The United Nations Joint Programme on HIV/AIDS (UNAIDS) and WHO estimate that 2.7 million people were newly infected in 2008, 2.3 million of whom were adults . All of these people will eventually need long-term HAART; a burden that will continue to stretch or outpace global health systems, especially in the low and middle-income, high-prevalence countries of East and Southern Africa, in addition to the burden on the individuals concerned, their families, and communities. UNAIDS estimate that 1.1 million HIV-positive individuals were newly started on treatment in 2008, whereas 2.0 million HIV-positive people died that year (1.7 million adults) . So, HIV treatment programmes, despite their huge efforts and amazing achievements, were overall only making a rather marginal gain on the future HIV burden, and that only because of the substantial number of HIV-positive people who died, largely because of lack of access to HAART. Clearly, primary prevention of HIV infection must remain a top priority for public health in the second decade or the 21st century, especially in the high HIV prevalence countries of Eastern and Southern Africa.
HIV prevention efforts should target all three modes of transmission of HIV: parenteral (mainly through injections contaminated with HIV-infected blood); vertical (from an HIV-infected mother to her child either in utero, during delivery, or through breastmilk); or sexual. But the balance of effort needs to be driven by a clear understanding of the HIV epidemic in the specific context of the programme. The great majority of new HIV infections are occurring in East and Southern Africa , where most new infections among adults are thought to be due to heterosexual transmission , and the great majority of infections among children are due to vertical transmission . The focus of this paper will be on the effectiveness of interventions to reduce risky sexual behaviours among adults and, hence, to prevent HIV, both among adults and, later, in children by fewer of their mothers being HIV-infected. Antiretrovirals given to the mother and the infant around the time of birth can also be used to substantially reduce vertical transmission.
Types of sexual behaviour change interventions
The dynamics of sexual HIV transmission are summarized by the standard formula for infectious diseases, which, in its simplest form is given by:
Equation (Uncited)Image Tools
where R 0 is the basic reproduction number, c is the rate of sexual partner change, β is the transmission probability per sexual partner, and D is the duration of infectiousness. New individuals can only become infected sexually if they are in discordant partnerships, in which one partner is HIV-positive and the other is HIV-negative.
At least in theory, behavioural HIV prevention interventions can focus on either c, the rate of sexual partner change, or on β, the transmission probability per sexual partner. Examples of behavioural interventions to reduce the rate of sexual partner change include interventions to increase the age of sexual debut (to reduce the total lifetime available to accumulate sexual partners), and interventions to reduce the number of partners. Behavioural interventions to reduce the transmission probability include those focussing on reducing the cofactor effects of other sexually transmitted infections (STIs), those focussing on the type of sex act, and those focussing on the number of sex acts. The first of these include educating and promoting the seeking of early, effective treatment for STIs. The second group of interventions includes encouraging condom use, discouraging unprotected anal sex in which the average risk of transmission per sex act is much greater than for unprotected vaginal sex, avoiding unprotected sex with partners who are more likely to be HIV-positive (e.g. ‘core groups’), avoiding having concurrent partners because of the increased likelihood of one of the partners having a recent infection when the risk of transmission is greatly elevated, and replacement of penetrative sex acts by nonpenetrative. The third group of interventions focus on reducing the number of unprotected, penetrative sex acts per sex partner.
In practice, the main sexual behaviour change interventions that have been implemented include:
1. Community-wide sexual health education
These include mass media, political speeches and rallies, and community health days.
2. Adolescent sexual health interventions
Interventions in schools and colleges, interventions to promote the use of youth-friendly health services, and community-wide youth interventions that include out-of-school youth.
3. Interventions among groups most at risk
For example, interventions among sex workers, intravenous drug users, and men who have sex with men (MSM).
4. Promotion of HIV testing and counselling (HTC)
Linked to counselling of HIV-negative individuals to help them to stay HIV-negative, and counselling of HIV-positive individuals to prevent onward transmission to their sexual (or injecting) partners.
5. Interventions among HIV-positive individuals
Including couples counselling and active condom promotion.
