SEARCH

SEARCH BY CITATION

Keywords:

  • neighbourhood effects;
  • networks;
  • HIV;
  • sexual health;
  • gay men

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Sexual risk behaviours
  5. Neighbourhood health effects
  6. Network influences
  7. Neighbourhood influences on networks
  8. Hypotheses
  9. Methods
  10. Results
  11. Discussion
  12. Acknowledgements
  13. References

Gay neighbourhoods have historically served as vital places for gay socialising, and gay social networks are important sources of social support. Yet, few studies have examined the influence of these forms of community on sexual health. Informed by theoretical frameworks on neighbourhoods and networks, we employ multi-level modelling to test hypotheses concerning whether gay neighbourhoods and social network factors are associated with five sexual risk behaviours: receptive and insertive unprotected anal intercourse (UAI), barebacking identity, recent internet use for finding sexual partners, and ‘Party and Play’ (PnP). Our analyses of a community-based sample of gay men in New York City reveal little evidence for the direct effect of gay enclaves on sexual risk with the exception of PnP, which was more likely among gay enclave residents. Having a network composed predominantly of other gay men was associated with insertive UAI, PnP, and internet use for meeting sexual partners. This network type also mediated the association between gay neighbourhoods and higher odds of insertive UAI as well as PnP. Our findings highlight the sexual health implications of two important facets of gay community and, in doing so, indicate the need to better contextualise the sexual health risks faced by gay men.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Sexual risk behaviours
  5. Neighbourhood health effects
  6. Network influences
  7. Neighbourhood influences on networks
  8. Hypotheses
  9. Methods
  10. Results
  11. Discussion
  12. Acknowledgements
  13. References

Rising HIV incidence rates among men who have sex with men (MSM) have generated significant concern. MSM account for over half (53%) of all new US HIV infections (CDC 2010); the rate of new diagnoses among MSM is over 44 times that of other men. Incident HIV infections have declined among heterosexuals and injection drug users, but have steadily risen among MSM (Center for Disease Control and Prevention (CDC) 2010). Coupled with the rising HIV incidence are increases in unprotected anal intercourse (UAI; Zablotska et al. 2009), and rates of sexually transmitted diseases (STD; Van de Marcus et al. 2006) indicating a pathway for changes in HIV incidence.

Though continued challenges remain, scholars have made important strides in expanding examinations of risk beyond the individual level. Recently, Frye and colleagues (2010) examined how gay residential concentration influences consistent condom use. Our paper extends such important work by examining neighbourhood and social network influences on sexual risk taking. Specifically, we assess whether gay enclave residence and gay social network composition influence sexual risk behaviours. In doing so, we consider the role of these important facets of community in the sexual health of gay men.

Sexual risk behaviours

  1. Top of page
  2. Abstract
  3. Introduction
  4. Sexual risk behaviours
  5. Neighbourhood health effects
  6. Network influences
  7. Neighbourhood influences on networks
  8. Hypotheses
  9. Methods
  10. Results
  11. Discussion
  12. Acknowledgements
  13. References

Unprotected anal intercourse

UAI heightens the odds of HIV transmission approximately five-fold relative to an act of vaginal intercourse (Varghese et al. 2002). Beyond the act of UAI, a greater number of sex partners heightens the risk for HIV transmission (Read et al. 2007). Given these risks, it is of concern that research indicates that MSM have become less likely to refrain from anal sex, decreased consistent condom use, and increased their number of discordant partners (Osmond et al. 2007).

The HIV epidemic enabled the introduction of risk reduction practices such as serosorting (Suarez and Miller 2001), negotiated safety (Davidovich et al. 2000), and strategic positioning (Van de Ven et al. 2002). Yet, these practices lower, but do not eliminate, the risk of HIV transmission due to the prevalence of undiagnosed infections (Wilson et al. 2010). Almost half of young MSM report no HIV testing within the previous year and may presume non-infection during such window periods (MacKellar et al. 2006). Additionally, recent data indicate MSM are increasingly being infected by primary partners (Sullivan et al. 2009). Furthermore, reducing community viral load levels remains critical to ensure that risk reduction practices reach maximum efficacy (Das et al. 2010). As such, while risk reduction methods provide some protection, they remain incompletely effective.

Related sexual risk behaviours

Although UAI is the primary pathway of HIV transmission among MSM, additional factors facilitate unprotected sex and corresponding HIV transmission. As such, we consider three additional aspects of sexual sociability among gay men that may elevate HIV risk: barebacking, party and play, and sexual internet use.

Acts of intentional UAI, known as ‘barebacking’, have emerged as a social phenomenon among MSM. Wolitski (2005: 14) defines barebacking as ‘intentional anal sex without a condom except when practiced by HIV negative primary partners who maintain a mutually monogamous or negotiated safety relationship with each other’. ‘Barebacker’ has now become a label or an identity within the gay community (Carballo-Dieguez et al. 2009). Men who identify as barebackers more often engage in HIV risk behaviours (Parsons and Bimbi 2007).

Beyond barebacking, the literature highlights linkages between drug use and sexual risk taking among MSM (Ross and Williams 2001). While this link is well established, the phenomenon of ‘party and play’ adds an additional layer of complexity to these practices. ‘Party and play’, PnP colloquially, involves the use of drugs to enhance sexual experiences. Some men seek sexual partners on the basis of whether or not they practise PnP (Halkitis and Parsons 2002). While not all drug use leads to risky sexual behaviour – or any sexual behaviour – some gay men associate drug use with sexuality via institutions in which the two intersect (McKirnan and Peterson 1989). Accordingly, these risk behaviours have become paired for those who seek PnP, which heightens HIV risk by enabling UAI.

Some MSM use the internet for sexual activities, including finding sex partners. The use of the internet for sex remains a concern because MSM who meet sexual partners online have a higher number of sex partners (Evans et al. 2007), are more likely to practise UAI (Berry et al. 2009), and to be at increased risk for STDs, including HIV (Evans et al. 2007). While the internet does not cause sexual risk taking per se, it facilitates access to a greater number of partners as well as access to risk subcultures.

The risk behaviours described above and their potential connections to rising HIV incidence rates suggest that analyses of meso-level social factors may enhance our understandings of these risk pathways. Drawing primarily from the sociological literatures on neighbourhoods and social networks, our examination of the risk behaviours described above is guided by theoretical perspectives on community. These allow us to test hypotheses concerning the role of gay neighbourhoods and social networks in the sexual lives of urban gay men.

