Objective To review the evidence for the effectiveness of a variety of approaches to preventing HIV infection.
Methods We reviewed what is known about the efficacy and effectiveness of a range of prevention approaches that are at various stages of research. These interventions attempt to induce behavioural change, apply technologies or modify social environments. Our intention was not to provide an exhaustive review of all types of HIV prevention strategies but rather to illustrate the landscape of interventions that have been developed and evaluated in different settings and that have the potential for widespread application among both adults and young people.
Findings There is a large quantity of evidence from experimental and observational research as well as from practical real-world experience in both developed and developing countries. This evidence supports the implementation and scale-up of a number of interventions and strategies. At the same time, there is a need to continue to develop new and more effective interventions while attending to a number of behavioural and social issues that cut across virtually all interventions designed to prevent the spread of HIV.
Conclusion We caution against confusing lack of implementation with lack of effectiveness and call for continual improvement in the quality and quantity of evidence. We have also identified a number of important directions for future HIV prevention research.
In the face of increasing rates of HIV infection around the world, there are those who doubt that HIV prevention strategies work, despite extensive evidence of the effectiveness of several interventions (1-4). In large part, this doubt is due either to a lack of understanding that prevention strategies may be working even when HIV infection rates are high or increasing (see a The Foundation for AIDS Research (amfAR), 1150 17th Street NW, Washington, DC, 20036,
USA. Correspondence should be sent to Dr Auerbach (email: [email protected]). b London School of Hygiene and Tropical Medicine, University of London, London, England.
chapter 2) or to a belief that only certain kinds of data constitute valid evidence of effectiveness (see chapter 4). Perceptions that HIV prevention "doesn't work" may also be a function of confusing the efficacy of interventions with not having taken effective interventions to scale.
Despite these challenges the field of HIV prevention is alive and well. In the course of the past 20 years, many approaches for stemming the spread of HIV have been developed, tested and evaluated (although relatively few have been evaluated systematically and rigorously) in different populations and settings, and a number have been widely adopted. The following is a brief review of what is known about the effectiveness of a range of interventions at various stages of research that attempt to induce behavioural change, apply technologies or modify social environments in order to prevent the spread of HIV. Some of these interventions have involved young people, and some interventions are more appropriate for young people than are others. Our intention is not to provide an exhaustive review of all types of HIV prevention strategies, nor to systematically evaluate the strength of evidence for them, but rather to illustrate the landscape of interventions that have the potential for widespread application among both adults and young people. Our review includes evidence from experimental studies (including quasi-experimental studies) and observational studies and related analyses that have used a range of designs and evaluation methods. The merits of different study designs, as well as criteria for assessing their quality and evaluating their evidence of effectiveness, are described in detail in chapter 4. Other chapters elaborate on the evidence for particular types of youth-focused interventions implemented in different settings.
The goal of interventions aimed at changing behaviours is to reduce the risk of HIV-related sexual and drug-use behaviours. Behavioural change interventions seek to delay the onset of sexual intercourse, reduce the number of sexual partners a person has and reduce the incidence of unprotected sex by increasing condom use. Behavioural change interventions also target drug use and seek to reduce or eliminate the incidence of drug injecting and the incidence of sharing needles, syringes and other drug-use equipment. True reductions in such behavioural risks would reduce the transmission and acquisition of HIV infection.
Interventions aimed at changing behaviours focus on counselling individuals, couples and small groups (and these interventions sometimes include HIV testing) and running workshops and other programmes that provide information and skills (including, for example, sex education, instructions on how to use condoms and other harm reduction strategies). These interventions may also aim to change social norms by seeking the involvement of opinion leaders or they may be peer-based, use social networks or be targeted at the community. Additionally, they may include social marketing, communications and mass media campaigns (2-6). These interventions are based on psychological and social science theories that emphasize the importance of knowing about the risks of HIV transmission, instilling motivation to protect oneself and others, changing expectations of outcomes, developing skills for engaging in protective behaviours and the ability to maintain protective behaviours, and providing social support for protective actions (5, 6). Evaluation designs have included experimental and observational studies.
