Levels of risky sexual behavior among men and women initiating antiretroviral therapy (ART) generally remained below baseline values during a three-year study in Uganda.¹ Although the proportion of participants who had had sex in the past three months increased from 28% to 41% during the study, the proportion of sexually active individuals who had recently had risky sex fell sharply at first and, despite a modest rebound, remained below initial levels. Overall, the estimated transmission risk to partners who were HIV-negative or of unknown status declined by 91%.

Because ART reduces the viral load of people with HIV, it should reduce the risk of transmission. However, one concern has been that as an ART user's condition improves, he or she might engage in risky sexual behavior with greater frequency, diluting or perhaps eliminating the reduction in transmission risk. To examine this issue, researchers analyzed clinical and behavioral data from a randomized trial conducted in Tororo, Uganda, in 2003–2007. Clients of a home-based AIDS program who lived within the program's catchment area were eligible for the study if they had never used ART and they had AIDS symptoms, a CD4 count of 250 or fewer cells per microliter or both. All participants received free ART and clinical care during the three-year trial, and during the first year they attended counseling sessions in which they developed plans for reducing their risk of transmitting HIV to others. Counselors discussed such options as abstinence, condom use, reducing the frequency of sex and fulfilling sexual desires without engaging in intercourse; they also provided free condoms on request. Additional informational sessions were provided to serodiscordant couples.

The main purpose of the trial was to compare outcomes across three clinical monitoring approaches. The current analysis, how-ever, focused on data from interviews conducted at baseline and every six months thereafter. Participants were asked about their sexual desire, opportunities to meet partners, sexual activity, condom use, partner type and partner HIV status. In addition, the investigators measured participants' plasma HIV levels. Respondents were classified as having had risky sex if they had used condoms inconsistently or not at all during vaginal sex with a partner who was HIV-negative or of unknown status. The investigators used logistic regression to identify predictors of having had risky sex, and estimated participants' transmission risk using the behavioral and clinical data in tandem with probability data from prior seroconversion studies. They assumed a condom failure rate of 20%.

Baseline data were available for 928 men and women. After exclusion of 94 respondents who died during the trial and 79 who did not provide complete behavioral data at the three-year follow-up, the analytic sample consisted of 755 respondents, about three-fourths of whom were female. Mean age at baseline was 37 for women and 41 for men. At baseline, most men were married or cohabiting (68%); the majority of women were widows (60%), though 24% were married or cohabiting.

Levels of sexual activity increased steadily during the trial. The proportion of respondents who had had sex in the past three months increased from 28% at baseline to 41% at 36 months; among respondents who were married, cohabiting or in a steady relationship, the proportion who were sexually active increased from 65% to 86% among men, and from 59% to 79% among women. At both the beginning and the end of the trial, only 6% of respondents had had more than one partner in the past three months; half of these nonmonogamous respondents were polygamous men whose only additional partner had been a wife.

Measures of risky sex often showed a different pattern, typically declining sharply at first (coinciding with the trial's behavioral interventions) and then increasing. For instance, the proportion of sexually active respondents who had had unprotected sex in the past three months with a partner who was either HIV-negative or of unknown status declined from 6% at baseline to 2% at six months; by 36 months, the proportion had returned to 6%. The only factor associated with risky sex in regression analyses was alcohol use, defined as either partner having used alcohol before having sex (odds ratio, 2.9).

Condom use varied according to partners' serostatus. At the end of the study, for example, 74% of respondents with HIV-negative partners reported always using a condom, compared with 46% of respondents with HIV-positive partners.

Clinical tests indicated that participants' virus levels declined sharply during the trial. For example, among respondents who had had sex with partners who were HIV negative or of unknown status, the proportion with undetectable viral loads increased from less than 1% at baseline to 98% at 36 months; the proportion with CD4 cell counts of 200 or higher increased from 20% to 95%. From such improvements, together with the changes in participants' risky sexual behavior, the researchers estimated that among sexually active individuals, the risk of transmission to HIV-negative partners and partners of unknown status fell from 47 to four transmissions per 1,000 person-years, a 91% decline. Of the 62 respondents whose partners underwent annual voluntary testing during the trial, only one transmitted the virus to a partner.

The study had several limitations, the ?authors say. It relied on self-reports of sexual activity, and respondents may have been reluctant to report unsafe behavior. In addition, the findings may not be generalizable to other settings, particularly countries where marriage rates are lower and levels of concurrent partnerships higher than in Uganda. Nevertheless, the study provides evidence that "integrated ART and prevention programs may help to reduce HIV transmission in Africa," according to the authors. Although the apparent benefits of the trial's behavioral interventions seemed to wane in the trial's later stages, rates of risky sex generally "did not increase above baseline levels," countering fears that new users might engage in risky behaviors as their condition improved and their sexual activity increased. Moreover, the investigators note that levels of risky sex were lowest in the earliest stages of treatment, when viral loads (and thus the potential for transmission) were highest, underscoring that behavioral interventions for ART users may be particularly important right after initiation of therapy. —P. Doskoch

REFERENCE

1. Apondi R et al., Sexual behavior and HIV transmission risk of Ugandan adults taking antiretroviral therapy: 3 year follow-up, AIDS, 2011, 25(10):1317–1327.