loveLife is an example of a national HIV prevention program implemented in a country experiencing a generalized epidemic. To a large extent, the political economy in which loveLife operates is conducive to its success. For this reason, it serves as an important case study for HIV prevention.

loveLife’s design responds to the fact that the sexual behaviour of teenagers drives the epidemic in South Africa. This assertion is based on the large proportion of the South African population that are under the age of 20 years (40%), a significant number of whom report high-risk sexual behaviour. The highest rates of new infection occur in late adolescence and early adulthood. The best prospect for changing the course of the HIV epidemic is in substantially reducing new infection rates among successive cohorts of 12 – 17 year olds.

The unrelenting course of HIV/AIDS in South Africa during the 1990’s necessitated a different approach to HIV prevention, particularly for young people, who express alienation from conventional public health messaging and associations with the red ribbon. Major predictors of high-risk sexual behaviour in this age group are coercion, peer pressure, transactional sex and abdication of sexual responsibility by boys. Further, the major challenge is not knowledge of HIV transmission, but widespread failure of individuals to internalize personal risk. It was thus critical to establish a program which attracted young people and with which they wanted to identify and talk about. Recognition that a major influence on post-liberation South African youth is the global youth culture of music, fashion, pop icons and commercial brands led to the positioning of loveLife – an aspirational lifestyle brand for young South Africans.

Underpinning the design construct of loveLife is a view that motivating young people to change sexual behaviour requires active experience of an alternative and positive lifestyle, while also making institutions responsive to growing demands from young people for friendly and appropriate health and social services. loveLife combines high-powered media with service delivery, institutional support and outreach to create a comprehensive national campaign aimed at halving the incidence of HIV infection among 15-20 year olds by 2007, and substantially reducing pregnancy and STI rates among teenagers.

Modeling projections, based on international precedents and South African baselines, suggest that a substantial number of HIV infections can be averted by an effective HIV prevention campaign through incremental improvements in key behavioural indicators over time. loveLife estimates that it requires $40 million dollars per annum (US $6.6 per 12-17 year old) to fund its national HIV prevention program at optimal levels, and a conservative estimate of economic benefit to economic costs of such an intervention is (1.4 – 1.9) : 1. Modeling further suggests that any success in beating the deadlines for projected behavioural outcomes will result in substantial numbers of lives saved, and dramatically change the course of the epidemic.

Although definitive proof of success is significant reduction in HIV rates among young people, we cannot wait for final confirmation. Intermediary indicators, namely: i) public response to interventions; ii) predictors of sexual behaviour; and iii) self-reported sexual behaviour known to mediate HIV reduction, can provide tentative evidence of positive change

By the end of year two, young people exposed to loveLife were reporting significant behaviour change. At the very least, these findings indicate a growing association by young people with positive lifestyle and responsible sexual behaviour. They suggest that loveLife is on the right track – enabling considerable numbers of young people to personalize and internalize the risk of HIV and teenage pregnancy. This is the first real evidence in South Africa that behaviour change among adolescents may be happening – potentially on a scale to change substantially the course of the HIV epidemic.

• Generalized Epidemics Can Be Stopped – But How?

Protagonists of greater investment in HIV prevention point to the cost-effectiveness of selected interventions like voluntary counselling and testing, STI control or female condom use. Evidence from the U.S. and more recently developing countries like Uganda and Thailand has thrown new light on the relative impact of different behavioural mediators of HIV reduction. In turn, this has refined epidemiological modeling to better simulate observed declines in HIV prevalence, reaffirming the potential of HIV prevention efforts. Yet HIV prevention cynics can still count the number of national success stories on one hand, maybe two if countries showing tentative evidence of declines are thrown in.

To a large extent, prevention advocates have been caught on the back foot, resorting to valid but clichéd defences: Not enough money dedicated to prevention; a piecemeal approach with little prospect of scale-up; scant political support; and the limitations of media-driven campaigns. All true, but without significant counter-evidence that adequately funded, large-scale, properly targeted comprehensive prevention programs, with solid political backing, work.

Deferring implementation of an effective prevention package even by a few years would reduce its global impact by millions of unaverted infections. Yet, while the strategies for effective prevention for concentrated epidemics are fairly clear, the same cannot be said for countries with generalized epidemics. Although some elements of a comprehensive prevention package are known – VCT and condom use, for example – effective strategies leading to primary behaviour change (delayed sexual debut, reduced rate of partner change, assertion of sexual limits) are less obvious.

For such countries, there is now no choice but to learn on the wing relying, in the short- to medium-term, on intermediary indicators of change to gauge progress towards reduction in HIV infection. These indicators include trends in: i) the public response to interventions; ii) predictors of sexual behaviour; and iii) self-reported sexual behaviour known to mediate HIV reduction.

This paper describes the implementation of loveLife, South Africa’s national HIV prevention program for young people. The experience of loveLife is instructive, in that it is unprecedented in scale and scope, and meets many of the criteria generally regarded as critical to the success of national prevention campaigns. The campaign is less than three years old. Annual surveys of nationally representative samples of South African teenagers have enabled early review of behavioural trends. These surveys are reinforced by a multi-year evaluation protocol that, over time should provide more rigorous substantiation of the apparent early trends (Appendix 1). There are strong indications of significant positive change in some of the key predictors of sexual behaviour, although young South Africans are generally still at very high risk for HIV infection. However, if the gains demonstrated are real and can be sustained over the next few years, it is likely that the HIV incidence among 15 – 24 year olds will be substantially reduced by 2007.

Most data on the South African HIV epidemic is currently derived from the anonymous, annual survey of pregnant women attending public pre-natal clinics. The most recent survey, conducted in October 2001, shows a national HIV prevalence of 24.8% (23.6% –26.1%) in women attending public sector prenatal clinics. This suggests that a total of 4.75 million South Africans were infected with HIV at the end of 2000, roughly one in seven adults aged between 20 and 64. Point prevalence rates, with 95% confidence intervals, for the nine provinces are shown in Table 1. These demonstrate marked provincial variation from 8.6% in the Western Cape to 33.5% in KwaZulu-Natal. While there are large provincial variations in HIV infection levels, the prenatal survey has thus far shown similar epidemic patterns for all provinces except the Western Cape, indicating that current differences can be attributed more to time lags than intrinsically lower risk of infection.

Stabilization of the prevalence rate of young people at around 16% is to be expected, and is consistent with incidence projections (Figure 1). However, these rates are still extremely high, and will continue to sustain the overall HIV prevalence among adults at over 20% unless interventions are successful.

A recent study has shed some light on mortality attributable to AIDS. This study traces trends in reported deaths until 1996, based on the statistics from Statistics South Africa and comparing the results for adults with more recent data obtained from the population register maintained by the Department of Home Affairs. The study also adjusted for the under reporting of deaths in South Africa. The empirical data were then compared with model estimates to assess the consistency of the empirical data with the model projections.

The first major finding of this study is that the reported deaths show clear evidence of a changing age pattern. There has been an increase in the numbers of young adult deaths relative to deaths at older ages. By 2000, the number of young adult deaths was higher than the numbers in older ages.

