Concurrency and the limited effectiveness of behavioural interventions on sexual risk behaviour of youth in sub-Saharan Africa

In our manuscript ‘Effectiveness of HIV prevention for youth in sub-Saharan Africa: systematic review and meta-analysis of randomized and nonrandomized trials’ we reviewed 28 interventions aiming to reduce sexual risk behaviour of youth in sub-Saharan Africa [1]. Of eight outcome variables studied (five behavioural and three biological outcomes), only condom use at last sex among men was found to be significantly influenced by the interventions (relative risk = 1.46; 95% confidence interval = 1.31–1.64). Helen Epstein, in her commentary to this review [2], posits that the lack of inclusion of concurrency in these HIV prevention programmes may account for their failure. In the studies reviewed, detailed information on the interventions was often lacking and concurrency might well have been left out. Although the majority of evaluations assessed the number of sexual partners (lifetime or in the last weeks/months), only five included outcome measures related to multiple partnerships, with only one having an adequately specific indicator of concurrency [3]. The measures used include: multiple partners in the last 6 months [4], multiple regular partnerships in the last 3 months [5], multiple sexual partners during the last 3 months [6], sticking to one partner as a prevention method [7] and ongoing nonprimary partners (makhwapheni) since last interview [3]. The study of Visser [6] in South Africa surprisingly showed that the number of learners who had multiple sexual partners during the last 3 months increased in both the intervention and control groups from pretest to posttest and in the experimental group this increase was statistically significant. By contrast, the intervention led to a reduction in multiple regular partnerships in the study in Zambia by Agha and Van Rossem [5], and Kim [7] reported that an intervention in Zimbabwe led to more youths sticking to one partner. A second study of Visser [4] did not detect a change in the number of partners, whereas Jewkes et al.[3] did not specifically report changes in nonprimary partners. One study assessed perceptions towards men having concurrent partnerships, asking respondents their opinion on the statement ‘Men need to have more than one sexual partner, often at the same time’. No effect was noted in the intervention arm [8]. Finally, in another study, the intervention group did not have an increased awareness that ‘not having multiple partners' was a prevention method for HIV transmission [9]. Overall, in these evaluations, as elsewhere, there is marked variation in how long-term concurrency is measured [10]. Hopefully, the recent UNAIDS consultation on concurrent sexual partnerships will assist in standardizing these indicators [11]. Mostly, the researchers dichotomized relationships as being either with a ‘steady’ or ‘casual’ partner, the latter being portrayed as considerably more risky and promiscuous. This distinction and the significance placed on it complicate HIV prevention efforts and is alluded to by Helen Epstein: an undue focus on the dangers of casual partnerships alone risks conveying a message that long-term steady relationships are safe. In reality, they may well be most hazardous. Though there is much evidence that concurrency is important in HIV transmission dynamics, there is much uncertainty about the portion of HIV infections attributable to this factor [10,12,13]. In our view, the projects failed not only because of a lack of focus on concurrency, but more importantly because they were often poorly implemented. Implementation difficulties were pervasive across studies and manifested as: refusal of teachers to talk about condoms, resource constraints and nonadherence to project design. In addition, interventions lacked grounding in conceptual frameworks of behaviour change. Moreover, notwithstanding the inherent complexities of behavioural research, overall, we could not detect a clear progression in the design of interventions, where subsequent interventions build upon the previous. Ultimately, it is possible that some interventions were doomed from the outset, simply because it is immensely difficult to alter individuals' sexual behaviour in the presence of static community norms. Even the two most well elaborated interventions and rigorous evaluations, including biological outcomes [3,14] resulted in small behaviour changes and no changes in HIV or pregnancy incidence. Jewkes et al.[3] did report a reduction in herpes simplex virus 2, but the only positive sexual behaviour change detected was a reduction in transactional sex in men. As Helen Epstein infers, long-term relationships– and we would add all types of sexual relationships– are difficult to alter through knowledge, raising awareness and skills learning, owing to the underlying social, emotional and economic dimensions of sex, which vary markedly between settings. Interventions, thus far, perhaps were not sufficiently cognisant of local community norms, be it concerning concurrency, condom use or other behaviours. Changing these community norms will require more than a behavioural intervention targeting individual youth; a shift in paradigm may be necessary, perhaps from a focus on the individual to a scenario of greater community responsibility.