Mimiaga, Matthew J ScD, MPH*†; Noonan, Elizabeth MS‡; Donnell, Deborah PhD‡; Safren, Steven A PhD*†; Koenen, Karestan C PhD§; Gortmaker, Steven PhD§; O'Cleirigh, Conall PhD*†; Chesney, Margaret A PhD‖; Coates, Thomas J PhD¶; Koblin, Beryl A PhD#; Mayer, Kenneth H MD†**
Previous studies in the United States have found that the prevalence of childhood sexual abuse (CSA) among men who have sex with men (MSM) is significantly higher than those in the general male population. Using a probability sample of urban MSM, Paul et al1 found that 20% of their sample reported CSA, whereas Lenderking et al2 and Doll et al3 reported rates of 35.5% and 37%, respectively. In comparison, Finkelhor4 estimated that between 5% and 10% of men in the general population have experienced CSA and that at least 20% of women were sexually abused as a child. Although cross-study comparisons are hampered by differing definitions of CSA (eg, noncontact experience like indecent exposure versus contact experiences like kissing, fondling, or intercourse to whether or not the individual considered the experience to be abusive), high prevalence rates of CSA have been consistently documented in samples of MSM, with rates of CSA closer to those of heterosexual women than those of heterosexual men.1
A predictive relationship between CSA and high-risk sexual behavior among MSM has been shown in a number of studies. Lenderking et al2 found that MSM who were sexually abused as children were more likely to have an increased number of male partners across their lifetime and were more than twice as likely to have had unprotected receptive anal intercourse in the past 6 months relative to MSM who were not sexually abused. Paul et al1 reported that MSM who experienced CSA were significantly more likely to engage in high-risk sex. In this study, men who experienced CSA 6 or more times (reported number of incidents without specifying same or new perpetrator) were nearly 2.5 times as likely to have unprotected insertive or receptive anal intercourse with a nonprimary partner and over 5 times as likely to do so with a serodiscordant male compared with those who experienced no CSA.1 Jinich et al5 reported similar findings, with 21.4% of abused men engaging in unprotected anal intercourse with a nonprimary partner in the previous 12 months compared with 15.0% of those who were not abused in childhood. Examining a sample of MSM of a Puerto Rican ancestry living in New York City, Carballo-Diéguez and Dolezal showed that men who experienced CSA and participated unwillingly and/or felt hurt by the experience engaged in unprotected receptive anal sex at higher rates (56%) than those who experienced childhood sexual contact with an older partner but identified as willing and not hurt (42%). Likewise, the willing/not hurt group exhibited a higher risk profile than a control group, which experienced no childhood sexual contact with an older partner (22%).6 O'Leary et al demonstrated that MSM who were abused in childhood not only placed themselves at risk through sexual behavior but also put others at risk. Examining a sample of HIV-infected MSM in New York and San Francisco, they found that those with a history of CSA were significantly more likely than those without a history of abuse to have unprotected anal (UA) sex with partners of HIV negative or unknown serostatus (33% versus 20%).7
In addition to the association of CSA to later sexual risk taking, CSA has been shown to be associated with other negative effects later in life. For instance, in a review of the literature, Browne and Finkelhor8 reported fear, anxiety, depression, anger, hostility, aggression, and sexually inappropriate behavior as initial effects of CSA. Long-term effects reported frequently were depression and self-destructive behavior, anxiety, feelings of isolation and stigma, poor self-esteem, difficulty in trusting others, a tendency toward revictimization, substance abuse, and sexual maladjustment. Additionally, Boudewyn and Liem9 showed CSA to be predictive in both men and women of depression and self-destructive behaviors such as acts of self-harm, suicidal ideation, and suicide attempts. Moreover, severity of CSA (use of force, genital contact versus noncontact exposure) has been associated with greater severity of sequelae.8
In light of the noted association between CSA and risky sexual behavior, MSM who were sexually abused in childhood have been shown to have higher rates of HIV and other sexually transmitted infections (STIs) compared with those with no history of CSA.2,3 Although these past studies have shown a cross-sectional association between higher rates of HIV and other STIs among MSM with a history of CSA, the current study is the first, multisite, longitudinal study to show the predictive association between a history of CSA and subsequent HIV infection among a large sample of HIV-uninfected MSM. We a priori hypothesized that MSM with a history of CSA would be more likely to engage in UA sex, more likely to engage in serodiscordant unprotected anal (SDUA) sex, more likely to become infected with HIV, and have a poorer response to a behavioral HIV risk reduction intervention because previous studies have suggested that coexisting psychopathologies or substance use disorders may interfere with HIV treatment effect. We further hypothesize that substance use, depression, self-efficacy, safer sex norms, and communication skills regarding sexual safety will mediate the relationship between CSA and (1) UA sex, (2) SDUA sex, and (3) HIV infection.
