Behavioral Health and Social Normative Influence: Correlates of Concurrent Sexual Partnering Among Heterosexually-Active Homeless Men

Abstract

Sexual concurrency poses significant HIV/STI transmission risk. The correlates of concurrency have not been examined among homeless men. A representative sample of 305 heterosexually active homeless men utilizing meal programs in the Skid Row area of Los Angeles reported on their mental health, substance use, and social network characteristics. Nearly 40% of men reported concurrency with one of their four most recent sex partners. Results indicated that HIV seropositivity (OR = 4.39, CI: 1.10, 17.46; p = 0.04), PTSD (OR = 2.29, CI: 1.05, 5.01; p = 0.04), hard drug use (OR = 2.45, CI: 1.07, 5.58; p = 0.03), and the perception that network alters engage in risky sex (OR = 3.72, CI: 1.49, 9.30; p = 0.01) were associated with increased odds of concurrency. Programs aimed at reducing HIV/STI transmission in this vulnerable population must take into account the roles that behavioral health and social networks may play in sexual concurrency.

Introduction

Concurrent sexual partnerships and HIV risk

Concurrent sexual partnerships are a significant risk factor for HIV and other sexually transmitted infections (STIs) because they facilitate more rapid disease transmission than monogamous or sequential sexual relationships (1–3). Concurrent sexual partnerships facilitate HIV and STI transmission by eliminating the time gap between potential transmission events that exist in sequential relationships (4). Concurrent sex is additionally problematic because individuals in such relationships are often unaware of their partner’s concurrency (5–8), and many of those engaging in sexual concurrency report having unprotected sex with their concurrent partners (9) or not disclosing their HIV status to casual concurrent partners (10).

Homelessness and HIV risk

Homeless men may be at particularly high risk for concurrent partnering. Men are more likely than women to have concurrent sexual partners (11). Additionally, the homeless population in Los Angeles is predominately African American (12), and prior research suggests that the higher prevalence of concurrency in non-white populations may help to explain why African Americans have higher rates of HIV than other racial/ethnic groups (13–16). Research has found that homeless populations experience rates of HIV infection from 2% to 10.5% depending on the study (17–19), much higher than the general population infection rate of less than ½ of 1% (17). Because of the increased level of HIV infection among homeless persons, concurrent partnering poses a serious risk of HIV transmission in this population.

Behavioral health and HIV risk

Behavioral health (i.e., substance use and mental health) may be associated with concurrent sexual partnering. Concurrent sexual relationships are more prevalent among those who use alcohol and other substances, particularly illicit drugs (11, 20). Although research is lacking among homeless men, a prior study involving homeless women found that more severe drug use was associated with having multiple sex partners (21). Prior studies have found depression and post-traumatic stress disorder (PTSD) to be associated with sexual risk behaviors. Depressed young adults are more likely to have multiple sexual partners, and black men suffering from depression are more likely to contract STIs (22). Symptoms associated with PTSD, such as detachment and perceiving a ‘foreshortened future,’ have been associated with unsafe sexual practices including unprotected sex and sex trade (23, 24), but have not been investigated in association with concurrency. Given the greater prevalence of lifetime traumatic experiences among homeless men (25), and thus the possibility of increased experiences of post traumatic stress (26, 27), PTSD is a critical factor to examine in relationship to concurrency.

Social networks and HIV risk

While the previously described characteristics of individuals may be important in understanding concurrent sexual partnering among homeless men, we know of no prior studies that have examined characteristics of homeless men’s social networks as they may be associated with concurrent sexual partnering. Consistent with social norms theory (28–31), the perceived behavior of alters (social network members) may be associated with homeless men’s sexual risk behavior. Prior studies have found associations between perceived sexual risk behavior within the social network and individual sexual risk. For example, among MSM, the perception that sexual risk behavior is normative among peers has been associated with unprotected sex (32), and more perceived sexual risk behavior in the social network has been associated with increased sexual risk behavior among men on methadone (33). Substance use and perceived norms of substance use behavior in social networks may also be associated with individual sexual risk behavior. Research has found that urban African American women are more likely to report recent multiple sex partners when they have network members who use heroin or cocaine (34), and crack and alcohol use in the social network have been associated with having multiple sex partners among intravenous drug users (35). These findings, while not specific to homeless men, indicate the importance of examining the potential normative influences of social network sexual risk and drug use norms in order to understand concurrent sexual partnering among homeless men.