Neighbourhood health effects

  1. Top of page
  2. Abstract
  3. Introduction
  4. Sexual risk behaviours
  5. Neighbourhood health effects
  6. Network influences
  7. Neighbourhood influences on networks
  8. Hypotheses
  9. Methods
  10. Results
  11. Discussion
  12. Acknowledgements
  13. References

Scholars have investigated ‘neighbourhood health effects’ to better understand how residential contexts influence health independently of residents’ individual characteristics (Robert 1999). Many studies have focused on the health implications of ethnic neighbourhoods (e.g. LaVeist and Wallace 2000, Frank et al. 2007). Few studies, however, have focused on the health implications of gay neighbourhoods. An exception is Frye and colleagues’ (2010) important study of neighbourhood effects – gay residential concentration, socioeconomic profile, ethnic heterogeneity, concentrated poverty, and residential stability – on consistent condom use among MSM. Notably, only gay residential concentration influenced consistent condom use (Frye et al. 2010). We extend this work by examining community-level influences on sexual risk behaviours. Importantly, our analyses differ by our focus on visible forms of community via gay institutions, which publicly distinguish gay enclaves from other locations with concentrations of gay residents and provide forums for social interactions that define gay life (Levine 1979).

The role of neighbourhoods in gay community

Gay neighbourhoods have long played a significant role in the lives of gay men (Kaiser 1997). Importantly, they may be considered as sites of resources and sanctuaries for the marginalised. In theorising how gay neighbourhoods matter for health, we must consider neighbourhood institutions that facilitate social behaviours by operating as ‘launch pads’ for gay life, that is, serving as places to meet other men, disseminate health information, and encounter aspects of gay socialisation (Green 2003). We contend that gay neighbourhoods potentially serve as important institutions themselves or, at a minimum, operate as critical locations for the presence of such institutions. Subcultural theory is useful to guide our consideration of such processes for sexual risk behaviours.

Subcultural theory

Subcultural theory posits that the concentration of individuals sharing particular lifestyles enables the development of subcultures that serve as meaningful community environments (Fischer 1984). Each subculture provides ‘a set of modal beliefs, values, norms, and customs associated with a relatively distinct social subsystem (a set of interpersonal networks and institutions)’ for those connected to them (Fischer 1975: 1323). Consequently, institutions emerge out of sustained interactions and influence both individual and collective behaviour and the maintenance of community (Fischer 1975).

From the perspective of subcultural theory, increasing concentrations of gay residents in particular locales contributed to the development of gay institutions and a local community culture of tolerance and openness to gay social practices – an environment that can positively contribute to the quality of life of its gay residents. This element of ‘safe space’ enables gay men to visibly display their identity, avoid having to justify themselves to others, and develop relationships without fear (Levay and Nonas 1995). Additionally, by living in such communities, gay men create solidarity and collective experience, which allows for the development of a shared history, support, friendship and trust – all of which have implications for the creation and maintenance of community norms.

Neighbourhood mechanisms

Based upon this theoretical perspective, we can consider two contrasting neighbourhood hypotheses. First, we can conjecture that, via subcultural norms, gay neighbourhoods foster a local culture that promotes healthy behaviours. For sexual risk taking, the ‘safe space’ culture of a gay neighbourhood may buffer the impact of anomie of gay residents, who elsewhere may engage in risky sex to cope with stigmatisation. Additionally, gay neighbourhoods are frequent sites for HIV education efforts and sexual health initiatives. As such, these neighbourhoods provide an environment facilitating efforts to discourage sexual risk behaviours. For the purpose of parsimony, we call this the ‘gay neighbourhood as health promoter hypothesis’.

A second consideration is that the institutionalisation of subcultural norms and practices within the neighbourhood may promote unhealthy activities (Portes 1998, Carpiano et al. 2011). Such a subculture may enable sexual risk behaviours that cohere with potentially harmful norms, particularly for those isolated from countervailing norms. Thus, it is important to consider that men who reside in gay enclaves may be at greater risk of immersing themselves in a subculture that promotes risk taking. In addition, the concentration of such individuals in a neighbourhood may enable a problematic pooling of high risk partners, and thus heighten exposure to poor sexual health outcomes for all in the neighbourhood through a concentration of risk. We term such potential harmful neighbourhood effects as the ‘gay neighbourhood risk subculture hypothesis’.

Network influences

  1. Top of page
  2. Abstract
  3. Introduction
  4. Sexual risk behaviours
  5. Neighbourhood health effects
  6. Network influences
  7. Neighbourhood influences on networks
  8. Hypotheses
  9. Methods
  10. Results
  11. Discussion
  12. Acknowledgements
  13. References

Networks have been shown to be key influences on health behaviours (Luke and Harris 2007). Accordingly, understandings of community influences on health must also account for the networks of social relations in which gay men are embedded. Although gay enclaves remain important social spaces for urban gay men, the experience of gay community may be more diffuse and negate the role of neighbourhood community. Such forms of community involve interactions that extend beyond neighbourhood relations and include friends and others with shared attitudes and desires (Woolwine 2000). Scholars have shown that friendships are an important part of social development for gay men, particularly for those who experience homophobia (Banks 2003). This conception of community merits consideration of the social networks of urban gay men – both in terms of intensity of interaction with other gay men and social network diversity.

Our consideration of network influences on sexual risk draws upon Granovetter’s (1985: 487) theory of embeddedness and action; he argues that rather than conceptualising actors as either individuals isolated from social context or ‘oversocialised’ automatons determined by structural factors, scholars must recognise that ‘attempts at purposive action are instead embedded in concrete, ongoing systems of social relations’. Such theoretical concerns situate sexual risk behaviours within the context of network socialisation, forcing the need to empirically examine characteristics of individuals’ networks. Yet, as O’Donnell and colleagues (2002: 459) argue, such embeddedness can be a ‘two-edged sword’: a gay network may reinforce norms about healthy behaviours, but may also introduce men to relationships that diminish their ability to negotiate healthy behaviours. We must consider how networks of social relations unfold, given that patterns of social interaction are neither random nor equal among all network members. As such, we examine not simply if one has gay social ties, but instead consider (a) sexual diversity of social ties and (b) intensity of socialising with other gay men.