Behavioural change interventions have been tested in a range of social settings, including health-care systems, HIV/AIDS service organizations, schools, churches, community centres, commercial establishments, workplaces, correctional facilities, the military and in homes. Outcomes related to HIV/AIDS that were assessed in these interventions generally fall into three categories:
• psychosocial (such as self-efficacy, perceived risk, personal or interpersonal skills, HIV/AIDS knowledge, intentions to adopt risk-reduction behaviours, communication with partners)
• behavioural (such as the safe use of injected drugs, reducing the incidence of sharing drug paraphernalia, encouraging the use of male or female condoms, reducing the number of partners and frequency of unprotected sexual activity and encouraging HIV testing), and
• biological (such as the incidence or prevalence of HIV or other sexually transmitted infections [STIs], hepatitis and, sometimes, pregnancy, particularly in studies with young people) (5, 6).
In fact, most behavioural interventions target a number of risk reduction outcomes.
Hundreds of studies of behavioural change interventions have been conducted since the early 1980s, both in the developed and the developing world. Until recently, these have almost entirely targeted people who are not infected with HIV, although there is a growing body of studies of interventions focusing on people who are HIV positive (7). Several systematic reviews and meta-analyses have summarized findings from these studies (5-18).
Most meta-analyses have included only experimental studies, and so have only reported on a subset of all studies of behavioural interventions, most of which have been conducted in North America and western Europe. Studies in these meta-analyses and systematic reviews of experimentally designed behavioural interventions have focused on HIV-negative heterosexual adults and adolescents, injecting drug users and men who have sex with men. They have found that such interventions can result in as little as no reduction in risk behaviours to as much as a 40% reduction in risk behaviours among different population groups and exposure categories over periods that generally have ranged from 3 months to 2 years. Of those interventions that have shown efficacy in reducing risk, most have had small to moderate effect sizes (not all of which are statistically significant) (5, 7-13, 19, 20), although substantial effect sizes were found in some studies included in reviews of interventions among men who have sex with men (14, 18). Within the overall category of behavioural change interventions, those considered to work best in reducing sexual risk include small-group cognitive behavioural interventions, educational interventions and face-to-face counselling and skill-building programmes (for example, teaching proper condom use, negotiation and refusal skills). Those that work best for reducing risks from drug use include outreach programmes, needle exchange activities, addiction treatment programmes and face-to-face counselling (21).
Beyond these meta-analyses, other reviews have provided additional evidence of the efficacy and effectiveness of behavioural interventions in reducing the risk of HIV infection in developing countries among commercial sex workers, adolescents, injecting drug users and men who have sex with men (14-16, 18, 22, 23).
There are some important caveats to meta-analyses and the behavioural interventions they assess. First, most behavioural intervention studies measure multiple outcomes and many report a composite risk-reduction outcome so one would have to tease out the data for each outcome from each study to know exactly what had been achieved, and this has often not been done. Indeed, it is possible that published studies tend to emphasize the one outcome that is significant, leading to reporting bias. Second, behavioural outcomes are not operationalized or measured consistently across studies. For example, condom use is measured as "never, sometimes, always used", "number of unprotected acts of intercourse" or "condom used at last act of intercourse", to name just a few. So we cannot be certain that like outcomes have been pooled.
Additionally, most studies of behavioural intervention are population-specific, with the reference group being defined variously by age, sex of the participant, sexual orientation, ethnicity, cultural community, geographical setting or exposure category. Thus, most summary reviews of behavioural interventions are specific to these particular social groups (7, 8, 10-19). Consequently, it takes careful sifting to determine which of the effective interventions ought to and could be replicated and scaled-up for different populations and settings.