The structure of social networks may also influence individual risk behavior among homeless men. Density (the number of connections between network members) has been associated with increased injection-related HIV risk behavior among inner-city drug users (36), and with the condom use norms of social network alters (37). Network density is hypothesized to relate to individual behavior via a ‘spiral of silence,’ wherein behavior that runs counter to perceived acceptable network norms in tightly-connected groups is avoided for fear of being ostracized (37). Risky behavior may be even more prevalent where there is high density in risky subgroups (e.g. among those who use substances or participate in risky sex), as common risky behavior is likely to be the norm perpetuated by the ‘spiral of silence’ in these groups.

The Present Study

Heterosexual transmission was responsible for 31% of all new HIV cases in 2009 (38), and homeless persons are disproportionately affected by HIV/AIDS (17); however, we are aware of no studies that have examined the context of sexual concurrency among heterosexually-active homeless men. This study expands prior research by utilizing a probability sample of heterosexually-active homeless men, and examining both individual and social network characteristics that may be associated with concurrent sexual partnering. We hypothesize that individual behavioral health (i.e., substance use and mental health), as well as specific social network features, including density and perceived network norms of risky sexual behavior and substance use, will be associated with concurrent sexual partnering among men in this study. If these hypotheses are confirmed, results may have important implications for service provision for homeless men, including increased attention to meeting behavioral health care needs in a social environmental context.

Methods

Participants

Participants in this study were 305 homeless men randomly sampled and interviewed in 13 meal programs in the Skid Row area of Los Angeles. This area is home to the highest concentration of homeless persons in Los Angeles County. Men were eligible if they were at least age 18, could complete an interview in English, and had experienced homelessness in the past 12 months (i.e., stayed at least one night in a place like a shelter, abandoned building, voucher hotel, vehicle, or outdoors because they didn’t have a home to stay in). As this sample was collected as part of a study of heterosexual risk behavior, all participants reported having vaginal or anal sex with a female partner in the past 6 months. Of the 338 men who screened eligible for the study, 320 men were interviewed (18 refusals). Of these 320: 4 named fewer than 20 network members (alters), 7 were partial completes/break-offs (when interviews could not be completed because the respondent had to leave suddenly, refused to finish the interview, or otherwise did not fully complete the interview process), and 4 were later found to be repeaters. The final sample size was 305, for a completion rate of 91% (305/334). For the data collection, computer-assisted personal interviews were conducted with the software EgoWeb (http://egoweb.github.com), an open source software designed specifically for the collection, analysis, and visualization of personal network data. Men were paid $30 for participation in the interview, which lasted on average 83 minutes. The research protocol was approved by the institutional review boards of the University of Southern California and the RAND Corporation.

Sample Design

To obtain a representative sample of heterosexually active homeless men from the Skid Row area of LA, we implemented a probability sample of men recruited from meal lines in the area. The list of operating meal lines in Skid Row was developed using existing directories of services for homeless individuals and performing interviews with services providers. Our final list contained 13 meal lines: 5 breakfasts, 4 lunches and 4 dinners offered by 5 different organizations. Each of the meal lines was extensively investigated to obtain an estimate of the average number of men served daily. This information was used to assign an overall quota of completes to each site, approximately proportional to the size of the meal line. We then drew a probability sample of homeless men from the 13 distinct meal lines. When the assigned quota could not be achieved in a single visit, the quota was divided approximately equally across the number of visits for each meal line. The interview team randomly selected potential recruits for screening by their position in line using statistician-generated random number tables. Tables were generated such that the site-daily quota could be achieved before the meal line was exhausted. Once the field director selected a potential recruit, an interviewer would wait for him to finish his meal before screening him. Interviews were conducted in any area that afforded privacy: these included corners in dining rooms, chapels, hallways, empty rooms, and in one case, the sidewalk.

The adopted sample design deviates from a proportionate-to-size stratified random sample because of changes in sampling rates during the fielding period, differential response rates of men across meal lines, and variability in how frequently men use meal lines. This last factor means that some men are more likely to be included in the sample. We accounted for the differential frequency of using meal lines by asking respondents how often they had breakfast, lunch and dinner at a meal line in the Skid Row area in the past 30 days, and how much of the past 6 months they had been homeless. This information was used to develop and implement sampling weights to correct for departures from a proportionate-to-size stratified random sample and potential bias due to differential inclusion probabilities (39).

