Our study has described the sexual risk-taking behaviour profile of potential Phase I and II HIV vaccine trial volunteers undergoing prescreening in Soweto, South Africa for the first HIV vaccine trials being conducted in Southern Africa. We designed a survey appropriately adapted for local use, which is an important component of measuring risk behaviour in international HIV vaccine trials.2,20
We assessed risk behaviour by self-reported condom use and self-reported numbers and types of sex acts, which have been shown to be appropriate, reliable and accurate indicators for the risk of acquiring an STD.21,22
We administered this survey as a component of the prescreening protocol, a generic protocol applied to all HIV-negative potential volunteers expressing an interest in participating in various Phase I/II HIV vaccine trials at the vaccine trial site in Soweto.
South Africa is extraordinarily diverse, with 11 national languages and a variety of cultures. Although risk behaviours have been well-documented on a national level, little has been published on Soweto in particular, and even less is known about the residents of Soweto who actually volunteer for HIV prevention trials. In our study, we showed this to be a young, predominantly single and unemployed, multi-ethnic population, of whom half reside in households with incomes below the poverty line.
Our analyses demonstrate that men in this study reported higher levels of risk behaviour than females overall during the previous six months, as indicated by the higher numbers of total sex partners, casual or anonymous sex partners, sex partners known or suspected to be HIV-positive and ‘one night stands’, as well as the higher levels of heavy alcohol and recreational drug use. In adjusted models, men were five times more likely than women to have casual or anonymous sex partners and 12 times more likely to have three or more sex partners. While it is possible that men throughout Soweto have higher levels of risky sexual behaviour than women, it is also possible that men with higher risk behaviour or women with lower risk behaviour were more likely to attend VCT services and subsequently enrol in this screening protocol for future HIV vaccine trials. Regardless of the explanation for these findings, the trend for higher risk behaviours among male participants may indicate that they should be targeted for increased risk-reduction counselling during the course of future African HIV vaccine trials.
Male gender was not the only predictor of higher levels of reported risk behaviour. Reporting any incidence of heavy alcohol use, choosing a partner without regard to personal safety or HIV status, and reporting at least one partner who travelled away from home for employment were also significant predictors of having had sex with a casual or anonymous partner. Any incidence of heavy alcohol use, reporting one or more partners with an age difference of at least 15 years, and reporting at least one partner who travelled away from home for employment were additional significant predictors of having had three or more sex partners. These outcomes are consistent with research showing that South African youth with partners with an age difference of only five years were at increased risk of HIV infection23
and that rural South African migrant men and their partners were at an increased risk of contracting STDs compared with non-migrant men.24
In adjusted models, reporting any recreational drug use significantly predicted having had any incidence of unprotected vaginal or anal sex, as well as having had any unprotected sex with a known or suspected HIV-positive partner, in the previous six months. Female gender and age greater than 25 years were also significant predictors of having unprotected sex. However, female gender and older age are not necessarily associated with increased risk behaviour, because this relationship might indicate married older women having unprotected sex with monogamous HIV-negative partners. But this association may be an indication of a form of ‘passive’ high-risk behaviour, in which married older women, for example, are forced into having unprotected sex with husbands who subsequently infect them with HIV,25
a concept reinforced by studies showing that married older women are more likely to use condoms inconsistently.26
Refinement of the survey tool to gather information regarding whether participants engaged in unprotected sex of their own volition or by force would help to clarify this association further.
Although some of these predictors occurred more frequently in male participants, they were independently associated with higher risk behaviour and might also be used to target vaccine trial participants for increased risk-reduction counselling. In particular, the substance use predictors were associated with all four classifications for increased risk behaviour and may be particularly useful in risk stratification of vaccine trial volunteers: participants reporting any history of recreational drug use were twice as likely to have had unprotected sex and five times as likely to have had unprotected sex with a known or suspected HIV-positive partner, while participants reporting any history of heavy alcohol use were three times as likely to have had sex with casual or anonymous partners and three times as likely to have had more than two partners in the previous six months.
A study of the risk behaviour patterns in another South African cohort of participants, observed for evaluating HIV incidence and study retention rates in preparation for HIV vaccine trials, showed that the cohort engaged in higher levels of sexual risk behaviour than similar participants from the same community.11
It is difficult to ascertain how closely our sample represents the sexual risk-taking behaviour of the general population of Soweto. There may be a potential bias in participant enrolment, as participants in our study were initially recruited because they had sought HIV VCT and were subsequently found to be HIV-negative. Compared with the general population, our methods may have selected a group of participants who practice riskier sex, because individuals at greater risk for HIV infection sought VCT services, or a group of participants who practice safer sex, because individuals at lower risk for HIV infection sought assurance that they were not infected. Regardless, it underscores the need to characterize the baseline characteristics of the actual populations that are participating in HIV vaccine trials, which was the intent of this study, as well as to understand how this population differs from the potential target population for an eventual vaccine.
There are several limitations that are inherent to the survey procedures used in this study. Both the assessment of personal information in a face-to-face interview and the risk-reduction counselling offered with VCT services prior to risk behaviour survey administration may have influenced the social desirability bias of reported behaviours. In addition, although the counsellors administering the survey were periodically trained and monitored, the administration of the survey by multiple interviewers over time and in a mixture of languages (English, Sotho, Zulu) might have varied between participants. These problems could be resolved with the future use of audio computer-assisted self-interviewing (ACASI) technology. ACASI has been shown in a longitudinal randomized trial within an HIV risk behaviour study to improve behaviour data collection, likely by decreasing social desirability bias, removing interviewer bias and error, and standardizing the interview questions.27
Another limitation is that participant recall bias might prevent accurate risk assessment, although studies have shown that structured interviews to recall information for the previous three to six months are preferable to recalling shorter or longer time periods.28
A further limitation is that behaviour may be accurately reported by the participant, but may not completely represent true risk. For example, a person might report 100% condom use but not have used condoms correctly during that time, which would under-represent risky sexual behaviour.
Suitable methods for assessing risk are particularly important in communities where baseline sexual practices and customs are not well-known. Research on risk behaviour and HIV vaccines must proceed together, and risk behaviour education must be an integral part of any future HIV prevention programme that includes an HIV vaccine.3,4
Our study provides the first sexual risk behaviour data for an African population participating in HIV vaccine trials to date as well as a baseline reference for determining whether sexual risk behaviour might change in response to participation in South African HIV vaccine trials. In addition to increased risk-reduction counselling for the prevention of HIV transmission in targeted groups, the predictors of higher risk behaviour described in this study might also be used to differentiate between participants more suitable for Phase I and II safety and immunogenicity studies, which require volunteers at relatively low risk for HIV infection, or Phase III efficacy studies, which require volunteers at relatively high risk for HIV infection.