Network diversity

In considering network sexual diversity, we conjecture that a more restricted range of ties poses greater sexual risk. Homophily may limit ‘people’s social worlds in a way that has powerful implications for the information they receive, the attitudes they form, and the interactions they experience’ (McPherson et al. 2001: 415). We consider that having predominantly gay social ties means exposure to a more limited range of norms than those with diverse networks. While a homophilous network is not necessarily a bad thing, it may have implications for gay men; lesbians and heterosexuals typically have fewer casual sex partners than gay men and normative tendencies towards monogamy (Patterson 2000). As such, more diverse networks may inhibit sexual risk taking. Considering these perspectives, we posit that gay men with homophilous networks – networks composed predominantly of gay men, termed hereafter ‘gay-centric networks’– are associated with higher odds of sexual risk behaviours.

Intensity of socialising

With respect to intensity of socialising, the more intensely one is embedded in a community, the more likely one may engage in certain behaviours normative to it, regardless of the person’s overall proportion of social ties that are part of that subculture. Sexual risk may occur as gay men conform to risk subcultures surrounding them. Hence, we hypothesise that intense socialising will present greater exposure to sexual risk behaviours.

Neighbourhood influences on networks

  1. Top of page
  2. Abstract
  3. Introduction
  4. Sexual risk behaviours
  5. Neighbourhood health effects
  6. Network influences
  7. Neighbourhood influences on networks
  8. Hypotheses
  9. Methods
  10. Results
  11. Discussion
  12. Acknowledgements
  13. References

If gay neighbourhoods, as institutions, are characterised by features specific to gay life, then such neighbourhoods should matter for the organisation of residents’ social networks. We evaluate this argument by examining network factors as mediators of the neighbourhood-risk associations. Our analyses will give rise to evidence that may support one of two explanations. If evidence for mediation is found, it indicates that gay neighbourhoods remain important influences on the sexual sociability of gay men. If no evidence for mediation is present, but network factors are associated with sexual risk, such findings are consistent with the notion that risk is influenced more by network communities than neighbourhood communities.

Hypotheses

  1. Top of page
  2. Abstract
  3. Introduction
  4. Sexual risk behaviours
  5. Neighbourhood health effects
  6. Network influences
  7. Neighbourhood influences on networks
  8. Hypotheses
  9. Methods
  10. Results
  11. Discussion
  12. Acknowledgements
  13. References

We have posited several hypotheses for the influence of neighbourhoods and networks – two important forms of community – on sexual risk-taking among gay men. Our study hypotheses are as follows:

Neighbourhood health promoter: Gay enclaves will be associated with lower odds of sexual risk behaviours.

Neighbourhood risk subculture: Gay enclaves will be associated with higher odds of sexual risk behaviours.

Network diversity: Men with a gay-centric network will have higher odds of sexual risk behaviours.

Intensity of socialising: Increased socialising with gay men will be associated with higher odds of sexual risk behaviours.

Homophily mediation: Possession of a gay-centric network will mediate the association between gay enclaves and sexual risk behaviours.

Intense socialising mediation: Increased socialising with gay men will mediate the association between gay enclaves and sexual risk behaviours.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Sexual risk behaviours
  5. Neighbourhood health effects
  6. Network influences
  7. Neighbourhood influences on networks
  8. Hypotheses
  9. Methods
  10. Results
  11. Discussion
  12. Acknowledgements
  13. References

Dataset

We test hypotheses using data from the 2005 Sex and Love survey, a repeated cross-sectional survey that examines aspects of gay, lesbian and bisexual (GLB) life. The data were collected at expo events in New York City (NYC). Though the expos, which were not gay pride events, required paid admission, discount passes were widely available and free passes were provided to various organisations. Intercept survey research has been used in numerous studies, including ones focused on GLB persons (Kalichman et al. 2001), and has been shown to provide valid and reliable data. Although the data were collected for purposes other than this paper, the breadth of information and sample size make it a novel dataset for testing our hypotheses.

The survey took approximately 15 minutes and contained items concerning perceptions of GLB life, the GLB community and health-related topics. Individuals were invited to complete the self-administered survey by a research assistant and encouraged to complete it in a nearby area to enhance privacy. They received a movie pass once the survey was returned. The response rate was 83 per cent.

Sample

We limited our analyses to respondents based on gay identity, geographic locale, and HIV status. First, we excluded men who identified as ‘bisexual’ or ‘straight’. Second, we limited our analyses on respondents’ residential locale. Respondents in our analytic sample lived either within NYC or New Jersey areas served by the PATH train routes, which allow frequent, rapid, and inexpensive passage to NYC comparable to the subway. This access is unique compared to other suburbs. Third, we limited our analysis to men who reported HIV negative or unknown HIV status, since HIV status may influence sexual behaviours; our sample of HIV positive men was too small to conduct stratified analyses. From the 710 respondents who fit these criteria, we created five analytic samples (one for each outcome), which ranged in size from 584 (82.3%) to 661 (93.1%) respondents who had non-missing data.

Measures

Dependent variables  We computed dependent variables for unprotected anal intercourse (UAI), barebacking, PnP, and sexual internet use – each coded dichotomously (0 = no and 1 = yes). UAI within the previous three months was examined for any unprotected receptive (RAI) and any unprotected insertive anal intercourse (IAI). Barebacker identity, was based on a respondent’s (yes/no) answer to the statement ‘I consider myself a barebacker’. Party and play (PnP), was derived from items asking respondents if they had used any of the following drugs during sexual encounters in the past three months: cocaine, methamphetamine, ecstasy, GHB and Ketamine. Met a sexual partner via the internet was based on an item asking respondents the total number of times they engaged in such behaviour during the past three months.

Neighbourhood-Level Variables  To test our hypotheses regarding sexual risk and community factors, we operationalised these constructs using neighbourhood- and respondent-level variables. The survey obtained information on respondents’ zip codes, thus our neighbourhood measures are assessed at the zip code level level. The size of zip codes in this region allows them to be acceptable approximations of local neighbourhood areas. With gay neighbourhoods we assessed gay enclaves using two data sources: local knowledge obtained via ethnographic social mapping as well as census data (US Census Bureau, 2011).

