Sex Trade, Sexual Risk, and Nondisclosure of HIV Serostatus: Findings from HIV-Infected Persons with a History of Alcohol Problems
- First Online:
- Cite this article as:
- Raj, A., Cheng, D.M., Levison, R. et al. AIDS Behav (2006) 10: 149. doi:10.1007/s10461-005-9050-x
- 91 Views
The purpose of this study was to assess the relationships between disclosure of HIV serostatus to sex partners and recent sexual risk behavior, substance abuse, and violence among sexually active HIV-infected patients with a history of alcohol problems. Participants (n = 124) were 79% males; age 25–61 years; 49% Black; and 35% with less than a high school education. Separate logistic regression models were used to assess relationships between each independent variable of interest and nondisclosure. Results demonstrate that buying sex and having more than one sex partner in the past 6 months were significantly associated with nondisclosure of HIV serostatus to a sex partner. Findings from this study underscore the ongoing need for behavioral interventions with HIV-infected individuals concerning disclosure. Programs that emphasize serostatus disclosure and/or consistent condom use in the context of sex trade and with multiple sexual partners will be particularly important.
KEY WORDS:HIVserostatus disclosuresex risksex tradedrug risk behaviorsviolence
Prevention targeting people living with HIV/AIDS is important in reducing the spread of HIV in the United States (Jaffe and Janssen, 2003). Prevention efforts include promotion of sexual risk reduction behaviors such as condom use and serostatus disclosure to sex partners (Centers for Disease Control and Prevention [CDC], 2003). The need for such efforts is supported by research documenting that the majority of people living with HIV are sexually active (Stein et al., 1998) and many of these individuals maintain high risk sexual behaviors including unprotected sex and multiple sex partners (Kalichman and Nachimson, 1999; Kline and VanLandingham, 1994; Marks and Crepaz, 2001; McGowan et al., 2004; Stein et al., 1998). These high-risk sexual behaviors are particularly common among seropositive individuals who are substance users (Kline and VanLandingham, 1994; Marks and Crepaz, 2001; Vanable et al., 2001) and/or involved in sex trade (McGowan et al., 2004).
Studies have shown HIV serostatus disclosure to sexual partners is inconsistent across population groups (Hays et al., 1993; Mansergh et al., 1995; Marks et al., 1991, 1994; Niccolai et al., 1999; Perry et al., 1994; Schnell et al., 1992; Simoni et al., 1995a, b; Sowell et al., 1997; Stein et al., 1998; Stempel et al., 1995; Wenger et al., 1994). U.S. laws often make nondisclosure of HIV serostatus to sex partners a felony (Wolf and Vezina, 2004), but these laws do not appear to be associated with greater disclosure (Duru et al., 2003). More work is needed to identify factors associated with nondisclosure to better support growing prevention efforts.
Consistent condom use among serodiscordant couples has been found to be associated with lower rates of seroconversion for the HIV-negative partner (Weller and Davis, 2002). However, unprotected sex combined with nondisclosure of serostatus to sex partners can result in HIV transmission to those denied the opportunity to make an informed decision regarding use of protection. Nondisclosure and unprotected sex may be particularly likely in the context of a more casual sexual relationship, substance abuse, and/or violence, where one might feel less obligated, able and/or safe to disclose to sex partners (Bedimo et al., 1998; Klitzman and Bayer, 2003). Research has documented increased sexual risk for HIV among those at the intersection of sex trade, substance abuse and violence, where lack of control of one's behavior or sexual encounter can increase the likelihood of unprotected sex and multiple sex partners (e.g., Brown, 1998; Pisani et al., 2003; Robertson et al., 2004; Romero-Daza et al., 2003; Surratt and Inciardi, 2004; Vanwesenbeeck, 2001; Weber et al., 2001). Research has also shown that sex trade, substance use, violence, unprotected sex, and STD continue subsequent to infection for many people living with HIV (Cohen et al., 2000; McGowan et al., 2004; Turner et al., 2001).