But perhaps the most important limitation of studies of behavioural interventions is that virtually all behavioural outcomes are self-reported, which raises questions about their veracity and validity. It is both difficult and often undesirable to directly observe and measure HIV risk and protective behaviours related to sexual intercourse and drug use, so we must rely chiefly on the indirect measures of self-reporting (24-26). Given the sensitivity of these behaviours, there is the possibility that people will consciously or subconsciously misreport them in ways they consider to be socially desirable. This has been demonstrated by studies that have compared self-reported data and biological markers; and it may be especially severe in studies of adolescents and young people. For example, a study of adolescents in the United Republic of Tanzania found substantial discrepancies in reported behaviour using five different methods of data collection. On self-completed questionnaires or during structured interviews, most young women denied having engaged in any sexual activity but many had biological markers of activity (such as pregnancy or an STI), and during in-depth interviews most admitted to engaging in sexual activity (26). Also, reporting bias may differ between the intervention and control arms of a study or between those exposed or unexposed to the intervention in an observational study, thereby distorting the effects of the intervention. Thus although there have been significant advances in developing techniques to optimize the validity of self-reports (such as through the use of computer-assisted survey instruments and carefully designed questionnaires), questions remain about the validity of study results based exclusively on self-reported behaviours. This has led to the increasing interest in including biological outcomes (such as STI or HIV incidence) in studies of behavioural interventions as complementary measures and sometimes as primary endpoints.
There are only a few published experimental studies testing the effectiveness of an intervention to reduce behavioural risk using both behavioural and biomedical endpoints (incidence of STIs or HIV, or both), and these have found mixed results. For example, two multisite intervention studies among heterosexual men and women in the United States found significant positive effects both on outcomes of behavioural change and STI incidence (27, 28). As a result, one of the protocols (Project RESPECT) has been widely replicated in the United States. But a study in London among men who have sex with men found only modest positive change in reported behaviours and an unexpectedly higher rate of STI acquisition among the intervention group than the control group, although this difference attenuated over time (29). A large multisite behavioural intervention trial in the United States among men who have sex with men, and which included HIV incidence as an outcome measure, found an 18.2% lower rate of HIV infection (15.7% after adjustment for baseline covariates) in the intervention group compared with the control group and a 20.5% lower incidence of unprotected receptive anal intercourse with partners who were HIV positive or of unknown serostatus in the intervention group compared with the control group. Although the behavioural outcome was statistically significant, the HIV incidence outcome was not (30). A randomized community trial of a multicomponent adolescent sexual health programme in rural Mwanza, United Republic of Tanzania, assessed both behavioural and biological outcomes, including HIV incidence and STI prevalence. The intervention had a significant impact on knowledge about HIV, reported attitudes towards HIV and some reported behaviours, with variations occurring by the sex of the participant, but it did not have a consistent impact in either direction on STI outcomes (31, 32). Finally, in Masaka, Uganda, the effects of a community-wide behavioural intervention, with and without improved STI treatment services, were assessed in a three-arm community-randomized trial. Comparison of the behavioural intervention and control arms showed an increase in condom use with casual partners, but there was no significant impact on HIV incidence, possibly because the trial was carried out at a time when incidence was already falling as a result of larger changes toward safer behaviour in Uganda (33).
In addition to experimental data, surveillance and other observational data provide evidence of behavioural change at the population level that is plausibly related to behavioural interventions, including information and education provided by nongovernmental organizations, social institutions, peers and the media. However, if we look at the observational data from countries and communities that have documented behavioural change, it is difficult to ascertain exactly what produced the change; this is especially true for specific behavioural interventions. This leaves us in a quandary and makes it difficult to determine what really works and how it might be replicated elsewhere. For example, documented declines in HIV prevalence in Uganda have been attributed to the promotion of a strategy known as "ABC": "abstain" (chiefly aimed at delaying sexual debut among young people); "be faithful" (aimed chiefly at reducing the number of partners); and "use condoms" (chiefly aimed at use with non-regular partners). But the actual interventions (scientifically tested or otherwise) that produced these behavioural changes have not been specified (34). In the absence of clear data, a controversy emerged - and continues - about the relative importance of delayed sexual debut, partner reduction and condom use in influencing the dynamics of the epidemic in Uganda (35). Similarly, a population-based study in Zimbabwe has shown evidence of a decline in HIV prevalence, particularly among young people, that was accompanied by changes in reported sexual behaviour. Again it is not possible to attribute this change to specific interventions (36).