Participants and Procedures
The EXPLORE study was a large-scale randomized HIV prevention trial among MSM conducted in 6 US cities, with a total of 4295 participants over a 48-month period.10 Inclusion criteria for the EXPLORE study were men who were uninfected with HIV, were 16 years or older (although none of the sample was under age 18 at enrollment for the present analysis), had anal sex with another man during the past year, and had not been involved in a mutually monogamous relationship in the past 2 years with a male partner who was HIV uninfected.11 Men were randomized to receive a behavioral intervention versus standard risk reduction counseling. The experimental intervention, described in detail by Chesney et al,12 consisted of 10 core counseling modules delivered at one-on-one counseling sessions. In the main EXPLORE trial, participants in both arms had HIV testing every 6 months and completed a sexual risk behavior and psychosocial assessment battery using audio computer-assisted self-interviewing. Recruitment strategies varied by city but included advertizing; street outreach and outreach at clubs, bars, bathhouses, sex clubs, health clubs, and video arcades; referrals from other cohort studies, current study participants, and community agencies and clinics; and use of Internet sites targeting MSM, community forums, mailings, and a recruitment video. Additional methods for data collection of the EXPLORE cohort have been described previously.10-12
History of CSA was defined as: (a) having had a sexual experience before the age of 13 years with someone who was 5 years older and/or (b) having had a sexual experience between 13 and 17 years old with someone who was 10 years older. Specifically, the following questions were asked of all participants at their baseline enrollment visit: (1) “before you turned 13 years old, did you have any sexual experience with someone who was 5 years, or more, older than you?” and (2) “between the time you turned 13 and your 17th birthday, did you have any sexual experiences with someone who was 10 years or more older than you?.” If a participant answered yes to either or both of these questions, he was considered to have had a history of CSA. The rationale for this comes from the fact that much of the literature on CSA defines it as an underage sexual experience with an adult.1-9
HIV Risk Outcomes
Three of the following HIV risk outcomes were utilized: (1) HIV infection: HIV antibodies were detected by enzyme-linked immunosorbent assay. Sera shown to be reactive after a first test were retested in duplicate. Repeatedly reactive samples were confirmed through Western blot assay or immunofluorescence assay. Participants were tested for HIV at each 6-month follow-up visit over the study's duration; (2) UA sex was defined as self-reported insertive or receptive anal sex without a condom with any HIV status partner in the previous 6 months and asked at each 6-month follow-up visit; and (3) SDUA sex was defined as self-reported insertive or receptive anal sex without a condom with either an HIV-infected partner or a partner of unknown serostatus in the previous 6 months and asked at each 6-month follow-up visit for which the participant remained HIV uninfected.
Adjustments were made for geographical location of study site (Boston, Chicago, Denver, New York, San Francisco, and Seattle), age at enrollment, education, and race/ethnicity. Due to significant multicollinearity among education, employment status, and household income, we only retained and adjusted for education in all models. When the intervention effect was not of direct interest in the model, adjustments were also made for study arm (intervention or control).
Potential mediators of relationship of CSA and the 3 separate HIV risk outcomes in this cohort were baseline measures of nonprescription drugs, alcohol, and psychosocial measures. Drug use was classified as “ever used” versus “never used” in the last 6 months and asked at each 6-month follow-up visit. Nonprescription drug types considered were as follows: (1) marijuana or hashish; (2) poppers or inhaled nitrites (including ampoules); (3) crack or rock cocaine (smoked) or cocaine (snorted or sniffed); (4) amphetamines, such as speed, crystal, or crank (swallowed, snorted, or smoked); (5) hallucinogens (such as phencyclidine [PCP], Special K, angel dust, acid, d-lysergic acid diethylamide [LSD], mushrooms, or Ecstasy); and (6) any injected nonprescription drugs. Heroin use was also assessed but occurred in less than 1% of participants.
Alcohol use in the previous 6 months was categorized as “light” (3 or less drinks/d on no more than 1-2 d/wk), “moderate” (4 or 5 drinks/d on no more than 1-2 d/wk, 1-5 drinks/d on 3-6 d/wk, or 1-3 drinks/d on a daily basis), or “heavy” (4 or more drinks every day or 6 or more drinks on a typical day when drinking). By a drink, we specified “a 12-oz can or glass of beer, a 4-oz glass of wine, a 1.5-oz shot of liquor, or a mixed drink with that amount of liquor.”
The EXPLORE questionnaire asked participants to qualify symptoms of depression on a scale of 1-4 based on a shortened version (7 items) of the Center for Epidemiologic Studies Depression scale.13 As a cutoff point has not been established for this shortened version, the score was divided in quartiles for analytic purposes, with a higher score indicating more depressive symptoms. Measures of self-efficacy for adopting safer sexual behaviors, communication skills for safer sex, and higher safe sex norms were 3 variables created from an exploratory factor analysis of 22 baseline questions on the EXPLORE questionnaire.12 These variables consisted of 9, 6, and 5 items, respectively. Factor-based scales were constructed using scores for each question (based on the 1-6 scoring) rescaled to 0-100. The scales were scored so that higher values reflected higher self-efficacy, higher safe sex norms, and better communication skills. Our analysis dichotomized these scores at 50, the midpoint of the range, to create variables measuring poorer safe sex norms, so that “1” corresponded to poorer self-efficacy, poorer communication skills, and poorer safe sex norms.
