Determinants of High-Risk Sexual Behavior during Post-Exposure Prophylaxis to Prevent HIV Infection
- First Online:
- Cite this article as:
- Golub, S.A., Rosenthal, L., Cohen, D.E. et al. AIDS Behav (2008) 12: 852. doi:10.1007/s10461-007-9286-8
- 177 Views
Men who have sex with men (MSM) receiving non-occupational post-exposure prophylaxis (NPEP) to prevent HIV transmission completed interview-assisted questionnaires regarding high-risk behavior in the 6 months prior to NPEP and during the 28-day NPEP period. 21% of participants reported unprotected sex during NPEP, and 11% reported unprotected sex with HIV-positive or HIV status unknown partners. In univariate analyses, unprotected sex during NPEP was associated with prevention fatigue, depression, loss of loved ones to HIV, and a history of engagement with HIV/AIDS service organizations, e.g., receiving services from an HIV-related agency, donating money to HIV-related causes, and reading HIV-related magazines. Logistic regression analyses revealed that the strongest predictor of risk-taking during NPEP was HIV engagement. These data underscore the importance of combining chemoprophylaxis with behavioral interventions that support risk-reduction. Such interventions should not assume that those most engaged with HIV/AIDS service organizations are less likely to engage in risk behavior.
KeywordsHIV preventionSexual behaviorGay and bisexual menPost-exposure prophylaxisCognitive disengagementRisk factors
National guidelines for the use of antiretroviral medication to prevent HIV infection following a high-risk sexual or needle-sharing exposure were published in the United States in 2005 (CDC 2005). Data on the effectiveness of non-occupational post-exposure prophylaxis (NPEP) come from case-control and other observational studies suggesting that the use of antiretrovirals after a high risk exposure can reduce the likelihood of HIV infection by 79–81% (Cardo et al. 1997; CDC 1995; Schechter et al. 2004). Analyses suggest that NPEP is cost-effective when targeted to high-risk populations (Herida et al. 2006; Pinkerton et al. 2004). Despite the effectiveness of NPEP, both researchers and clinicians are ambivalent about its wide-scale use because of concerns about behavioral disinhibition, as well as potential drug toxicities (Blower et al. 2000; Kelly et al. 1998; Lurie et al. 1998; Richens et al. 2005).
Data from several longitudinal studies of sexual behavior following NPEP indicate that taking NPEP does not lead to increases in risk-taking. In the San Francisco PEP study, over 75% of men who have sex with men (MSM) reported a decrease in the number of high-risk acts with high-risk partners, compared to baseline, at both 6 months and 12 months post-NPEP (Martin et al. 2004). Another longitudinal study found comparable rates of risk-taking among NPEP and non-NPEP users at 6-month follow-up (Schechter et al. 2004). Knowledge of the availability of NPEP was not associated with increased rates of unprotected sex among MSM (Waldo et al. 2000) or among heterosexual serodiscordant couples (van der Straten et al. 2000). And attitudes toward NPEP—including perceiving less HIV/AIDS threat due to NPEP availability—were not associated with increased incidence of STDs or new HIV infection over time (Van Der Snoek et al. 2005).
Estimates of repeat NPEP use vary from 8% over 5 years (Sonder et al. 2007) to 28% over 24 months (Schechter et al. 2004). In the San Francisco PEP Study, 17% of participants received at least one repeat course of NPEP within 1 year (Martin et al. 2004). However, the authors estimate that this rate of repeat exposure is significantly lower than that which would be expected to occur among individuals with similar risk profiles at baseline. As such, repeat NPEP use has been attributed to individuals’ maintenance of their usual patterns of risk-taking behavior, rather than increased risk due to behavioral disinhibition (Martin et al. 2004).