Attribution issues notwithstanding, observational data from these and other countries provide evidence that delaying sexual debut, reducing the number of sexual partners and increasing the use of condoms has been achieved through the implementation of various behavioural change interventions -both formal and informal - and that these behavioural changes appear to have contributed to 50-90% reductions in HIV incidence and prevalence in a number of populations in the 1990s, as measured by epidemiological surveillance (37-42). Most analysts have concluded that it was the combination of behavioural intervention strategies and involvement by high-level political and community leaders that produced significant behavioural changes and, as a result, reductions in HIV incidence and prevalence in these countries (3, 4, 34, 36, 42).
Some of the observed trends in HIV incidence and prevalence may be due to the natural dynamics of the epidemic, including an "exhaustion of susceptibles" and people modifying their behaviour in the light of knowing others who have AIDS; and behavioural trends may reverse over time, erasing some of the gains made in earlier years. An analysis by researchers leading the Rakai cohort study in Uganda illustrates these possibilities. HIV prevalence, reported age at first sexual intercourse, reported number of sexual partners and condom use had probably all improved in this cohort in the late 1980s and early 1990s, as they had in other areas of southern Uganda. However, although HIV prevalence declined by 6.2% between 1994 and 2003, most of this decline (about 5%) was attributable to HIV-related mortality. In this community, between 1995 and 2002 the age at first sexual intercourse actually dropped among both males and females, and from 1994 to 2003 the proportion of adolescent girls aged 15-19 reporting having had two or more non-marital sexual partners increased significantly. But despite these behavioural changes, HIV incidence remained stable, which the study authors attributed to an increase in condom use with casual partners occurring among both males and females (43).
These observations underscore the need for, and the usefulness of, combining epidemiological and behavioural data for the same time periods. They also emphasize the usefulness of prevalence data for young people, which can be used as a marker of recent change and lead to a better understanding of what is really occurring in terms of HIV transmission and prevention and to what behavioural changes epidemiological outcomes may be attributed (36, 44).
The establishment of voluntary counselling and testing programmes has been a feature of national HIV prevention strategies in a number of countries. The key goals of these programmes are to provide people with the opportunity to learn their HIV status, to counsel people about how to avoid becoming infected or spreading HIV and to refer people to appropriate medical and psychosocial care. A number of studies in developed and developing countries have been undertaken to determine if, in addition to its diagnostic and referral benefits, voluntary counselling and testing may lead to reductions in risky behaviours and reductions in HIV infection rates (45). Overall, data from randomized controlled trials and observational studies show mixed results in this regard (46). For example, in a randomized trial of an intervention for couples, which was conducted in Kenya, Trinidad and the United Republic of Tanzania, both male and female participants who had counselling and were tested were significantly more likely to report a reduction in the incidence of unprotected intercourse with a non-primary partner than those in the control group, who received only basic health information. Among those in the intervention group, HIV-positive men were more likely than HIV-negative men to report a reduction in the incidence of unprotected intercourse with both primary and non-primary partners (47). (This study also looked at STI endpoints, but it was underpowered and the effect was non-significant.) Other research from sub-Saharan Africa has found that behavioural changes induced by voluntary counselling and testing programmes among couples whose HIV status is discordant (that is, one partner is HIV positive and one is not) varied by the sex of the participant: condom use within these couples was more frequent and consistent when men were the HIV-negative partners (48-50). Thus, it appears that the effectiveness of these programmes in preventing HIV infection is limited to certain individuals and couples.
