Abstract
The objective of the study was to assess the level at which individuals in HIV discordant couples engage in concurrent sexual partnerships and factors associated with this risk of transmitting HIV. By using data from a group of HIV discordant heterosexual couples (n = 145), we examined the frequency of concurrent sexual partners and factors associated with such partnerships. The prevalence of concurrent partnerships with heterosexual partners was 16%. Fewer than half of individuals with concurrent partnerships reported that their main study partner knew about these relationships. Of individuals involved in concurrent partnerships, 30% reported inconsistent condom use in these relationships. Unmarried individuals in new HIV sero-discordant relationships were particularly at risk for concurrent partnerships. In conclusion, more frequent HIV testing may complement increased attention to communication, disclosure and condom use in this population that is especially vulnerable to acquiring and transmitting HIV infection.
Keywords
INTRODUCTION
Several researchers have examined population patterns of sexually transmitted infections (STIs) empirically, as well as conceptually and methodologically, and have found concurrent partnerships to be an especially efficient way to transmit STIs. 1–5 Although the prevalence of concurrency has been examined in several populations, it has not been considered in a population of HIV discordant heterosexual couples. 6,7 HIV-negative partners in concurrent partnerships, if unknowingly infected by one partner, may spread the infection more readily to other partners because primary HIV infection (i.e. infection within the last 18 months as compared with chronic HIV infection) is associated with much higher infectiousness. 8–11 On the other hand, HIV-positive partners who have unprotected sex pose transmission risk to all their sexual partners, even those who are HIV positive who may put themselves at risk for super-infection. 12
This study examines aspects of concurrent sexual partnerships with heterosexual partners among individuals in a group of HIV discordant heterosexual couples, drawing from cross-sectional baseline data collected as part of the California Partners' Study II, conducted in San Francisco from 1996 to 1999.
METHODS
Study population
The California Partners' Study II was a randomized behavioural risk reduction intervention trial. Sexually active HIV sero-discordant heterosexual couples in the San Francisco Bay Area were recruited between November 1996 and July 1999 through the media, advertising at HIV-related agencies and clinical venues. Couples were eligible for enrolment if partners were of the opposite sex and HIV serostatus; both were 18 years and older and spoke English or Spanish. The University of California, San Francisco Committee on Human Research and University of California, Berkeley Institutional Review Board approved the protocol.
Measures
Outside partners
If a participant reported at least one episode of vaginal or anal sex in the previous six months with members of the opposite sex who were not their study partner, he or she was considered to have a concurrent sexual partnership. All individuals who had an outside sexual partner engaged with an outside partner after the date of first sex with the main partner. Same sex concurrent relationships were not included in our analyses because of small numbers, only two men and one woman had same sex concurrent partnership, and the fact that different risk factors for concurrent partners may exist with these types of relationships. The median number of concurrent heterosexual partners was determined by taking the median value of responses to the following question: ‘In the last six months, not including your study partner, how many [partners of the opposite sex] have you had vaginal, anal or oral sex with?’ Although oral sex was included in the measurement of median number of concurrent partners due to the way the questionnaire was structured, we included only data from questions regarding anal or vaginal sex in our analysis as we were considering sexual behaviour, which significantly increased HIV transmission risk.
Condom use in the previous six months
To assess sexual risk behaviour, respondents were asked the following questions separately about vaginal and anal sex with both study partners and concurrent partners: ‘How often have you used condoms during sex in the last six months: never; less than half the time; about half the time; greater than half the time; or always?’ The outcome variable ‘inconsistent condom use’ was created by combining the two variables on vaginal and anal sex, and was made dichotomous as ‘always use condoms during anal or vaginal sex’ versus ‘less than always’.
Other variables that characterize study participants
Eleven attributes, either demographic characteristics or factors potentially linked with HIV risk, of participants in our study were used in our analyses. Basic demographic categories included gender and race/ethnicity (Caucasian, African American, Hispanic/Latino, Asian, Native American or mixed). Marital status (married to the study partner or not) was also included. Individuals reporting at least one sexual partner of the same sex in his or her lifetime were classified as having a history of bisexual behaviour. The variable ‘history of exchanging sex for money and goods’ was defined as whether or not the individual reported ever trading sex for drugs, goods or money. Substance use was assessed for the previous six months and was classified as frequent alcohol use (defined as consumption of more than two glasses of alcohol at least twice a week); crack use; and injection drug use. Three variables (age, educational status and duration of the primary relationship) were made into dichotomous variables in the following manner. Age was made into a dichotomous variable: age <40 or age ≥40. Educational status was presented as high school graduate and above, or less than high school graduate. Duration of relationship (with study partner) was also made into a dichotomous variable that considered whether the relationship was of one year or greater or less than one year duration.