Measures

Concurrency

We asked respondents to name (first name only) the sex partners they had during the past 6 months. We then asked respondents a series of questions about their four most recent sex partners during that past 6 month period. We had limited questioning about the four most recent sex partners to reduce respondent burden. For each partner named, the respondent was asked, “Around the time that you last had sex with [NAME], were you also having sex with other people?” Concurrency was indicated by a ‘yes’ response to this question for any of the four partners named.

The majority of the 305 men in this study (79%, N=241) reported having no more than four sex partners during the past 6 months; the remaining 21% of the men (N=64) named more than four sex partners during that time period. If any of these 64 men who named more than four sex partners had not reported concurrency with their four most recent partners, their rate of concurrent sexual partnering for the past 6 months may have been underestimated. Of these 64 men who named more than four sex partners, however, the large majority (91%, N=58) reported that they had engaged in concurrent sexual partnering around the time that they last had sex with one of those four recent partners. Thus, for 98% (299) of the total sample of 305 men, concurrency with any of the four most recent sexual partners is an inclusive indicator of concurrent sexual partnerships in the past six months, while the remaining 2% of the sample may have engaged in concurrent sexual partnering with a partner not captured by this measure.

Background characteristics

Control measures included background characteristics that we have utilized in previous studies (40, 41): age, race/ethnicity, education (having at least a high school education or General Equivalency Diploma (GED)), being currently married, and having been in jail, prison or on parole in the prior 6 months. We also controlled for the total number of female and male sex partners that men reported having during the past 6 months; more sexual partners increases the opportunity for overlapping sexual relationships. We controlled for HIV status as well. Although there is evidence that knowing one’s HIV status (as opposed to being unaware of HIV infection) may reduce sexual risk behavior (42), studies comparing HIV-negative and HIV-positive populations have found HIV seropositivity to be associated with increased sexual risk behavior (43, 44). As in previous research with a population of homeless women (45), men were asked whether they had ever been tested for HIV and whether they had found out their test result the last time they were tested, as well as whether a health care provider had ever told them that they had HIV/AIDS or if they had ‘reason to believe’ that they might be infected with HIV. A measure of self-reported HIV seropositivity was coded as yes for all men who reported that a health care provider had told them they had HIV, and as yes for those who had ‘reason to believe’ they were HIV positive and had also received their last HIV test result.

Mental health

Past 12-month depressive disorder (Y/N) was measured using a 3-item screening instrument (46) that was tested on community residents, medical patients and mental health patients, and that has demonstrated sensitivity between 83% and 94% and specificity of 90% in these samples (47). Items from this instrument were drawn from the Diagnostic Interview Schedule (48) and the CES-D (49). This screening instrument has been used in several recent studies with homeless persons (50, 51). PTSD was measured with the Primary Care PTSD Screen, a 4-item screener originally designed for use in primary care settings (52). The four items reflect four underlying factors specific to PTSD: re-experiencing, numbing, avoidance, and hyper-arousal. Respondents in this study are defined as screening positive for PTSD if they answer ‘yes’ to any three of four items. A cut point of 3 on the Primary Care PTSD Screen has been shown to maximize sensitivity and specificity of this measure in primary care patients (53). In a primary care setting, persons identified as having at least 3 of the 4 PTSD symptoms would then be administered a structured interview to formally diagnose PTSD (52).

Substance use

Hard drug use was coded as any use of heroin, crack, cocaine, methamphetamine, or hallucinogens in the prior six months. Men were also asked whether they had ever injected illegal drugs. Binge drinking was assessed through a question asking men how often during the prior six months they had 5 or more drinks containing any kind of alcohol within a 2 hour period (0 = ‘not at all’ to 9 = ‘every day’). Binge drinking items and response scales were modified from National Institute on Alcohol Abuse and Alcoholism (NIAAA) Task Force recommendations (54), and all substance use items have been previously vetted with a population of homeless women (55).