6. Interventions to promote the use of high-quality STI treatment services.
Targeting of behaviour change interventions by age and sex
HIV prevention programmes have tended to focus primarily on interventions to increase the chances of an uninfected person remaining uninfected. To be most effective, behaviour change programmes need to take both perspectives: changing behaviours in HIV-negative partners to reduce their risk of HIV acquisition and changing behaviours in HIV-positive partners to reduce their risk of onward HIV transmission. The latter is sometimes known as ‘positive prevention’.
Measures of HIV incidence will be critical to understanding where interventions to reduce HIV acquisition need to be targeted, with interventions starting before HIV incidence rates peak and then being sustained and reinforced during the period of high incidence, and being most intensive in populations such as the urban, and those with behaviour patterns that put them at particularly high risk such as sex workers, MSM, the highly mobile, and injecting drug users. Figure 1 shows national estimates of HIV incidence by age and sex from a recent HIV study  in South Africa. These imply that behaviour change interventions against HIV acquisition must be particularly effective within 20–29-year-old women and 20–39-year-old men because these are the age ranges in which HIV incidence peaks in South Africa. Whereas Fig. 1 showed HIV incidence rates, Fig. 2 shows the HIV prevalence data from the same national study . Interventions to reduce HIV transmission must focus on the age groups in which HIV prevalence is high. In the context of South Africa, this would imply that the key age range for such interventions would be among the much wider age range of 20–49-year-old women and 30–49-year-old men.
Effectiveness of behaviour change interventions for HIV prevention
The results of a systematic review of phase IIB and III randomized controlled trials (RCTs) of interventions to prevent sexually transmitted HIV in nonpregnant populations with laboratory documented HIV as an outcome by Wasserheit  was presented at the International Society for Sexually Transmitted Diseases Research (ISSTDR) Conference in London in June 2009. She identified 36 RCTs of 38 different interventions. The author classified eight of these as being trials of sexual behaviour interventions. These were the Masaka Trial in Uganda (Masaka) , the MEMA kwa Vijana Trial in Mwanza, Tanzania (MkV) [5–8], the Regai Dzive Shiri Trial in Zimbabwe (RDS)  (Cowan FM, et al., unpublished mimeo), the Stepping Stones Trial in South Africa (Stepping Stones) [10,11], the Zimbabwe VCT Trial (VCT) , the SHAZ Trial (SHAZ) [13,14], the EXPLORE Trial in USA (EXPLORE) , and the Mexico Female Sex Worker Trial (Mexico) . An additional trial of a multicomponent intervention was identified in discussions with Richard Hayes (personal communication): the Manicaland Trial in Zimbabwe (Manicaland) . A further trial in Tanzania (the SATZ Trial) has also evaluated the impact of a multicomponent behaviour change intervention among young people in South Africa and Tanzania [18,19], but has not yet reported the results of the impact on HIV and other biomedical outcomes.
Selected intervention design characteristics of these nine trials are summarized in Table 1. Seven trials were conducted in Africa [four in Zimbabwe (Manicaland, RDS, VCT, and SHAZ), and one each in South Africa (Stepping Stones), Tanzania (MkV), and Uganda (Masaka)], whereas two were conducted in the Americas [one each in USA (EXPLORE) and Mexico (Mexico)]. All but one (SHAZ) of the seven African trials were targeted to members of the general population, rather than to groups at particularly high risk of HIV. Four of the African trials (MkV, RDS, Stepping Stones, and SHAZ) had young people as their primary target group, of which three (Masaka, MkV, and RDS) had a specific intervention component for young people attending schools, and one of these (MkV) restricted the evaluation to young people who had attended these schools. The Stepping Stones Trial intervention required young people to volunteer for an intensive series of learning activities. The Masaka and Manicaland trials targeted all young to middle-aged adults in whole communities, whereas the Zimbabwe VCT trial included employees of 22 businesses in the capital city. The SHAZ trial specifically targeted the interventions and evaluated the outcomes in a group known to be at particularly high risk of HIV: out-of-school orphan adolescent girls. Similarly, both trials in the Americas were among groups at high risk: urban MSM in the EXPLORE trial and female sex workers in the Mexico Border Cities trial. Five of these trials, all in Africa, evaluated multicomponent interventions (Masaka, Manicaland, MkV, RDS, and SHAZ).