The literature shows mixed results with regard to whether serostatus disclosure is associated with condom use. Niccolai et al.’s (1999) study showed a significant relationship between condom use and disclosure for both men and women. Kalichman and Nachimson (1999) found similar findings for men but not for women, and other studies have shown no relationship between disclosure and condom use (Marks and Crepaz, 2001; Stein et al., 1998). In contrast, research has consistently demonstrated that HIV-infected people are more likely to disclose their serostatus to longer term/committed partners as compared with casual partners (Duru et al., 2003; Perry et al., 1994), and disclosure to all sex partners is significantly more common among those with fewer sex partners (Niccolai et al., 1999; Stein et al., 1998). Although sex trade is associated with more and casual sexual partners (Vanwesenbeeck, 2001) as well as less consistent condom use among people living with HIV (McGowan et al., 2004), it has not been examined in terms of its association with disclosure. Given the gender and power differences between those buying as compared with selling sex (Pitts et al., 2004; Vanwesenbeeck, 2001), buying and selling sex must be assessed separately in terms of its relation to sexual risk and disclosure.
Substance abuse is a major issue for HIV-positive individuals; a population-based probability sample of HIV-positive individuals in care found that 46% reported drug abuse in the past year and 36% reported binge alcohol use in the past 4 weeks (Turner et al., 2001). Among HIV-positive samples, those with substance abuse problems are more likely to report unsafe sex practices, such as multiple sex partners, a characteristic related to nondisclosure (Duru et al., 2003; Marks and Crepaz, 2001; Kalichman and Nachimson, 1999; Kline and VanLandingham, 1994; Stein et al., 1998). Additionally, cognitive impairment due to substance abuse could impede ability to disclose. However, prior research has consistently demonstrated no relationship between current substance use and disclosure (Duru et al., 2003; Kalichman and Nachimson, 1999; Perry et al., 1994; Stein et al., 1998). Nonetheless, as previous research has not been restricted to a population with known substance use problems, there remains a need to ascertain whether this relationship exists within an HIV-infected cohort with a history of alcohol problems.
Studies demonstrate that HIV-infected individuals, particularly women, are significantly more likely than those uninfected to have experienced violence and related trauma in their lives, including physical and sexual victimization (Bedimo et al., 1997; Cohen et al., 2000; Department of Justice [DOJ], 2004; Gielen et al., 2000; Liebschutz et al., 2000; Sowell et al., 1999; Zierler et al., 2000). Some researchers have suggested that disclosure may result in violence from the partner or others (Rothenberg and Paskey, 1995), inhibiting individuals’ willingness to disclose. Although Stein et al.(1998) found that among the HIV-infected women, those disclosing to all sex partners were significantly more likely to have had a history of victimization, this violence was not assessed for its temporal relationship to disclosure. Duru et al.(2003) found no relationship between disclosure and partner violence.
Understanding why disclosure to sex partners does or does not occur is an important step in promoting disclosure, as well as sexual risk reduction, among people living with HIV. Overall, our review of the literature points to the need for examination of correlates of nondisclosure for HIV-infected individuals with substance abuse problems. Thus, the purpose of this study was to assess the relationships between disclosure of HIV serostatus to sex partners and behavioral risk (i.e., unsafe sex behavior, substance abuse, and recent exposure to violence) in a cohort of HIV-infected patients with a history of alcohol problems.