Personal network characteristics

We used procedures for conducting personal network interviews that are well-established (56, 57) and have been used in a prior study of homeless women (50, 55). We asked respondents to provide the first names of 20 individuals that they knew, who knew them, and that they had contact with sometime during the past six months (alters had to be at least 18 years or older). Contact could be face-to-face, by phone, mail or through the internet. We constrained network size to be the same across respondents to maximize comparability of network structure measures (58). Twenty alters has been shown to capture structural and compositional variability present in personal networks (59); four men who were not able to name 20 alters were excluded from the sample to maintain comparability across cases.

Men were asked separate questions about which of their network alters had, in the past six months, “drunk alcohol to the point of being high, drunk, or buzzed,” “used drugs like the ones we talked about earlier,” or “had multiple sex partners, had sex with someone they didn’t know, or didn’t use a condom with a new partner.” Because perceived alcohol and drug use were highly correlated, they were combined into a single continuous measure indicating the number of network alters who used drugs or alcohol. The measure of perceived risky sex was skewed, in that many respondents had no network alters meeting this criteria, so this measure was included as an indicator of having one or more network alters whom the respondent believed had participated in risky sexual behavior.

Network density is an index varying between 0 and 1 that represents the proportion of ties among a group of alters relative to the total number of possible ties. For the purposes of calculating density, respondents were asked how often each pair of named network alters “had contact with each other sometime during the past year or so – either face-to-face, by phone, mail, or e-mail. Never, rarely, sometimes or often?” Measures of network density were calculated among all network alters, and separately among those alters who were reported by the respondents as having engaged in risky sex and drinking/drug use.

Analysis

Logistic regression models were used to predict the odds of having concurrent sexual relationships with at least one of the prior four sex partners (analyses were conducted in STATA 9.2). Each predictor associated bivariately at p<.10 with concurrency was retained in the multivariate model (60). Individual demographic characteristics (age, race/ethnicity, education, marital status, and jail/prison/parole) and the total number of sex partners were retained in the multivariate model as control variables.

Results

Sample characteristics

As shown in Table 1, nearly 40% of the sample reported concurrent sex with any of their most recent four sex partners. Most men self-identified as African American (71.69%), followed by white (non-Hispanic, 11.52%), Hispanic (10.43%) and other or multiracial (6.35%). Most respondents (73.31%) had a high school diploma/GED. Few were currently married (6.08%) and 37.31% had been incarcerated or on parole in the prior six months. More than 7% reported that they had been told, or had reason to believe, that they were HIV positive. Men had, on average, nearly four sex partners in the past six months. Nearly 43% of the men reported symptoms of PTSD, while over 46% reported experiencing depression. About half of the men had used hard drugs in the prior six months (48.44%), 38% reported binge drinking, and nearly 20% of men had ever used injection drugs. More than two-thirds of men (68.35%) had at least one network alter who engaged in risky sexual behavior, and the average number of alters who used drugs or alcohol was 2.19. The average density of men’s personal networks was 0.13, which can be interpreted as a network in which 13% of all possible ties are present (a network where every person knows one another would have a density of 1.0, or 100% of possible ties). The average percentage of possible ties present in the subgroups was 8% in the drinking or drug-using group and 6% among those who are perceived to participate in risky sex.

Table 1.

Descriptive statistics (weighted): Concurrent sex, demographic characteristics, sexual risk behavior, HIV status, mental health, substance use and social network features, N = 305.

Variables	%	Mean	S.E.
Concurrent sex with any of most recent sex partners	39.25	---	---
Background Characteristics
Age	---	45.56	10.33
Race/ethnicity
African American	71.69	---	---
White	11.52	---	---
Hispanic	10.43	---	---
Other or multiracial	6.35	---	---
Education
Less than high school	26.69	---	---
>=HS/GED	73.31	---	---
Married	6.08	---	---
Jail/prison/parole in past 6 mos	37.31	---	---
Total number of sex partners	---	3.67	0.29
Self-reported HIV-positive	7.42	---	---
Mental Health
PTSD	42.85	---	---
Depression	46.36	---	---
Substance Use
Binge drinking (6 mos)	38.09	---	---
Hard drug use (6 mos)	48.44	---	---
Injection drug use - ever	19.55	---	---
Social Network Characteristics
Any alters who engage in risky sex	68.35	---	---
Number of alters who use drugs or alcohol	---	2.19	0.21
Total network density	---	0.13	0.01
Risky sex ingroup density	---	0.06	0.01
Substance use ingroup density	---	0.08	0.01