Only one attribute of an individual was not determined from his or her self-report. This was the variable ‘non-monogamous study partner’, which was created from the report of an individual's partner on the number of concurrent partners in the past six months.
Variables that characterize outside partners
Characteristics that related to concurrent sexual partnerships were determined solely from information asked of participants. The variable ‘exchanging sex for money and goods with outside partner’ was ascertained by asking individuals ‘have you ever had sex with X [outside partner] in exchange for money, food, clothes, drugs, or a place to stay?’ The variable ‘sex with the outside partner for greater than 1 year’ was determined by subtracting the year of first sex with X [outside partner] from the study date and creating a dichotomous variable. The variable ‘high during sex with outside partner’ was taken directly from a question asking ‘Did you ever have sex with outside partner when intoxicated or high?’ The variable ‘will have sex with the outside partner again’ was created with responses from the same question. The variable ‘know that outside partner is infected with HIV’ was created from ‘Is [outside partner] infected with HIV?’, and was created by separating positive responses to this question from all other responses.
Analytic methods
Baseline frequencies of key variables relating to concurrent heterosexual partnerships were measured. Two-by-two frequency tables and associated odds ratios (OR) were used to examine demographic attributes and behaviours associated with reporting concurrent heterosexual partners. Multivariate analyses were completed using logistic regression models, controlling for potentially confounding variables.
For those individuals reporting more than one concurrent heterosexual partner, information about the second concurrent partner was analysed as a separate observation along with the information from the first concurrent partner for these individuals (detailed information was collected on up to two concurrent partners for each individual). Such repeated observation was accounted for using generalized estimating equations. All analyses were conducted using SAS version 8.2 (SAS Inc, Cary, NC, USA).
RESULTS
Demographic characteristics of the 290 (145 couples) study participants are summarized in Table 1. Forty-seven (16% of our study group) participants reported having a sexual relationship with a partner of the opposite sex in the previous six months who was not their study partner (Table 2). No significant difference was discerned between men and women in the reporting of concurrent partners. Twenty-six of these individuals (55% of those reporting concurrent partners) reported more than one concurrent partner. The median number of concurrent partners was one for women and two for men.
Baseline characteristics of individuals in the study populations
Characteristics of concurrent heterosexual partnerships within the previous six months (No. = 47)
A number of interesting aspects of these concurrent partnerships are further revealed in the descriptive data shown in Table 2. Of individuals with concurrent partners, 31% reported exchanging sex for drugs, goods or money with the concurrent partner. More than half (57%) of individuals with concurrent partners reported having sex with a specific concurrent partner while high or intoxicated. Of individuals with concurrent partners, 55% reported having sex with a concurrent partner for more than one year, although only 15% thought that they would probably have sex again with that same concurrent partner. Of individuals with concurrent partners, 40% had non-monogamous study partners as determined from the partner's report, although fewer than half (48%) of these individuals knew that their main partner was non-monogamous, according to the report of the non-monogamous partner.
Of the individuals in the study reporting concurrent partnerships, 69% reported having one or more heterosexual concurrent partners and reported consistent condom use during vaginal or anal sex with these partners. However, only 41% of individuals with concurrent partnerships disclosed their HIV status or risk (i.e. the fact of being in a main partnership with an HIV positive partner) to their concurrent partners prior to engaging in sex with these partners. Of individuals with concurrent partnerships, 19% reported knowing that their concurrent partner was HIV positive.
As shown in Table 3, the odds of an individual having a concurrent partner were increased by having a ‘history of bisexual behaviour’ (OR, 3.5; 95% confidence interval [CI], 1.7, 7.1), by having a ‘study partner relationship for <1 year’ (OR, 5.2; 95% CI, 2.5, 10.9) and by having ‘a non-monogamous study partner’ (OR, 4.0; 95% CI, 2.0, 8.0). ‘History of bisexual behaviour (OR, 2.4; 95% CI, 1.3, 4.6), and ‘study partner relationship for <1 year’ (OR, 3.1; 95% CI, 1.4, 7.1), remained statistically significant after adjusting for potential confounders in the multivariate model. The odds of an individual having a concurrent partner were decreased by ‘being married’ (OR, 0.11; 95% CI, 0.03, 0.36) and by having more education (secondary education and up) (OR, 0.33; 95% CI, 0.11, 0.97). Of these variables, only ‘being married’ remained statistically significant after adjusting for potential confounders in the multivariate model (OR, 0.18; 95% CI, 0.05, 0.62).