Gay enclave is a dichotomous item coded 1 if the zip code was part of a gay enclave and coded 0 if not. As described by Levine (1979), gay enclaves have defining characteristics: a substantially gay population, gay institutions, and conspicuous gay subcultures. We first measured gay residential concentration using census data on the percentage of male same-sex partnered households in a neighbourhood. However, numerous gay men live in non-gay neighbourhoods, but such locations do not have the visible gay institutions that gay enclaves do. Given our concern with visible gay life to identify gay enclaves, the development of ‘social maps’ included the identification of gay institutions throughout these neighbourhoods. ‘Social mapping’ can delineate patterns of socialising and thus enable the development of a ‘social map’ through the use of participant-observation (Clatts et al. 1995). We used participant-observation to document clusters of gay institutions to form social maps of visible gay life. The ethnographic social maps were corroborated by a content analysis of gay institutions documented in two city-wide gay publications. The presence of multiple gay institutions was necessary, and thus an isolated gay institution was insufficient to indicate a gay enclave. The presence of both residential concentration and gay institution clusters were used to code for gay enclaves (Carpiano et al. 2011). The point biserial correlation between gay enclave and solely male same-sex partnered households (rpb = 0.49) was nearly identical across analytic samples, and indicated neighbourhood distinctions. Across these five analytic samples, respondents were nested within 125–132 neighbourhoods, averaging 4.7 to 5.0 men per neighbourhood depending on the specific sample.

With neighbourhood-level confounders we included measures of neighbourhood concentrated affluence/disadvantage and residential stability to control for other neighbourhood-level processes. To assess concentrations of poverty and affluence, we used the index of concentration at the extremes (ICE) (Casciano and Massey 2008), a census-derived measure calculated as: [(number of affluent households) – (number of disadvantaged households)]/total number of households. The ICE can range from –1 (all households below the poverty level) to +1 (all household incomes > $75,000), with 0.0 indicating a balance of affluent and disadvantaged households. The ICE is advantageous since it allows the use of a single indicator for both concentrated poverty and affluence – two neighbourhood socioeconomic constructs that have differing implications, but are often highly correlated, thus problematising the assessment of their influences (Carpiano et al. 2009). We assessed residential stability through a single census measure: the percentage of residents who lived in the neighbourhood for at least ten years.

Respondent-level Variables  With social network variables we assessed two aspects of networks. Socialises with gay men was measured with a single item asking respondents the degree to which they socialise with gay men and used a four-point scale ranging from 0 = never to 3 = always.

Network diversity was calculated using a composite item created from four variables similar to the above variable, but included socialising with lesbians, bisexuals, and heterosexuals in addition to gay men. The response scales were collapsed into the binary categories of 0 = never/sometimes and 1 = frequently/always. Using these variables, we coded network composition by whether one had a gay-centric network (1, socialising frequently/always with only gay men) or a more diverse network (0, socialising frequently/always with other groups as well). Even though our two variables use a similar measure, their point biserial correlation, was no larger than 0.125, indicating that they capture two different aspects of social networks: intensity and diversity of network ties.

Our models also included control variables for attachment to the gay community and a variety of demographic factors: individual-level confounders. Given that our sample was recruited from GLB expos, which could present biases by including respondents more highly involved in the gay community, we controlled for attachment to the gay community using three items for the degree to which they have: (a) a ‘strong sense of belonging to the GLB community’; (b) ‘a lot of pride in the GLB community’; and (c) ‘a strong sense of attachment towards the GLB community’. The Cronbach’s alpha across each analytic sample was high (≥0.90). Age was coded as three categorical variables: 18–30 (referent), 31–40, and 41 or older. Education was coded as a binary variable: Less than a Bachelor’s degree = 0; Bachelor’s degree or higher = 1. Income was a binary variable coded less than $80,000 = 0 and $80,000 or more = 1. Race/ethnicity was measured using dummy variables for Black, Latino and other race, with White as the referent category. We also controlled for relationship status (0 = single and 1 = partnered) and data collection site. Our decision to include respondents who reported being partnered was due to several factors. First, some respondents’ current relationship status did not perfectly correspond with this prior three month time period, hence there was potential for risk prior to the relationship. Second, the living situations among this subset were quite varied with respect to relationship length, co-habitation, and sex outside of the relationship. Descriptive statistics for all variables are found in Table 1.

Table 1. Descriptive statistics of the analytic sample
 Unprotected anal intercourse in the last 3 monthsBarebacker identityParty and play in the last 3 monthsMet a man via the internet for sex in the last 3 months
Any insertiveAny receptive
  1. SD = standard deviation

Neighbourhood-level variables, sample size125125132131129
Gay enclave6.406.406.066.116.20
Index of concentration at the extremes, mean (SD)0.04 (0.28)0.04 (0.28)0.04 (0.28)0.04 (0.29)0.05 (0.27)
Residential stability, mean (SD)53.11 (10.98)53.11 (10.98)53.31 (10.94)53.21 (10.91)53.27 (11.03)
Individual-level variables, sample size585584661641621
Prevalence of dependent variable24.2721.7510.8910.7632.53
Gay-centric network9.579.599.539.369.34
Socialises with gay men, mean (SD)2.44 (0.62)2.44 (0.62)2.44 (0.63)2.44 (0.63)2.44 (0.61)
Gay community attachment, mean (SD)3.20 (0.75)3.20 (0.75)3.21 (0.76)3.21 (0.75)3.19 (0.75)
Age
 18–3031.1131.5131.6231.5131.08
 31–4035.9035.7935.8536.0436.55
 41 and older32.9932.7132.5332.4532.37
Race/ethnicity
 White62.5662.6760.9761.1562.48
 Black10.2610.1010.4410.45 9.34
 Latino15.3815.5815.8915.9115.14
 Other11.7911.6412.7112.4813.04
Bachelor’s degree or higher 71.7972.0972.0171.7672.79
Income $80,000 or higher19.8320.0320.8820.4421.42
Partnered/non-single45.8145.7242.0642.1242.67
Data collection site
 Gay erotic expo44.4444.5244.6344.7744.61
 Gay life expo55.5655.4855.3755.2355.39

Analyses

Based on the nested structure of the data and our multilevel theoretical rationale (Luke 2004), we conducted analyses using multilevel logistic and linear regression models in Stata 11. We specified a two-level random intercept model for each outcome, with respondent-level network and demographic variables specified at level 1 while neighbourhood-level variables are specified at level 2. These models, common in neighbourhood health effects research, include fixed effects for all parameters and a single random effect for the intercept, which is specified to vary across neighbourhoods (Luke 2004). For each logistic regression model, we report the odds ratios and 95% confidence intervals; for linear regression models, we report unstandardised slope coefficients and standard errors.