Multivariate Logistic Regression Model

Only the social network density variables were not bivariately associated with sexual partner concurrency and were not included in the multivariate model. Mental health, substance use, and number of alters perceived to engage in risky behaviors remained statistically significant predictors of concurrent sex with any of the four recent partners in the multivariate model (Table 2). In terms of background characteristics, self-reported HIV seropositivity was associated with increased risk of recent sexual partner concurrency (OR = 4.39, CI: 1.10, 17.46; p = 0.04), as was having more sex partners (OR = 2.62, CI: 1.85 – 3.71; p = 0.00). Having recent symptoms of PTSD increased the odds of concurrency (OR = 2.29, CI: 1.05, 5.01; p = 0.04), as did hard drug use (OR = 2.45, CI: 1.07, 5.58; p = 0.03). Perception of social network sexual risk behavior was also associated with concurrency, as having at least one network alter who was perceived to engage in risky sex was associated with increased odds of individual concurrency (OR = 3.72, CI: 1.49, 9.30; p = 0.01). An alternative version of this variable excluding the respondent’s own sex partners was also examined in these analyses, but did not alter the direction or statistical significance of this relationship. An additional sensitivity analysis was conducted that excluded the subset of men with more than four sex partners (the group for whom the measure of concurrency may be an underreporting of total concurrency); this also did not result in any changes to directionality or statistical significance in the multivariate results.

Table 2.

Multivariate binomial logistic regression predicting any concurrent sex with most recent sex partners (weighted), N = 305.

Variables	OR (95% CI)	p>|z|
Background Characteristics
Age	0.99 (0.96 – 1.03)	0.64
Race (white is omitted)
African American	2.77 (0.57 – 13.38)	0.20
Hispanic	1.32 (0.19 – 9.52)	0.78
Other	2.31 (0.39 – 13.75)	0.36
>=HS/GED	1.48 (0.64 – 3.43)	0.36
Married	0.55 (0.07 – 4.64)	0.58
Jail/prison/parole in past 6 mos	0.65 (0.29 – 1.45)	0.29
Total number of sex partners	2.62 (1.85 – 3.71)	0.00
Self-reported HIV-positive	4.39 (1.10 – 17.46)	0.04
Mental Health
PTSD	2.29 (1.05 – 5.01)	0.04
Depression	0.93 (0.43 – 2.01)	0.86
Substance Use
Binge drinking (6 mos)	0.80 (0.37 – 1.71)	0.56
Hard drug use (6 mos)	2.45 (1.07 – 5.58)	0.03
Injection drug use - ever	2.36 (0.86 – 6.52)	0.10
Social Network Characteristics
Any alters who engage in risky sex	3.72 (1.49 – 9.30)	0.01
Number of alters who use drugs or alcohol	1.04 (0.97 – 1.12)	0.26

Discussion

Concurrent sexual partnering was relatively common among the men in this study, with nearly 40% reporting concurrency with a recent partner. Prior studies have suggested that symptoms associated with PTSD may be related to sexual risk behavior (23, 24), and men in this study were more than twice as likely to have had a recent concurrent sexual partnership if they suffered from symptoms of PTSD. Further research into the mechanisms by which PTSD is associated with sexual concurrency could help to inform interventions to improve mental health status and reduce co-occurring sexual risk behavior. To our knowledge, this is the first study to find a relationship between PTSD and concurrency, and studies of PTSD in relationship to HIV risk are still few in number. Because PTSD is a disabling condition that this study and other studies (61, 62) have shown to be prevalent among homeless persons, our findings highlight the importance of treating PTSD in homeless men to reduce the risk of contracting STIs such as HIV.