Study participants were from the HIV-Alcohol Longitudinal Cohort (HIV-ALC), a prospective, observational study of HIV-infected patients with past or current history of alcohol problems. HIV-ALC participants (N = 349) were recruited and followed-up between July 1997 and July 2001; participants were followed every 6 months for up to seven observations (median number of observations in sample = 2; Ehrenstein et al., 2004). The eligible HIV-ALC participants were also participants of the ADHERE randomized controlled trial (ADHERE RCT), an evaluation of an intervention to increase medication adherence among HIV-infected people with a history of alcohol problems. Data for the current analyses were from the third observation point of HIV-ALC participants and included ADHERE RCT study participants; this observation was the first time point in which disclosure of serostatus to sex partners was assessed. Only participants reporting sexual activity within the past 6 months and responding to the disclosure question at this time point were included in the current cross-sectional analyses (n = 124). Of these, 32% were not in the ADHERE RCT, 31% were in the treatment group of the ADHERE RCT, and 37% were in the ADHERE RCT control group.
Recruitment and Eligibility
Recruitment of HIV-ALC participants occurred by multiple methods and from several sites: Boston Medical Center HIV Diagnostic Evaluation Unit 56%; posted flyers 16%; Boston Medical Center Primary Care Clinic 13%; respite facility for homeless persons 5%; methadone clinic 4%; participant referrals 4%; and Beth Israel Deaconess Medical Center 2%. The majority were recruited from medical settings that addressed HIV-related issues.
All HIV-ALC participants were required to be HIV-infected, age 18 years or older, and with a history of alcohol problems. A history of alcohol problems was defined as having two or more positive responses to the CAGE questionnaire (Ewing, 1984), or, if the patient was recruited from the Boston Medical Center HIV Diagnostic Evaluation Unit and did not meet CAGE criteria, having a diagnosis of alcohol abuse or dependence from one of two attending physicians. Ninety percent of participants (313/349) met CAGE eligibility, and 10% (36/349) met eligibility based on clinical assessment. Diagnostic interviews for alcohol problems in a sample of participants meeting CAGE criteria (N = 141) confirmed that alcohol was a problem for these individuals in that it revealed a lifetime history of alcohol dependence in 80% (113/141) or abuse in 15% (21/141) of cases (Samet et al., 2004).
Other entry criteria for the HIV-ALC study included the following: fluency in English or Spanish; Mini-Mental State Examination (MMSE) score greater or equal to 21 (Folstein et al., 1975); and no plans to move from the Boston area in the subsequent two years. As chronic alcohol use is associated with cognitive impairment, we used the MMSE cut-off of 21 to exclude participants in whom such impairment may preclude obtaining informed consent, an accurate and complete interviewer-administered questionnaire, or a follow-up interview (Smith et al., 2003).
After screening for eligibility and obtaining informed consent, a research associate interviewed participants in either English or Spanish using a standardized instrument to ascertain baseline information on demographics, health and medical care, HIV risk behaviors, and drug and alcohol use. For the Spanish interview, standardized scales in Spanish were used when available; the remainder of the questionnaire was translated from English into Spanish, back-translated to check for accuracy, and then corrected. Participants were then followed-up with surveys every 6 months for up to 36 months. Although biological data were not included in current analyses; they were collected for use in this study. Attempts were made to obtain CD4 cell counts and HIV RNA (viral load) levels from all participants. These data were obtained via laboratory tests performed within 6 months of the interview, as part of patients’ clinical care; if these tests were not available, blood samples were obtained from participants and tested for CD4 cell count and HIV RNA using the Boston Medical Center Clinical Laboratory.
Demographic characteristics, including age, gender, race, and sexual orientation (dichotomized as gay/lesbian/bisexual vs. heterosexual) were measured via single items.
Nondisclosure of Serostatus to sex partners was measured via a single item created for use in this survey, “Have you told any of the following people that you are HIV-infected?’’ For the response related to sexual partners of the past 6 months, data were dichotomized as nondisclosure to one or more sex partners of the past 6 months versus disclosure to all sex partners of the past 6 months.
Sexual Risk Variables were measured via items from the Risk Assessment Battery (RAB), which is designed to measure sexual and drug use risk for HIV (Navaline et al., 1994) and has been used with substance using and HIV-infected populations in previous studies (Ehrenstein et al., 2004; Rees et al., 2001). Only sexual risk items from the RAB were included in analyses. All items assessed risk in the past 6 months and were dichotomized. These items included inconsistent/no condom use (vs. consistent condom use), number of sexual partners (one vs. two or more), selling sex for money or drugs, and buying sex with money or drugs. Additionally, STD in the past 6 months was measured via self-report.
