complex and sophisticated statistical model that predicts the impact of HIV infection based on a series of information sources. A recent report models the impact of this disease over the next 10 years24. Over the last 5 years, the Human Sciences Research Council (HSRC) has undertaken two very large household surveys, which have been remarkable in their approximate agreement in estimating the overall number of infections31, 32. A further study of the youth presented both infection data and attitudes to HIV33.
Death certificate data, i.e. gender and whether the death is ‘natural’ or not (and hence an autopsy is required) are relatively accurately recorded and collected. While death certificate data do not allow for specific analysis of the causes of death, the steady increase in deaths in people in the 20 to 50 age group over the last decade is very striking. The classic peak seen after the age of 60, due to chronic disease, is now equaled by an earlier peak in the 20 to 50 age group. This peak is correlated strongly with the increased HIV seroprevalence observed in the antenatal surveys. Deaths in this group of young people have traditionally been due to trauma (so-called ‘unnatural deaths’ and more commonly affecting men), while the new peak comprises largely ‘natural’ deaths, and seems to affect largely women. It is probable that HIV infection is causing this new peak30, 34.
HIV infection was first noted in South Africa in 1982, with the death of a flight attendant who died of pneumocystis pneumonia35. The epidemiology followed that of the US, with the virus infecting predominantly homosexual men and haemophiliacs.
Haemophiliacs, their treatment revolutionised by the wider availability of pooled clotting factor for treatment, were further exposed to a product imported from San Francisco in the United States during a countrywide shortage. This product, pooled from a large number of donors was contaminated with HIV, resulting in the infection of a large number of South African patients. In other southern African countries the epidemic appears to have been spread hetereosexually from the beginning.
The virus appears to have spread rapidly to the heterosexual community in the late 1980s. It was firmly established as a generalized epidemic by 1991, when the first Department of Health – African National Congress survey demonstrated infection of 1% of pregnant women attending antenatal clinics, thus defining the epidemic as being
“generalized”36. Prevention strategies at this time focused largely on information-based media campaigns.
The epidemic continued to grow unabated over the next 15 years, although in a geographically inconsistent way. KwaZulu-Natal suffered explosive growth in the epidemic in the 1990s, for reasons ascribed to social disruption due to the political conflict in the province. Other factors included migration, lack of access to condoms, low circumcision rate, poverty resulting in sex work and female economic reliance on men, and gender violence. Other provinces such as Mpumalanga, Free State, Gauteng
and North-West province also demonstrated high prevalences, while Limpopo and the Eastern Cape, both very poor provinces, demonstrated significantly lower HIV infections. It appears, however, that many of these lower infection provinces are now rapidly approaching the levels in the other provinces, with the exception of the Western Cape, where the epidemic is predicted to stabilize at a lower level24.
Currently, the Department of Health estimates that 5.54 million South Africans (approximately 10.8% of the population) have been infected with HIV. The estimate by ASSA is of 5.4 million people infected, out of a population of 48 million, giving a prevalence rate of 11%, an incidence rate of 1.2%, and 600 000 people living with AIDS, in mid-2006.
The HSRC estimates from its household survey a prevalence of 4.8 million in those over the age of 2 years, Statistics South Africa estimates 4.5 millions, and UNAIDS, using its specific model, between 5.3 and 5.5 million. This level of concordance is remarkable, considering that the information sources in the case of the HSRC are completely independently collated, while the others rely on overlapping data sources24, 30–33.
The ASSA model suggest that prevalence rates are attaining a plateau across all the provinces, albeit at different levels, with KwaZulu-Natal estimated to have the highest prevalence, and the Western Cape the lowest. This plateau refers to the situation where death rates approximate new infection rates. Deaths from AIDS complications in SA are expected to peak in 2015. Life expectancy in South Africa in 2006 was estimated to be 49 in men and 53 in women, 13 years lower than they would have been without AIDS.
HIV infection further impacts on demographics by removing many women of child- bearing potential from the population pool, as well as causing a steady decrease in fertility in women who have HIV infection. The model suggests that after 2011, population growth will settle at 0.5% annually, although a UN report actually suggests that the country would enter negative population growth in 200536. The proportion of deaths directly attributable to the virus varies from province to province according to prevalence. While 47% of overall deaths in the country in 2006 were HIV-related, this was as low as 30%
in the Western Cape, while approaching 60% in higher prevalence (and more highly populated) provinces such as KwaZulu Natal, Mpumalanga and Gauteng24.
