: JOHANNESBURG (Reuters) - South Africa has more people living
: with HIV than any other country in the world, the United
: Nations said on Thursday.

: The bleak assessment came after Pretoria earlier this week
: released data that showed 4.7 million South Africans, or
: one in nine of the population, were living with the deadly
: disease.

Well, if the UN said it, it must be true...right? :)

On the other hand:

-----

South Africa's HIV/AIDS statistics problem
by Anita Allen

The Citizen (forthcoming), Johannesburg, SA

Are HIV/AIDS statistics shocking or are they meaningless? After an analysis of the only survey and the cornerstone of all HIV/AIDS statistics in South Africa, I argue the latter.

An annual survey of HIV prevalence among pregnant women attending ante-natal clinics has been conducted by the Department of Health (DoH) since 1990. Its latest survey results for 2000 were released this week. Since this is the most credible source for HIV/AIDS statistics, including those from UNAIDS/WHO, it is important to fully understand the nature of the beast.

It is not possible to relate HIV infection with AIDS or "AIDS deaths" because HIV testing is not mandatory, HIV infection is not notifiable and AIDS cannot be recorded as a cause of death. Also, it is impossible to plot the so-called AIDS epidemic over time since, prior to 1994, the ante-natal clinic surveys were based on former provincial boundaries and mostly did not include data from former self-governing states. Nor is it possible to compare the surveys 1994 to 1997 with subsequent years because of the lack a consistent sampling frame.

What is left are the surveys from 1998 to 2000 which are comparable because only then was an identical protocol, methodology and interpretation used. These results show no statistical difference. This means the estimated HIV prevalence in South Africa is within the margins of error of previous calculations: 22,8% (1998), 22,4% (1998) and 24,5% (2000). Thus the DoH has concluded that the epidemic is stablising and that 4,7 million people in South Africa are now infected by HIV.

The department arrived at this figure by taking a sample of about 16,000 pregnant women at 400 clinics countrywide which contribute 40 samples each. These ACTUAL samples are then used via complicated formulae to calculate ESTIMATES of HIV prevalence for the entire population based on the 1991 census figures. This involves making a number of assumptions:

1. The prevalence of HIV in all pregnant women is the same as for pregnant women attending ante-natal clinics.

2. The prevalence of HIV in all women aged 15-49 years is the same as for pregnant women.

3. The male:female ratio of HIV positivity is 0,73:1

4. 30% of babies born to HIV- positive women will be infected with HIV.

5. HIV tests are specific for detecting HIV antibodies, and testing positive indicates HIV infection.

Assumptions 1 and 2 have no scientific evidence to support them, but the bias is clear. In SA 80% of all pregnant women, of whom 85.2% are black, attend public ante-natal clinics. Since few pregnant white or Indian women attend public health ante-natal clinics, the surveys are representative of pregnant black women. Therefore the bias would be towards an overestimate.

Assumption 2 relies on three premises:

- HIV tests are specific for detecting HIV antibodies and nothing else;

- HIV antibody positivity indicates HIV infection; and

- pregnancy does not affect the specificity of the test.

Several studies have shown that pregnancy is one condition which leads to false-positives. This is known by the manufacturer, whose instructions on the HIV test kits being used in South Africa, namely ELISA Abbott Axysm System for HIV1/HIV2, advises against using its ELISA on sera drawn from pregnant women.

Also, many studies have added to the list of more than 70 conditions in addition to pregnancy which may lead to false-positives. This includes viral infections of all kinds, alcoholic liver disease, hepatitis, blood transfusions, flu, herpes, TB, malaria. Since many of these are common in Africa, a significant number of its people would have antibodies to them, another reason to suppose that the bias in the annual ante-natal clinic surveys would be towards an overestimate.

The assumption that for every HIV positive pregnant woman there is an HIV positive man is founded on the assumption that HIV is sexually transmitted. Daft as it may sound, there is no evidence in the scientific literature of sexual transmission of HIV. On the contrary, the longest ongoing study by Padin, NS et al, has consistently reported that,. on average, a women must have 1000 unprotected sexual contacts with an HIV-positive man to develop antibodies to HIV. For a man, the number is 8000-9000 contacts with an HIV positive woman. Effectively, this means that sexual transmission of HIV is zero.

Even if one assumes sexual transmission then women would be 8 to 9 times more likely to be HIV positive than men so an extrapolation based on a 0.73:1 ratio would also be biased towards an overestimate.

Assumption 4, mother-to-child transmission (MTCT) of HIV, is hugely controversial. Even researchers who believe it is possible don't agree on the rate of transmission from mother to child, which is variously reported as between 10-43%. A recent study comparing mother/child pairs by Coutsoudis, A et al at the University of Natal compared MTCT from mothers testing HIV antibody positive who exclusively breast fed, exclusively formula fed and mixed breast/formula fed. The results showed no increased risk of HIV infection from mothers exclusively breastfeeding when compared to exclusive formula feeding - the rate being 19.4% at six months and 22,7% at 15 months.

Despite this, the annual ante-natal clinic surveys extrapolations are based on a 30% MTCT rate of HIV infection - the basis for the mathematically manipulated and much quoted estimate of 1600 babies in South Africa being born each month with HIV infection.

None of the ante-natal surveys or accompanying comments and analyses addresses the issue of Assumption 5. HIV tests, including the ELISA used in the ante-natal surveys, test for the presence of antibodies to HIV and not the virus itself. No convincing evidence on whether these are specific to HIV this has been published, nor tabled in the Presidential Aids Advisory Panel proceedings. The manufacturer of the ELISA acknowledges this in a disclaimer included on the instructions with its test kits: "At present, there is no recognised standard for establishing the presence or absence of antibodies to HIV-1 and HIV- 2 in human blood"!

In most western countries, the ELISA test used in South Africa is only a screening test and a person is diagnosed positive only if they test positive on a Western Blot test as well. Western Blot tests are not used in the ante-natal surveys and since 1996 no confirmatory tests have been done. In 1994 the false positive rate was nearly 18%. In 1991 it was 26%.

Since the surveys in 1998-2000 accepted a single ELISA test without confirmation, they definitely included false-positives in their results. A higher bias in these years towards an overestimate is therefore a given.

All of this implies that the annual survey - and all the other HIV/AIDS statistics - could be meaningless.

Fortunately, the isolation of HIV and the related pinpointing of any specific protein antibodies will be one of the tests conducted by the Presidential Aids Advisory Panel. Its Interim Report is now complete. According to Health Minister Dr Manto Tshabalala-Msimang it will "go to Cabinet at its next meeting on March 28, and will be released at a suitable occasion thereafter".

-----