Consistent with prior research findings that substance use is associated with sexual risk behavior (11, 20, 21), hard drug use was also associated with increased odds of having concurrent sexual partnerships among the homeless men in this study. These findings support the importance of employing evidence-based practices to reduce sexual risk behaviors within substance abuse treatment programs that serve homeless men. Homeless persons have high levels of unmet need for substance abuse treatment (63, 64); access to HIV prevention services through substance abuse treatment may therefore be hindered. In Los Angeles, although providers of shelter services offer little in the way of evidence-based HIV prevention programming (65), it may be prudent to promote HIV prevention in low-barrier settings such as homeless shelters. The findings of this study suggest that addressing both mental health and substance use disorders among homeless men is a critical aspect of HIV prevention. Consistent with previous research (43, 44), HIV seropositivity was also associated with sexual risk taking, in this case, concurrent sexual partnering. Concurrency places people at risk of HIV, but HIV infection reported by the men in the present study may have occurred prior to our data collection period. Qualitative research involving homeless women suggests that the condition of homelessness may lead to a prioritizing of more immediate needs (e.g. intimacy, a place to stay), resulting in diminished concern for reducing engagement in high risk sexual activity among seropositive individuals (66).

Consistent with social norms theory, and confirming the relationship between risky social networks and personal sexual risk behavior found among other vulnerable populations (MSM and men on methadone) (32, 33), having network members perceived to engage in risky sex was associated with concurrent sexual partnering among homeless men in this study. Among homeless youth, prior research has found that HIV risk behavior decreases over time when social networks change to include more pro-social peers, as opposed to peers who engage in “deviant” behaviors such as risky sex (67). Although changes in homeless men’s networks have not been investigated, these findings from the population of homeless youth suggest the potential importance of helping homeless men to minimize ties with other persons in their network who engage in risky sexual behaviors.

The finding of an association between concurrency and sexual risk behavior in men’s social networks has special relevance for the population of homeless men and highlights the unique challenge in facilitating change in the social environment for persons who are entrenched in extreme poverty and who therefore have limited options for changing their physical or social surroundings. Research conducted among people living with HIV/AIDS has determined that HIV risk behaviors are reduced when housing stability is improved (68). Although the mechanisms through which risk reduction occurs when housing is provided have not been investigated, it is possible that change in the physical environment (e.g. permanent supportive housing away from Skid Row) is accompanied by change in the social environment, such that “deviant” network ties are reduced.

Although respondents’ own substance use (specifically, hard drug use) was predictive of concurrency, neither the measures of substance using alters nor measures of density were significantly associated with concurrent sexual relationships among homeless men. It is possible that social norms regarding substance use exert their influence more strongly on an individual’s substance use, and that substance use, in turn, increases the likelihood of risky sexual behavior. It appears that for homeless men, the density of the ties among sexual risk-taking alters is not as important as the simple presence of these individuals in the network. This suggests that the normative influence of sexually risky network alters may be related to exposure to these individual alters, rather than being a phenomenon unique to tightly-connected (dense) groups (as is hypothesized by the ‘spiral of silence’ theory (37)).

There were several limitations in this study. Because we focused on heterosexually active homeless men, there was only limited representation of men who have sex with men and women (7.26% of the sample); these results may therefore not be generalizable to populations of homeless MSM/W or MSM. Further research must address risk of concurrency in the highly vulnerable population of homeless MSM/W and MSM. Additionally, these data are cross-sectional, so it is difficult to make inferences about the causal direction of the findings. For example, the association between social network members’ risky sexual behavior and respondent concurrent sexual partnering could be due to sexual norms in men’s social groups that encourage engagement in risky sex, as hypothesized. Alternatively, the association between risky sex among network members and respondents’ concurrency could be a product of respondents choosing to associate with similar men (i.e., homophilous network selection), or a combination of normative influence and homophilous selection working in concert.

Despite these limitations, this study has identified key individual and social network features that are associated with concurrent sexual partnering among homeless men, and that therefore speak to the increasing interest and effort among federal agencies to address HIV from a perspective that encompasses behavioral health and the social and physical environments of vulnerable populations such as homeless persons (69, 70). To our knowledge, this is the first study to identify that PTSD and the perceived presence of sexually-risky social network members are concomitant to concurrent sexual partnering, and the first study to examine concurrency among homeless men. Homeless men are a highly vulnerable population, with notable rates of HIV seropositivity and substantial levels of concurrent sexual partnering. Because of the high risk that concurrent sexual partnering poses for the transmission of HIV and other STIs, research and programs aimed at reducing HIV and STI transmission among homeless men must take into account the potential role of multiple influences on concurrent sexual risk behavior.