The empirical death data show that there has been a steady increase in adult mortality rates during the 1990s. The mortality of young adult women has increased rapidly in the last few years with the mortality rate in the 25-29 year age range in 1999/2000 being some 3.5 times higher than in 1985 (Figure 2). Mortality in the 30-39 year age range in 1999/2000 was nearly twice that in 1985, but this off a much higher base rate.

The second finding of the study estimated the extent of AIDS mortality, without explicitly requiring cause of death. While there is inevitably some degree of uncertainty because of the assumptions underlying both the model and the interpretation of the empirical data, the authors’ best estimate on current data is that about 40% of the adult deaths aged 15 -49 that occurred in the year 2000 were due to HIV/AIDS and that about 20% of all adult deaths in that year were due to AIDS. Combined with the excess deaths reported among children, it is estimated that AIDS accounted for 25% of all deaths in the year 2000, and has thus become the single biggest cause of death in South Africa.

The number of South Africans with HIV infection is expected to continue to rise over the next ten years – unless major behaviour change occurs to significantly alter the course of the epidemic. Projections from various models, including two South African models (the Metropolitan-Doyle model, and the Actuarial Society of Southern Africa model) and two international models (the US Bureau of the Census, and the UN model used for their world population projections), indicate that the number of HIV infections could reach 6.4 – 12.1 million by 2010. This is between 15 and 25 per cent of the total South African population. It is estimated that over 200 000 South Africans are currently living with AIDS. This will rise rapidly over the next decade to almost a million people living with AIDS by the year 2010. But even then AIDS cases will not have reached their peak. Figure 3 shows the projected number of AIDS cases under the best and worst infection scenarios.

Driving the predictions described above are two factors: First, the majority (53%) of South Africans are under the age of 25 years, and second, young people are at highest risk for HIV infection.

HIV in South Africa mostly affects younger people with around 60% percent of all adults who acquire HIV becoming infected before they turn 25 (Figure 4). Gender differences are also quite pronounced, with women at highest risk between the ages of 15 and 20, whereas men probably achieve their highest incidence at 20 to 24 years of age.

• Sexual behaviour among young South Africans

This profile of youth hardest by HIV infection is borne out by patterns of adolescent sexual behaviour.

The proportion of South African teenagers who have had penetrative sex increases exponentially between ages 12 and 17. Among all 12-17 year olds, the median age of first penetrative sex is 16-17 years (Figure 5). Among sexually active 12-17 year olds, half (51%) say that their first penetrative sex occurred aged 14 years or below (Figure 6).

Defining this risk profile helps clarify communication objectives: While 42% of all 15-17 year olds have had penetrative sex, only 13% of all 12-14 year olds have, suggesting that opportunity exists to encourage young people to delay sexual intercourse and to equip them to make safe sexual choices in the future.

Sexually active South African teenagers report a high number of sexual partners (Figures 7 and 8). About half of both 12-14 year olds (n=99) and 15-17 year olds (n=523) who are sexually active report more than one sexual partner in the last year, while one eighth (12%) of sexually active 12-14 year olds and a quarter (23%) of sexually active 15-17 year olds report three or more sexual partners in the past year. One in seven (14%) sexually active 17 year olds report ten or more sexual partners

ever.

Source: Africa Strategic Research Corporation (2001). National probability sample survey of 2204 12-17 yr olds, Johannesburg. Commissioned by the Henry J. Kaiser Family Foundation (USA)

Source: Africa Strategic Research Corporation (2001). National probability sample survey of 2204 12-17 yr olds, Johannesburg. Commissioned by the Henry J. Kaiser Family Foundation (USA)

Although South African teenagers report relative high condom use (60%) during the last time they had penetrative sex (Figure 9), less than one third (30%) say they have always used a condom during the past year (Figure 10). Younger teenagers, in particular, report high-risk sexual intercourse, with 38% of sexually active 12 - 14 year olds saying they never used a condom in the past year. Young girls, likely engaging in transactional (R² = -1.0, p<0.001) or coerced sex (R² = -0.68, p=0.008) report the lowest rates of condom use.

Source: Africa Strategic Research Corporation (2001). National probability sample survey of 2204 12-17 yr olds, Johannesburg. Commissioned by the Henry J. Kaiser Family Foundation (USA)

Together, these three mediators of HIV infection (early sexual debut, high rates of partner change and erratic condom use) create a risk profile for young South Africans heavily skewed towards high-risk sexual behaviour (Figure 11). Projections are that, in the absence of effective prevention, the cumulative lifetime probability of HIV infection for the 16 million South African less than 15 years of age now exceeds 50%.

On the other hand, the skewed risk profile among young people creates opportunity for rapid reduction in HIV incidence if effective HIV prevention is achieved. A relatively modest downshift in a continuous, but skewed distribution of sexual behaviour can effect a marked contraction of the high-risk pool. And the first step in designing effective risk reduction strategies is to understand the determinants of adolescent sexual behaviour.

Source: Africa Strategic Research Corporation (2001). National probability sample survey of 2204 12-17 yr olds, Johannesburg. Commissioned by the Henry J. Kaiser Family Foundation (USA)

• Predictors of sexual behaviour among South African teenagers

Multivariate regression conducted on two national surveys of self-reported sexual behaviour (Y2000: n=1999; Y2001 n=2204) demonstrates predictors of sexual behaviour among teenagers. (It should be noted that in both surveys, random probability sampling is done within a subset of 65 enumerator areas regarded as generally characteristic of all enumerator areas. This may be a threat to external validity of the findings).

Demographic and socio-economic factors, namely male gender (male, p=0.046), older age (p=0.005) and less education (p =0.000) play some part in determining whether teenagers have penetrative sex (R²=0.028). Income is also a predictor of age of sexual debut (R²= - 0.09, p=0.001).

Personal attitudes predict sexual behaviour. Perceived happiness was associated with higher age of sexual debut among boys (R²= -0.3, p=0.05), possibly due to variable social discount rates applied to future health benefits. In other words, a sense of fatalism may place young people at higher risk.

Within boy-girl relationships, reported coercion is strongly associated of sexual debut among girls (R² = -0.68; p = 0.008), as is transactional sex (R² = -1.139, p <0.000).

Open communication with parents about HIV, sex and sexuality appears to reduce risk. Parents' responses to questions related to open discussion were correlated with self-reported sexual behaviour by their children. Talking with parents about HIV (coeff. = 5.15, p=0.038) is associated with lower rates of partner change. Young people whose parents report that they talk to their children about dealing with the pressure to have sex (coeff. = 0.512, p=0.038) and the risks of unprotected sex (coeff. = 0.485, p=0.032) are more likely to say they have changed their sexual behaviour as a result of HIV.

These findings help define intermediate indicators that can gauge progress towards HIV reduction. Systematic trends analysis of both self-reported sexual behaviour and the underlying determinants may illustrate changes not yet translated into declining HIV prevalence rates.