Substance Abuse in this study was defined as binge alcohol consumption or illicit drug use in the past 30 days. Both of these variables were calculated using the alcohol and drug use frequency questions from the Addiction Severity Index (ASI; McLellan et al., 1992). Illicit drug use was defined as having used heroin, other opiates/analgesics, barbiturates, cocaine, or amphetamines in the past 30 days. Binge alcohol consumption in the past 30 days was defined as having five or more drinks on an occasion for males or four or more drinks on an occasion for females.
Victimization from interpersonal violence within the past 6 months was measured via two yes/no items, one assessing physical abuse and the other assessing sexual abuse. Physical abuse or assault was defined as having been “kicked, hit, choked, shot, stabbed, burned, or held at gunpoint by a stranger, a family member or someone you know.” And sexual assault was defined as “unwanted sexual touching anywhere on your body, touching of genitals and/or breast, or made to have oral sex or vaginal or anal intercourse against your will by force or the threat of force by a stranger, a family member or someone you know” (Liebschutz et al., 2000).
Two sample t tests and chi-square tests were used to assess bivariate relationships between demographics and disclosure. Separate logistic regression analyses were performed to evaluate the associations between each independent variable of interest (sexual risk behaviors, STD, victimization, and substance use) and disclosure. Unadjusted models and adjusted models controlling for age, gender, race, sexual orientation, and RCT involvement and group assignment (as some of this cohort participated in the ADHERE RCT study) were fit to the data. The ADHERE RCT study involved evaluation of a medication adherence intervention that was not designed to affect HIV disclosure to sex partners or high risk sexual behaviors. The intervention did not show increased medication adherence among treatment as compared with control participants (Samet et al., in press), but there was the possibility that the intervention or participation in the RCT could affect nondisclosure. A chi-square analysis was used to assess the relationship between RCT involvement/group assignment and nondisclosure; a significant association was not observed. However, in an additional effort to avoid potential confounding, we controlled for RCT involvement and group assignment in all adjusted analyses. All initial analyses were conducted with two-sided tests of hypotheses at the .05 significance level. In addition, the Bonferroni method was used to adjust for multiple comparisons.
Participants (n = 124) were aged 25–61 years (mean 42 years), predominantly male (79%), racially mixed (49% Black), one-third without high school graduation (35%), and a minority homeless (14%). Almost one-third of participants (n = 38) identified as gay, lesbian, or bisexual (n = 28 gay or lesbian, 10 bisexual), only five of these (13%) were females.
Cohort Characteristics: Nondisclosure, Sexual Risk Behaviors, Victimization, and Substance Use
Demographics for Total Sample of HIV-Infected Persons with a History of Alcohol Problems (N=124) and Frequency of Reporting Nondisclosure vs. Disclosure by Demographics
Nondisclosers (n = 40)
Disclosers (n = 84)
χ2 or t-test value (df)a
0.2 (1, 123)
1.6 (1, 123)
0.003 (1, 123)
Non-high school graduate
Sexual orientation (%)
Gay or bisexual
Age, years, M (SD)b
0.8 (1, 122)
Victimization was less common than sexual risk behaviors. Recent sexual victimization was reported by 3% of the cohort, and recent victimization from physical abuse by 8%. Over one-third of the sample reported binge drinking (39%) and illicit drug use (40%) in the past 30 days.