Six of the seven African trials used a cluster-randomized design (Masaka, Manicaland, MkV, RDS, Stepping Stones, and VCT); the exception was the recent SHAZ trial where participants were individually randomized. Both trials (EXPLORE and Mexico) in the Americas used individual randomization.
Three of the trials (Manicaland, MkV, and RDS), all in Africa, implemented no additional interventions in the comparison arm over and above preexisting standard programmes. In the Masaka trial, community-based interventions that were not thought likely to affect HIV incidence were implemented in the comparison arm communities, whereas in the Stepping Stones trial, volunteers were offered a single 3-h session, whereas those in the intervention were offered a total of 48 h of sessions in addition to a community meeting. However, in the SHAZ trial, the comparison group were offered a substantial intervention. They received the same 4 weeks of life skills education and health training given to those in the intervention arm, but did not receive the additional 6 months of vocational training and business start-up grant. The Zimbabwe VCT Trial randomly allocated businesses. In the intervention arm, all staff members were offered free on-site, same-day voluntary HIV counselling and testing (VCT), whereas those in the comparison arm were offered a voucher to obtain free off-site VCT. In practice, 71% of staff in the intervention (on-site testing) businesses accepted testing and received their results, whereas only 5% did so in the comparison (off-site testing) businesses. Thus, this trial was able to compare the impact of a high uptake of VCT with a low uptake. The EXPLORE trial in USA compared an intensive (intervention arm) with a much less intensive series of counselling sessions (comparison arm), whereas the Mexico Border Cities Trial compared a single 35-min behavioural session including motivational interviewing with a single didactic session of the same duration, and, therefore, tested the difference between interventions of equal duration but different approaches.
It is clear from this summary that the trials studied different interventions, compared them against different comparison interventions, used different designs (e.g. cluster vs. individual randomization), and took place in different settings. A meta-analysis would, therefore, be inappropriate.
Table 2 summarizes the duration and size of follow-up and the main trial results in terms of the impact on HIV. The duration of follow-up varied from 6 months (Mexico) to over 8 years (MkV), with most studies falling within the 1.7–4-year range. There was also a very wide range in terms of the number included in the evaluation, ranging from 315 (SHAZ) to 13 814 (MkV), Although it is not possible to calculate the effective power that each trial had to detect a given reduction in HIV, it is clear that at least three trials (VCT, SHAZ, and Mexico) were not designed to be able to detect even a 50% reduction, and very few (perhaps only MkV) would have had adequate power to detect a true reduction of 40% or less. Yet, it is arguable that even a reduction of 25% would be of public health importance. In any event, none of the trials showed a statistically significant impact on HIV, either in terms of a reduction or an increase; in fact, none approached statistical significance with the possible exception of the EXPLORE trial [adjusted odds ratio = 0.82, 95% confidence interval (CI) 0.64–1.05]. The point estimates for the impact all varied between 0.82 and 1.27, with the exception of the Zimbabwe VCT Trial which had a somewhat worryingly increased HIV incidence in the intervention (high VCT uptake) arm (adjusted rate ratio = 1.49), but with very wide 95% CI (0.79–2.80).
It should be noted that although these trials were disappointing in terms of the impact on HIV of the behavioural interventions that were evaluated, some trials showed an impact on other sexual and reproductive health outcomes. The most important of these are noted in Table 2. For example, the MkV Trial showed significant improvements in sexual and reproductive health knowledge, reported attitudes, and some, but not all, reported sexual behaviours both in the follow-up conducted about 3 years after the start of the interventions and after more than 8 years. However, it did not demonstrate any consistent impact on herpes simplex virus 2 (HSV2) or other STIs. On the contrary, the Stepping Stones Trial showed a statistically significant decrease in HSV2 of one-third, and the Masaka Trial also reported a borderline significant reduction in HSV2.
