infection), and the three mediator variables (racial/sexual identity incongruence, internalized homophobia, and interpersonal anxiety; Table 4). All correlations were significant, thereby supporting the further testing of these variables as mediators in the path model. The hypothesized SEM model shown in Figure 1 (Table 5; Model A) had a nonsignificant chi-square indicating that there was not a statistically significant difference between it and the data generated model. Other indices of model fit also suggested that the model fit the data well (Table 5). In this model, the path from interpersonal anxiety to HIV/STI behavioral risk was not statistically significant. After establishing relevant theoretical justifications, the path from interpersonal anxiety to HIV/STI behavioral risk was removed and the model was reestimated. This resulted in a trimmed model (Figure 2) that continued to demonstrate acceptable fit (Table 5; Model B), which was a better fit with the data. Because Model B was more parsimonious, it was selected as the final model.

Figure 1.

Theoretical model for path from childhood religious affiliation (weighted) to adulthood HIV infection.

Note. CIAI = condomless insertive anal intercourse; CRAI = condomless receptive anal intercourse; NUM SEX PTR = number of sexual partners. Latent constructs are show in circles, and observed variables are displayed in squares.

Table 4.

Correlation Between Study Variables.

Variable		1	2	3	4	5	6	7
1	Childhood religious affiliation, weighted for prevalence discouraging homosexuality	—	.03	.01	.06*	−.05	.05*	.01
2	HIV behavioral risk: CIAI		—	.01	−.06*	.02	.11**	.04
3	HIV behavioral risk: CRAI			—	.12**	.18**	−.16**	.07*
4	HIV infection				—	.03	−.10**	.01
5	Racial/sexual identity incongruence					—	−.29**	−.07*
6	Internalized homophobia						—	.27**
7	Interpersonal anxiety							—

Note. CIAI = condomless insertive anal intercourse; CRAI = condomless receptive anal intercourse.

^*p < .05. **p < .01.

Table 5.

Goodness of Fit Indicators for Prevalence of Childhood Religious Affiliation Discouraging Homosexuality to Adult HIV Risk.

Model	Discrepancy
	χ2	df	p	RMSEA	WRMR
Model A
All paths from percent discouraging homosexuality to HIV risk–related outcomes	410.5	68	.05	0.07	2.26
Model B
Parsimonious removal of theoretical path	479.5	70	.05	0.07	2.47

Note. χ² = minimum fit function test; df = degrees of freedom; RMSEA = root mean square standard error of approximation; WRMR = weighted root mean square residual.

Figure 2.

Structural model and standardized coefficients for paths from childhood religious affiliation (weighted) to adulthood HIV infection.

Note. CIAI = condomless insertive anal intercourse; CRAI = condomless receptive anal intercourse; NUM SEX PTR = number of sexual partners. Latent constructs are show in circles, and observed variables are displayed in squares.

^ap < .10. ^bp < .05. ^cp < .01.

The final model demonstrated that prevalence of childhood religious affiliation discouraging homosexuality was positively associated with HIV/STI behavioral risk and HIV infection through mediators that included increased internalized homophobia and increased racial/sexual identity incongruence in an adult sample of Black MSM. Internalized homophobia was associated with increased racial/sexual identity incongruence. Racial/sexual identity incongruence was associated with increased HIV/STI behavioral risk. HIV/STI behavioral risk was independently associated with HIV infection. The final model accounted for 1% of the variance in HIV/STI behavioral risk and 37.7% of the variance in HIV infection. Most paths were statistically significant with p < .05 and p < .01, while some paths coefficients were at the .1 level (Figure 2).

Discussion

This study tested a path model through which childhood exposure to religious environments with high prevalence of beliefs that homosexuality should be discouraged in society was associated with adulthood HIV/STI behavioral risk and HIV infection in a nonprobability sample of Black MSM. This is among the first reports in the literature to examine the associations between potentially nonaffirming childhood religious environmental exposures, internalized homophobia, racial/sexual identity incongruence, interpersonal anxiety, HIV/STI behavioral risk, and HIV infection. This is also one of the few reports that link religious affiliation to HIV/STI behavioral risk among a large sample of Black MSM. Although the impact of negative childhood experiences on adult sexual risk has been primarily studied with regard to sexual abuse (S. D. Fields, Malebranche, & Feist-Price, 2008; Lloyd & Operario, 2012; Phillips et al., 2014; Sweet & Welles, 2012), these results indicate that Black MSM may be vulnerable to internalized homophobia, racial/sexual identity incongruence, and HIV risk if exposed to religious environments with a high prevalence of members who discourage homosexuality.