We tested mediation hypotheses using a three step process (Baron and Kenny 1986, Kenny 2010). In step one, the direct association between gay enclave and sexual risk is assessed. In step two, we introduce network factors to determine the degree to which this relationship changes. In step three, we assess the association between network factors and gay enclave. If mediation is present, the association between gay enclave and the outcome should be reduced in magnitude when a network variable is entered into the model, and gay enclave should be associated with that network variable. Interpretation of mediation is based on the magnitude of observed coefficients in each step and not simply statistical significance (Kenny 2010). A mediating path may be implied if step two and step three are successfully executed (Kenny 2010). We tested for the statistical significance of mediating effects using the PRODCLIN2 program for estimating appropriate confidence intervals (Fritz and MacKinnon 2007, MacKinnon et al. 2007) as well as procedures outlined by Herr (2010) and Kenny (2010) for mediation with binary logistic models.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Sexual risk behaviours
  5. Neighbourhood health effects
  6. Network influences
  7. Neighbourhood influences on networks
  8. Hypotheses
  9. Methods
  10. Results
  11. Discussion
  12. Acknowledgements
  13. References

Tables 2 and 3 detail results from our models. For each outcome, two models are presented. Model one includes only neighbourhood factors, while model two adds network factors. Both models adjust for all neighbourhood- and individual-level control factors.

Table 2. Odds ratios (95% confidence intervals) for recent unprotected anal intercourse
 Unprotected anal intercourse
InsertiveReceptive
1212
  1. ≤ .10; **≤ .01. All models adjusted for neighbourhood- and individual-level controls.

Neighbourhood factors
 Gay Enclave1.42 (0.82, 2.45)1.25 (0.71, 2.19)1.45 (0.83, 2.52)1.44 (0.82, 2.53)
Social network factors
 Gay-centric Network  3.47** (1.64, 7.32) 2.11 (0.97, 4.59)
 Socialises with Gay men 1.08 (0.73, 1.59) 0.73 (0.49, 1.09)
 Intercept Variance Component0.000.000.000.00
Table 3. Odds ratios (95% confidence intervals) for barebacker identity, party and play, and internet use for sex
 Barebacker identityIn Last 3 Months
Party and playMet a man via the internet for sex
121212
  1. *≤ .05; **≤ .01. All models adjusted for neighbourhood- and individual-level controls.

Neighbourhood factors
 Gay enclave0.86 (0.45, 1.67)0.82 (0.42, 1.60)2.50** (1.37, 4.57)2.08* (1.11, 3.90)0.91 (0.54, 1.54)0.85 (0.51, 1.42)
Social network factors
 Gay-centric network  1.77 (0.81, 3.86) 2.50* (1.22, 5.16) 2.45** (1.35, 4.42)
 Socialises with gay men 0.89 (0.59, 1.35) 2.25** (1.32, 3.84) 1.27 (0.93, 1.75)
 Intercept variance component0.000.000.000.000.100.08

Across most outcomes, gay enclave residence was not directly associated with sexual risk behaviours. Enclave residence, however, was associated with PnP. In model 1, it is associated with 2.5 times higher odds of engaging in PnP. In model 2, this estimate was reduced to 2.08 when network factors were controlled, but remained statistically significant.

With respect to social network factors, having a gay-centric network was associated with engaging in three outcomes. Specifically, it was associated with higher odds of engaging in unprotected IAI (OR = 3.47), engaging in PnP (OR = 2.50), and using the internet to find sexual partners (OR = 2.45). For unprotected RAI, having a gay-centric network trended toward a similar relationship, but the magnitude of the association did not achieve statistical significance. Socialises with gay men was only associated with PnP, indicating a greater than two-fold increase in odds for each unit increase in socialising intensity. Barebacking identity was related neither to enclave residence nor network factors.

We tested for mediation for three outcomes (unprotected IAI, PNP, and internet for sex) because (a) one or both network factors are associated with the outcome and (b) the magnitude of the estimate for gay enclave decreases from model 1 to model 2, per procedures outlined by Kenny (2010). Table 4 shows results for analyses whereby gay-centric network and socialises with gay men are regressed on neighbourhood and network factors. For each network factor, model 1 assesses the association with neighbourhood factors and model 2 introduces network factors.

Table 4. Predictors of social network factors
 Unprotected insertive anal intercourse
Gay-centric network1Socialises with gay men2
1212
Neighbourhood factors
 Gay enclave2.22* (1.20, 4.11)2.13* (1.14, 3.97)0.075 (0.075)0.066 (0.075)
Social network factors
 Gay-centric network   0.159 (0.084)
 Socialises with gay men 1.78* (1.01, 3.15)  
 Intercept variance component0.000.000.010.01
 Party and play
Gay-centric network1Socialises with gay men2
1212
Neighbourhood factors
 Gay enclave2.18* (1.08, 4.41)2.09* (1.02, 4.29)0.085 (0.081)0.075 (0.081)
Social network factors
 Gay-centric Network   0.192* (0.083)
 Socialises with gay men 1.99* (1.14, 3.47)  
 Intercept variance component0.040.040.02*0.02*
 Met a man via the internet for sex
Gay-centric network1Socialises with gay men2
1212
  1. ≤ .10; *≤ .05.

  2. 1 Odds Ratios (95% CI)

  3. 2 Unstandardised Slope Coefficients (Standard Errors)

  4. All models adjusted for neighbourhood- and individual-level controls.

Neighbourhood factors
 Gay enclave2.04 (0.99, 4.23)1.93 (0.92, 4.05)0.085 (0.072)0.076 (0.073)
Social network factors
 Gay-centric Network   0.185* (0.083)
 Socialises with gay men 1.93* (1.10, 3.40)  
 Intercept variance component0.030.030.010.01

For unprotected IAI, gay enclave is associated with having a gay-centric network, but not increased socialising with gay men. Given that gay-centric network was found to have a direct association with IAI and it reduces the magnitude of the estimate for gay enclave, this collective evidence indicates partial mediation of the association between gay enclave residence and IAI. The estimated confidence interval of this indirect effect indicated that this mediating effect is statistically significant.

For PnP, gay enclave is associated with gay-centric network, but not socialising with gay men. When these results are considered with respect to the results in Table 3, these findings indicate that the association between gay enclaves and PnP is partially mediated via gay-centric networks. The confidence interval of this indirect effect indicated that this mediating effect is statistically significant.