SAS statistical software was used to perform each analysis. Although 4295 enrolled participants were included in this study, 51 participants were missing the baseline assessment of CSA, and a further 182 men were excluded from the longitudinal analyses because they were lost to follow-up before their first visit (6-month postbaseline assessment). Data for this study were collected every 6 months, over a 48-month period, providing 8 waves of data collection.
Descriptive statistics to assess the association between CSA on demographic variables and hypothesized mediators use logistic regression analysis.
Cox Proportional Hazards Regression Models
To assess if a history of CSA had a measurable effect on the rate of HIV infection, Cox proportional hazards regression modeling14 was adopted on the discrete timescale of twice-yearly visits over the course of study follow-up. These models were adjusted for age at enrollment, education, race/ethnicity, randomization arm, and study site.
To assess the effect of the EXPLORE intervention on the rate of HIV infection for the CSA status (CSA versus no CSA), an individual Cox proportional hazards model with HIV infection as the outcome was conducted, with CSA status by treatment condition categories as the independent variable. Of particular interest was assessing if history of sexual abuse was an effect modifier of the EXPLORE intervention effect.
Generalized Estimating Equation Modeling
Generalized estimating equation models15,16 were used to determine the relationship between (1) a history of CSA and UA sex and (2) a history of CSA and SDUA sex. These equations account for the within-subject correlations of the repeated measures across time. These models were adjusted for age at enrollment, education, race/ethnicity, randomization arm, and study site.
Because potential mediators are hypothesized to be on the causal pathway between our primary exposure (CSA) and HIV risk outcomes (UA sex, SDUA sex, and HIV infection), it is not appropriate to control for these variables in the final regression models.17
We followed the procedures for testing mediator effects as outlined by Baron and Kenny18 and updated by Kraemer et al.19 According to Kraemer et al, a mediational relationship exists in the present analysis if: (1) the proposed mediator is associated with CSA, (2) the mediator has either a main or an interactive effect on the HIV risk outcome, and (3) changes in the mediator variable precede changes in the dependent variable. The test for mediator effects (ie, variables that actually account for the relationship between CSA and outcome variables) involves 3 steps. First, separate regression analyses (generalized estimating equation models for UA sex and SDUA sex and Cox proportional hazards model for HIV infection) were conducted to test for the relationship between CSA and each of the hypothesized mediator variables. Second, both the CSA and the mediator variables (separate models for each mediator) were included in a regression model as predictor variables for the 3 HIV risk outcomes (separate models for each outcome). Finally, the mediator must be a significant predictor of the given HIV risk outcome in an equation including both the mediator and CSA variables and attenuate the magnitude of effect for the CSA indicator.
Of the 4244 participants enrolled with baseline assessment of CSA, 1686 of the overall EXPLORE cohort (39.7%) reported a history of CSA. Additional characteristics of the overall EXPLORE cohort and main outcome analysis have been described previously.10-12
When comparing a history of CSA with no history of CSA, the men who reported CSA were significantly more likely to be Hispanic or black, have less than a college degree, be unemployed, and have an income of less than $30,000 (Table 1). Assessing the potential mediator outcomes, those with CSA were significantly more likely to have symptoms of depression, use alcohol, use nonprescription drugs, have lower self-efficacy for adopting safer sexual behaviors, have poorer communication skills regarding safer sex, and have poorer safer sex norms (Table 2).
Cox Proportional Hazards Regression Models
There were 258 HIV-1 infections in the EXPLORE study, with an overall rate of 2.1 per 100 person-years. Notably, the rate of HIV-1 infections differed by abuse history: 1.8 per 100 person-years among those with no history of CSA and 2.5 per 100 person-years among those with a history of CSA. Participants who reported experiencing CSA had an increased risk of HIV infection in the trial relative to those who had no history of CSA (adjusted hazards ratio [AHR] = 1.30, 95% confidence interval (CI): 1.02 to 1.69) (Table 3).
Overall, the EXPLORE intervention had a modest effect on reducing the hazard of HIV infection (AHR = 0.82, 95% CI: 0.64 to 1.05).10 Our subset analysis of the EXPLORE intervention's effect on HIV infection by CSA status indicated that compared with participants with no CSA history who did not receive the EXPLORE intervention (referent group; AHR = 1.0), participants who reported no CSA but received the intervention had a decreased rate of HIV infection (AHR = 0.93, 95% CI: 0.66 to 1.30); however, participants with a history of CSA who received (AHR = 1.06, 95% CI: 0.74 to 1.52) and did not receive (AHR = 1.46, 95% CI: 1.04 to 2.04) the EXPLORE intervention were at increased risk for HIV infection over study follow-up.
Generalized Estimating Equation Models
A significant association was observed between UA sex and SDUA sex and a history of CSA as compared with participants who had no history of CSA: adjusted odds ratio (AOR) = 1.24 for UA sex (95% CI: 1.12 to 1.36) and AOR = 1.30 for SDUA sex (95% CI: 1.18 to 1.43) (Table 4).