Furthermore, these findings illustrate that, while socio-economic and demographic factors may determine whether or not teenagers engage in penetrative sex, the risk associated with being sexually active is amenable to intervention in the short- to medium term. Personal lifestyle, and positive peer and parental interaction can reduce risks of sexual behaviour. These insights have helped craft the design and implementation of loveLife -- South Africa’s national HIV prevention program for young people

loveLife - A New Lifestyle Brand For Young South Africans

Launched in September 1999, loveLife is the largest and most ambitious effort ever undertaken in South Africa to positively influence adolescent sexual behaviour in order to reduce teenage pregnancy, HIV/AIDS and other sexually transmitted diseases. Organised under the auspices of a national advisory board of leading South Africans chaired by Mrs. Zanele Mbeki, loveLife is the product of more than two years of research internationally and among young South Africans investigating the barriers to, and opportunities, for more effective HIV/AIDS prevention among young South Africans.

loveLife combines traditional commercial marketing techniques with service delivery, institutional support and outreach to create a lifestyle brand with which young people associate healthy, positive living. In order to create this level of association among loveLife’s 12-17 year target group, the loveLife brand is positioned as part of popular youth culture.

loveLife meets many, though not all, of the criteria generally regarded as essential for the success of HIV prevention. It enjoys considerable (though not universal) public and political support. Public approval ratings in a number of national surveys commissioned by loveLife are consistently around 85 - 90%. Political support is demonstrated both through public identification with loveLife by the Presidency and national Ministries, and by a formal public-private financing partnership between the South African Government and the Henry J. Kaiser Family Foundation in support of loveLife.

In addition to the above, considerable funding is received from the Bill & Melinda Gates Foundation and UNICEF. South African corporate sector partnerships effectively double the value of media exposure and reduce the costs of some aspects of service delivery. Recently, the Global Fund on HIV/AIDS, TB and Malaria announced substantial funding for the public clinic component of loveLife. loveLife’s current budget is roughly US $20 million per annum – US 50 cents per capita, or US $3.33 per capita in the target group of 12-17 year olds. In relative terms, these levels of funding for HIV prevention are considerable, and funding needs are projected to double over the next three years to fully implement the national campaign.

loveLife has achieved accelerated roll-out of adolescent services and support programs throughout South Africa over a short period of time (two and a half years), and now has in place most of the critical elements to effect behaviour change. loveLife's high-profile multi-media strategy encompasses television, radio, billboard and taxi advertising, print and website (Box 1).

• 649 South African press articles on loveLife or covering loveLife events
• 123 radio interviews or inserts on loveLife
• 24 TV interviews or inserts on loveLife

• loveLife World AIDS Day One-Hour Special on public television’s biggest station – 2.8 million viewers
• S’camto groundBREAKER television series (26 weeks)
• loveLife Games TV series (13 weeks)
• Fortnightly lifestyle magazines (S’camtoPRINT and thethaNathi) in newpapers with the largest weekly and daily circulations (Sunday Times and Independent Newspapers respectively)
• Radio programs on one national and 6 regional radio stations, with direct national and regional broadcasts by "loveLife radio" through special event licences
• Parents’ campaign "Love Them Enough to Talk about Sex" initiated on 3 TV stations, 10 radio stations and nine newspapers and 1000 billboards. These public service announcements featured national leaders endorsing the call, and included Nelson Mandela, Desmond Tutu, Deputy President Jacob Zuma, the Minister of Health, and sports and entertainment personalities
• Outdoor media campaign on 2000 billboards and 800 taxis.
• 750 schools participating in the schools program, combining loveLife-branded school signage (including toll free line) and weekly distribution of loveLife’s lifestyle magazines (S’camtoPRINT and thethaNathi)

loveLife has the national infrastructure in place to expand service delivery, institutional support and outreach (Box 2).

• Counsellors on the toll free thethaJunction and parent lines respond to over 40 000 calls a month
• Twelve loveLife Y-Centres fully functional, with four more under construction
• National positive lifestyle and debating programs operate across Y-Centres, loveLife franchises and adolescent friendly clinics
• Radio production and broadcast training program operates across 8 Y-Centres
• Sixty public sector clinics active in the National Adolescent Friendly Clinic Initiative
• Fifty seven non-government organizations participate in the loveLife Franchise network
• Three hundred and fifty new young people are recruited to loveLife’s year-long groundBREAKER volunteer service
• Over a thousand school teachers trained nationally and provincially in preparation for the loveLife Games 2002
• Two hundred and fifty thousand young people taking part in the loveLife Games 2002: 44 regional games, nine provincial games, one national event
• 2002 Love Tour program (centred around outdoor radio broadcast units) in full swing in KwaZulu and Eastern Cape
• Love Train program operating on schedule, run by ten groundBREAKERS under Love Train management

A national random probability survey conducted at the end of year two (November 2001) found that 62% of young South Africans (12-17 yrs) had heard of loveLife (Figure 12). Knowledge of loveLife is fairly evenly spread across all race groups and rural/urban divide. Sexually active young people report greater exposure to loveLife, although this is largely explained by higher awareness of loveLife among older teenagers. Of note is the lower knowledge of loveLife by poor or very poor young people (58%) compared to those who have enough to live on (63%). This is further demonstrated by the fact that knowledge of loveLife is lowest in the poorest provinces, reflecting the paucity of media (particularly TV) compared to wealthier provinces (range across provinces 50% - 85%) (Figure 13).

These findings indicate a remarkably widespread awareness of loveLife by young South Africans in a relatively short time. However, lower access by poor or very poor households reiterates the need for even greater penetration by media, service expansion, and wider outreach. Considerable opportunity still exists to influence large proportions of young South Africans not yet reached.

loveLife has been relatively successful in positioning itself as a positive lifestyle brand focusing particularly on sex, sexuality and gender relations, as opposed to being narrowly defined as an "HIV program" (Figure 14).

Furthermore, those young people who know of loveLife generally have a positive impression of it, although the challenge of communication with youth is illustrated in the indifference expressed by one quarter of respondents (Figure 15).

All of the above suggests that loveLife is well placed to effect large-scale behavioural change among young South Africans, and to demonstrate the nature and impact of a truly national HIV prevention campaign in a country experiencing a generalized epidemic. The question remains: How to effect this behavioural change?

• Behind the design of loveLife

While there is still no guarantee of success, there are a number of insights based on the loveLife experience that may be helpful to other HIV prevention initiatives.

The first insight is that maximum returns on cost-beneficial prevention could only be achieved by targeting young people. The reasons for this are two-fold:

First, the epidemiology of HIV epidemic described above clearly demonstrates that interventions targeting even the late teens miss the opportunity to avert a substantial proportion of infections.

Second, behavioural studies show that young people are more amenable to intervention than older age groups, and that sexual behaviour change achieved early on is more effective and sustained for longer.^,

At the margin, the greatest returns on investments in HIV prevention in South Africa will be achieved through an intervention targeting 12 –17 year olds. Clearly, this does not imply that there should no prevention aimed at younger or older age groups, and the preventive potential of treatment, care and support should not be discounted either. The point is that HIV prevention efforts should be segmented, and investments concentrated to achieve the greatest reduction in burden of disease.

A major influence on post-liberation South African youth is the global youth culture of music, fashion, pop icons and commercial brands, communicated principally through commercial media. A series of 24 focus groups held across South Africa in both rural and urban areas found consistently high awareness of and loyalty to commercial brands. High intensity brand association is facilitated by a relatively sophisticated marketing milieu, and high media penetration (99% radio, 75% TV coverage). Another critical insight of the focus group series was that young people were alienated by traditional HIV messaging (such as "ABC"), and were turned off by the red AIDS ribbon, equating that symbol with disease and death. Further, national studies have repeatedly shown that the primary problem is not a lack of information, but rather a failure to personalise and internalise risk. For example, despite more than 90% having correct knowledge of modes of HIV transmission, over half (54%) of sexually experienced youth rate their chance of infection as low or very low, with a further 18% not knowing or refusing to answer.