Age, gender, race, and education were not associated with disclosure of HIV serostatus (Table I). However, participants identifying as gay, lesbian, or bisexual were significantly more likely not to disclose to all sex partners than those identifying as heterosexual (53% and 22%, respectively; p = .002). Further analyses were conducted to assess whether effects were attributable to bisexual (n = 10) as compared with gay/lesbian (n = 28) participants by conducting a Fisher's exact test to assess differences; these groups did not significantly differ in terms of probability of nondisclosure. Additional analyses stratifying prevalence of nondisclosure among gay, lesbian, and bisexual participants by gender demonstrate that two of the five lesbian or bisexual women reported nondisclosure as compared to 16 of the 33 gay or bisexual men.
Nondisclosure of Serostatus—Multivariate Associations
Participants reporting multiple sex partners in the past 6 months, ORadj = 8.9, 95% CI = 3.4–23.4, and sex bought with drugs or money, ORadj = 1.5, 95% CI = 2.6–50.9, were significantly more likely than those not reporting these behaviors to not disclose to a sex partner (Table II). The association between sex bought and nondisclosure pertains only to men as sex purchase was only reported by men in our sample. Adjusting for multiple comparisons using the Bonferroni method did not change the results of the logistic regression analyses as multiple sex partners and buying sex remained significant predictors of nondisclosure.
Prevalence and Crude and Adjusted Odds Ratios Assessing Associations Between Nondisclosure of HIV Serostatus by Sexual Risk Behaviors and STD, Substance Abuse, and Victimization (Independent Variables)
Nondisclosers (n = 40)
Disclosers (n = 84)
Crude OR (95% CI)
Adjusted OR (95% CI)a
Sexual risk (%)
Multiple sex partners
One sex partner
Inconsistent/no condom use
Consistent condom use
No sex sold
No sex bought
Substance abuse (%)
No binge drinking
Illicit drug use
No illicit drug use
No sexual abuse
No physical abuse
To assess whether sex trade was significantly associated with nondisclosure after controlling for multiple sex partners, a logistic regression model was created that included both of these variables as well as demographics (age, gender, race, sexual orientation) and RCT involvement and group assignment. Multiple sex partners (ORadj = 6.7, 95% CI = 2.4–18.4) remained significantly associated with nondisclosure and sex bought (ORadj = 5.0, 95% CI = 1.0–23.5) became marginally significant in this multivariate model.
Approximately one-third (32%) of HIV-infected persons with a history of alcohol problems did not disclose their HIV serostatus to one or more sex partners. The percentage of nondisclosure is comparable to that previously reported by HIV clinic samples using similar questions and timeframes (Duru et al., 2003; Stein et al., 1998). The finding that one-third of HIV-infected patients did not disclose their HIV serostatus to all sex partners is problematic, particularly given the fact that 39% of nondisclosing participants did not use condoms consistently. Similar to previous studies (Duru et al., 2003; Stein et al., 1998), condom use was not related to disclosure. Given that most of our sample was tied into HIV clinical care for 6 months or more at assessment, results from this study support the need for better sexual risk reduction programs promoting condom use and disclosure of serostatus to sex partners within the clinical care setting.
Findings from this study additionally demonstrate that the odds of nondisclosure of HIV serostatus were significantly higher among participants who had multiple sex partners in the past 6 months and participants who had purchased sex in the past 6 months. Although previous studies have demonstrated that nondisclosure is more common among those reporting multiple sex partners and within the context of more casual sexual relationships (Duru et al., 2003; Niccolai et al., 1999; Perry et al., 1994; Stein et al., 1998), this study offers the first demonstration that those buying sex with drugs or money are significantly more likely to report nondisclosure. Further, an association between buying sex and nondisclosure is observed even after controlling for multiple partners. Notably, this association is only indicative of men, as only men in our sample reported buying sex. Nonetheless, these results, as well as the trend suggesting an association between nondisclosure and selling sex, support the idea that disclosure may be less likely in the context of more casual sexual encounters in which one may feel less obligated or safe disclosing to their sex partner (Klitzman and Bayer, 2003).