The most recent Centers for Disease Control and Prevention (CDC) NPEP guidelines (CDC 2005) encourage HIV prevention counseling during NPEP use, and suggest that clinicians should use the provision of medication as an “educable moment” within which to assess their patients’ needs for behavioral intervention and other services (Merchant and Mayer, 2005). Behavioral interventions to accompany NPEP have been developed (Roland et al. 2005), and a randomized trial comparing risk behavior and seroconversion between participants assigned to two versus five sessions of risk-reduction counseling is currently being evaluated (Roland 2006). However, to date, there have been no published reports examining risk behavior during the 28-day NPEP treatment period. Understanding patients’ risk behavior during NPEP treatment is important for several reasons. First, in instances in which NPEP fails to prevent infection (either because the medications are not 100% effective or because of non-adherence to treatment) transmission of drug-resistant virus is possible (CDC 2005). Second, coinfection with other sexually transmitted diseases (STDs) during NPEP may enhance susceptibility to HIV infection (Rottingen et al. 2001), making prophylaxis less effective. And third, recent interest in the use of antiretroviral medication for pre-exposure prophylaxis (PREP) may contribute to patients’ perceptions that they are protected from HIV-infection while using antiretroviral medications. While Phase II/III clinical trials of PREP are currently underway in five countries, including the United States, there have been anecdotal newspaper reports that some individuals are already experimenting with PREP, buying or selling antiretrovirals at clubs or bars immediately prior to high-risk sex (Cohen 2006; Costello 2005). Data collected from a sample of attendees of minority gay pride events in four cities in 2004 revealed that 25% of participants had heard of PREP, and seven-percent of attendees in San Francisco and Baltimore reported having used PREP at least once (Kellerman et al. 2006). In order to meet the continued challenge of combining emerging biomedical prevention strategies with behavioral interventions that will enable and support them, it is imperative to understand the psychosocial and behavioral vulnerabilities that might be associated with high-risk sexual behavior in the context of NPEP. This study was designed to examine patterns of risk-taking among MSM undergoing NPEP treatment and identify potential determinants of high-risk behavior during the treatment period.
This study was conducted as part of two open-label, non-randomized clinical trials designed to evaluate the safety and tolerability of two different antiretroviral NPEP regimens: tenofovir disoproxil fumarate (TDF) and lamivudine (06/2003-11/2004) and TDF plus emtricitabine (03/2005-06/2006). Support for the study was provided by an educational grant from Gilead Sciences (Foster City, CA), who supplied their medications at no cost to the patients. The open-label trials were approved by the Institutional Review Board of Fenway Community Health, and secondary analysis of self-report data was approved by the Institutional Review Board of Queens College, City University of New York. Procedures were identical for both studies, and the data are combined for these analyses in order to maximize statistical power. There were no differences between participants across study samples in any of the measures described below.
Potential participants were identified by medical providers at Fenway Community Health (FCH), a free-standing care facility serving the gay and lesbian communities of greater Boston. A FCH physician is available 24-h a day through an NPEP pager system, and all health care providers and support staff are trained to refer patients for NPEP evaluation and screening if they present to or call the clinic about a recent exposure to HIV. Patients were offered NPEP if they: (a) met criteria for a high-risk exposure, and (b) were able to begin the medication within 72-h of the exposure incident. High-risk exposure was defined as unprotected anal, vaginal, oral, penile, or musocal exposure to ejaculate, cervicovaginal secretions, or anorectal secretions from a partner who is HIV infected or of unknown HIV status. Unprotected exposure could include condom failure, breakage, or slippage. All patients who met criteria for NPEP were recruited to participate in this study, unless they met one of the following exclusion criteria: (a) pregnancy (the medications used in this study were not considered standard of care during pregnancy), (b) chronic hepatitis B infection (since the drugs used in this study are active against hepatitis B infection and patients might experience a flare-up upon discontinuing medication), (c) active psychiatric illness that could prevent compliance with study procedures, (d) unwillingness to participate in study procedures, or (e) known intolerance or allergy to medications in the NPEP treatment regimen. Participants who were considered ineligible for this study were offered alternative NPEP regimens. Patients who called after-hours were provided with an initial dose of medication through their local pharmacy, and were scheduled for a clinic visit the next day. During this study period (06/03–06/06), 117 individuals were deemed eligible for participation in this study by the NPEP provider on call, and all (100%) agreed to participate in the open label trial. In the analyses below, we included only MSM (n = 108), defined as men whose NPEP-related exposure was with another man or who reported sex with other men in the self-report measures.
Study participation entailed six visits over a 3-month period. At Visit One, participants gave a detailed medical history, underwent a physical examination (including testing for pregnancy, HIV, and other STDs), and received risk-reduction counseling, appropriate referrals, and a 14-day supply of NPEP medication. At Visit Two (approximately 10–14 days later), participants reported medication adherence, discussed any symptoms/side effects, and received the remaining 14-day supply of NPEP medication. Visit Three occurred at the completion of the NPEP medications (approximately 28 days after the initial visit) and repeated the procedures in Visit Two. Visits Four through Six occurred at 6 weeks, 12 weeks, and 14 weeks after the initial visit and consisted of symptom reports and follow-up HIV counseling and testing. Data presented in this study were collected as part of an interviewer-assisted questionnaire completed at Visits Two (14 days) and Three (28 days). Questionnaire data were collected at Visit Two rather than at baseline to reduce participant burden during the emotionally taxing enrollment visit. Participants were given no financial compensation for their participation.