Marijuana/hashish, poppers/inhaled nitrites (including ampoules), crack/rock cocaine and powder cocaine, amphetamines (such as speed, crystal, or crank), depression (SDUA sex only), lower self-efficacy for adopting safer sexual behaviors, poorer communication skills regarding safer sex, and poorer safer sex norms were found to significantly mediate the primary relationships between (1) CSA→UA and (2) CSA→SDUA, supporting the hypothesis that these mediators are on the causal pathway of these 2 HIV risk outcomes (all P's < 0.001). (Data from the mediation analyses are available upon request.)
Similarly, marijuana/hashish, poppers/inhaled nitrites (including ampoules), crack/rock cocaine and powder cocaine, amphetamines (such as speed, crystal, or crank), depression, lower self-efficacy for adopting safer sexual behaviors, poorer communication skills regarding safer sex, poorer safer sex norms, UA sex, and SDUA sex were found to significantly mediate the primary relationship between CSA→HIV infection, supporting the ultimate hypothesis that these mediators are on the causal pathway between having a history of CSA and subsequent HIV infection (all P's < 0.001).
This is the first study of which we are aware to demonstrate a predictive relationship between a history of CSA and higher rates of HIV infection among HIV-uninfected MSM, when controlling for study arm, study site, age at enrollment, education, and race/ethnicity. This finding extends previous research that has documented higher rates of CSA among HIV-infected MSM by identifying CSA as a risk factor for HIV infection in an HIV-uninfected sample of MSM. These findings are particularly noteworthy as no other study to date has documented this increased hazard ratio, particularly among a large-scale multisite sample of HIV-uninfected MSM enrolled in a longitudinal cohort study. In addition, MSM with a history of CSA were more likely to engage in UA sex and SDUA sex. These results corroborate findings from other studies of MSM. For example, in a study of over a thousand adult MSM, Bartholow et al20 reported a significant association between CSA and HIV risk behavior.
Another pertinent finding from the current study is that among participants reporting CSA, the EXPLORE intervention had no effect in reducing HIV infection rates. It is not clear that this study had sufficient power to detect a significant effect; however, these findings provide good initial evidence that the presence of CSA history in MSM interferes with their ability to derive benefit from traditional HIV prevention interventions. This finding speaks to the importance of conducting postintervention analyses in intervention trials to understand why interventions do not work or for whom they do not work.
These findings are particularly relevant for HIV prevention efforts as the observed rates of CSA in this cohort were very high. The rate of CSA (39.7%) among the MSM in this study is similar to the rates found in other studies of MSM1-3 but substantially higher than the rates found in the general male US population.4 This is particularly concerning due to the harmful consequences of CSA that have been demonstrated specifically in MSM, including mental health counseling and hospitalization, psychoactive substance use, depression, and suicidal thoughts or actions.
In this study, depression was significantly more prevalent among MSM with a history of CSA relative to those without. In addition, MSM with a history of CSA versus those without were more likely to use illicit substances and alcohol. These findings support the notion that sexual risk taking among MSM, particularly those with a history of CSA, is often occurring in the context of “intertwined syndemics.”21 The coexistent untreated depression and increased substance use that are often associated with sexual risk-taking behavior in this population need to be addressed in future HIV prevention interventions to decrease further infection, acquisition, or transmission of HIV and other STIs.
Notably, individuals with a history of CSA compared with those without reported (1) lower self-efficacy for adopting safer sex behaviors, (2) poorer communication skills regarding safer sex, and (3) lower social norms that favor safer sex.12 The cognitive and behavioral consequences of CSA can affect many of the psychosocial variables that have been used to explain HIV risk taking. Three of the most common psychosocial models that have been employed to explain HIV risk-taking and/or HIV prophylactic behavior are the health belief model, the theory of reasoned action/planned behavior, and the social cognitive theory (ie, self-efficacy models).22,23 Health belief models emphasize the role of perceived benefits and barriers to condom use and perceived vulnerability to and consequences of acquiring HIV.24 In the theory of reasoned action/planned behavior,25,26 health behavior, such as condom use, is a function of intentions to use condoms, and, in turn, intentions to use condoms are a function of variables such as attitudes and norms regarding HIV and condom use. Social cognitive models (ie, self-efficacy) explain condom use as a function of an individual's knowledge about HIV, his or her expectancies about the outcomes of using condoms (ie, pleasure reduction versus disease prevention), and his or her own self-efficacy that he or she will be able to use a condom in different sexual situations.27,28 These psychosocial models of HIV prophylactic behavior have been tested both cross sectionally and longitudinally in a variety of populations, including MSM, HIV-infected and HIV-uninfected individuals, and heterosexual men and women.29 All these models include social and cognitive variables that can become distorted in individuals who have a history of CSA. CSA and the resulting negative emotions and negative beliefs about oneself, for example, can interfere with self-efficacy, perceived social norms about sexual risk taking, and the perceived costs and benefits of condom use and norms and attitudes.