Interpretation and potential explanations
There are three main potential reasons for these disappointing findings: the interventions tested were inherently ineffective, they were inadequately implemented, or there were problems with the measurement of effectiveness. It is clear that at least two of the studies suffered from inadequate implementation. In the Masaka Trial, some aspects of the in-school intervention, including the teaching on condoms, were not delivered in the majority of schools, at least partly because the teachers were expected to teach the classes outside normal school hours . A different problem affected the Regai Dzive Shiri Trial, in which it appears that the interventions per se were both well designed and well implemented. However, the original intention was to evaluate the effectiveness of the interventions in a cohort of young people recruited at baseline and then particularly targeted for the interventions. However, an interim follow-up survey after about 2 years showed that attrition from the cohort was likely to be unacceptably high, so the design was changed to a cross-sectional survey of all young people living around the centre of the intervention community. This meant that the primary trial outcome had to change from HIV incidence to HIV prevalence, but, more importantly, subsequent analysis showed that the exposure to the various components of the intervention had been rather low in the evaluation group (Cowan et al. 2008, Table 1). Any analysis restricting to those with adequate exposure might introduce bias and would negate the advantages of the randomized design. These two examples demonstrate that large behavioural intervention trials face important challenges wherein the realities of intervention and evaluation implementation can sometimes lead to compromises. Such trials necessarily tend towards the evaluation of effectiveness rather than efficacy.
It should also be remembered that none of the trials evaluated the intervention in a vacuum. HIV prevention efforts have been in place in all countries since the early stages of the epidemic, though of varying natures and intensities, so all participants in both arms of the trial will have been exposed to some other HIV prevention interventions. In some trials (e.g. Stepping Stones, VCT, SHAZ, EXPLORE, and Mexico), these were further reinforced in the comparison arm by additional HIV prevention interventions, though either of a different intensity or approach to those in the intervention arm. Behaviour change interventions may take several years to have a major impact. Also, their effect may not be linear. There may even be a tipping point before which there is very little impact and after which there is a major impact. Yet, most of these trials have evaluated impact at one or a limited number of points in time, usually with follow-up being limited to 2–3 years .
Despite these important caveats, it is clear that the results of these trials of behavioural interventions are not encouraging in terms of their effectiveness against HIV. However, we know that true behaviour change (such as always using a condom with a discordant partner) would reduce HIV incidence. Also, population-level changes in behaviours have been associated with reductions and increases in HIV incidence in observational studies (such as in Uganda ). The problem appears to be two-fold. First, it is likely that none of the behavioural interventions that have been tested to date within large-scale trials with HIV as an outcome have had a substantial enough impact on HIV risk behaviours. But also, several of the trials that have been performed have either suffered from problems with intervention implementation, coverage, or have been underpowered to be able to detect a realistic effect size on HIV.
In general, trials that evaluated behaviour change against behavioural outcomes have had more encouraging results than those using HIV or other biological outcomes . Yet, it is essential that, if interventions are to recommended because of their effectiveness for HIV prevention, then trials must include HIV and, wherever possible, other biological outcomes such as other STIs and pregnancy. Trials with reported sexual behaviours as their outcome are insufficient for three main reasons:
1. Because of the limited validity of reported sexual behaviour, particularly in young people [24–26]. In some settings, immaculate conceptions appear to be a common event when reported sexual behaviours are validated against biological markers of pregnancy, and many sexually transmitted infections would have had to have been acquired nonsexually .
2. Because social desirability bias in reporting of sexual risk behaviour is likely to be greater at follow-up in the intervention arm. For example, if one has been telling people in the intervention arm that they should use a condom each time they have sex, they are more likely to misreport having used a condom at follow-up than those in the comparison arm.
3. Because the incidence or prevalence of other STIs, let alone reported sexual behaviours, may not be good surrogates for HIV .
The way forward: research
Despite the discouraging results of the trials reviewed above in terms of the impact of the behavioural interventions evaluated against HIV, effective ways of preventing HIV and STIs are urgently needed, and research to develop and evaluate such interventions should remain a high priority, including research on potential behavioural interventions. Several of these trials and others, which did not measure HIV as an outcome , have shown that, at the least, such interventions can increase knowledge on HIV, its modes of transmission, and methods of prevention, and can impart the skills needed to change behaviours such as negotiation of safe sex and sexual refusal skills.