Through SEM testing, the study hypothesis was largely supported. However, the hypothesized path proposing an association between interpersonal anxiety and HIV/STI risk was not supported by the data and was subsequently removed. Prior to removing the construct from the model, considerations were made of the reasons for why this path was not significant and whether it made theoretical sense to remove the path and then reestimate the model. Prior research has been inconclusive regarding whether negative affective states such as anxiety are associated with increased HIV/STI behavioral risk (Crepaz & Marks, 2001). More recent evidence suggests that interpersonal anxiety could lead to fewer sexual partners, less frequent sexual encounters, and by extension less frequent CRAI or CIAI (Kashdan et al., 2011; Ramrakha et al., 2013). Also, anxiety is commonly comorbid with depression, which is associated with decreased sexual behavior (Kashdan & Roberts, 2011; Laurent & Simons, 2009). Given that the items selected to measure the latent factor “interpersonal anxiety” were drawn from the CES-D, it may be more sensitive to depression-related anxiety.

The finding that childhood religious environments that discourage homosexuality were associated with increased numbers of sexual partners for adult MSM contributes to a growing evidence base for how experiences distal to the sexual encounter may nonetheless affect current HIV/STI risk-related behaviors. For example, using longitudinal data collected from MSM (N = 4,244) across six U.S. cities in a study evaluating the efficacy of a behavioral intervention for reducing HIV risk behavior and infection outcomes (Koblin, Chesney, Coates, & EXPLORE Study Team, 2004), Mimiaga et al. (2009) conducted a secondary analysis to test their hypothesis that a history of childhood sexual abuse was associated with condomless anal sex. Mimiaga et al. (2009) identified that there were increased odds for condomless sex among MSM with a history of childhood sexual abuse and that the association was not moderated by the effects of the EXPLORE HIV behavioral risk reduction intervention. This highlights the importance of early intervention to mitigate the impact of childhood exposures to negative events and environments on adult HIV/STI risk.

Early interventions could leverage mobile technology—popular with youth—to provide young Black MSM with access to smartphone-based religious/spiritual support that is developmentally appropriate, consistent with their faith tradition(s), and affirms their sexualities. Religious/spiritual support interventions would not decrease exposure to nonaffirming religious environments; however, such interventions can be designed to build psychosocial assets that promote resilience and positive youth development, both of which are identified as protective from negative social and health outcomes in adulthood (Furrow, King, & White, 2004; Gavin, Catalano, David-Ferdon, Gloppen, & Markham, 2010; Lerner, Lerner, & Benson, 2011; Romeo & Kelley, 2009; Schmid & Lopez, 2011). Research on religious/spiritual support interventions in health is primarily focused on helping to understand (Gomez-Castillo et al., 2015) and buffer (Balboni et al., 2013; Chochinov & Cann, 2005; Corwin, Wall, & Koopman, 2012) spiritual pain, anguish, and conflict during severe illness or end of life. Even though there is evidence that religious/spiritual support mitigates psychological distress (Maton, 1989; Sulmasy, 2006), no known studies have examined whether religious/spiritual support interventions were protective against negative psychosocial influencing factors (e.g., internalized homophobia and racial/sexual identity incongruence) that are antecedent to HIV risk for Black MSM. Future research is needed to develop and test mental health promotion and HIV risk reduction outcomes of asset-based religious/spiritual support interventions for young Black MSM.

In addition to documenting the associations between childhood religious environments that discourage homosexuality, adulthood HIV/STI behavioral risk and HIV infection, the findings from this study also provide information on pathways that potentially explain the mechanisms by which these outcomes occur. In this study, childhood religious environments that discourage homosexuality were directly associated with internalized homophobia. The association that was identified between internalized homophobia and HIV/STI behavioral risk was mediated by racial/sexual identity incongruence. These results contribute to a growing evidence base with regard to internalized homophobia, other related constructs, and HIV/STI behavioral risk (Barnes & Meyer, 2012; Foster et al., 2011; Jeffries et al., 2014; Lassiter, 2015; Pitt, 2009). Nonetheless, the literature is inconsistent with regard to what variables mediate the relationship between internalized homophobia and HIV/STI behavioral risk (Newcomb & Mustanski, 2011). The results from this study help address this current gap in the scientific literature by articulating at least one path (internalized homophobia → racial/sexual identity incongruence) by which childhood religious environments may affect adult HIV/STI behavioral risk and HIV infection.