Findings from this study suggest that the reasons why MSM continue to engage in risky sexual practices may differ depending on whether individuals experienced CSA or not. Analyses of the baseline EXPLORE data12 demonstrated that the 4295 men enrolled in the study had widely divergent responses to the screening questions associated with sexual risk taking, and about 60 different patterns of responses were seen. For some of the men, unsafe sex was highly correlated with alcohol or other substance abuse, whereas for others, condoms decreasing sexual pleasure seemed more important and others reported low levels of self-efficacy and depression. The data in the present study suggest that the EXPLORE intervention, which included some skills building but was predicated on participants' perceptions that they could change their behavior, might not have been robust enough to change patterns of internalized anger, depression, and lack of self-efficacy that may have been long standing in the participants who experienced CSA.30-32 Future behavioral interventions for this group of at-risk MSM may need to incorporate counseling and skills building that address the traumatic memories and coping strategies that ensue after young men are abused, given the high prevalence of these childhood experiences and their role in potentiating sexual risk-taking behavior.
The pathways to sexual risk behavior in MSM are complex, particularly among MSM with CSA histories. Catania et al33 examined mediators of the relationship between CSA and subsequent HIV risk taking among MSM using cross-sectional data and concluded that CSA contributes to the ongoing HIV epidemic among MSM by distorting or undermining critical motivational, coping, and interpersonal factors that, in turn, influence adult sexual risk behavior. Similarly, in this study, substance use, depression, lower self-efficacy for adopting safer sexual behaviors, poorer communication skills regarding safer sex, poorer safer sex norms, UA sex, and SDUA sex were found to significantly mediate the primary relationship between CSA and subsequent HIV infection, supporting our hypothesis that these mediators are in fact on the causal pathway. Future research could usefully examine the social, behavioral, and cognitive sequelae of CSA that contribute to adult sexual risk taking and increased risk for HIV infection. Specifying these pathways may also lead to the development of HIV prevention technology to specifically target the unique prevention needs of this at-risk group.
The limitations of this study include reduced generalizability to a wider population of MSM. Moreover, the enrollment criteria for EXPLORE may have biased this study toward the very high prevalence of CSA. This seems likely as the study excluded MSM who were in a mutually monogamous relationship. The primary exposure-having a history of CSA-relied on retrospective reports, as all other studies of MSM and CSA do, which may potentially introduce recall bias. In addition, interpretation of statistical significance of findings from subset analyses needs to be performed with caution, given the nominal nature of the computed P values and CIs.
Another limitation important to consider has to do with our operationalization of CSA using age differentials. Given that the long-term negative mental and physical health effects of CSA have been shown to be associated most robustly with high severity of trauma,1 using age differentials alone to categorize CSA mixes those MSM who experienced less severe and more severe trauma and may occlude important differences that exist by type or severity of trauma. Future research is warranted to not only examine trauma by age but also explore differences along other dimensions of trauma including severity, type of CSA-associated sexual contact, number of contacts, and use of violence/threat of violence.
Furthermore, several unique conceptual issues are introduced when using age differentials to operationalize CSA for MSM in particular. Age discrepancy is often used to operationalize CSA such that the power dynamic in the sexual relationship between child or teen and older adult inherently constitutes abuse.2,34,35 For young adolescent or teenage MSM, however, this approach may be problematic due to unique developmental challenges associated with being gay or bisexual or questioning that further complicate the relationship between age and power. For example, young MSM may have a difficult time finding sexual partners or romantic relationships with same-sex same-age peers, and the process of finding same-sex age-mates during adolescence may itself be risky in terms of facing harassment and violence from peers.36-38 As a consequence, it is usual to find that many MSM report that some of their first sexual partners during adolescence included sexual encounters with older partners.34,39 Thus, further data on consensual versus forced sexual encounters with older partners are important to consider for future studies.33,34 Taken within this context of “normal” MSM adolescent development, therefore, it is possible that the current study overestimates the prevalence of CSA and that the statistical relationships between CSA and health outcomes may have been attenuated (ie, the relationships between CSA and mediators or outcomes may be stronger than the current study).
Finally, other developmental issues and stressors should be considered while interpreting findings. Sexual trauma is situated amid other developmental factors that may acerbate or confound the CSA-adult outcome relationships. For example, Friedman et al40 examined the role of sexual orientation-related abuse during adolescence in relation to CSA and adult health outcomes, and Paul et al1 examined the role of other types of childhood abuse (eg, physical abuse) that may co-occur with CSA. Thus, it is important to understand that for some MSM, we may be observing the broader effects of multiple developmental traumas/stressors on later life health outcomes and not just a singular traumatic event. Interventions with MSM with CSA histories should likewise consider the possible role of multiple developmental traumas/stressors in addressing HIV risk behavior.
Of note, despite the high rates of CSA in the sample, these data do not suggest anything causal regarding the association of abuse to later homosexuality. In fact, one interpretation of the existing data is that children may be more likely to be sexually abused or victimized because of the perception that they might be gay/homosexual. Regardless, this study did not collect or analyze data that would be able to confirm or reject this hypothesis in either direction. Furthermore, the majority of men in EXPLORE did not report any type of CSA, so there is little to suggest that CSA “causes” homosexuality, but rather, CSA makes MSM more likely to engage in HIV risk behaviors. Similarly, we did not collect data on adult revictimization in this study and hence cannot confirm or reject whether MSM with CSA histories were more or less likely to experience further abuse in adulthood.
