Diagnosed HIV-infected people form an increasingly large sub-population in South Africa, one that will continue to grow with widely promoted HIV testing and greater availability of antiretroviral therapy (ART). Factors that have been identified as being associated with unsafe sexual behaviours in HIV-infected people include: age, income, depression, drug and alcohol use[3, 4], low self-efficacy, attitudes to condoms and suprainfection, conception, duration of relationship, intimate partner violence[8, 9], perceived stigma, and disclosure to family and partner(s)[10–12]. The relative influence of risk factors may change over time in a population of HIV-infected people, for example, as a result of changes in sexual behaviour norms. The determinants of unsafe sexual behavior can also change over time for a HIV-infected person depending on factors including the length of time since HIV diagnosis, the quality of counselling and support, and the HIV status of their partner. Many HIV-infected people are in HIV discordant couples for whom practicing safer sex consistently over a long period may be challenged by numerous factors including emotional and sexual intimacy issues, poor communication, and stresses including concerns about family disclosure and childbearing[2, 14–16].
For HIV-infected people, experiences associated with morbidity and HIV treatment may have important influence on their family and partnering relationships, and sexual behaviours. ART reduces infectivity, however, it also increase the duration of potential infectiousness by reducing morbidity and mortality, and may also result in risk compensation. Thus, efforts to promote safer sex (for example, reduction of concurrent partnerships, reduction of the number of partnerships, condom use, abstinence) in people receiving HIV treatment and their partners remains an important public health goal.
Several direct and indirect effects of ART have been identified on psycho-sexual functioning [18–20]. Side effects, in particular lipodystrophy, may result in sexual dysfunction. Low self-efficacy is associated with poor adherence[22–24], as well as, a higher risk of unsafe sexual behaviours. However, there is a pressing need to monitor sexual risk behaviours of HIV-infected people over time to measure the full extent of long-term ART on sexual behaviour and HIV transmission[26–28]. Studies in Europe, US and Africa have reported that HIV-infected individuals receiving ART do not have increased sexual risk behaviour[29–34]. Yet, in order to inform secondary prevention efforts, greater knowledge about family and partnering contexts, issues related to stigma and HIV disclosure, and characteristics associated with unsafe sexual behaviours of HIV-infected people is needed. This information is especially important in sub-Saharan Africa where the HIV epidemic has been most severe and public access to HIV treatment is relatively recent compared to developed countries.
A group of HIV-infected people little studied with respect to their sexual behavior are HIV-infected individuals who have accessed ART services but who are not yet ART-eligible. This group are difficult to prospectively study as they have high attrition from HIV clinics, and the period of follow-up after diagnosis of HIV may be short as individuals initiate ART or die. Nonetheless it is important to learn more about this group for two reasons. First, HIV prevention strategies may need to be specifically tailored for people who know their HIV positive status but do not yet require treatment. Second, this group provides an appropriate comparison for studying the impact of ART initiation on sexual behaviour.
In Africa, the few studies that have sought to measure the impact of ART initiation on sexual behavior vary widely in their design. In a study by Bunnell et al in Uganda, there was no comparison group with which to compare the observed changes in sexual behaviour in HIV-infected people receiving ART. One South African study used two separate cross-sectional samples of HIV-infected people waiting to start ART to control for sexual behaviour trends in a cohort of people initiating ART. A limitation of repeated samples is that each sample may differ in important ways that are unmeasured. Consequently, a lack of difference in sexual behaviours between the two cross-sectional samples may be an artifact of the sampling process. Other cohort studies in Africa have compared sexual behavior pre-ART with sexual behavior reported after ART initiation[31, 32]. In these studies, the same individual may therefore contribute to pre- and post-ART exposure groups. A limitation of this approach is that for each individual, the pre-ART period will always be earlier than their post-ART period, and thus, any secular changes in sexual behaviours during the period of study might erroneously be interpreted as an impact of ART. Adjusting for time in the cohort or conducting sensitivity analyses provides information with which to assess whether study results may have been influenced by factors other than ART. In our study of long-term sexual behaviors of participants in a rural ART programme in South Africa, we used a different study design; that of prospectively observing sexual behavior of pre- and post-ART groups simultaneously. Our control group consists of people not yet eligible for ART who are followed longitudinally at the same time as a group starting ART.