Numerous studies highlight benefits of religious affiliation on primary and tertiary HIV prevention outcomes for various populations (Dowshen et al., 2011; Gillum & Holt, 2010a, 2010b; Ironson et al., 2011). A recent substantive review of 137 studies examined associations between religion (religious affiliation and/or religiosity) and sexual HIV risk (Shaw & El-Bassel, 2014). The results indicated that 57 of the studies in the review reported that particular religious affiliations were associated with sexual HIV risk. Over half (54%) of those studies reported that religious affiliation was associated with decreased HIV risk. The remaining 46% indicated that religious affiliation was associated with increased HIV risk. Although these studies provide useful information regarding the role of religious affiliation on HIV, a number of important gaps remain unaddressed. For example, as research has explored mechanisms by which religious affiliation protects against sexual risk, limited studies have examined potential mediators through which religious affiliations that discourage homosexuality may contribute to HIV behavioral risk for Black MSM (Shaw & El-Bassel, 2014). There is some evidence that that even in the midst of nonaffirming religious affiliations, Black MSM may sometimes be able to filter out messages that do not fully align with their own understandings of their sexualities and other aspects of their personhood (Pitt, 2010). These studies notwithstanding, there is not yet a clear body of evidence with regard to the effects of affiliation with nonaffirming religions on increased HIV risk for adult Black MSM.

Limitations

The results from the present study should be considered in light of several limitations. The analysis in this study was based on cross-sectional data collected at baseline; therefore, cause-and-effect relationships or temporal changes over a period of time cannot be assessed. Also, as secondary data were used to conduct the analysis, the study was limited to the use of variables that were included in the database. For example, although the HPTN 061 data largely fit with the theorized model, it is likely that there may be other unmeasured constructs in the model. As such, other HPTN 061 sample characteristics that could affect this analysis are also unknown, such as length of childhood religious affiliation, frequency of attendance to religious activities, and the degree that MSM were practicing HIV serosorting with their sexual partners (Wilton et al., 2015). Thus, these results should only be considered a partial depiction of the mediators by which childhood religious affiliations that discourage homosexuality are associated with adult HIV/STI behavioral risk and likelihood of HIV infection.

Although the study data were weighted using representative proportions of religions in the United States that discourage homosexuality, data on the actual individual-level childhood experiences that the men had in church are not available in the HPTN 061 data. Consequently, it is not known how belonging to religions that discourage homosexuality as measured at the national level was borne out in the interactions between the individual MSM and their local religious environments, or whether the men necessarily perceived any interaction or exposure to such environments as negative. Therefore, the possibility of spurious associations due to misclassification bias cannot be ruled out (Hatzenbuehler, Pachankis, & Wolff, 2012). There is little research that investigates the perceptions of childhood religious affiliations and its effect on Black MSM. Therefore, while novel and provocative, the results of this study should be validated by future studies that obtain data on the experiences and perceptions of Black MSM who were exposed to various religious affiliations in childhood.

Recent calls by public health scientists recommend that research move beyond the dominant focus on individual-level factors toward increased attention to social and structural factors that influence vulnerability to HIV infection (Lynam & Cowley, 2007; Millett et al., 2010; Millett, Jeffries, et al., 2012; Millett, Peterson, et al., 2012; Young, 2006). The integration of aggregate data from USRLS into the current study contributes to the advancement of scientific knowledge regarding how religious climates—as social-/structural-level phenomena—are associated with HIV risk for Black MSM. The results of this study also contribute to the broader scientific literature on associations of national- (MacInnis & Hodson, 2015) and state-level (Hatzenbuehler et al., 2012) religion data with sexual and sexuality-related behaviors. Future studies are needed that investigate both individual and social-/structural-level religious variables and their associations with HIV risk in Black MSM.

Conclusion

Early life experiences that Black MSM have with their religious affiliations can have negative impacts on adulthood HIV/STI behavioral risks and HIV infection. While most HIV prevention interventions target sexually active adult and adolescent MSM, more research is needed to develop upstream interventions that mitigate the potential harm of exposure to religious environments that may structure young Black MSM’s trajectory toward the development of internalized homophobia, racial/sexual identity incongruence, and HIV/STI behavioral risk. Future research should explore structural interventions that target environments, such as places of worship, and mobile technology–based religion/spiritual interventions that have the potential to facilitate the development of psychosocial assets for young Black MSM that can serve as protective factors against negative health outcomes in adulthood.