South African survey lifts lid off HIV epidemic, shows progress on prevention

A block-buster report on South Africa's HIV epidemic, published yesterday, challenges many existing perceptions of the scale of the problem and how South Africans are responding to it.

The report, backed by Nelson Mandela's Foundation and his Children's Fund, was carried out by the Human Sciences Research Council - a highly-respected social research agency - in partnership with the South African Medical Research Council, the Centre for AIDS Development Research and Evaluation, and the French Agence Nationale de Recherches sur le Sida (ANRS). The project was also funded in part by the Swiss government.

The main finding is that "HIV is a generalised epidemic in South Africa that extends to all age groups, geographic areas and race groups". 11.4% of South Africans are estimated from this survey to be living with HIV (95% confidence intervals 10.0-12.7%), excluding only those under the age of 2 and some significant populations such as those living in military camps, university hostels, boarding schools, prisons and hospitals. Among those aged 15-49, the figure is 15.2% (CI 13.9 - 17.5%).

Among the findings which may come as more of a surprise are:

• KwaZulu Natal's prevalence is not, in fact, the highest in South Africa - although the total number of people with HIV is large since it is the most populous province. It appears that antenatal surveys in the province have overstated the adult HIV rate since the centres surveyed treat highly mobile populations. Three provinces do now stand out as having higher prevalence than the others: these are Gauteng (which includes Johannesburg and Soweto), Free State and Mpumalanga. There is one province - Western Cape - where the overall figures, including among women, are higher than those previously reported from antenatal clinic surveys. (Antenatal clinic populations would be expected to give higher HIV rates, since the women seen there are more likely to be sexually active.)
• The impact of AIDS on South Africa's children is already massive. 13.0% of children aged 2-14 had lost one or both parents, though by no means all due to AIDS. 3.0% of households are described as "child headed" (i.e. with the oldest member of the household aged 12-18).
• Among children aged 2 to 14, the HIV prevalence was higher than expected, at 5.6% (CI 3.7-7.4%) and was nearly the same for boys and girls. For 27 (out of 86) who could be matched with a biological parent, 20 of those parents were HIV negative. This requires further explanation and may raise serious questions about the extent to which medical procedures have transmitted HIV in South Africa.
• The relationship between poverty and HIV risk appears to be complicated and differs between racial groups in South Africa. It is strongest for white, coloured and Indian populations, and weakest for black Africans, although there is a clear relationship between the kind of settlement in which people are living and the likelihood of their being HIV positive. Those living in urban areas in "informal" housing (squatter camps) are far more likely to be HIV positive and this accounts for much of the variation between and within provinces.

• The proportion of white South Africans aged 15-49 who are living with HIV is estimated at 6.2%, almost identical to the rate among coloured communities (mixed-race, mainly Afrikaans-speaking South Africans). This implies a generalised epidemic among whites and coloureds, beyond anything seen in Europe or the Americas. While there is a substantial degree of uncertainty attached to these figures, due to lower participation of white households in the survey, it seems unlikely that the resulting bias would understate HIV prevalence. Rates of HIV are lowest among Indians, and highest - at 18.4% in the 15-49 age group - among black Africans.

Survey methods

The report includes detailed descriptions of how the survey was carried out, with multiple levels of checking for the quality of the work undertaken, which is important for the credibility and value of its findings.

The household survey on which the report is based was a massive undertaking, in which 1,000 of the 80,000 "census enumerator areas" used in the national census of 2001 were chosen for sampling. Households were mapped using aerial photography to ensure that people weren't missed because of the lack of official records. Eleven visiting points were identified in each area, with three people to be interviewed at each of them. The aim was to reach 10,000 households: 1,200 Indian, 1,800 coloured, 2,200 white and 4,800 black African.

Four questionnaires were developed: for adults (25+); youth (15-24); carers of children (2-11); and for children (12-14). These covered, in appropriate ways:

• demographic information (poverty, education, religion, parental survival/orphan status)
• knowledge and communication about sex and HIV/AIDS in families, communities and the media
• sexual experience and behaviour including use of condoms, number of partners, etc
• traditional practices and experiences, e.g. circumcision
• general health status

All questionnaires were carefully piloted and translated from English into eight different South African languages. 171 recently retired nurses, all but 4 of them women, carried out the work on the ground, organised in 34 teams. They were recruited as people known to have a high level of trust across many communities in South Africa's still very divided society. Wherever possible, both race and language were matched between the field workers and the households contacted.

Of 14,450 potential participants, 13,518 were actually visited. 9,963 agreed to be interviewed of whom 8,840 agreed to give a saliva sample for HIV testing, of which 8,428 were adequate for testing. Overall, this is comparable, for example, to the UK National Survey on Sexual Attitudes and Lifestyles carried out in the year 2000. The response rate varied and was lower in some populations, which is acknowledged to have led to bias (although what matters is the direction of that bias).

All HIV tests were carried out in one of three hospital laboratories which are experienced in carrying out HIV antibody tests, run regular internal quality control procedures and participate in external quality assurance schemes, such as that run by the UK's Public Health Laboratory Service. Specificity for the system used (OraSure collection device with Vironostika HV Uni-Form II plus O antibody test) is 99%, which implies that less than 100 tests would have been false positives.

Sexually transmitted infections

"Although only 2.6% of participants said that they had been diagnosed with an STI during the last three months, 38.9% of these were found to be HIV positive, compared with 13.2% amongst those who had not been ..."

Voluntary counselling and testing

"Among respondents aged 15 years or more ... 18.9% said that they had previously had an HIV test and were aware of their HIV serostatus."

"When respondents who had not had an HIV test were asked if they would consider going for testing, 59.4% reported that they would consider a test if confidentiality was maintained, whilst 28.5% would be motivated by the accessibility, cost and quality of services."

Most South Africans (61.4%) see VCT as being accessible. The main reasons for not testing were perceived low personal risk and then lack of services for people who test positive.

Patterns of behaviour

40.2% of adult and youth participants claimed to have changed their behaviour in the last few years. Among sexually active youth (15-24), 57.1% of males and 46.1% of females reported using a condom the last time they had intercourse. This compares to a figure of 19.8% among 15-19 year old women in 1998 and is a remarkable change. There are also striking changes in reported sexual behaviour in terms of numbers of partners at one time.

"Our findings compare favourably with those found in other countries such as Brazil, Senegal and Uganda."

Knowledge, attitudes and political opinions

There is continuing uncertainty among as many as a quarter of the population about the link between HIV and AIDS and there are still large areas of uncertainty and ignorance, for example, concerning the risks of transmission of HIV through breast feeding. The great majority of people profess support for people living with HIV although, as the report points out, the hostile minority may still be very threatening to people with HIV and AIDS. 96.5% of the population agree antiretrovirals should be provided to prevent mother-to-child transmission of HIV and 95% agree that they should be provided to treat people with HIV and AIDS.