In examining mediation with respect to internet use for sex, gay enclaves are neither associated with gay-centric network nor socialising with gay men. Thus, these results indicate that the odds of internet use to find sex partners are attributable solely to network factors.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Sexual risk behaviours
  5. Neighbourhood health effects
  6. Network influences
  7. Neighbourhood influences on networks
  8. Hypotheses
  9. Methods
  10. Results
  11. Discussion
  12. Acknowledgements
  13. References

Motivated to contextualise the sexual risks faced by gay men, we investigated the roles of two important sources of gay community – neighbourhood enclaves and social networks – in sexual risk behaviours. Our analyses reveal an interesting pattern of findings with respect to the roles of neighbourhoods and networks. Overall, while some role was found for neighbourhood conditions, the majority of effects are attributable to social networks.

Unprotected insertive anal intercourse: neighbourhood influences networks

The odds of engaging in unprotected IAI or RAI were not found to be higher among gay enclave residents. Gay neighbourhoods, however, were indirectly associated with increased odds of unprotected IAI via having a gay-centric network – a factor that is independently associated with unprotected IAI, but not unprotected RAI – providing evidence for the homophily mediation hypothesis for this outcome. These results may be due to the social organisation of strategic positioning practices (Van de Ven et al. 2002). In other words, gay neighbourhoods may be indirectly organising risk reduction practices among HIV-negative men because of a subcultural understanding of a lower risk of IAI for HIV transmission.

Our observed mediation indicates that residential location still plays a role in shaping risk by influencing social networks. This finding is important to consider in light of the reported decline of gay neighbourhoods and their contested importance in the experience of gay community (Rosser et al. 2008). Whereas prior studies have highlighted nightclubs (Green 2003), and gay enclaves generally (Carpiano et al. 2011), as important institutions for facilitating drug use, our indirect findings suggest that gay neighbourhoods (via social networks) also shape sexual behaviours by serving as ‘launch pads’ for gay life (Green 2003: 440) – aspects of which, similar to any other subculture or collective lifestyle (Frohlich et al. 2001, Cockerham 2007), may serve to promote or harm health.

Barebacker identity

Identifying as a barebacker was not significantly associated with neighbourhood or network forms of gay community. Thus, no evidence supported any hypotheses for this outcome. It may be possible that there are more specific drivers of this subcultural identity within the gay community. Yet, we find it surprising that neither network variable was significantly associated with this outcome as we would expect such a subculture to be network-based.

Party and play: neighbourhoods and networks

Gay enclaves were associated both directly and indirectly with higher odds of engaging in PnP. Having a gay-centric network was directly associated with PnP and partially mediated the gay enclave-PnP relationship. In addition to mediation, these results for PnP lend support for our ‘neighbourhood risk subculture’ hypothesis as well as both network hypotheses. These findings are interesting given previous findings of the role of gay enclaves in organising ecstasy and crystal meth use among gay men (Carpiano et al. 2011).

With regard to social networks, PnP is multifaceted; motivated not simply by the proportion of one’s network that are gay, but also by how intensely one socialises with gay men. Men who are more intensely associated with only gay men may have greater access to such subcultural participation and may participate in PnP as part of a risk subculture within the community without the influence of countervailing norms from more diverse networks.

Internet use for meeting sexual partners: the role of networks

For internet use for finding sexual partners, network factors are particularly important; gay enclaves play no organising role. Beyond the notable finding of gay enclave residents not being more likely to use the internet to find sex partners, it is interesting to note that they are not less likely either. We might expect that men living in more diffuse locations would find the internet more critical for such purposes than those who reside amidst a concentration of gay venues. Yet, there remain no differences by neighbourhood, highlighting the extent to which the internet plays a role in the sexual lives of a broad cross-section of gay men.

Our results suggest that internet use for meeting sexual partners is shaped by networks. Yet, using the internet to find sex partners is solely associated with having a homophilous network – a finding consistent with our network diversity hypothesis. From the perspective of subcultural theory, men with gay-centric networks, in the absence of countervailing norms, may be more likely to become enmeshed in subcultures. Thus, a limited network may shape such participation via internet use.

Limitations and strengths

While our study fills several gaps in the literature, our findings must be considered with respect to some limitations. In terms of theory, our analysis can only examine the association between neighbourhood factors and individual risk behaviours. We were unable to examine neighbourhood versus non-neighbourhood social ties. Certainly, such information would provide useful insights regarding potential mechanisms. Yet, as noted above, few prior studies have examined whether gay enclaves are associated with individual health. Therefore, our study is important for examining the role of neighbourhood community for gay men’s sexual health.

In terms of methodology, it is important to note several issues. First, with respect to self-report of information, though efforts were made to provide quiet spaces, the survey was completed at public events, which may have facilitated socially desirable responses or led to recall biases. Outcomes were, however, assessed for the past three months, thereby limiting the potential for recall bias. Second, data collection occurred at GLB themed events. Therefore, with respect to external validity, the data may not fully generalise to the urban gay population. With respect to internal validity, we controlled for attachment to the gay community, a potential bias introduced via sampling individuals from GLB themed expos. Models with and without gay community attachment included showed an equivalent pattern of findings, thereby increasing our confidence that our results are not completely due to sampling bias. Third, the cross-sectional nature of our data limits our ability to make causal statements. We are unable to determine if social networks influence sexual behaviours or if individuals who are more likely to practise certain behaviours seek out certain types of networks. With respect to factors that might select people into gay enclaves versus other neighbourhoods, supplemental analyses (not shown here) revealed that no individual-level variables examined in this study were associated with gay enclave residence. Thus, these findings give us greater confidence that the gay enclave effect we found is not simply attributable to demographic factors that differentially select individuals into gay enclaves.

Despite these limitations, our study extends research on gay men’s health and neighbourhood health effects in several ways. First, we utilised a multilevel conceptual approach to consider the influence of gay neighbourhoods on individuals’ sexual risk practices. Second, we improved upon the dearth of neighbourhood effects research on gay health issues by relying on social mapping in addition to census data. Third, we improved upon prior research of gay men’s networks by examining both the intensity and diversity of social network interactions for individual sexual risk practices. Fourth, we examined the interplay of neighbourhoods and social networks, which is often overlooked in the neighbourhood effects literature.

