Portal de Periódicos da CAPES

Community randomized trials for HIV prevention

2003; Lippincott Williams & Wilkins; Volume: 17; Issue: 18 Linguagem: Inglês

10.1097/00002030-200312050-00014

1473-5571

Michel Alary, Catherine M Lowndes, Marie‐Claude Boily,

HIV Research and Treatment

In this issue, Orroth et al. [1] present a simple and plausible hypothesis that could explain the apparently contradictory results of the Mwanza [2], Rakai [3] and Masaka [4] community randomized trials of sexually transmitted infections (STIs) treatment as an intervention to prevent HIV transmission. This hypothesis emerged from a careful re-analysis of the baseline data from each trial population, showing higher risk behaviour and higher rates of curable STIs in Mwanza compared to Rakai and Masaka. An important next step would be to model the three trials on simulated populations mimicking those of each site, to verify if these differences could explain the contradictory results. The hypothesis that a more mature epidemic in Uganda than in Mwanza accounts for the observed differences was put forward early on in this debate [5,6]. The study by Orroth et al. [1] suggests a mechanism through which secular changes in the population brought about by a maturing epidemic will affect the capacity of STI control to impact the HIV epidemic. Indeed, the progress of the epidemic may lead per se to a reduction in risky behaviour and STI prevalence through: (1) genuine changes in sexual behaviour due to a combination of increased awareness and fear when seeing significant numbers of people dying of AIDS in one's surroundings; (2) selective morbidity and mortality among high-risk individuals due to AIDS. The respective roles of these two mechanisms still have to be studied adequately and may well vary with the ongoing evolution of the epidemic and population studied. Nevertheless, both will effectively result in reductions in risky behaviour, which will likely lead to a decrease in short-duration curable STIs. Although the role of the latter in HIV transmission will remain important at the individual level, it will become less important at the population level, as shown by secondary analyses of the Mwanza and Rakai studies on population attributable fractions of HIV associated with treatable STIs [7,8]. Consequently, the potential impact of STI control on HIV transmission at the population level will be greatly reduced. Several lessons can be drawn from the controversy surrounding the results of the Mwanza, Rakai and Masaka trials and from the analyses that have been carried out to explain these differences. These lessons relate to issues such as the preparation and design of community-based randomized trials, the role of observational and modelling data in monitoring the impact of interventions and the importance of tailoring HIV preventive interventions according to the local epidemiological context. First, the execution of a community-based randomized trial is a major endeavour, involving huge human and financial resources. It is thus very important to understand the potential mechanisms through which a given intervention to be tested may be effective. In addition to the collection of high quality descriptive data on the study population, which is already the rule in the preparedness phase of trials, the use of mathematical modelling prior to the conduct of the trial could prove useful to assess the potential impact of the intervention studied given the local epidemiological context, and to improve study design. Such an approach may be needed because of the very complex nature of the transmission dynamics of infectious diseases in general, and of HIV in particular, at the population level [9–11]. Mathematical modelling has already contributed to our understanding of the differences between the Mwanza and Rakai trials [12–14] and to improvements in the design, analysis and interpretation of phase III HIV vaccine trials [15–18]. Second, since it is not feasible to test all interventions – or to test a given intervention in many different settings – and since there is a need to monitor the impact of the global effort, alternatives to randomized community-based trials are needed. Sentinel surveillance of HIV infection implemented in most African countries has proved of limited utility to monitor the impact of interventions and, recently, WHO and UNAIDS have proposed the use of second generation surveillance [19], incorporating data from multiple sources and including data on HIV and STI prevalence as well as on behaviour. We argue that the use of detailed descriptive data over time combined with mathematical modelling could greatly enhance our capacity to monitor both the epidemic itself and the impact of interventions put in place. Indeed, in Cotonou, Benin, where comprehensive descriptive data are available on the general population [20] as well as on core [21] and bridging [22] populations, it has been possible to show empirically that the majority of cases of HIV infection in the general population are linked to transmission within the core, and between the core and the general population, through the bridge [23]. Mathematical modelling applied to this population has strongly suggested that an intervention specifically targeting the core and bridging populations could greatly reduce overall HIV transmission at the community level [24]. To enhance the value of the modelling approach, better communication between modellers and field epidemiologists would improve the relevance of data collection for models and thus lead to more accurate estimates of parameters of particular importance for the models, such as rates of partner change, sexual mixing patterns [25] and concurrency [26], in different segments of the population. Finally, one of the most important lessons we can draw from the Mwanza, Rakai and Masaka trials is the importance of tailoring interventions according to the stage of the HIV epidemic. The general concept of phase specific strategies for the prevention and control of STIs was first put forward by Wasserheit and Aral in 1996 [27] and has been widely discussed subsequently [28]. Although multiple strategies are needed and should be carefully balanced in every setting, it may well be that the main priority in contexts similar to that described above for Cotonou, which prevail in many West African and Asian urban centres, would be core group preventive interventions, including STI care and intensive condom promotion. Although core group interventions remain important at all stages of the epidemic, additional actions may however be needed in situations similar to that found in Mwanza in the early 1990s, where transmission within the general population is increasing and/or where core groups are difficult to identify, as in many rural settings. Here the priority may be the introduction of syndromic management of STIs at the primary health care level, including appropriate risk-reduction counselling for STI patients, combined with behaviour change approaches at the community level. In settings with considerably more mature epidemics, as in Uganda, interventions to prevent HIV transmission in stable relationships, such as voluntary counselling and testing with special emphasis on couples, male circumcision, herpes simplex virus-2 suppression and antiretroviral therapy, may also become necessary, while awaiting effective HIV vaccines and vaginal microbicides. Unfortunately, in some areas of the world with exploding and non-levelling HIV epidemics combined with high rates of STIs in all segments of the population, as in many parts of Southern Africa, it may be extremely difficult to prioritize as the implementation of comprehensive interventions using all possible strategies is urgently needed.