Future trials to evaluate the effectiveness of behavioural interventions should have HIV as an outcome and be realistically powered to be able to detect a difference of 25% in HIV incidence, allowing for the fact that HIV incidence may well be lower in the comparison arm that predicted by preliminary studies. This means that they will need to be both large and expensive. This has not proved to be an insurmountable problem for funders of trials of biomedical interventions such as HIV vaccines, microbicides, and male circumcision; behavioural trialists must convince funders that their trials are of similar importance and merit adequate funding with long enough follow-up to produce reasonably unequivocal results. They must also include detailed process evaluation and collect data on secondary outcomes on the theoretical pathway by which the behavioural intervention is thought to have its effects on HIV. Very careful thought needs to be given to the choice of what will be done in the comparison arm of the trial. If this is excessive, then the chance of the trial being able to show a significant difference will be reduced.
Although such RCTs will continue to have an important place, especially if large enough and with long enough follow-up, intervention research should not be limited to trials. Studies of processes of change that aim to advance knowledge of the mechanisms by which interventions may work to change behavioural risk-taking, and which attempt to identify key mediators of behaviour change will also be important .
In terms of the interventions themselves, particular attention needs to be given to the social and sexual norms of the general population among whom the particular target group (such as young people or groups most at risk) live and interact. For young people, this should include the people who either control or influence their behaviours such as their parents and other adult relatives, older role models and influential people in the community, and external influences such as radio, films, television, and songs.
Also, most interventions evaluated to date have targeted HIV acquisition among the HIV-uninfected individuals; equal attention needs to be given to evaluating interventions that focus on prevention of HIV transmission from HIV-infected individuals. As discussed above, interventions to reduce HIV acquisition will need to target young people, and then be maintained thereafter, whereas interventions to reduce HIV transmission will need to primarily target young and middle-aged adults rather than adolescents. The latter will include interventions to reduce their number of sexual partners, and particularly their concurrent partners, and to increase their use of condoms even within long-term relationships. It will also be important to evaluate multicomponent behavioural interventions that target both HIV acquisition and HIV transmission.
Research is also needed on how the more widespread availability of HAART and antiretroviral therapy (ART) for prevention of mother-to-child transmission (PMTCT) affects the effectiveness of behavioural interventions among HIV-positive adults and discordant couples to reduce HIV transmission, and hence how best to design these interventions. Changing access to HAART will also complicate evaluations that use HIV prevalence as an outcome, as effective HAART programmes will lead to increasing HIV prevalence as HIV-related mortality declines, unless this is coupled with reductions in HIV incidence due to reduced viral load in those on treatment.
For technical reasons, mass media approaches have not been evaluated within the RCTs reviewed above. Yet this does not mean that they cannot be evaluated, at least in terms of their impact on knowledge, report attitudes, and reported self-efficacy, using carefully designed before-after studies .
Another area needing further research is the impact of HTC on HIV incidence, given the worrying findings from the Zimbabwean trial (VCT) described above and the results from many other studies in Africa and elsewhere which have shown that although reported sexual risk-taking tended to be reduced after VCT among those who tested HIV-positive, it tended to either be unchanged or was increased in those who tested HIV-negative [31–33]. A recent meta-analysis  of seven intervention studies (most of which were before–after or time series studies) of the impact of VCT on condom use and/or reported number of partners reached more encouraging conclusions, but data on the individual studies included in the meta-analysis again shows that although risk behaviours were more likely to decrease after VCT among those who tested HIV-positive or among discordant couples who tested together, the results were not clear cut among those who tested HIV-negative . Could it be that HTC is good for persuading HIV-positive individuals to reduce their risk of onward transmission of HIV to their partners, but leads to relative disinhibition among those who test HIV-negative along the lines of ‘I have taken risks in the past and have not been infected, so maybe I can continue with my previous behaviours’? But perhaps it is also that the posttest counselling for those who test HIV-negative is often too perfunctory or is not heard by the client who has just been told that they are HIV-negative?
Finally, we need to explore the potential for alternative study designs to the RCT with HIV as the primary outcome, exploiting the potential for careful plausibility evaluations within ‘natural experiments’ when programmes are being started, and continuing the search for cheaper but valid surrogate markers of HIV. Although RCTs will remain the most rigorous and convincing intervention study design, they are not always feasible either technically (e.g. to evaluate mass media interventions) or because of funding, and other study designs have a potential contribution to make as long as they are carefully designed to try to rule out extraneous reasons for the effects observed, and their limitations are acknowledged .