In order to reach young people in South Africa, it was concluded that loveLife needed to be dramatically different from conventional HIV prevention efforts. Its mindset had to become that of young people’s, effectively competing with Nike, Diesel, Coke and other major brands for their attention and brand association. In order to have impact, loveLife had to become accepted as part of popular youth culture. For this reason, loveLife is positioned as a new lifestyle brand for young South Africans.

Despite the HIV epidemic, young South Africans express a strong sense of confidence in the future. In a nationally representative sample survey (n=1999) conducted in November 2000, three quarters of respondents stated that they were "positive about the future of South Africa" (73%) and feel that "South Africa has a lot to offer young people" (76%). About 8 in ten teenagers report being "very happy" (61%) or "somewhat happy" (22%) about their present lives. Top priority for most young people is getting a good education (63%). Yet, young South Africans also report major concerns – HIV/AIDS (33%), teenage pregnancy (30%), crime (17%), sexual abuse (12%), and violence (8%) head up their list of worries. And as illustrated earlier, a sense of pessimism in the future predicts high-risk sexual behaviour.

For loveLife, the implication was that it had to capitalize on the general optimism of young South Africans, tapping in to their aspirations and motivating young people who express an attitude of fatalism. While the causal links between high risk behaviour, HIV and resultant morbidity and mortality need to be made explicit, campaigns run largely on scare tactics may in fact reinforce the sense of fatalism and hopelessness.

Underpinning all of loveLife’s communication is a strong motivational element, engineering the dialectic between young people’s perceptions of the future and their current sexual behaviour and broader lifestyle.

The first goal of loveLife is to get all South Africans – and particularly 12-17 year olds – talking more openly about sex, sexuality and the HIV epidemic. This goal derived from the insight of international experience that open communication about sex, and early sex education is essential to delaying the onset of adolescent sexual activity, reducing teenage pregnancy, increasing condom usage and reducing HIV/AIDS and other sexually transmitted diseases., Furthermore, it is becoming increasingly clear that positive change requires more than a passive process of information transmission; rather it is active engagement with messaging, contextualised through personal experience and interaction across social networks, which enables individuals to personalise risk and internalise communication.

Although loveLife’s initial effort was to get young people talking more openly, it was soon realised that poor parent-child communication was a major barrier to success. Despite 40% of parents citing HIV/AIDS as their major concern for their children, and a further 21% stating that fear of sexual abuse caused them greatest anxiety, only half (46%) often talked to their children about HIV/AIDS. Worse still, much of the discussion about HIV/AIDS fails to get to its root cause. Only one in five talked often with their children about dealing with pressure to have sex (19%) and how to decide when they are ready to have sex (18%). Yet as shown earlier, children of parents who report open communication about sex and HIV are twice as likely to say that they have changed sexual practices due to the epidemic.

These findings led to the implementation of a focused "bookend campaign" for loveLife, exhorting parents to "Love Them Enough to Talk about Sex". Parents exposed to the campaign say that it has provided them with an opportunity to talk with their children about difficult issues like HIV/AIDS and sex (Figure 16).

The intensity of the campaign has been ratcheted up since these survey results in November 2001, with national leaders such as Mr Nelson Mandela, Archbishop Desmond Tutu and Deputy President Jacob Zuma featuring on television, radio and in print endorsing the campaign and exhorting parents to "Love them enough to talk about sex". While it is too early to gauge impact, there are strong indications that South African parents have responded positively to the campaign. For example, counsellors on loveLife’s toll free parents line have seen an increase in calls from an average of 1 500 calls answered per month before the inception of the endorsement campaign in February 2002, to roughly 10 000 calls now answered every month.

The predictors of adolescent sexual behaviour described above demonstrate that HIV will only be reduced by substantive normative change in youth culture. Skewed gender and power relationships are the major drivers of the epidemic. Simply telling a 15 year-old girl in a coercive relationship to "abstain" does little to protect her. Helping her to understand and negotiate sexual limits and encouraging non-penetrative forms of sex may save her life.

loveLife’s communication strategy is non-prescriptive, encouraging debate and thought, explaining risks and consequences, rather than telling young people what to do. The content of communication is framed by three fundamental values, namely Informed choice, shared responsibility and healthy sexuality. In the context of these values, specific sexual behaviour and the associated spectrum of risk are discussed.

Young people have generally found the non-didactic approach of loveLife intriguing, provoking them to think, talk about it and act. Four out of five (77%) of young people exposed to loveLife say it has caused them to take personal action. Of all South African youth, almost half (48%) say they have taken personal action as a result of loveLife (Figure 17).

Most commonly, young people report discussion among family and friends, though a significant proportion is prompted to seek additional information (Figure 18).

While all of the above are subject to systematic bias associated with self-report and attribution, at a minimum, these findings indicate strong association with risk-reducing behaviour and the positive lifestyle promoted by loveLife. On their own, they are not enough to confirm that real change has happened, but they seem to validate the communication strategy adopted by loveLife.

Traditional behaviour change models focused on individual behaviour and were based on the assumption of direct correlation between knowledge and action. The theory was that preventative behaviour is the result of rationale decision-making informed by the seriousness of the threat, perceptions of the effectiveness of risk reducing measures (such as condoms), the value individuals put on good health, as well as the self-confidence, social and communication skills of the individual. The problem with this model is that it assumes a logical pattern of decision-making and a degree of personal control that is unrealistic in the complex terrain of sexual relations. There is growing recognition that individual risk reduction and behaviour change requires change in the attitude, values and norms of society generally. Achieving sexual behaviour change is a complex task requiring integrated approaches implemented at all levels of society and sustained over a considerable number of years. And prevention initiatives need to succeed in creating a social consciousness and environment that facilitates appropriate personal action. Simply put, young South Africans need to see and hear the message of positive lifestyle, but also experience it in their own lives. And societal institutions – public clinics, for example – need to gear up to respond to the new demands from young people. loveLife’s integrated strategy combines high-powered media with service delivery, institutional support and outreach.

• loveLife’s media campaign

The media campaign is itself differentiated, with various media fulfilling different functions. Outdoor advertising (on 2000 billboards, 800 minibus taxis and 180 rural water tanks) is used to position the brand, provoke debate and point to thethaJunction, a free sexual health helpline for young people. Television, watched at least three times a week by four-fifths (82%) of all teenagers, serves as the flagship around which sub-brands are developed. For example, loveLife’s reality TV series S’camto groundBREAKERS is reinforced by E’scamtweni on a national commercial radio station, S’camtoPRINT in a national weekly newspaper, and lends its name to loveLife’s national volunteer youth service, groundBREAKERS (S’camto is colloquial for "let’s talk").

Radio is critical in reaching out, particularly to the poorest households. Over 99% of South Africans has access to radio, and analysis of loveLife’s coverage shows that the largest gaps are in low-income areas not served by radio stations on which loveLife has programming. Currently, a high priority is to secure new program partnerships with such stations, because the marginal benefit of reaching poor households is even greater than reaching better off households (income is a significant predictor of sexual behaviour). Radio also allows greater interaction with the audience than television, mainly through phone-in talk shows – a program format used to good effect by loveLife. For example, Youth Crossfire on YFM, attracts a weekly audience of 400 000 listeners. loveLife currently has radio programs on seven commercial and public radio stations, while its parent campaign commercials are aired on many more.