Race was measured in the HSRC survey, and despite sampling problems, has allowed fairly accurate predictions of racial vulnerability, with black Africans at highest risk, but with all race groups being heavily affected. Again, both biological and social issues have been advanced as reasons for this, and the likelihood is that complex interplay of the two accounts for the differences31, 32.
Gender and age dynamics play a large role in the spread of HIV infection. Surveys have indicated that young women, especially in the 15 to 24 age groups, are several-fold more vulnerable to HIV infection than their age-matched male counterparts (16.9% vs.
3.7%)24, 33. Deaths of women aged 20–49 years increased 150% between 1998 and 2003
according to analysis of death certificates, despite adjusting for population growth and possible improvement in registration30, 34. The vulnerability of women to HIV infection may aggravate other areas of gender inequality. For instance, illness due to HIV infection may make entry into the already difficult labour pool even more complex. Both social and biological theories to account for this vulnerability have been advanced, and prevention strategies targeting the youth have been a major focus of government programmes.
This group is of epidemiological interest, as it represents the most accurate pool of new infections (i.e. incidence), and a decrease in this number may indicate a decrease or delay in the age of infection. This was one of the significant changes that were noted in Uganda, which is often cited as a success in the prevention of HIV infection, where age at sexual debut was significantly elevated amongst women. Disappointingly, however, the HIV infection rate in this group in SA has remained relatively stable in the last few years, probably demonstrating a plateau in the number of new infections24, 30.
Gender violence undoubtedly plays a role in the transmission of HIV. Women who are exposed to violence at home have a higher HIV infection prevalence than those who do not. Sexual assault is associated with a high risk of HIV exposure, with 50% of rapists estimated to be HIV-infected in one provincial police survey. Poor coverage with post- exposure prophylaxis has meant that this seemingly effective measure is only available to a small number of assaulted people37, 38.
The South African government has committed significant resources to educational HIV prevention and support programmes over the last decade, and surveys indicate that knowledge about the transmission of the virus is relatively sophisticated, although perceptions of risk are disappointingly low31–33. This failure to effect significant behaviour change has been a major challenge. A further frustration has been the improvement in free male condom distribution throughout the country, again with seemingly little impact on the epidemic24.
Circumcision is a major protective factor in the epidemic, and there is strong epidemiological evidence that it plays a role in SA, and has been used to partially explain the provincial differences noted in HIV prevalence, specifically between KwaZulu Natal and the Eastern Cape26, 39.
Finally, there would be an estimated 505 000 deaths annually from AIDS complications by 2010, in the absence of antiretroviral therapy24. However, the increased accessibility of this highly effective but complex intervention has made this assumption more complex.
The implementation of the National Comprehensive Plan of the Department of Health (now strongly reinforced by the HIV/AIDS and Sexually Transmitted Infections Strategic Plan for South Africa, 2007 to 2011) is intended to be focused on provision of antiretroviral therapy (ART), which is highly effective in reversing the immunological deficiency induced by HIV infection. This will annually defer death in approximately
100 000 people with advanced HIV. While the lifespan in people on ART is not yet apparent, mathematical models (based on cohort data that have been available now for over 12 years of ART treatment) suggest average lifespans of over 20 years from the time of initiating medication. This presupposes an uninterrupted drug supply, however, and consistent coverage and adequate health facilities. The original model assumed that over 200 000 people would be on ART by the middle of 2006, and that just over 700 000 were in need of ART. It further suggested that without the provision of widespread ART, life expectancy would drop 19 years by 2015, while with ART it would drop by 16 years24, 40–42. New mass interventions being considered and researched include mass circumcision programmes, HIV vaccines and vaginal microbicides; the last two interventions are years away from demonstrating efficacy, while the first will be complex and costly, although undoubtedly effective. More targeted interventions such as chemical prophylaxis for ‘high risk’ groups are also being explored.
Children
In 2004, the antenatal HIV infection prevalence amongst pregnant women attending government antenatal clinics was 29.5%43. Table IV summarizes HIV prevalence by province and year indicating the lack of apparent leveling-off of the epidemic in reproductively active women. Of great concern is the increase in prevalence in adolescent and young pregnant women, with rates as high as 16% in women less than 20 years of age (Table V).