HIV-uninfected MSM with CSA histories are at greater risk for HIV infection, report higher rates of HIV sexual risk behavior, and derive less benefit from prevention programs. The efficacy of HIV prevention interventions may be enhanced by incorporating treatment components designed to address the specific mental health concerns of individuals with a history of CSA.
We gratefully acknowledge the contributions of the EXPLORE study participants, Sari Reisner, MA, for his assistance with the literature review for this article, and the EXPLORE Study Team (Site Principal Investigator; staff listed in alphabetic order within each group): Protocol co-chairs: Beryl Koblin, Margaret Chesney, and Thomas Coates. Boston's Fenway Community Health Center and the Latin American Health Institute: Kenneth Mayer (Site Principal Investigator), Felipe Agredano, Eduardo Aguilu, Rodrigo Barahona, Keith Bell, Christine Borges, Manual Burnias, Mark Cayabayab, Dan Church, Allison Cohn, Yvonne Colon, Janet Dargon, Nancy DeSousa, Judy Erdman, Josh Gagne, Eliza Goodhue, Juan Jimenez, William Johnson, Robert Knauz, Wilfred Labiosa, Ana Lara, Darren LeBlanc, Vin Longo, Marc Manseau, Marshall Miller, Matthew Mimiaga, Elie Mohns, Arnel Montenegro, David Pantalone, Oscar Patino, Tracey Rogers, Edual Ruiz, Steve Safren, Liz Salomon, Julio Silva, Laura van der Leeden, Rodney VanDerwarker, and Curt Weber. Chicago's Howard Brown Community Health Center: David McKirnan (Site Principal Investigator), Althea Batticks, Jason Bird, Liz Bradshaw, Robert Brown, Tom Buckingham, Toni Buckingham, Kelly Carson, Irene Chubinsky, Scott Clark, Scott Cook, Jeff Eichholz, Erica Gaffold, Sanford Gaylord, Dale Gluth, Mark Hartfield, David Henry, Brent Hope, Shane Gosselink, Jenny Hopwood, Laura Hosto, Jennifer Howard, D. J. Jacques, Heather Jandura, Susan Killelea, Andy Knight, Simone Koehlinger, Melissa Kohnke, Felicity LaBoy, Han Lee, Kandis Martin, Nicole Martin, Michele McGrady, Cheron McNeal, Denise Miles, Gino Moore, Michael Munn, Jose Narvaez, Aisha Nawab, Arlette Oblaza, Kevin O'Keefe, Liz Perez, Elisse Pertiller, Kelly Picketts, Borris Powell, Chris Powers, Bart Ramey, Ingrid Rodriguez, Laurez Rutledge, Porfirio Sanchez, Michael Saven, Chris Schmidt, Mark Schulze, Jim Skinner, David Snyder, Al Sorrese, Justin St. Andre, Gerry Taranzo, Ted Taylor, Sonia Torres, Kristin Vanfossan, Gregory Victorianne, and Erik Wetz. Denver Public Health: Franklyn Judson (Site Principal Investigator), Misty Aas, Ramon Armendariz, Chloe Bailey, Brian Bost, Julie Caine, David Cline, Stuart Cooper, Kent Curtis, Beth Deyo, John Douglas, Michael Furhman, Rene Gonzalez, Jeff Hiller, Paul Huber, Sharon Huber, Ken Miller, Philip Osteen, Laurie Peter, Doug Robinson, Dave Ward, Tim Wright, and Andrew Yale. New York Blood Center: Beryl Koblin (Site Principal Investigator), Anne Aldrich, Louise Austin, Lynne Bartell, Jane Bensel, Roberta Bernet, Damian Bird, Adam Bonilla, Carolyn Booher, Michael Camacho, Bradley Clark, Kent Curtis, Nikki Englert, Tonya Flores, George Gates, Corinne Geller, Octavio Gonzalez, Denise Goodman, Krista Goodman, Joshua Hinson, Sean Lawrence, Thomas Lee, Jay Loeffel, Angelo Luna, Larry Metzger, Carolle Morris, Patrick O'Quinn, Eric Ortiz, Ofiji Parris, Alfredo Perez, Terrence Precord, Alberto Rodriguez, Jason Santiago, Craig Siulinski, Leah Strock, Paul Teixeira, Eric Torres, Francesca Valenti, Curt Weber, Avery White, and Jess Zimmerman. San Francisco Department of Public Health: Susan Buchbinder (Site Principal Investigator), Grant Colfax (Site Co-Principal Investigator), Jonas Abella, Mike Ahern, Ari Bacharat, Alba Barreto, Christopher Boyden-DeShazer, Jesse Brooks, Meredith Broome, Tony Buckman, David Colbert, Emily Cole, Alfonso Diaz, Michael Edgar, Beth Faraguna, Paige Fratesi, Vincent Fuqua, Reggie Gage, Anjali Garg, Dale Gluth, Ted Guggenheim, Gavin Hall, Thomas Knoble, Rachel Langdon, Irene Lee, Jennifer Lessard, Nicole Lightburn, Tim Matheson, Corvette