Print partnerships with the largest daily and weekly newspaper groups in South Africa have established a monthly circulation in excess of 2 million for loveLife’s two lifestyle magazines for young people, thethaNathi and S’camtoPRINT. Print media allows for considerable interaction with readers, and also provides a durable product for widespread dissemination. loveLife’s schools project currently disseminates copies of these magazines to 750 predominantly rural schools, in conjunction with loveLife-branded school name signage, also pointing learners to thethaJunction.

A minority (10%) of South African teenagers have regular access to the internet, and while loveLife’s website does enhance interactivity with its television programs, it is principally used to communicate with a broader national and international audience.

• Service delivery and institutional support

As Figure 20 suggests, greater access to relevant information generates a demand for new knowledge. Young people want to know what the information means for them personally.

Roughly 90% of young people have access to public or private telephones, and the first port of call for further information is often loveLife’s toll-free sexual health helpline for young people (thethaJunction 0800 121 900).

Twenty-four counsellors are on duty during the 50 hours of operation per week (after school hours). In addition to on-line counselling, callers receive publications posted to them after the call and may be referred to local organizations for further help. thethaJunction receives between forty and sixty thousand calls a month (equivalent to roughly 1% of the target population per month, although roughly a third are "hoax calls"). Despite rapid upscaling of the call centre, demand for thethaJunction has consistently outstripped supply, and service levels remain at about 70% of calls answered. Of young people who know of loveLife, 9% say that they have contacted thethaJunction. This means that approximately 335 000 different individuals have contacted thethaJunction since its inception less than two years ago.

Fortunately, South Africa has a relatively good telephone service. However, the poorest young people still do not have access to telephones. Radio communication is the most accessible form of mass media, while most young people (>95%) attend schools. New partnerships with regional radio stations and greater engagement through schools are thus key strategies to reach poor or very poor young people. However, such engagement cannot happen from Johannesburg, and a fundamental element of loveLife’s service infrastructure is the regional network of loveLife Y-Centres.

loveLife Y-Centres form the backbone of loveLife’s national infrastructure, providing hubs for regional networks of franchise holders, adolescent friendly clinics and outreach programs. They serve as best-practice sites for youth leadership development and adolescent clinical services; training venues for groundBREAKER and positive lifestyle programs; provide trained facilitators to support the loveLife Games and other outreach program; and sounding boards for testing and reviewing loveLife media products. All loveLife Y-centres meet a consistently high level of quality in the provision of all activities, including clinical and counselling services, training and support, sports and recreation, computer training and facilities. A number of cross-site initiatives enhance their activities, including sports and debating leagues, and the positive lifestyle and radio training initiatives. There are currently 13 loveLife Y-Centres fully operational, with a further four under construction.

Growth in demand for sexual health services in response to behaviour change initiatives – particularly for young people – is constrained by service delivery factors including: limited opportunity for youth participation in clinic activities and outreach; attitudes of health personnel; lack of training in adolescent care; and management systems better geared to paediatric or chronic care. In addition to addressing these factors, a sustained program aimed at improving sexual and reproductive health services for young people will simultaneously improve general management systems for effective information for prevention of HIV; prevention of mother-to-child transmission of HIV; syndromic management of sexually transmitted diseases; voluntary counselling and testing (or referral); treatment of opportunistic infections; and management of concomitant infections like TB.

The National Adolescent Friendly Clinic Initiative is an ambitious plan to transform public sector clinics, placing equal emphasis on improving both supply and demand of youth-friendly health care. Systematic quality improvement is implemented by clinic managers, assisted by external facilitation in full partnership with national and provincial departments of health. Initial clinic assessment against 40 criteria for adolescent-friendly services provides the baseline against which progress is assessed. The quality improvement process combines systems analysis, management and clinical training and problem-solving, and is supported by NAFCI specific resource material. All health workers and clinic support staff participate in values clarification processes. Direct involvement in loveLife programs has provided considerable incentive to health workers who have felt demoralized and overwhelmed by the extent of HIV/AIDS. In each clinic, a brightly painted "chill room" for young people is established, creating a friendly environment to talk, listen to music, read loveLife’s fortnightly lifestyle magazines S’camtoPRINT and thethaNathi, and have easy access to condoms. A "loveLife kiosk" placed in each clinic waiting room provides information for parents and young people and is run by two young volunteer groundBREAKERS. Young people participate in sexuality education, motivational development, debating and recreational challenges, and implement outreach programs into surrounding schools.

NAFCI clinics collaborate with loveLife franchise holders in their neighbourhood, promoting concerted, community-wide responses to HIV/AIDS. There are currently 57 non-government organizations participating in the loveLife franchise, in which NGOs committed to communicating positive lifestyle through popular youth culture receive training and access to loveLife’s resource and promotional material. The franchise also provides the opportunity to further extend loveLife’s national motivational, debating and recreational programs. The loveLife franchise was a response to a demand from NGOs to participate in loveLife’s national campaign, and has helped mobilize South African civil society.

Young people are pivotal in all loveLife’s efforts at societal mobilization, and loveLife has established a national volunteer program of groundBREAKERS" - 18-25 year olds, otherwise unemployed, mobilizing local action to prevent HIV/AIDS, STI’s and teenage pregnancy. These groundbreakers (currently numbering 350) form the mainstay of loveLife national team, implementing most of the programs in Y-Centres, clinics, franchise holders and outreach. They receive both leadership training and training specific to the tasks assigned to them, and are supervised and supported through the loveLife institutions in which they are based. For young South Africans, the hiatus in employment (about 60% of 17-25 year olds are unemployed), lack of educational, recreation and other opportunities create a vicious cycle of poverty and HIV infection. Fostering new leadership among young people in communities heavily affected by HIV/AIDS and demoralized by poverty may be a key strategy in helping break the vicious cycle.

• Outreach

Together, the service initiatives described above (sexual helplines, loveLife Y-Centres, adolescent friendly clinics, loveLife franchise and groundBREAKER volunteers) create a national institutional platform for loveLife. While all elements of this platform will continue to develop and expand, many young people across South Africa will not have direct, regular interaction with any of them despite their outreach into surrounding communities. Given that most young people (>95%) receive at least 8 years formal education, schools appeared to be an ideal venue for a national outreach strategy.

However, loveLife elected not to use the formal school curriculum as a means, nor to use the classroom as a primary venue for communication. This decision was based on the historically dysfunctional education system that is taking considerable time to transform; slow and politically-charged process of curricula reform in South Africa; high rates of sexual and emotional abuse of school children, especially by male teachers; and loveLife’s commitment to popular youth culture as its vehicle for communication. For these reasons, loveLife uses school sports as its primary entry point into schools.