Table IV. HIV infection antenatal prevalence by province 2002–200443 Province
HIV prevalence (95% CI)
2002 2003 2004
KwaZulu-Natal 36.5 (33.8–39.2) 37.5 (35.2–39.8) 40.7 (38.8–42.7) Gauteng 31.6 (29.7–33.6) 29.6 (27.8–31.5) 33.1 (31.0–35.3) Mpumalanga 28.6 (25.3–31.8) 32.6 (28.5–36.6) 30.8 (27.4–34.2)
Free State 28.8 (26.3–31.2) 30.1 (26.9–33.3) 29.5 (26.1–32.9)
Eastern Cape 21.7 (19.0–24.4) 23.6 (21.1–26.1) 28.0 (25.0–31.0)
North West 26.2 (23.1–29.4) 29.9 (26.8–33.1) 26.7 (23.9–29.6)
Limpopo 15.6 (13.2–17.9) 17.5 (14.9–20.0) 19.3 (16.8–21.9)
Northern Cape 15.1 (11.7–18.6) 16.7 (11.9–21.5) 17.6 (13.0–22.2)
Western Cape 12.4 (8.8–15.9) 13.1 (8.5–17.7) 15.4 (12.5–18.2)
South Africa 26.5 (25.5–27.6) 27.9 (26.8–28.9) 29.5 (28.5–30.5)
CI: confidence interval. The true value is estimated to fall within the two confidence limits.
Table V. HIV infection prevalence by age group among antenatal clinic attendees, South Africa: 2002–200443
Age group (Years)
HIV prevalence (95% CI)
2002 2003 2004
< 20 14.8 (13.4–16.1) 15.8 (14.3–17.2) 16.1 (14.7–17.5)
20–24 29.1 (27.5–30.6) 30.3 (28.8–31.8) 30.8 (29.3–32.3)
25–29 34.5 (32.6–36.4) 35.4 (33.6–37.2) 38.5 (36.8–40.3)
30–34 29.5 (27.4–31.6) 30.9 (28.9–32.9) 34.4 (32.2–36.6)
35–39 19.8 (17.5–22.0) 23.4 (20.9–25.9) 24.5 (21.9–27.2)
40+ 17.2 (13.5–20.9) 15.8 (12.3–19.3) 17.5 (14.0–21.0)
The first national population-based survey conducted in South Africa in 2002 showed that the prevalence of HIV infection in children aged 2–14 years was 5,6% (Table VI)43. A second national survey conducted in 2004 showed reported rates of HIV infection in children aged two to four years to be 4.9 % in boys compared with 5.3 % in girls24. There are no national surveillance data available for the prevalence of HIV infection in infants.
Based on the prevalence of HIV among pregnant women, however, and assuming varying access to interventions to prevent transmission, it is estimated that between 45 000 to 60 000 children are newly infected annually in South Africa.
Table VI. HIV infection prevalence by age group, South Africa, 200243
Age (Years) N HIV-infected (%) 95% CI
Children (2–14) 2 348 5.6 3.7–7.4
Youths (15–24) 2 099 9.3 7.3–11.2
Adults(>25) 3 981 15.5 13.5–17.5
Total 8 428 11.4 10.0–12.7
Perinatally acquired HIV infection or HIV transmitted from mother to child accounts for the majority of paediatric HIV infections occurring in South Africa32. Mother-to-child
transmission (MTCT) of HIV infection can occur in utero, during labour and delivery or through breastmilk, with the bulk of transmission occurring in the intra-partum period44. Transmission rates vary from less than 2% in the developed world due to the use of highly active antiretroviral therapy (HAART), elective caesarean section and safe replacement feeding, to more than 30% in the developing world without access to antiretroviral prophylaxis and with prolonged breast-feeding45. MTCT rates vary in South Africa from 2% to 30% depending on use of interventions to prevent MTCT and the duration and method of breast-feeding46.Factors that are associated with MTCT transmission are as follows:
Maternal risk factors
These include advanced maternal disease or surrogate markers thereof, including viral load and CD4+ count, viral phenotype and genotype, smoking or other substance abuse, lack of ARV therapy or ARV resistance, sexually transmitted infections (STIs) or other co-infections and sexual behaviour.
Obstetric risk factors
These include vaginal or pre-term delivery, prolonged rupture of membranes, placental disruption or abruption, chorio-amnionitis, invasive fetal monitoring, and episiotomy or use of forceps.
Neonatal risk factors
Major factors include prematurity, oral thrush, gender and exposure to infected breastmilk.