Additionally, utilization of voluntary counselling and testing remains low in most communities with a high prevalence of HIV, chiefly because access to services is often limited and stigma and discrimination continue to surround HIV infection. To address this situation, a large community-randomized intervention study of voluntary counselling and testing is under way in South Africa, Thailand, the United Republic of Tanzania and Zimbabwe. The intervention aims to increase the availability of services in community settings, to engage communities through outreach and to provide post-test support -all of which are intended to change community norms, mitigate stigma and reduce the risk of HIV infection among all community members regardless of whether they participate directly in the intervention (51).
Numerous studies have shown that substance abuse treatment programmes can have a significant effect on HIV transmission among injecting drug users (52, 53), although few of these studies are experimental and most have occurred in the United States. These drug treatment programmes usually involve both opiate substitution (particularly with methadone) and the provision of behavioural counselling; they generally attempt to help injecting drug users decrease the number and frequency of injections or to cease injecting altogether. Reducing drug-related risk behaviours leads to fewer potential exposures to HIV. Some of the earliest data demonstrating the efficacy of drug treatment programmes to prevent the spread of HIV came from an observational study in the United States of in-treatment and out-of-treatment intravenous opiate users randomly recruited from a methadone maintenance programme and its surrounding neighbourhood in Philadelphia, Pennsylvania. At follow up at 18 months, 3.5% of the methadone maintenance patients who had been in treatment continuously were HIV positive compared with 22% of out-of-treatment injecting drug users, representing a 6-fold difference in the rate of seroconversion (54). Another prospective study showed that at 36 months 8% of injecting drug users in treatment had become infected compared with 30% of those not in treatment (55). Although these data come from observational studies rather than experimental studies, the large differences observed suggest that such interventions can be effective. But the availability of methadone maintenance programmes is limited throughout the world, not least because drug substitution is illegal or highly regulated in many countries, including those with HIV epidemics strongly associated with the use of injected drugs.
3.3 Biomedical interventions: evaluating technologies
Research into biomedical interventions to prevent HIV infection involves testing the effectiveness of physical and chemical technologies to prevent the transmission or acquisition of HIV. The goal of such interventions is to moderate the influence of biological or physiological factors that may increase infectiousness or susceptibility to HIV or to prevent infection from progressing after actual exposure. Some biomedical strategies have been well tested and implemented; others are still at the early stages of development. Although most studies of biomedical interventions assess outcomes at the individual level, some have been designed to effect community-level change.
As soon as it was understood that AIDS was caused by a bloodborne pathogen, the importance of the health services taking steps to avoid parenteral iatrogenic transmission became apparent. Beginning in the mid-1980s, several strategies were introduced to reduce the likelihood of patients acquiring HIV infection from blood transfusions both in developed and developing countries; these included the widespread use of diagnostic assays for detecting HIV in blood and imposing restrictions on donors known to be at risk of HIV infection (56-59). Other strategies that have been used to reduce iatrogenic transmission include educating the public about the risks of engaging in non-sterile medical practices, such as sharing needles, and demonstrating to patients that the medical care they are receiving is safe by taking a new auto-disable syringe out of a sealed package and using single-dose phials (60). Although rigorous studies testing the effectiveness of these interventions are lacking, implementing such practices in developed countries has been effective in virtually eliminating HIV transmission through blood donation and medical practices (61, 62). However, some epidemiologists continue to argue that unsafe injection practices account for a greater proportion of HIV transmissions in resource-poor settings than sexual intercourse does (63). But this has been countered by a thorough assessment of the epidemiological data from across sub-Saharan Africa and by a cohort study in rural Zimbabwe both of which unequivocally concluded that sexual interaction and not unsafe injections remains the primary mode of HIV transmission in that region (64, 65). Despite these conflicting perspectives, all agree that there is a need to effectively monitor injection practices at health facilities and to ensure the scale-up of safe procedures.