The loveLife Games is a partnership between loveLife and the United Schools Sports Association of South Africa (USSASA) – mandated by government to develop schools sports. The loveLife Games has the active support of the Ministries of Education and Sports and Recreation and the National Olympic Committee of South Africa. Sport is the main activity of the loveLife Games, but participants are also introduced to new opportunities, skills and information. Positive lifestyle finds expression on the sports fields and graffiti walls, in vigorous debating competitions and motivational and leadership development programs, and in groundbreaking action sports. For many young people, some of whom have never before traveled on a tar road, the loveLife Games means far more than simply a few days’ outing. It’s an introduction to a lifestyle that is positive and attainable. A high level of professionalism has been achieved through a six-month training program for 1 800 enthusiastic teachers from all regions of South Africa, responsible for organizing the Games. The knock-on effect of this training is illustrated in Figure 21. Inter-school fixtures held in March and April see over ten thousand schools competing for regional honours. Forty-four Regional Games in May and June decide which schools participate in the Provincial Games held in July and August. The national loveLife Games take place in Durban from in October 2002. The Games are covered by a thirteen-week television series (loveLife Games TV).

Overall, both teachers and young people rate the loveLife Games very highly (Table 2).

The success of the loveLife Games was further demonstrated in the National Impact Survey 2001. Seven in ten young people who knew of loveLife (43% of total) knew of, or had participated in the loveLife Games.

This year, loveLife has teamed up with polar explorer Robert Swan and Mission Antarctica, leading up the World Summit on Sustainable Development. Six groundbreakers that spent three weeks in the Antarctic are leading a road tour (together with the yacht "2041") to most regional loveLife Games, illustrating that both our personal and global future depends on choices made today.

Over the next five years, loveLife will continue to develop the loveLife Games as one of its flagships for outreach. At present, much of our effort is directed at regional, provincial and national levels, with the teacher training designed to effect trickle-down into schools. There is huge demand to extend the high-intensity program down to district level, the feasibility of which is largely a function of finances.

Two other innovative outreach programs extend loveLife’s reach even further, and to young people in its target group that have left school.

The Love Tour is a program of events centred around an outdoor broadcast unit. These events include an arts program dealing with issues of self-actualization and identity; an extreme sports challenge (e.g. Artificial wall climbing) illustrating life challenge and overcoming obstacles; information and discussion with young people; and music and other live performances. School-based events are held through the week, and the Love Tours presence in a community culminates in a large public event over the weekend. The Love Tours are almost entirely run by groundBREAKERS, and operate a year-long schedule through KwaZulu-Natal and the Eastern Cape respectively. Roughly 30 000 people are reached every week through the Love Tours.

The bright purple Love Train has proven to be a major draw card for young people. This train, also equipped with a radio studio, provides a five-day long program of activities in small towns and rural sidings. The success of both the Tours and the Train depends in part on collaboration with adolescent friendly clinics and franchise holders in the vicinity of each stop. Part of their function is to seed new franchise holders and friendly clinics where none exist.

All of the above initiatives aim to give young South Africans an ongoing experience of the positive lifestyle promoted by loveLife. Significantly, of those exposed loveLife more than 80% knew of loveLife from more than 4 sources, implying some success in loveLife’s efforts to develop multiply reinforcing contact with young people (Figure 22).

Similarly, a majority (60%) of youth who have heard of loveLife report that they have seen or heard of at least four different loveLife products (Figure 23). Almost a third (30%) who report having heard of loveLife are aware of seven or more loveLife products.

It is early days yet. Almost four out of ten (38%) of young people had not been exposed to loveLife by end Year Two. Despite rapid rollout, neither was the extensive infrastructure described above in place from the start. And program management and quality is an on-going challenge. Nevertheless, the campaign is now of such magnitude and depth that projected effect over the medium term, based on realistic assumptions of impact on sexual adolescent behaviour, is possible. The validity of these assumptions may, to some extent, be gauged by actual changes observed to date. Following are projections of the impact of loveLife on HIV incidence for a single cohort of 15 year olds over the next three decades, using international precedents and South African baselines of behavioural mediators of HIV reduction.

• Projected impact of a national HIV campaign targeting SA youth

Table 3 shows the assumptions of behavioural change over the period 2003 to 2007 for 15 to 19 year olds as a result of an effective national intervention. Baselines have been set according to levels estimated for South African youth in 2002.

These changes were then matched as closely as possible to key input variables (5 risk group categories) of an existing HIV/AIDS demographic model to produce projections for HIV incidence, prevalence and AIDS mortality.

The model used (ASSA2000) was developed by the Actuarial Society of South Africa (ASSA). The ASSA models are modes of "the third kind" as they are neither limited to a simple functional extrapolation of the past development of the epidemic (e.g. Epimodel), nor do they involve complex micro simulations of individual behaviours. They can best be described as heterosexual behavioural component population projection models, meaning that they are typically used to project a population forward in time, allowing for future fertility, mortality and migration - to which has been added a model of heterosexual behaviour. The outcome is projection of the proportion of the population infected or likely to get infected with HIV in future.

This is primarily a demographic model, not readily designed to accommodate assumptions of successful intervention, although these can be incorporated to a limited extent. Notwithstanding these limitations, we believe that it can provide initial indication of the potential gains of effective national intervention programs.

The following graphs illustrate the cumulative impact of the positive behavioural changes outlined in Table 3, for a single cohort of 15 year olds. Effecting similar change in successive cohorts of 15 year olds will have an impact greater than the sum of single cohort effects, because of the compound effect on the size of the infective pool. Further details related to modeling are presented in Appendix 2.

• Projected Incidence

Without effective intervention, the incidence rate of HIV infections climbs rapidly to 7% in females aged 17, with a 2% reduction in incidence at this age as a result of the combined interventions. Incidence of HIV infection in males reaches a peak of 6.6% in 2012 when the cohort is 24 years of age compared to 4.6% as a result of the interventions (Figures 24 and 25).

The numbers of infections over time (2003 to 2032) generated by these cohorts with and without interventions are shown in Figures 28 and 29. Infections averted for females and males in this cohort for the period 2003 to 2032 number 38 578 for females and 37 182 for males.

Benefit-cost analysis is done on the single cohort trends described above, given the number of assumptions underlying semi-dynamic models of impact on successive cohorts. This analysis will thus give a conservative estimate of benefit-cost. The graphs show the projected impact of five years of effective intervention (2003 – 2007) for a single cohort of 15 year olds, and assume that behaviour changes occurring during that period are then sustained through life. Figure 30 shows the projected outcome at five years compared to the current situation.

Assuming that the impact of the campaign is uniform across the entire target group (12-17 years), the full cost of the HIV prevention campaign in any single year may be assigned to a single 15 year-old cohort. loveLife estimates that it requires a maximum of US $40 million per annum for at least another decade to implement an effective national HIV prevention campaign in South Africa. These estimates are based on its current budget of $20 million per annum and projected expansion in 2003 – 2007.

In the absence of effective intervention, the total number of infections projected for this single cohort from 2003 – 2032 is 419 190. The total number of infections decreases by 75 760 (18%) to 343 430 through effective intervention. Cost per infection averted is therefore estimated at $528.

The South African gross domestic product for 2001 was US $ 104.2 billion. Without effective intervention, HIV/AIDS is expected to reduce the South African gross domestic product by 0.3% - 0.4% per annum, i.e. by $312 – 417 million p.a. ($746 – $994 per infection). Based on the above, and using economic cost comparisons alone, the benefit-cost ratio of an effective HIV prevention campaign (versus no intervention) is 1.4:1 to 1.9:1.