Data on participants’ sexual behavior were collected at two time-periods. At visit two, participants were asked to report their sexual behavior in the 6 months prior to beginning NPEP treatment. At visit three, participants were asked to report their sexual behavior during the 4 week NPEP treatment period. Psychosocial and sociodemographic variables were collected at visit two only.
Sexual Risk Behavior
Sexual risk behavior questions were divided by partner-type: known HIV-positive, known HIV-negative, HIV status unknown. Participants were first asked to estimate their total number of partners per type within the assessment period. For each partner type, participants were asked to estimate the number of times they had anal receptive and anal insertive sex, both with and without a condom (Chesney et al. 2003).
Participants were asked two questions (r = .46) to assess HIV Anxiety, similar to measures used in other studies (Crosby and Holtgrave, 2006; MacKellar et al. 2007). Participants rated these items on a five-point scale (ranging from never to all the time): “I worry about becoming infected with HIV” and “I feel like it’s only a matter of time before I get infected.” We assessed HIV Burden by four face-valid items, using the same five point scale (alpha = .70). Items asked participants to rate the extent to which they feel HIV limits them sexually or socially, the extent to which they think about HIV while they are having sex, and the extent to which they feel HIV is “too present” in their lives.
A measure of participants’ Personal Experience of HIV was created by asking participants three questions (alpha = .76): “How many people in your close circle of friends/family are HIV-positive?”; “How many people in your extended circle of friends/family are HIV-positive?”; “How many people have you lost to HIV?” Responses were summed to create a composite score.
Depression was assessed as the mean of seven items (alpha = .92) taken from the Center for Epidemiologic Studies Depression Scale (CES-D; Radloff 1977). These items have been used in similar studies of HIV prevention behaviors (Colfax et al. 2004), and have been found to correspond to full-scale CES-D scores (Santor and Coyne 1997). Participants rate how often in the past month they have experienced a variety of symptoms, including loneliness, restlessness, poor appetite, etc.
To measure HIV Engagement, an eight-item scale (alpha = .72) was adapted from a measure of gay identity/affiliation used in previous studies of NPEP (Kalichman 1998). The scale presents participants with eight activities (e.g., attending an HIV-related event; receiving services from an HIV-services agency; reading an HIV-related magazine) and asks them to estimate the frequency with which they engaged in these activities in the past year.
Finally, participants were given two face-valid items designed to measure attitudes toward NPEP and toward HIV prevention. To measure Prevention Fatigue, participants rated the following statement on a six-point likert scale: “People I know are engaging in riskier sex because they are tired of trying to be safe.” To measure PEP as a Prevention Tool, participants rated the following statement on the same likert scale: “PEP should only be used when other prevention efforts fail (e.g., condom breaks)”.
We first conducted descriptive analyses to determine the prevalence of risk-taking behavior during NPEP. Because over 50% of the sample reported no high-risk behavior during NPEP treatment, all our continuous measures of sexual risk-taking were strongly positively skewed. Therefore, we created two dichotomous variables reflecting participants’ reports of sexual risk-behavior during NPEP treatment: (a) any unprotected anal sex with an HIV-positive or HIV status unknown partner; and (b) any unprotected anal insertive or receptive intercourse, regardless of the partner’s HIV status.
Bivariate analyses (t-tests or Mann–Whitney U tests for continuous variables, Chi-Square for categorical variables) were used to examine differences between participants who did or did report unprotected anal sex during NPEP in terms of demographic and psychosocial characteristics. Variables that were significantly associated with unprotected sex in bivariate analyses were included into a multivariate logistic regression model to assess their relative predictive power.
Participant demographics (n = 89)
H.S. Equivalent or less
Some college or college graduate
Some graduate school or degree
Income (yearly household)
The total number of partners reported during NPEP ranged from 0 to 25 (median = 1.0; IQR 0–2), with 34% of participants reporting no sexual partners in the 28-day period. Twenty one percent (n = 19) of participants reported having unprotected anal insertive or receptive intercourse during NPEP. The number of high-risk (HIV-positive or HIV status unknown) partners ranged from 0 to 21 (median = 0; IQR 0–1), with 43% (n = 38) of participants reporting at least one high-risk partner and 20% (n = 18) reporting between two and ten high-risk partners. Of those participants who reported high-risk partners, 94% reported having anal insertive or receptive intercourse, and 26% of these (11% of total participants) reported having unprotected anal sex with a high-risk partner during NPEP treatment.