The way forward: programmes
Given the burden of the HIV epidemic in high-prevalence countries, decisions about the implementation of programmes, including interventions that aim to reduce behavioural risks of HIV, will not wait for further research results to become available.
In the meantime, programmes should continue to promote in-school interventions to increase the knowledge and skills of young people, not because they are likely to have a direct impact on HIV incidence in the short-to-medium term, but as a long-term investment as they will help to provide the background level of knowledge and may influence the general population attitudes related to sexual risk and HIV. Similarly, interventions to improve the quality of the sexual and reproductive health services provided by health facilities and to increase their utilization by clients deserve continued support, as they have been shown to be capable of success at least in terms of these limited outcomes .
Condom uptake remains pitifully low in most countries, especially those with high HIV prevalence . Much more strenuous efforts need to be made to both promote their use and to make them ubiquitously available.
Although the importance of concurrent partnerships is widely acknowledged, interventions that aim to specifically target this pattern of sexual relationship are still largely in their infancy, and their introduction should be linked to careful evaluation.
It is widely acknowledged that HIV prevention is likely to require multiple strategies, leading to calls for ‘combination prevention’ approaches. These are discussed in detail in the paper by Hankins and de Zalduondo in this issue .
Finally, although this article has focused exclusively on interventions that solely or largely rely on the promotion of low-risk sexual behaviours among the population at risk of either HIV acquisition or HIV transmission, it is important to realize that biomedical and structural interventions will also require behaviour change. For example, for male circumcision, a biomedical intervention, to be successful in prevention of HIV, men will need to come to health services to request the procedure, to abstain from sex for a period of several weeks after the operation, and then not to risk compensate by either increasing their number of sexual partners or reducing their condom use, for example. Similarly, ARTs for the PMTCT of HIV prevention will require high adherence to the medications by the clients.
This paper draws heavily on a literature review that was reported by Judy Wasserheit at the ISSTDR Conference in London in June 2009. It also draws on discussions with my colleagues in the Mwanza NIMR/LSHTM Research Team and collaborators over many years, and discussions held at the Mwanza Intervention Trials Unit (MITU) Scientific Symposium in Mwanza in July 2009.
This paper is dedicated to the people of Mwanza Region who have participated in the many HIV prevention research studies that have taken place over the past 20 years, and which have advanced our knowledge on what works, or, all too often, what does not work in terms of interventions to reduce behavioural HIV risk.
David Ross is a staff member of the London School of Hygiene and Tropical Medicine. No specific external funding was used to support the preparation of this paper, though his attendance at the MITU Symposium was supported by the MkV Trial Long-term Evaluation Project funded by the UK Department for International Development and Irish Aid. (Irish Aid is the name of the Irish Government's Aid Agency.)
1. UNAIDS/WHO. AIDS epidemic update 2009 [UNAIDS/09.36E/JC1700E]. Geneva, Switzerland: UNAIDS and WHO; 2009.
2. Rehle T, Shisana O, Pillay V, Zuma K, Puren A, Parker W. National HIV incidence measures: new insights into the South African epidemic. S Afr Med J 2007; 97:194–199.
3. Wasserheit JN. STI/HIV prevention research: the agony, the ecstasy, and the future. Plenary Presentation at the 18th International Society for STD Research Conference; 30 June 2009; London, UK.
4. Kamali A, Quigley M, Nakiyingi J, Kinsman J, Kengeya-Kayondo J, Gopal R, et al. Syndromic management of sexually-transmitted infections and behaviour change interventions on transmission of HIV-1 in rural Uganda: a community randomised trial. Lancet 2003; 361:645–652.
5. Obasi AI, Cleophas B, Ross DA, Chima KL, Gavyole A, Plummer ML, et al. Rationale and design of the MEMA kwa Vijana adolescent sexual and reproductive health intervention in Mwanza Region, Tanzania. AIDS Care 2006; 18:311–322.
6. Hayes R, Changalucha J, Ross DA, Gavyole A, Todd J, Obasi AIN, et al. The MEMA kwa Vijana Project: design of a community-randomised trial of an innovative adolescent sexual health intervention in rural Tanzania. Contemp Clin Trials 2005; 26:430–442.