Moore, Mario Moreno, Paul O'Malley, Jennifer Owen, Jesus Perez, Robin Rifkin, Chris Rubino, Mateo Rutherford, Jennifer Sarche, Georgia Schreiber, Rob Schwarz, Craig Siulinski, John Stryker, Jason Tomasian, Jim Touchstone, Seth Watkins, Sarah Wheeler, Belinda Van, and Allison Zerbe. Seattle's University of Washington: Connie Celum (Site Principal Investigator), Scott Britt, Fransing Daisy, Aline Dang, Tennessee Dickenson, Niles Eaton, Terry Elliott, Raymond Evans, Paul Farley, Mark Fleming, George Froehle, Jerome Galea, Hal Garcia-Smith, Patrick Gonzalez, Bruce Gooding, Krista Goodman, Justin Haines, Keifa Herzog, Rick Hieb, Eric Hildebrandt, Damon Jameson, E. J. Janson, Thom Kelty, Bill Krutch, Erin Lennon, Matt Leidholm, Alfredo Lopez, Paul Louey, Matt Meko, Jenny Melmed, Dany-Paul Mucha, Shelley Ozscuro, Joe Picciano, Jim Price, Monica Rayne, Alex Rodriguez, Barbara Steele, Nancy Stoaks, Jason Stucky, Matthew Swank, Stephen Tabet, Jeff Thompson, Dennis I. Torres, John Torres, Patrick Tschumper, Paul B. Verano, Ken Wheeler, and Robert Yoon. Abt Associates, Inc: Dana Benet, Sam Bozeman, Sam Clark, Anne Coletti, Michelle Culp, Kirsten Firla, Michael Iatesta, Maria Madison, Sean McKee, and George Seage. Center for AIDS Prevention Studies: Margaret Chesney and Thomas Coates (Principal Investigators), Patrick Barresi, Kevin Filocamo, Cliff Leonardi, Scott Stumbo, and Matthew Troy. Statistical Center for HIV/AIDS Research and Prevention: Neil Albright, Geetha Beauchamp, Rana Bonnice, Lynette Browne, Claire Chapdu, Maya Covarrubias, Martina Deseyve, Lynda Emel, Alice Fisher, Eileen Hess, Sarah Holte, Yijian Huang, Marla Husnik, MaryAnn Klotz, Craig Magaret, Wolfe Maykut, Peter McDonnell, Barbara Metch, Geoff Minerbo, Lisa Ondrejcek, Jennifer Schille, Steve Self, and Al Williams. Central Laboratory: Karen Anderson, Rhonda Canotal, Yao Tsing Chow, Naana Cleland, Dale Dondero, Barryett Enge, Eileen Liu, Chip Sheppard, Brent Sugimoto, Sean Watson.
1. Paul JP, Catania J, Pollack L, et al. Understanding childhood sexual abuse as a predictor of sexual risk-taking among men who have sex with men: the Urban Men's Health Study. Child Abuse Negl. 2001;25:557-584.
2. Lenderking WR, Wold C, Mayer KH, et al. Childhood sexual abuse among homosexual men. Prevalence and association with unsafe sex. J Gen Intern Med. 1997;12:250-253.
3. Doll LS, Joy D, Bartholow BN, et al. Self-reported childhood and adolescent sexual abuse among adult homosexual bisexual men. Child Abuse Negl. 1992;16:855-864.
4. Finkelhor D. Current information on the scope and nature of child sexual abuse. Future Child. 1994;4:31-53.
5. Jinich S, Paul JP, Stall R, et al. Childhood sexual abuse and HIV risk-taking behavior among gay and bisexual men. AIDS Behav. 1998;2:41-51.
6. Carballo-Diéguez A, Dolezal C. Association between history of childhood sexual abuse and adult HIV-risk sexual behavior in Puerto Rican men who have sex with men. Child Abuse Negl. 1995;19:595-605.
7. O'Leary A, Purcell D, Remien RH, et al. Childhood sexual abuse and sexual transmission risk behaviour among HIV-positive men who have sex with men. AIDS Care. 2003;15:17-26.
8. Browne A, Finkelhor D. Impact of child sexual abuse: a review of the research. Psychol Bull. 1986;99:66-77.
9. Boudewyn AC, Liem JH. Childhood sexual abuse as a precursor to depression and self-destructive behavior in adulthood. J Trauma Stress. 1995;8:445-459.
10. The EXPLORE Study Team. Effects of a behavioural intervention to reduce acquisition of HIV infection among men who have sex with men: the EXPLORE randomized controlled study. Lancet. 2004;364:41-50.
11. Koblin BA, Chesney MA, Husnik MJ, et al. High-risk behaviors among men who have sex with men in 6 US cities: baseline data from the EXPLORE study. Am J Public Health. 2003;93:926-932.
12. Chesney MA, Koblin BA, Barresi PJ, et al. An individually tailored intervention for HIV prevention: baseline data from the EXPLORE study. Am J Public Health. 2003;93:933-938.