Comparison of psychosocial factors in participants who reported risk-taking during NPEP versus those who did not
No unprotected anal sex (n = 70)
Unprotected anal sex
All partners (n = 19)
HIV + or unknown-status partners (n = 10)
Unprotected acts pre-NPEP
Personal experience of HIV
PEP as last resort
We first examined differences between participants who reported any unprotected anal sex during NPEP (regardless of partner’s status) to those who did not. Participants who reported unprotected sex during NPEP reported a greater number of unprotected acts in the 6 months prior to NPEP, Z = 3.70, P < .001, r = .31. Risk-taking participants also reported significantly greater engagement in the HIV care system (Z = 2.75, P < .01, r = .32), significantly higher levels of prevention fatigue and higher depression scores, compared to those who did not report unprotected sex during NPEP.
Second, we compared the subset of participants who reported unprotected sex with high-risk partners (HIV-positive or HIV status unknown) during NPEP to all other participants. Results were similar to those found among the full sample of participants who engaged in unprotected anal sex during NPEP. Participants who reported unprotected sex with high-risk partners during NPEP reported a greater number of unprotected acts in the 6 months prior to NPEP (Z = 3.26, P < .001, r = .41), reported greater engagement in the HIV care system, (Z = 3.12, P < .01, r = .37), and reported higher levels of prevention fatigue and depression scores, compared to those who did not report unprotected sex with HIV-positive or status unknown partners. Participants who reported sex with HIV-positive or status unknown partners also scored significantly higher on the measure of personal experience of HIV (Z = 2.0, P < .05, r = .22).
Three of the 89 participants who completed second interview measures later returned for a second course of NPEP during the study period. Two reported unprotected anal sex with an HIV-status unknown partner during NPEP; one reported no unprotected anal sex during NPEP. Although the small sample size limits interpretation, these data suggest that individuals who report unprotected sex with high-risk partners during NPEP are significantly more likely to be repeat NPEP users, compared to those who do not report unprotected sex with high-risk partners during NPEP (Fisher’s exact P < .05).
Logistic regression analyses of predictors of risk-taking during NPEP Treatment
Any unprotected sex
Unprotected sex with a high-risk partner
Unprotected acts pre-NPEP
Personal experience of HIV
Among this sample of MSM receiving NPEP treatment, 21% percent reported unprotected anal insertive or receptive intercourse during their 28-day NPEP treatment period. Thirty-eight percent of participants reported engaging in anal sex with HIV positive or HIV-status unknown partners during NPEP, and 26% of these participants (11% of total participants) specifically reported having unprotected anal sex with HIV-positive or HIV status unknown partners during NPEP. It is critical for clinicians to recognize the potential for high-risk sexual behavior during NPEP with both HIV-positive/HIV status unknown and HIV-negative partners. Unprotected sex with HIV-positive or status unknown partners presents a risk of re-exposure; currently, little is known about the risk of HIV-infection during NPEP or the possibility of extending the course of medication in the event of a secondary exposure. Although unprotected sex with an HIV-negative partner during NPEP does not present the same level of risk for HIV transmission, this behavior is still potentially risky for three reasons. First, since NPEP is not 100% effective in preventing transmission, the individual receiving NPEP may be placing his HIV-negative partner at risk of infection. Second, unprotected sex with an HIV-negative partner still presents a risk of transmission of other STDs, which may enhance susceptibility to HIV infection (Rottingen et al. 2001) and make NPEP less effective. Third, patients may be mistaken in believing their partner to be HIV-negative, either because status was assumed rather than discussed (Gold et al. 1999; Parsons et al. 2006) or because a partner failed to accurately disclose his status (Parsons et al. 2005, Sullivan 2005). The prevalence of unprotected anal sex with partners of any status underscores the importance of combining the use of NPEP with behavioral interventions that promote risk-reduction practices (Merchant and Mayer 2005; Roland 2006). Behavioral interventions designed to accompany NPEP traditionally stress the maintenance of risk reduction behavior after completion of the regimen. These data suggest that interventions are needed that specifically address the psychosocial factors that might make maintaining risk reduction difficult during the 28-day NPEP period.
In univariate analyses, a series of psychosocial variables were associated with high-risk behavior during NPEP, including: risk-taking in the 6 months prior to NPEP, loss of loved ones to HIV, depression, prevention fatigue, and engagement in the HIV care system. In multivariate testing, HIV engagement—defined as receiving services from an HIV-related organization, donating money to or volunteering for an HIV-related cause, and/or reading HIV-related magazines and web sites—emerged as the strongest predictor of unprotected sex during NPEP and the only significant predictor of unprotected sex with an HIV-positive or status unknown partner during NPEP. Why would individuals most connected to HIV care and prevention be most likely to engage in risk behavior during NPEP?