7. Ross DA, Changalucha J, Obasi AI, Todd J, Plummer ML, Cleophas-Mazige B, et al. Effects of an adolescent sexual health intervention: a community-randomised trial in rural Tanzania. AIDS 2007; 21:1943–1955.
8. Doyle AM, Ross DA, Maganja K, Baisley K, Masesa C, Andreasen A, et al., for the MEMA kwa Vijana Trial Study Group. Long-term biological and behavioural impact of an adolescent sexual health intervention in Tanzania: a community-randomised trial. PLoS Med (in press).
9. Cowan FM, Pascoe SJS, Langhaug LF, Dirawo J, Chidiya S, Jaffar S, et al. The Regai Dzive Shiri Project: a cluster randomised controlled trial to determine the effectiveness of a multicomponent community-based HIV prevention intervention for rural youth in Zimbabwe – study design and baseline results. Trop Med Int Health 2008; 13:1235–1244.
10. Jewkes R, Nduna M, Levin J, Jama N, Dunkle K, Khuzwayo N, et al. A cluster randomized-controlled trial to determine the effectiveness of Stepping Stones in preventing HIV infections and promoting safer sexual behaviour amongst youth in the rural Eastern Cape, South Africa: trial design, methods and baseline findings. Trop Med Int Health 2006; 11:3–16.
11. Jewkes R, Nduna M, Levin J, Jama N, Dunkle K, Puren A, Duvvury N. Impact of Stepping Stones on incidence of HIV and HSV-2 and sexual behaviour in rural South Africa: cluster randomised controlled trial. BMJ 2008; 337:a506.
12. Corbett EL, Makamure B, Cheung YB, Dauya E, Matambo R, Bandason T, et al. HIV incidence during a cluster-randomized trial of two strategies providing voluntary counselling and testing at the workplace, Zimbabwe. AIDS 2007; 21:483–489.
13. Kang M, Dunbar MS, Assefa H, Laver S, Nhamo D, Padian N. ‘SHAZ!’ Phase II: potential efficacy of an economic and life skills intervention to mitigate structural risk among female orphans in Zimbabwe [abstract p286]. 18th ISSTDR Conference; 28 June–1 July 2009; London, UK; 2009.
14. Dunbar M, Kang M, Assefa H, Laver S, Mudekunye I, Nhamo D, Padian N. ‘SHAZ!’ Phase II: potential efficacy of a combined economic and life skills intervention to prevent HIV/STI infection among female orphans in Zimbabwe [abstract p287]. 18th ISSTDR Conference; 28 June–1 July 2009; London, UK; 2009.
15. The EXPLORE Study Team. Effects of a behavioural intervention to reduce acquisition of HIV infection among men who have sex with men: the EXPLORE randomised controlled study. Lancet 2004; 364:41–50.
16. Patterson TL, Mausbach B, Lozada R, Staines-Orozco H, Semple SJ, Fraga-Vallejo M, et al. Efficacy of a brief behavioral intervention to promote condom use among female sex workers in Tijuana and Ciudad Juarez, Mexico. Am J Public Health 2008; 98:2051–2057.
17. Gregson S, Adamson S, Papaya S, Mundondo J, Nyamukapa CA, Mason PR, et al. Impact and process evaluation of integrated community and clinic-based HIV-1 control: a cluster-randomised trial in Eastern Zimbabwe. PLoS Med 2007; 4:e102.
18. Aarø LE, Flisher AJ, Kaaya S, Onya H, Fuglesang M, Klepp K-I, Schaalma H. Promoting sexual and reproductive health in early adolescence in South Africa and Tanzania: development of a theory and evidence-based intervention programme. Scand J Public Health 2006; 34:150–158.
19. Flisher AJ, Klepp K-I. School-based HIV/AIDS prevention in Sub-Saharan Africa [editorial]. Scand J Public Health 2009; 37(Suppl 2):4–6.
20. Kinsman J, Harrison S, Kengeya-Kayondo J, Kanyesigye E, Musoke S, Whitworth J. Implementation of a comprehensive school-based AIDS education programme in Masaka District, Uganda. AIDS Care 1999; 11:591–601.