Findings from this study further suggest that less disclosure among those reporting sex trade and multiple sex partners is not counteracted by increased condom use. As mentioned above, disclosure is not related to condom use; additionally, study findings suggest higher odds of nondisclosure among those reporting recent STD. Those reporting a recent STD had almost 6 times the odds of nondisclosure as compared with those not reporting a recent STD. The lack of statistical significance of this finding despite the large effect size is likely due to the small number of participants reporting recent STD. Overall, these results speak to the need for prevention programs that focus on promotion of condom use during sex trade and within casual sexual relationships, regardless of willingness or comfort disclosing to these partners.
Consistent with previous research (Duru et al., 2003; Stein et al.1998), recent drug and alcohol use were not related to disclosure. Victimization from physical abuse was also not related to disclosure, but victimization from sexual abuse was. Participants reporting sexual abuse had about six times the odds of nondisclosure of HIV serostatus. Again, these findings were not statistically significant likely due to small numbers of participants reporting victimization from sexual violence. Although previous studies have identified high rates of sexual abuse among people living with HIV/AIDS (Gielen et al., 2000; Sowell et al., 1999; Zierler et al., 2000), research has not specifically assessed sexual abuse and nondisclosure. Given the lack of choice given to people being victimized in a sexual attack, opportunity for disclosure may not occur or may place the individual at greater risk.
Overall, these results support the need for interventions to facilitate disclosure and condom use among HIV-infected persons with a history of alcohol problems, particularly those reporting sex trade and multiple sex partners. Given recent study demonstrating that a second infection with a different HIV-1 strain can occur and accelerate disease progression among people living with HIV (Allen and Altfeld, 2003), increasing consistent condom use with this population will reduce risk for both the infected individual and their partner.
There are certain limitations to the current study. These include reliance on self-report, potential social desirability and recall biases, and use of a modest sample size; cross-sectional data analyses also preclude assumptions of causality. Additionally, our sample cannot be generalized to HIV-positive individuals as a whole; however, it does appear to be generalizable to sexually active HIV-infected persons with alcohol problems (Samet et al., 2004). These limitations speak to the need for further longitudinal studies of nondisclosure to sex partners among HIV-infected individuals with and without a history of alcohol problems. Additionally, use of larger samples in future research would allow further exploration of the effects of both gender and sexual orientation.
An additional limitation of this study was the use of a dichotomous measure of nondisclosure. Participants reporting nondisclosure to one sex partner is categorized the same as those reporting nondisclosure to all sex partners, but associations between disclosure and sex sold may differ between these groups; our current measure of disclosure precludes our ability to assess such associations. Our measures also did not assess whether nondisclosure was specifically occurring in the context of non-condom use, sex trade, substance abuse or violence; further research is needed to assess the contexts of nondisclosure. Recent qualitative research with gay men in San Francisco indicates that HIV-infected men disclose their HIV serostatus in adherence with community norms, but unprotected sex was more rather than less likely in the context of their disclosure (Sheon and Crosby, 2004). Further research is needed to assess the context of nondisclosure in more detail.
Data management was provided by DM-STAT, Inc., Medford, MA. The authors appreciate the contributions of the clinical staff of the HIV Diagnostic Evaluation Unit at Boston Medical Center, including Colleen LaBelle, RN, and Jennifer Doyle. We also acknowledge Dr. Anita Palepu and Dr. Theresa Kim for their feedback on this work. Finally, the authors appreciate the contributions of Howard Libman, MD, at the Beth Israel Deaconess Medical Center. This research was conducted in part in the General Clinical Research Center at Boston University School of Medicine, USPHS grant M01 RR00533. Development of this work was supported by National Institute on Alcohol Abuse and Alcoholism (NIAAA): RO1-AA13766 (Clinical Impact of HCV and Alcohol in HIV-Infected Persons), RO1-AA11785 (Medication Adherence in Alcohol Abusing HIV Patients), RO1-AA10870 (Enhanced Linkage of Alcohol Abusers to Primary Care).