One possibility is that individuals who are most involved with HIV-related activities are also most informed about the data concerning chemoprophylaxis, and are more likely to believe themselves protected from new infection during their NPEP treatment. The majority of these data were collected prior to wide-spread publicity about the potential for using antiretrovirals for pre-exposure prophylaxis (i.e., PREP), but it is still possible that participants who scored high on HIV engagement were more optimistic about NPEP efficacy. This interpretation is a reminder that patients—especially those with the most sophisticated knowledge of the field—experience a given intervention as one option in an array of existing and emerging prevention strategies. As new biochemical prevention strategies emerge, it will be important to reexamine existing prevention messages and develop educational and counseling materials that stress the integration of biochemical and behavioral approaches to risk reduction. These data highlight the importance of understanding HIV prevention techniques in relationship to each other.
Another possible explanation for the relationship between HIV engagement and high-risk behavior during NPEP is a spurious association driven by both variables’ relationship to overall risk-taking. Higher risk MSM may be more likely to engage in HIV-related services because they are concerned about their own HIV status. Were this the case, however, multivariate testing would have revealed a larger role for baseline risk behavior in predicting risk taking during NPEP. It is possible that general patterns of risk taking behavior do play a role in fostering HIV engagement, but data from this study suggest the importance of HIV engagement in predicting high-risk behavior during NPEP above and beyond baseline risk taking patterns.
Another possible explanation for the association between HIV engagement and sexual risk-taking during NPEP can be drawn from theories of cognitive disengagement (McKirnan et al. 2001). According to this model, individuals who engage in ongoing cognitive restraint to control undesirable behaviors (e.g., overeating, binge drinking, high-risk sex) must internalize and maintain high performance standards in order to control the “forbidden” behavior (Heatherton and Baumeister 1991; McKirnan et al. 1996). Over time, and especially during period of high-stress, these standards become burdensome and aversive, and individuals are motivated to cognitively disengage from them, resulting in overindulgence in the very behaviors they have been trying to avoid. Models of cognitive disengagement have been used to explain the association between alcohol and drug use and unsafe sexual behavior especially among individuals who are experiencing prevention “burnout” (McKirnan et al. 2007; Vicioso et al. 2005). As applied to our data, individuals most engaged in the HIV care system might also have the greatest commitment to and highest standards for risk-reduction, prior to the high-risk exposure that precipitated their NPEP treatment. However, these same individuals would be at greatest risk for lapses into risk-taking “binges” in response to external stressors, including the emotional strain of NPEP itself. Many patients report that taking medication serves as a constant reminder of the threat of illness (Golub et al. 2006), and daily adherence to NPEP may serve as a constant reminder of a prevention “failure” that led them to seek NPEP, increasing the need for cognitive escape in relation to current risk behavior. In the present study, we did not collect data that would allow us to test this hypothesis, but further research is needed to explore psychosocial factors that impact changes in risk taking behavior in response to external events or stressors.
This study is restricted to MSM, and is limited by its relatively small sample size and its lack of ethnic and racial diversity. All of these factors decrease the potential generalizability of our findings. It is also possible that individuals most engaged with HIV-related activities were most comfortable answering questions about their sexual risk-taking, and therefore were most honest in their responses. This study did not specifically include measures of participants’ perceptions regarding HIV-related risk during NPEP, which may have been a critical factor in determining behavior. Additional research is needed to examine these relationships among broader populations, and such research might include biological markers of high-risk behavior, such as screening for sexually transmitted infections (STIs). Previous research has demonstrated that it is clinically important to screen MSM who present for NPEP for infection with other STIs (Hamlyn et al. 2006); these data suggest that it might be worthwhile to screen patients at the end of the treatment period as well.
Most importantly, this study underscores the importance of developing effective behavioral interventions to accompany biomedical prevention strategies. As HIV prevention research and practice move toward biomedical strategies such as vaccines, topical microbicides, and pre-exposure prophylaxis, it will be critical to maintain a focus on the complex psychosocial factors that motivate and sustain traditional prevention behaviors.
This work was supported in part by an educational grant from Gilead Sciences, Inc. and by a grant from The City University of New York PSC-CUNY Research Award Program. The authors gratefully acknowledge the contributions of Chris Grasso, Hilary Goldhammer, Anna Collado, Ronald Bill, two anonymous reviewers for their insightful comments on an earlier draft of this manuscript, and the patients and medical providers of Fenway Community Health.