21. Hallett TB, Garnett GP, Mupamberiyi Z, Gregson S. Measuring effectiveness in community randomized trials of HIV prevention. Int J Epidemiol 2007; 37:77–87.
22. Shafer LA, Biraro S, Nakiyingi-Miiro J, Kamali A, Ssematimba D, Ouma J, et al. HIV prevalence and incidence are no longer falling in southwest Uganda: evidence from a rural population cohort 1989–2005. AIDS 2008; 22:1641–1649.
23. Ferguson J, Dick B, Ross DA. Steady, ready, go! The WHO, UNICEF, UNFPA, UNAIDS Technical Report on the Evidence for Policies and Programmes to Achieve the Global Goals on HIV and Young People: conclusions and recommendations. WHO Tech Rep Ser 2006; 938:317–341.
24. Catania JA, Gibson DR, Chitwood DD, Coates TJ. Methodological problems in AIDS behavioral research: influences on measurement error and participation bias in studies of sexual behavior. Psychol Bull 1990; 108:339–362.
25. McFarlane M, St Lawrence JS. Adolescents' recall of sexual behavior: consistency of self-report and effect of variations in recall duration. J Adolesc Health 1999; 25:199–206.
26. Wight D, West P. Poor recall, misunderstandings and embarrassment: interpreting discrepancies in young men's reported heterosexual behaviour. Cult Health Sex 1999; 1:55–78.
27. Plummer ML, Ross DA, Wight D, Changalucha J, Mshana G, Wamoyi J, et al. ‘A bit more truthful’: the validity of adolescent sexual behaviour data collected in rural northern Tanzania using five methods. Sex Transm Infect 2004; 80(Suppl 2):ii49–ii56.
28. Ross DA, White RG. Using surrogate outcomes for HIV incidence: handle with care! [Abstract TUAC0204, abstract book 1]. IAS Conference; 3–8 August 2008; Mexico City, Mexico; 2008. p. 312.
29. MacKinnon DP. Introduction to statistical mediation analysis. New York, New York, USA: Taylor & Francis; 2008.
30. Bertrand JT, Anhang R. The effectiveness of mass media in changing HIV/AIDS-related behaviour among young people in developing countries. WHO Tech Rep Ser 2006; 938:205–241.
31. Metcalf CA, Douglas JM, Malotte CK, Cross H, Dillon BA, Paul SM, et al. Relative efficacy of prevention counselling with rapid and standard HIV testing: a randomized, controlled trial (RESPECT-2). Sex Transm Dis 2005; 32:130–138.
32. Weinhardt LS, Carey MP, Johnson BT, Bickham NL. Effects of HIV counselling and testing on sexual risk behavior: a meta-analytic review of published research, 1985–1997. Am J Public Health 1999; 89:1397–1405.
33. Denison JA, O'Reilly KR, Schmid GP, Kennedy CE, Sweat MD. HIV voluntary counseling and testing and behavioral risk reduction in developing countries: a meta-analysis, 1990–2005. AIDS Behav 2008; 12:363–373.
34. Cremin I, Nyamukapa C, Sherr L, Hallett TB, Chawira G, Cauchemez S, et al. Patterns of self-reported behaviour change associated with receiving voluntary counselling and testing in a longitudinal study from Manicaland, Zimbabwe. AIDS and Behavior 2010; 22:708–715.
35. Bonell CP, Hargreaves JR, Cousens SN, Ross DA, Hayes RJ, Petticrew M, Kirkwood B. Alternatives to randomisation in the evaluation of public-health interventions: design challenges and solutions. J Epidemiol Community Health 2009. DOI: 10.1136/jech.2008.082602. [Epub ahead of print]
36. Dick B, Ferguson J, Chandra-Mouli V, Brabin L, Chatterjee S, Ross D. The health services contribution to achieving the global goals on HIV and young people: a review of the evidence for interventions to increase young people's use of health services in developing countries. WHO Tech Rep Ser 2006; 938:151–204.
37. Hankins CA, de Zalduondo BO. Combination prevention: a deeper understanding of effective HIV prevention. AIDS 2010; 24(Suppl 4):S70–S80.
© 2010 Lippincott Williams & Wilkins, Inc.