These estimates demonstrate that incremental change effected through a five year HIV prevention campaign costing $40 million is cost-beneficial even for a single-year age cohort. While difficult to quantify from static or semi-dynamic models, the benefit-cost ratio is obviously considerably higher for loveLife’s entire target group (12-17 year olds). Furthermore, long-term gains will be greatly enhanced if significant behaviour change can be achieved earlier than predicted. Figure 31 illustrates the decline in new HIV infections achieved by a one-off change in key indicators of sexual behaviour in 2003.

Two important observations may be made from this chart. First, gains are most impressive in the first five years after successful behaviour change as a significant proportion of young people select themselves out of high-risk groups. Unless a substantial proportion of the cohort shifts en masse to lower risk categories, the infective pool will remain too large to trigger sharp reductions in HIV prevalence. This illustrates the need for a high-impact campaign of size and scope big enough to effect behaviour change now. There is no room for half-measures.

Second, the marginal gains in averting HIV infection as a result of one-off change (in 2003) decline over time, as young people who remain in medium- to high-risk categories get exposed to cumulative probabilities of infection. In order for HIV prevalence to change substantially, the new patterns of sexual behaviour need to be replicated through successive cohorts of young people. In addition, incremental improvements in key indicators of sexual behaviour off the new baseline are required to keep reducing the risk profile of individuals over the course of their lifetimes.

While an HIV prevention program producing incremental change is cost-beneficial over both short and long-terms, beating the deadlines for the behaviour change modeled above will substantially increase the number of infections averted.

The only definitive proof of successful intervention is a substantial decline in HIV prevalence deviating sharply from projections of the natural course of the epidemic. Although the time lag between declining incidence and prevalence rates among 15-20 year is relatively short, it will still take several years to say with certainty that a national HIV prevention campaign has worked.

Attributing change to a national campaign such as loveLife is even more difficult.

In the first instance, countless local initiatives and several national media efforts contribute to behavioural change. While a comprehensive national campaign may create and sustain the momentum for change, that campaign works largely by leveraging existing institutional capacity.

Second, there is likely to be systematic bias in surveys of self-reported sexual behaviour, as young people may choose to associate with positive behaviour change.

Third, survey costs associated with attaining enough statistical power to compare changes between young people exposed and not exposed to loveLife are considerable, and sample sizes required will only increase over time as the fraction of young people not exposed to loveLife diminishes. At the end of year two for example, there was a positive 14-percentage point difference (p<0.05) between those exposed and not exposed to loveLife in response to the question whether their sexual behaviour had changed as a result of HIV/AIDS. Given that up to 80% of young South Africans may be exposed to loveLife by the end of 2002, future surveys may be hard-pressed to show further gains in this regard.

Nevertheless, while the findings presented above are tentative indicators of change, they indicate at least a growing association by young people with positive lifestyle and responsible sexual behaviour. They provide some reassurance that loveLife is on the right track – enabling considerable numbers of young people to personalize and internalize the risk of HIV and teenage pregnancy. This is the first real evidence in South Africa that behaviour change among adolescents may be happening – potentially on a scale to change substantially the course of the HIV epidemic.

The study design is that of a nationally representative population based household survey of sexual behaviours, communications, HIV prevalence and other related behavioural and biological outcome indicators among youth aged 15-24, which will measure the effectiveness of loveLife’s interventions nationally over a five-year period.

Within the overall nationally representative population survey, "nested" pre–post evaluations of two of loveLife’s intensive community interventions (adolescent friendly clinics and Y-Centres), compared to control communities with exposure to elements of loveLife’s national campaign, including media and service interventions like the loveLife Games. The use of common questionnaires and biological testing procedures will assure comparability of outcome indicators from all surveyed communities.

The survey is powered (0.8) to detect a 50% decline in HIV prevalence among the age group 15-24 yrs (n=13 700).

Indicators measured are:

• Biologic endpoints: All study participants will provide oral specimens for HIV testing and those in the intensive sites communities (11 NAFCI, 11 Y-Centres and 11 Controls) will also provide urine specimens for testing of STI testing (chlamydia and gonorrhoea).
• Demographics: Key demographic indicators will be collected such as age, gender, ethnicity, race, type of housing, socio-economic status, religion, and community characteristics.
• Awareness/attitudes: Knowledge and attitudes with respect to HIV/AIDS, STI’s, pregnancy, gender norms, media, loveLife messaging.
• Communications: Nature of personal communication networks related to HIV/AIDS, and sexual health.
• Behaviours: Social influences, sexual history, sexual behaviours and change over time, condom, contraceptive use, and health seeking behaviours.
• Exposure to interventions: Both loveLife and other HIV/AIDS sexual health targeted media campaigns including call centre, radio, and TV print, billboards outreach and peer education. Service utilization includes Y-centre attendance, adolescent clinic attendance, and participation in other loveLife activities.

The survey will use a cross-sectional design based on a probability sample of the population in their homes. It is however critical that consecutive follow-up surveys in years 3 and 5 are able to establish sentinel trends. Therefore, a master sample will be developed on the basis of fixed enumeration areas (and their characteristics from the census). This will form the basis of subsequent surveys.

The prevention assumptions presented above were matched to the model in the following way.

This group is assumed to be for the most part beyond risk (5% are assumed to be at low risk rather than no risk). The growth in consistent use of condoms is accommodated by growing the NOT risk group. At 2002, the NOT group is assumed to be unaffected by intervention. The portion in the group as a result of consistent condom usage is assumed to be the same size as the "always use condom" group. The rest of the NOT group is assumed to be in this group for other reasons: abstinence, faithful to one partner, etc. The growth in this group is entirely as a result of the growth in those using condoms all of the time.

"Always use condoms" was assumed to be a subset of this group. It was thus assumed that for the purposes of the model, those that "use condoms come of the time" (say 30% of the time), would be the difference between "use condom last sexual act" and "always use condom". The ASSA2000 model has an intervention for this built in. The probability of condom use is not however applied equally to all risk groups. Those in the RSK group are assumed to be twice as likely to use condoms than all others. The condom usage applied to the model is thus a function of the size of the risk groups and the mean condom usage across all groups.

The ASSA2000 model has this specific intervention built in. This reduction is applied to all risk groups (this may be debatable). A question that arises here is the total sexual activity of an individual. We assumed that total sexual activity does decrease, but not to the extent that partners decrease. It has been assumed for the purposes of modeling that the total number of sexual acts decreases by half the decrease in partners (e.g.: partner decrease of 60% is accompanied by a 30% decrease in total sexual acts). This means, of course, that sexual acts per partner will increase while partner numbers decrease.

The given intervention refers to those in the 15-19 year old age group as opposed to the cohort of 14 year olds in 2002. This was modeled by changing the position of the sexual activity curve. It was assumed to predominantly be as a result of delayed sexual activity, affecting especially younger people. By slightly delaying sexual activity, the total of those sexually active in this younger group is reduced.

This would ideally have been modeled by slightly flattening the curve of sexual activity by age (i.e. those that become active early withdraw from sexual activity in larger numbers, thus lowering the activity later more so than at the beginning). This was assumed to be largely included in the modeling of sexually actives, and was thus not explicitly modeled.