In conclusion, this study highlights the importance of two forms of gay community contexts – gay neighbourhoods and social networks – for the sexual health of gay men. Improved understanding of the health risks faced by this vulnerable population requires a better contextualisation of these risks – one that moves beyond a focus on individual-level factors in their lives and also considers the role of the social contexts in which their health behaviours take place.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Sexual risk behaviours
  5. Neighbourhood health effects
  6. Network influences
  7. Neighbourhood influences on networks
  8. Hypotheses
  9. Methods
  10. Results
  11. Discussion
  12. Acknowledgements
  13. References

The Sex and Love Project was supported by the Hunter College Center for HIV Educational Studies and Training, under the direction of Jeffrey T. Parsons. The authors acknowledge the contributions of members of the Sex and Love research team: Michael Adams, Anthony Bamonte, David Bimbi, Lauren DiMaria, Christian Grov, Catherine Holder, James Kelleher, Juline Koken, Julia Tomassilli, Brooke Wells, and Anna Levy-Warren. Richard Carpiano also acknowledges receiving funding from Investigator Awards from the Michael Smith Foundation for Health Research and the Canadian Institutes of Health Research.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Sexual risk behaviours
  5. Neighbourhood health effects
  6. Network influences
  7. Neighbourhood influences on networks
  8. Hypotheses
  9. Methods
  10. Results
  11. Discussion
  12. Acknowledgements
  13. References
  • Banks, C. (2003) The cost of homophobia: literature review on the human impact of homophobia on Canada. Saskatchewan: Community-University Institute for Social Research.
  • Baron, R.M. and Kenny, D.A. (1986) The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations, Journal of Personality and Social Psychology, 51, 6, 117382.
  • Berry, M., Raymond, H.F., Kellogg, T and McFarland, W. (2009) The Internet, HIV serosorting and transmission risk among men who have sex with men, San Francisco, AIDS, 22, 6, 78789.
  • Carballo-Dieguez, A., Ventuneac, A., Bauermeister, J.A., Dowsett, G.W., Dolezal, C., Remien, R.H. and Rowe, M. (2009) Is ‘bareback’ a useful construct in primary HIV prevention? Definitions, identity, and research, Culture, Health & Sexuality, 11, 5165.
  • Carpiano, R.M., Kelly, B.C., Easterbrook, A. and Parsons, J.T. (2011) Community and drug use among gay men: the role of neighbourhoods and networks, Journal of Health & Social Behavior, 52, 1, 7490.
  • Carpiano, R.M., Lloyd, J.E.V. and Hertzman, C. (2009) Concentrated affluence, concentrated disadvantage, and children’s readiness for school: a population-based, multi-level investigation, Social Science & Medicine, 69, 3, 42032.
  • Casciano, R. and Massey, D.S. (2008) Neighbourhoods, employment, and welfare use: assessing the influence of neighbourhood socioeconomic composition, Social Science Research, 37, 2, 54458.
  • Center for Disease Control and Prevention (CDC) (2010). HIV and AIDS among gay and bisexual men. Available at http://www.cdc.gov/nchhstp/newsroom/docs/FastFacts-MSM-FINAL508COMP.pdf. Date last consulted 29 July 2010.
  • Clatts, M.C., Davis, W.R. and Atillasoy, A. (1995) Hitting a moving target: the use of ethnographic methods in the development of sampling strategies for the evaluation of AIDS outreach programs for homeless youth in New York City, NIDA Research Monograph, 157, 117135.
  • Cockerham, W.C. (2007) Social Causes of Health and Disease. Malden, MA: Polity Press.
  • Das, M., Chu, P.L., Santos, G.M., Scheer, S., Vittinghoff, E., McFarland, W. and Colfax, G.N. (2010) Decreases in community viral load are accompanied by reductions in new HIV infections in San Francisco. PLoS ONE 5, 6, e11068. doi:10.1371/journal.pone.0011068.
  • Davidovich, U., de Wit, J.B.F. and Stroebe, W. (2000) Assessing sexual risk behaviour of young gay men in primary relationships: the incorporation of negotiated safety and negotiated safety compliance, AIDS, 14, 6, 7016.
  • Evans, A.R., Wiggins, R.D., Mercer, C.H., Bolding, G.J. and Elford, J. (2007) Men who have sex with men in Great Britain: comparison of a self-selected internet sample with a national probability sample, Sexually Transmitted Infections, 83, 3, 2005.
  • Fischer, C.S. (1975) Toward a subcultural theory of urbanism, American Journal of Sociology, 80, 6, 131941.
  • Fischer, C.S. (1984) The Urban Experience, 2nd edn. New York, Harcourt, Brace, and Jovanovich.
  • Frank, R., Cerda, M. and Rendon, M. (2007) Barrios and burbs: residential context and health-risk behaviors among Angeleno adolescents, Journal of Health & Social Behavior, 48, 3, 283300.
  • Fritz, M.S. and MacKinnon, D.P. (2007) Required sample size to detect the mediated effect, Psychological Science, 18, 3, 2339.
    Direct Link:
  • Frohlich, K.L., Corin, E. and Potvin, L. (2001) A theoretical proposal for the relationship between context and disease, Sociology of Health & Illness, 23, 6, 77697.
  • Frye, V., Koblin, B., Chin, J., Beard, J., Blaney, S., Halkitis, P., Vlahov, D. and Galea, S. (2010) Neighbourhood-level correlates of consistent condom use among men who have sex with men: a multi-level analysis, AIDS & Behavior, 14, 4, 974985.
  • Granovetter, M.S. (1985) Economic action and social structure: the problem of embeddedness, American Journal of Sociology, 91, 3, 481510.
  • Green, A.I. (2003) ‘Chem friendly’: the institutional basis of ‘club-drug’ use in a sample of urban gay men, Deviant Behavior, 24, 5, 42747.
  • Halkitis, P.N. and Parsons, J.T. (2002) Recreational drug use and HIV-risk sexual behavior among men frequenting gay social venues, Journal of Gay & Lesbian Social Services, 14, 1, 1938.
  • Herr, N.R. (2010) Mediation with dichotomous outcomes. Available at http://nrherr.bol.ucla.edu/Mediation/logmed.html. Date last accessed 10 August 10, 2010.