A large body of evidence accumulated from epidemiological and clinical studies has shown that the risk of sexual transmission of HIV is substantially increased in the presence of other STIs (66). The evidence shows that STIs - particularly those associated with genital ulceration - can enhance the infectiousness of people who are HIV positive as well as the susceptibility of people who are HIV negative. STI prevalence is high in many countries with substantial HIV epidemics, and this is often due at least partly to poor STI treatment services. Therefore, the diagnosis and treatment of STIs have the potential to be effective prevention strategies by reducing STI prevalence in couples whose HIV status is discordant. Three large randomized trials have measured the effects of community-wide STI treatment interventions in different populations living close to Lake Victoria in east Africa. The first, conducted in the Mwanza region of the United Republic of Tanzania, showed that improved STI treatment services, using syndromic management and delivered through government-run primary health-care units, reduced HIV incidence in the general adult population by an estimated 38% (67, 68). The second, conducted in the Rakai district of Uganda, evaluated the effect of periodic mass treatment ofSTIs. All adults living in intervention communities were treated for STIs at 10-monthly intervals, whether or not they reported STI symptoms. After 20 months of follow up, this intervention showed no significant effect on HIV incidence (69). The third trial, conducted in the Masaka district of Uganda, evaluated whether a combined behavioural and STI syndromic management intervention was more effective than a behavioural intervention alone in reducing HIV transmission; it found no significant difference in HIV incidence between study arms after 3.6 years of follow up (33). All three trials reported some significant reductions in the incidence and prevalence of STIs.
The contrasting results of these studies call for careful analysis and interpretation. A synthesis of data from these trials concluded that differences in the study populations - with respect to sexual risk behaviour and STI rates as well as the stage of the HIV epidemic - and not differences in intervention strategy, were the key determinants of the contrasting HIV outcomes in the three locations (70, 71). The investigators concluded that using STI control measures to prevent the spread of HIV is likely to be most effective in populations with sexually transmitted epidemics that are early and concentrated and in populations with a high prevalence of STIs and sexual risk behaviours. The results of the Mwanza trial have led to widespread implementation of syndromic STI management in many countries.
It has been recognized that infection with herpes simplex virus type 2 (HSV-2) plays a particularly important role in HIV transmission. A systematic review and a meta-analysis summarized evidence from observational studies ( 72, 73). The more recent review found that prevalent HSV-2 infection was associated with a 3-fold increase in the risk of HIV acquisition both in men and women in studies in the general population; weaker effects were seen in studies of at-risk groups, possibly because such groups also have a high prevalence of other STIs (73). It is likely that HSV-2 infection also increases the infectiousness of people who are HIV positive. Data from a randomized controlled trial of valacyclovir in Burkina Faso showed that suppressive therapy administered during a 3-month period reduced genital shedding of HSV-2 and HIV and also reduced HIV plasma load (74). Based on these observational data, and strengthened by the findings of the study in Burkina Faso, a number of phase III randomized trials are under way in Peru, South Africa, the United Republic of Tanzania, the United States, Zambia and Zimbabwe to measure the effects of episodic or suppressive antiviral therapy in people who are infected with HSV-2 on HIV acquisition or transmission (75).
3.3.3 Antiretrovirals to prevent HIV infection
One of the most significant developments in HIV prevention was the finding that certain antiretroviral drugs could be administered to pregnant women and their newborns in such a way as to significantly reduce the likelihood of HIV being passed from mother to child before, during and after delivery. Three key randomized controlled trials provided the proof. The first, conducted in the United States, involved a three-part regimen of zidovudine given to the mother and her newborn; it reduced HIV transmission by two thirds (76). The second, conducted in Thailand and Cote d'Ivoire, involved a shorter regimen of zidovudine, and it reduced transmission by 44-50% (77, 78). The third trial, conducted in Uganda, demonstrated that a single dose of nevirapine administered to the mother at the onset of labour and to the infant soon after birth reduced HIV transmission by nearly 50% (79). Together these studies demonstrate that long-course and short-course zidovudine and single-dose nevirapine are effective interventions for reducing mother to child transmission of HIV (80). Several studies are under way in resource-poor settings to assess the effects of using highly active antiretroviral treatment (HAART) during pregnancy (that is, treating the woman for her own infection) on the subsequent incidence of mother to child transmission. Preliminary results from one study conducted in Cote d'lvoire showed the HIV transmission rate among pregnant women on HAART was 1.45% compared with 3.89% for those on a standard short-course antiretroviral regimen (81).