13. Mirowsky J, Ross CE. Age and depression. J Health Soc Behav. 1992;33:187-205.
14. Therneau TM, Grambsch PM. Modeling Survival Data: Extending the Cox Model. New York, NY: Springer; 2000.
15. Liang KY, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika. 1986;73:13-22.
16. Hardin JW, Hilbe JM. Generalized Estimating Equations. Boca Raton, FL: Chapman and Hall/CRC Press; 2003.
17. Shadish WR, Cook TD, Campbell DT. Experimental and Quasi-Experimental Designs for Generalized Causal Inference. Chicago, IL: Rand McNally; 2002.
18. Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research: conceptual, strategic and statistical considerations. J Pers Soc Psychol. 1986;51:1173-1182.
19. Kraemer HC, Wilson T, Fairburn CG, et al. Mediators and moderators of treatment effects in randomized clinical trials. Arch Gen Psychiatry. 2002;59:877-883.
20. Bartholow BN, Doll LS, Joy D, et al. Emotional, behavioral, and HIV risks associated with sexual abuse among adult homosexual and bisexual men. Child Abuse Negl. 1994;18:747-761.
21. Stall R, Mills TC, Williamson J, et al. Association of co-occurring psychosocial health problems and increased vulnerability to HIV/AIDS among urban men who have sex with men. Am J Public Health. 2003;93:939-942.
22. Aggleton P, O'Reilly K, Slutkin G, et al. Risking everything? Risk behavior, behavior change, and AIDS. Science. 1994;265:341-345.
23. Carmel S. The health belief model in the research of AIDS-related preventive behavior. Public Health Rev. 1990-1991;18:73-85.
24. Kirscht JP, Joseph JG. The health belief model: some implications for behavior change, with reference to homosexual males. In: Mays V, Albee G, Schneider S, eds. Primary Prevention of AIDS: Psychological Approaches. Newbury Park, CA: Sage; 1989:111-127.
25. Fishbein M, Middlestadt S. Using the theory of reasoned action as a framework for understanding and changing AIDS-related behaviors. In: Mays V, Albee G, Schneider S, eds. Primary Prevention of AIDS: Psychological Approaches. Newbury Park, CA: Sage; 1989:93-110.
26. Fisher WA, Fisher JD, Rye BJ. Understanding and promoting AIDS-preventive behavior: insights from the theory of reasoned action. J Health Psychol. 1995;14:255-264.
27. Bandura A. Perceived self-efficacy in the exercise of control over AIDS infection. Eval Program Plann. 1990;13:9-17.
28. Wulfert E, Wan CK. Safer sex intentions and condom use viewed from a health belief, reasoned action, and social cognitive perspective. J Sex Res. 1995;4:293-305.
29. Wulfert E, Safren SA, Brown I, et al. Cognitive, behavioral, and personality correlates of HIV-positive persons' unsafe sexual behavior. J Appl Soc Psychol. 1999;29:223-244.
30. Kelly JA, Murphy DA, Bahr GR, et al. Factors associated with severity of depression and high-risk sexual behavior among persons diagnosed with human immunodeficiency virus (HIV) infection. Health Psychol. 1993;12:215-219.
31. Exner TM, Meyer-Bahlburg HFL, Ehrhardt AA. Sexual self-control as a mediator of high risk sexual behavior in a New York City cohort of HIV-positive and HIV-negative gay men. J Sex Res. 1992;29:389-406.
32. Quadland MC, Shattls WD. AIDS, sexuality, and sexual control. J Homosex. 1987;14:277-298.
33. Catania JA, Paul J, Osmond D, et al. Mediators of childhood sexual abuse and high-risk sex among men who have sex with men. Child Abuse Negl. 2008;32:925-940.
34. Arreola S, Pollack L, Paul J, et al. Childhood sexual experience and adult health sequelae among gay and bisexual men: defining childhood sexual abuse. J Sex Res. 2008;45:246-252.
35. Rind B, Tromovitch P, Baurserman R. A meta-analytic examination of assumed properties of child sexual abuse using college samples. Psychol Bull. 1998;124:22-53.
36. Bontempo DE, D'Augelli AR. Effects of at-school victimization and sexual orientation on lesbian, gay, or bisexual youths' health risk behaviors. J Adolesc Health. 2002;30:364-374.
37. D'Augelli AR, Pilkington NW, Hershberger SL. Incidence and mental health impact of sexual orientation victimization of lesbian, gay, and bisexual youths in high school. Sch Psychol Q. 2002;17:148-167.
38. Pilkington NW, D'Augelli AR. Victimization of lesbian, gay, and bisexual youth in community settings. J Community Psychol. 1995;23:34-56.
39. Rind B. Gay and bisexual adolescent boys' sexual experiences with men: an empirical examination of psychological correlates in a nonclinical sample. Arch Sex Behav. 2001;30:345-368.
40. Friedman MS, Marshal MP, Stall R, Gay-related development, early abuse and adult health outcomes among gay males. AIDS Behav. 2008;12:897-902.
© 2009 Lippincott Williams & Wilkins, Inc.