Introduction

Substance use is a driver of the HIV/AIDS epidemic among men who have sex with men (MSM).1–4 Stimulants (e.g., cocaine, crack, methamphetamine), poppers, erectile dysfunction medications (EDM), and alcohol, all substances commonly used in sexual situations, have been prospectively linked with HIV seroconversion among MSM.5–9 The disinhibiting effect of substances used during sex is theorized to increase engagement in risk behavior by facilitating a “cognitive escape” from the constant vigilance of practicing safe sex 10 and activating sexual outcome expectancies (e.g., “substance use enhances sexual pleasure”).6 The magnitude of the association between substance use and sexual risk behavior may be enhanced among individuals who exhibit “risk prone” psychological profiles.11 , 12 For example, prior research has demonstrated that MSM who experience difficulties managing sexual thoughts, urges, or behaviors (i.e., sexually compulsivity) are more likely to use stimulants, to engage in sexual risk behavior and to use stimulants during sex.13–16 The present study examined the associations among substance use, sexual compulsivity, and sexual risk behavior in a probability-based sample of MSM.

Methods

The Urban Men's Health Study 2002 was a probability-based sample of MSM in San Francisco. Of the 879 participants who completed a random digit dialing telephone interview, 711 completed a mail-in questionnaire assessing psychological factors and substance use. Details of the sampling frame and adjusted sample weights are described elsewhere.13 , 17

Measures

Sexual compulsivity was measured using the Sexual Impulsivity Scale (Cronbach's α = 66).18 This scale, which assesses perceived difficulties in controlling sexual behaviors, was centered (M = 0, SD = +1) to facilitate interpretation of the adjusted odds ratios (AOR). Substance use in the last 6 months was dichotomized (any use versus no use) into separate measures of stimulant, poppers, and EDM use. Alcohol use in the last 30 days was categorized into: abstainers (no use), light/social drinkers (1–2 drinks on average), moderate drinkers (3–4 drinks on average), and heavy/binge drinkers (5+ drinks on average). Sexual risk behavior was assessed by asking sexual behavior questions about the four most recent partners in the last 12 months.19

Statistical Analyses

Participants were classified into three groups as follows: (1) No unprotected anal intercourse (UAI), (2) any serodiscordant UAI (SDUAI), and (3) UAI only with partners of the same HIV status (i.e., serosorting). Compared to a reference group that reported no UAI (n = 415), separate multivariate logistic regression analyses examined correlates of SDUAI (n = 105) and serosorting among HIV-negative MSM (n = 185).20 The multivariate models included interaction terms for sexual compulsivity with stimulant, poppers, EDM, and alcohol use. Stratified AORs adjusted for variables that were significant in the full multivariate model (e.g., poppers, EDM use), were calculated to probe significant interaction terms.

Results

As described elsewhere, the majority of the 711 participants who responded to the mail-in survey were Caucasian (81 %), 40 years of age or older (63 %), and HIV-negative (73 %).17 Results of the multivariate logistic regression are presented in Table 1. Being HIV-positive and any stimulant, poppers, EDM or heavy/binge alcohol use were independently associated with higher odds of engaging in any SDUAI. Among HIV-negative men, significant correlates of serosorting during anal intercourse included: age, stimulant, and poppers use.

Table 1 Correlates of any serodiscordant unprotected anal intercourse (SDUAI) and any HIV-negative serosorting
Full size table

The independent association of sexual compulsivity with SDUAI was moderated by poppers and EDM use. Stratified AORs revealed that sexual compulsivity was associated with significantly higher odds of SDUAI only among EDM users [AOR 1.61; 95 % confidence interval (CI) 1.10–2.37; P <0.05] compared to MSM who did not report EDM use (AOR 1.24; 95 % CI 0.87–1.78; P = 0.25). In contrast, the association of sexual compulsivity with SDUAI was significant only for MSM who did not report popper use (AOR 1.83; 95 % CI 1.36–2.46; P <0.001) compared to MSM who reported using poppers (AOR 0.96; 95 % CI 0.66–1.40; P = 0.82).

Discussion

Consistent with prior research, alcohol and other substance use were independently associated with higher odds of SDUAI among MSM.5–9 Findings also indicated that HIV-negative MSM who use stimulants or poppers are more likely to engage in serosorting, perhaps as a method of reducing risk of HIV infection, than non-users.20 Although recent ground breaking randomized clinical trials have demonstrated that antiretroviral therapy (ART) decreases rates of HIV seroconversion and onward transmission, substance users were excluded from these well-controlled efficacy trials.21 , 22 Further clinical research is needed to examine whether behavioral interventions targeting substance use and sexual risk behavior optimize the effectiveness of ART for HIV prevention with substance-using MSM.23

Unique to our findings is the importance of sexual compulsivity as key moderator of the substance use-SDUAI association. The association of sexual compulsivity with SDUAI was significant only among EDM users compared to non-users. EDM use may increase the likelihood that sexually compulsive impulses are enacted by directly improving sexual performance and indirectly (via activation of expectancies) enhancing already difficult to manage sexual thoughts and urges. The treatment of erectile dysfunction among MSM should include HIV risk reduction counseling, particularly for those with elevated sexual compulsivity. In contrast, the association of sexual compulsivity and SDUAI was lower among poppers users. Unlike other substances, poppers are used almost exclusively in sexual contexts to facilitate anal intercourse.24 Because poppers are so closely paired with anal sex, this may increase the likelihood of SDUAI, irrespective of psychological factors.

While innovative, our cross-sectional analyses do not permit any conclusions about causality or temporality. Prospective studies that collect daily data to measure the temporal co-occurrence of substance use and sexual behavior would allow for a more fine-grained analysis of this association.25 Despite this limitation, results from this study underscore the importance of clinical research to examine how behavioral and psychological interventions can augment the effectiveness of biomedical HIV/AIDS prevention among MSM.