Interventions to reduce the incidence of HIV transmission that occurs through breastfeeding have also been tested. Where feasible and acceptable, the use of breast-milk substitutes has been shown to significantly reduce infection among infants (82, 83). Where implemented fully, antiretroviral treatment and breastfeeding interventions have had a remarkable effect: in the United States, for example, the number of children who acquired HIV perinatally decreased by 89% between 1992 and 2001 (84).
22.214.171.124 Antiretrovirals for post-exposure prophylaxis
In some countries health-care workers and prison officers have had access to treatment with antiretroviral drugs following assumed (chiefly parenteral) exposure to HIV infection in the workplace. A retrospective study of occupational post-exposure prophylaxis concluded that zidovudine monotherapy (administered within 24 hours of exposure and over the course of 28 days) reduced occupational HIV transmission (chiefly through needle-stick injuries) by approximately 81% (85). In more recent years, this approach has been extended to become a medical strategy for coping with non-occupational exposures, such as sexual assault or condom breakage, and it appears to be effective (86, 87). However, studies have not included data on untreated individuals, so these findings should be interpreted with caution. Also, adherence to the full regimen of post-exposure prophylaxis is difficult to achieve but it is likely to be important. In a programme in San Francisco, 40% of all eligible survivors of sexual abuse initiated post-exposure prophylaxis regimens but less than 10% completed them (88). Lack of complete adherence might compromise the effectiveness of using antiretrovirals for post-exposure prevention, as has been shown for incomplete adherence in HIV treatment.
126.96.36.199 Antiretrovirals for preventing sexual transmission
A key benefit of using HAART for treatment of HIV disease is that it reduces plasma viral load. Strict adherence to HAART regimens contributes to the sustained suppression of virus. Some data indicate that the risk of HIV transmission is directly associated with the infected individual's viral load (89). These findings have led many experts and laypeople to believe that the widespread use of HAART could have important effects on the sexual transmission of HIV. There are no prospective data yet to prove this, although a multisite randomized controlled trial is under way to determine whether an-tiretroviral treatment can prevent sexual transmission of HIV-1 in serodis-cordant couples (90, 91). The primary data establishing the relationship between lower plasma viral load and reduced sexual transmission of HIV come from a cohort in Rakai, Uganda, who were not taking HAART (92).
It is making a significant leap to assume that the effects of HAART at the individual level will translate into similar effects at the population level. Mathematical models have demonstrated that at the population level the beneficial effect of HAART on reducing sexual transmission of HIV could be offset by an increase in the circulation of drug-resistant strains of HIV and increases in risky sexual behaviour that might occur if people believe HIV transmission is unlikely or if HIV infection becomes less feared because treatment is available (93, 94). Furthermore, because HAART extends survival time, where it is widely used there is a larger pool of HIV-positive individuals who have the potential to spread the infection over a longer period of time. The net effect of all these competing factors is, therefore, difficult to predict. However, an observational study in Taiwan, China, concluded that the estimated HIV transmission rate in the country declined by 53% after the introduction of free access to HAART in 1997, while no significant behavioural change occurred. The researchers were careful to note that the widespread use of HAART can be an effective measure to control HIV epidemics in countries with a low prevalence, but may not be as effective, or may not be effective at all, in settings with an already high prevalence of infection, for a number of reasons having to do with the exponential course of HIV epidemics (95).
Considering these data together - data from studies in animals, of postexposure prophylaxis, of preventing mother to child transmission and data from Rakai on lower viral loads and lowered rates of transmission - has led to an interest in exploring the possibility of using antiretroviral treatment for pre-exposure prophylaxis. Randomized trials are planned or under way in Botswana, Ghana, Peru, Thailand and the United States to test the hypothesis that administering antiretroviral treatment (specifically tenofovir) to people who are HIV negative but at high risk will reduce HIV acquisition among them (96).
Was this article helpful?