Characteristics associated with individuals who have concurrent heterosexual partners
Race/ethnicity (no bivariate measure as multiple groups)
*Indicates statistical significance
DISCUSSION
The California Partners' Study II population is a distinctive United States urban population that showed a relatively higher level of several attributes linked with the risk of HIV infection compared with other populations: exchanging sex for money or goods; a history of bisexual behaviour; crack cocaine use; frequent alcohol use; and injection drug use. The sample had a high rate of unemployed, a majority of African Americans and Caucasians and most were married.
The prevalence of concurrent heterosexual partnerships was 16%. This is more than four times higher than the prevalence reported among married men and women in the United States population in the early 1990s, and close to 50% higher than reports on men in the United States. 13,14 Approximately one-third of concurrent partnerships involved exchanging sex for money or goods, and slightly more than one-half of individuals with concurrent partners reported having sex on at least one occasion with a concurrent partner when the individual was intoxicated or ‘high’. Fewer than half of concurrent partnerships were known to the individual's main partner, and more than half of these concurrent partnerships had gone on for more than one year. However, interestingly, the probability that the individual thought he or she would have sex with the concurrent partner again was only about 15%.
As with all such behavioural studies, certain limitations are inherent in the data and methods. One of the main limitations of this study was the primary variable examined here – outside relationships are subject to ‘social desirability bias’, which is when individuals report what they perceive to be the socially desirable response rather than what actually took place. This bias may be especially important when individuals discuss concurrent partners, as having concurrent partners may not be considered socially acceptable. To encourage more honest reporting, partners were interviewed separately, and they were ensured at each visit that none of their information would be shared with their study partner.
In our study of this urban, economically disadvantaged and ethnically mixed population, we have found a number of factors significantly associated with concurrent partnerships that appeared to affect the risk of further spread of HIV infection. A history of bisexual behaviour and being in the study partner relationship less than one year increased the risk. In contrast, being married was strongly associated with a reduced likelihood of having a concurrent partner.
Two final aspects of our findings suggest potential foci for further research and intervention. First, fewer than half of individuals with concurrent partnerships reported that their main study partner knew about these relationships. The heightened risks of HIV infection and other STIs posed by these concurrent partnerships make this observation of special concern. A partner's unsuspected concurrent partnership can put an unknowing individual at higher risk for a number of reasons, including, of course, that this individual may be less likely to take precautions in the primary partnership. As several authors have suggested, fostering and reinforcing that certain risk behaviours are not only personally unwise, but also raise the burden and effects of disease for all could help reduce these risks by decreasing the burden and effects of disease for all could help reduce these risks by decreasing the practice of concurrent partnerships. 15,16 Cassel et al. 15 noted that these intervention messages and efforts appear to have reduced the practice of concurrent partnerships in other settings such as Uganda. Recently, Dunkle et al. 17 have advocated accompanying HIV testing and counselling among cohabiting couples in sub-Saharan Africa as an added strategy to reduce heterosexual transmission.
Finally, the sharply heightened risk of HIV transmission present during primary HIV infection, also discussed in the Introduction, implies the need for special consideration of more frequent and repeated HIV testing among individuals in HIV sero-discordant couples and their concurrent partners. 18 In our population, this would apply especially to unmarried individuals in new HIV sero-discordant relationships, who show a higher rate of concurrent partners and who are thus doubly exposed. More generally, more frequent testing may complement increased attention to communication, disclosure and condom use, to help those most vulnerable in populations such as the one we have described in this paper to reduce their HIV risk.
Footnotes
ACKNOWLEDGEMENTS
This research was supported by cooperative agreement U64/CCU912270 from the US Centers for Disease Control and Prevention. The authors thank the California Partners Study II participants and staff for contributing to this research, to Nancy Padian and Cynthia Gomez for their contributions as co-PIs and to Steve Shiboski for his statistical consultation. For A van der Straten research work for this study was conducted at the University of California, San Francisco, Department of Obstetrics, Gynecology and Reproductive Sciences. For A Eaton work for this manuscript was conducted while at the University of California, Berkeley Department of Epidemiology, USA.