  • Kaiser, C. (1997) The Gay Metropolis, 1940–1996. New York: Houghton Miffin.
  • Kalichman, S.C., Benotsch, E., Rompa, D., Gore-Felton, C., Austin, J., Luke, W., DiFonzo, K., Buckles, J., Kyomugisha, F. and Simpson, D. (2001) Unwanted sexual experiences and sexual risks in gay and bisexual men: Associations among revictimization, substance use, and psychiatric symptoms, Journal of Sex Research, 28, 1, 19.
  • Kenny, D.A. (2010) Mediation. Available at http://davidakenny.net/cm/mediate.htm. Date last accessed 28 January 2010.
  • LaVeist, Thomas A. and Wallace, John M. (2000) Health risk and inequitable distribution of liquor stores in African American neighbourhood, Social Science & Medicine, 51, 4, 61317.
  • Levay, S. and Nonas, E. (1995) City of Friends: A Portrait of the Gay and Lesbian Community in America. Cambridge, MA, The MIT Press.
  • Levine, M.P. (1979) The gay ghetto, Journal of Homosexuality, 4, 4, 36377.
  • Luke, D.A. (2004) Multilevel Modeling. Thousand Oaks, CA, Sage Publications.
  • Luke, D.A. and Harris, J.K. (2007) Network analysis in public health: history, methods and applications, Annual Review of Public Health, 28, 1, 6993.
  • MacKellar, D., Valleroy, L.A., Anderson, J.E., Behel, S., Secura, G., Bingham, T., Celentano, D.D., Koblin, B.A., LaLota, M., Shehan, D., Thiede, H., Torian, L.V. and Janssan, R.S. (2006) Recent HIV testing among young men who have sex with men: correlates, contexts, and HIV seroconversion, Sexually Transmitted Diseases, 33, 3, 18392.
  • MacKinnon, D.P., Fritz, M.S., Williams, J. and Lockwood, C.M. (2007) Distribution of the product confidence limits for the indirect effect: program PRODCLIN, Behavior Research Methods, 39, 3, 3849.
  • McKirnan, D.J. and Peterson, P.L. (1989) Stress, expectancies, and vulnerability to substance abuse: a test of a model among homosexual men, Journal of Abnormal Psychology, 97, 4, 4616.
  • McPherson, M., Smith-Lovin, L. and Cook, J.M. (2001) Birds of a feather: homophily in social networks, Annual Review of Sociology, 27, 1, 41544.
  • O’Donnell, L., Agronick, G., San Doval, A., Duran, R., Myint-U, A. and Stueve, A. (2002) Ethnic and gay community attachments and sexual risk behaviors among urban Latino young men who have sex with men, AIDS Education & Prevention, 14, 6, 45771.
  • Osmond, D.H., Pollack, L.M., Paul, J.P. and Catania, J.A. (2007) Changes in prevalence of HIV infection and sexual risk behavior in men who have sex with men in San Francisco: 1997–2002, American Journal of Public Health, 97, 9, 16771683.
  • Parsons, J.T. and Bimbi, D. (2007) Intentional unprotected anal intercourse among sex who have sex with men: barebacking – from behavior to identity, AIDS and Behavior, 11, 2, 27787.
  • Patterson, C.J. (2000) Family relationships of lesbians and gay men, Journal of Marriage and Family, 62, 4, 105269.
  • Portes, A. (1998) Social capital: its origins and applications in modern sociology, Annual Review of Sociology, 24, 1, 124.
  • Read, T.R.H., Hocking, J., Sinnott, V. and Hellard, M. (2007) Risk factors for incident HIV infection in men having sex with men: a case-control study, Sexual Health, 4, 1, 359.
  • Robert, S.A. (1999) Socioeconomic position and health: the independent contribution of community socioeconomic context, Annual Review of Sociology, 25, 1, 489516.
  • Ross, M.W. and Williams, M.L. (2001) Sexual behavior and illicit drug use, Annual Review of Sex Research, 12, 1, 290310.
  • Rosser, B.R.S., West, W. and Reinmeyer, R. (2008) Are gay communities dying or just in transition? Results from an international consultation examining possible structural change in gay communities, AIDS Care, 20, 5, 58895.
  • Suarez, T. and Miller, J. (2001) Negotiating risks in context: a perspective on unprotected anal intercourse and barebacking among men who have sex with men – Where do we go from here? Archives of Sexual Behavior, 30, 3, 287300.
  • Sullivan, P.S, Salazar, L., Buchbinder, S. and Sanchez, T.H. (2009) Estimating the proportion of HIV transmissions from main sex partners among men who have sex with men in five US cities, AIDS, 23, 9, 115362.
  • US Census Bureau (2011) US Census Bureau. Available at http://www.census.gov. Date last consulted 22 November 2011.
  • Van de Marcus, U., Bremer, V., Hamouda, O., Kramer, M.H., Friewald, M., Jessen, H., Rausch, M., Reinhardt, B., Rothaar, A., Schmidt, W. and Zimmer, Y. (2006) Understanding recent increases in the incidence of sexually transmitted infections in men having sex with men: changes in risk behavior from risk avoidance to risk reduction, Sexually Transmitted Diseases, 33, 1, 1117.
  • Van de Ven, P., Kippax, S., Crawford, J., Rawstorne, P., Prestage, G., Grulich, A. and Murphy, D. (2002) In a minority of gay men, sexual risk practice indicates strategic positioning for perceived risk reduction rather than unbridled sex, AIDS Care, 14, 4, 47180.
  • Varghese, B., Maher, J.E., Peterman, T.A., Branson, B.M. and Steketee, R.W. (2002) Reducing the risk of sexual HIV transmission: quantifying the per-act risk for HIV on the basis of choice of partner, sex act, and condom use, Sexually Transmitted Diseases, 29, 1, 3843.
  • Wilson, D.P., Regan, D.G., Heymer, K.J., Jin, F., Prestage, G.P. and Grulich, A.E. (2010) Serosorting may increase the risk of HIV acquisition among men who have sex with men, Sexually Transmitted Diseases, 37, 1, 1317.
  • Wolitski, R.J. (2005) The emergence of barebacking among gay and bisexual men in the United States: a public health perspective, Journal of Gay & Lesbian Psychotherapy, 9, 1, 934.
  • Woolwine, D. (2000) Community in gay male experience and moral discourse, Journal of Homosexuality, 38, 1, 537.
  • Zablotska, I.B., Crawford, J., Imrie, J., et al. (2009) Increases in unprotected anal intercourse with serodiscordant casual partners among HIV-negative gay men in Sydney, AIDS & Behavior, 13, 4, 638644.