The AIDS Committee of the Actuarial Society of South Africa (ASSA) was set up in 1987 to assist the actuarial profession (and later the wider public) in estimating the impact of the AIDS epidemic in South Africa. As part of this work the Committee has produced various papers and monographs on the epidemic and constructed models (the ASSA500 model released in 1996, the ASSA600 model released in 1998 and most recently the ASSA2000 suite of models) to predict the impact of the epidemic at both the national and regional level.

The ASSA 2000 model can be described as a model of "the third kind" as it is neither limited to a simple functional extrapolation of the past development of the epidemic (e.g. Epimodel) nor involves a complex micro simulation of individual behaviour. It can best be described as a heterosexual behavioural component population projection model. By this is meant that it is the usual type of model used to project a population forward in time, allowing for future fertility, mortality and migration to which has been added a model of heterosexual behaviour (in terms of the number of partners, sex activity, condom usage, age of partner, etc.) in order to determine what proportions of the population are infected or likely to get infected with the virus in future.

The model divides the population into groups differentiated by their level of exposure to a heterosexual epidemic. These groups are:

• PRO: Sex workers and their clients. Male PROs are assumed to have 200 new partners per year and female PROs 190 new partners per year.
• STD: Individuals regularly infected with sexually transmitted diseases. STD males and females are assumed to have 10 new partners per year.
• RSK: Individuals exposed to risky sexual behaviour but not infected with STDs. RSK males and females are assumed to have 1 new partner per year.
• NON: Individuals who not at risk of HIV infection.

The distribution of the population across these various risk categories, with their relative fertilities, is shown in Table A1. Fertility of HIV infected women is assumed to be around 30% lower than in uninfected women. However, fertility impacts are assumed to differ for infected women in different age groups. This assumption influences birth rates and the degree to which antenatal survey rates are likely to under or over-estimate community prevalence.

"Sexual activity" in this context is a combination of "number of new contacts", and "probability of successful infection" with each new contact. Rather than model number of contacts and probability of infection during a single sexual encounter, the model uses a "force of infection" to model the spread of HIV within each risk group and the neighbouring risk group.

The populations are therefore divided into the following groups, with each group’s calculation done on a separate worksheet within the workbook:

• Young: All individuals aged 0 to 13. The only infections assumed are those arising at birth. On their 14^th birthday individuals are allocated into the various other risk groups according to the assumed proportions in the population as a whole.
• Fem PRO: Female members of the PRO risk group subdivided by duration since infection up to age 59 years
• Fem STD: Female members of the STD group subdivided by duration since infection up to age 59 years
• Fem RSK: Female members of the RSK group subdivided by duration since infection up to age 59 years
• Fem NOT: Female members of the NOT group up to age 59 years
• Fem OLD: On their 60^th birthday all individuals are allocated to the OLD class. The duration since infection classification still applies although no further infections or fertility occur beyond this age. The OLD worksheets are just a run-off of the population. No one is assumed to survive beyond age 90.
• Male XXX: The same structure as the Fem XXX but with no births.

The probabilities of male to female and female to male transmission of HIV infection are shown in the next Tables A2 and A3:

As far as possible the parameters in the model are set based on current best estimates from a review of literature or empirical data. In this regard data from the most recent national Demographic and Health Survey (DHS) was used to determine the size of the population regularly infected with STDs, condom usage, and the age of partners. However, it is the nature of the epidemic and the nature of the model, that a number of the parameters are not known with any degree of accuracy. These parameters are set so that the model reproduced, as closely as possible, the national antenatal prevalence as well as the number of deaths estimated by the recently released South African Medical Research Council Report. Table A4 provides estimates for various parameters used in the model.

Estimates of condom usage by risk group are shown in Table A5.

As far as possible, the parameters of the ASSA2000 model have been set by reference to studies of empirical evidence. This was possible, for example, in respect of the size of the STD group, the probability of transmitting the virus, age of the partners, and condom usage. Where it was not possible, parameters are set within bounds of reasonableness to produce output comparable with observations of antenatal seroprevalence levels and estimates of the actual number of deaths based on the registered deaths.

Calibration involves adjustments of parameters that have not been estimated independently so that the ‘results’ of the model more or less match observed reality. In particular, the model results should match the results of the annual ANC surveys both in terms of overall level and by age. The model results should also match the number of adult deaths estimated on the basis of those recorded by the Department of Home Affairs on the population register after adjusting for an estimate of under-

The procedure used in the original calibration of the model was first to fit the prevalence figures for the PRO group. This group is roughly equivalent to commercial sex workers (CSWs) but not identical in that some CSWs will be at less risk than PROs as defined in the model because, for example, they use condoms and have STDs treated. The next step after fitting the PRO group, was to fit the STDs and finally the ANC attendees. It was assumed that most sexual activity occurs between partners in the same risk group. Further, by definition, members of the RSK group do not have sexual contact with members of the PRO group, and members of the NOT group do not have unprotected sexual contact with members of any of the other groups.

Based on evidence that suggests that women only visit a clinic late in their second trimester, the projection intervals run from the middle of one year to the middle of the next year. This assumes that women are, on average, six months pregnant when they first visit the clinic.

In order to model the prevalence of women attending public antenatal clinics rather than that of all pregnant women, the ASSA2000 model increases the age-specific rates of all pregnant women in the population by a multiple to allow for the fact that the prevalence of women attending public antenatal clinics is likely to be higher than that of pregnant women in general.

Adopting this approach, it is impossible to get anything but a crude fit to the PRO and STD target points. Modeled prevalence plateaus at a far higher level than the observed and ‘target’ points suggest. One possible explanation for this mismatch is that the survey and target CSW prevalence may include people from the (lower risk) RSK and STD risk groups, while the model definition of the PRO group is narrower, including only people at higher risk.

The ANC Age Profile sheet compares the prevalence by age from the antenatal survey with that generated by the model. It can be accessed by clicking the ‘ANC Age profile’ button on the Assumptions sheet. Calibration should aim to ensure as close a fit to the age profile of the antenatal prevalence data as possible by changing such factors as the female curve of sex activity and the age distribution of partners.

The Reported deaths worksheets contain charts which look at the fit with deaths recorded by the Department of Home Affairs. There are two such worksheets, one for male deaths and one for females. The user can view these worksheets by selecting the appropriate tabs on the workbook after running a projection. A full screen view is obtained by choosing the ‘Full screen’ option on Excel’s View menu.

The charts show the estimated number of deaths in the country based on the registered deaths by age for each of the years 1996 (calendar), 1997/98, 1998/99 and 1999/2000, as well as deaths projected by the model for the last year of the projection. The aim in calibration is to match as closely as possible the estimates of the number of deaths based on the registered deaths.

A similar approach it used to fit the urban-rural models to other countries. However, in this case the data are far sparser. Where there is data but it isn’t available for individual ages, or individual years, there are macros, which interpolate or extrapolate the necessary data. Where data simply does not exist or is considered to be too unreliable, default assumptions (largely based on those used to model KwaZulu-Natal) are used. Because countries to which this model is fitted do not have national antenatal surveys, the urban and rural prevalence against which the model is calibrated was determined as the median of the sentinel site observations, and the national determined as a weighted average of the urban and rural measures.