AIDS stats, just how many of us are dying?

Author: Kerry Cullinan

Publishing date: 22.01.2004 00:00

Have AIDS deaths been exaggerated as South African writer Rian Malan claimed in a recent controversial article published in the British magazine The Spectator?

The current controversy over the accuracy of AIDS statistics sparked by Malan’s article is actually a debate about how many deaths we can expect in the next 10 years, but it does nothing to address our present reality in which at least 4.5 million South Africans are living with HIV.

Malan put both matters on the table in a recent article for the Spectator republished this month in the South African magazine Noseweek. He claims that AIDS deaths are "computer-generated estimates" that have been vastly and deliberately exaggerated.

He also says that anyone who raises doubts about the doomsday AIDS predictions is branded as "a moral leper bent on sowing confusion and derailing 100 000 worthy fundraising drives".

The controversy hinges on two issues: whether AIDS deaths have been exaggerated and whether there is a conspiracy to deliberately distort these deaths to ensure that huge resources are allocated to AIDS.

It is important to note that Malan and Noseweek editor Martin Welz have both flown the flag of the AIDS dissidents in the past, those who either dispute the existence of the Human Immunodeficiency Virus (HIV) or say that the virus cannot be the cause of the 29 illnesses that are now termed "AIDS-defining".

They thus have a political interest in the AIDS debate, which is perhaps why Malan and Welz, in the editorial of the latest Noseweek, imply that there is a conspiracy which includes the media to keep the AIDS figures high.

On the other hand, despite the many inaccuracies in Malan's articles which have been eloquently highlighted by the Treatment Action Campaign's Nathan Geffen, he is right about one important thing.

Actuaries concede that with more information about the epidemic, the prevalence in the population is probably lower than initially projected. The Actuarial Society of Southern Africa (ASSA) is currently revising its figures and expects its new model to be available in late February. This model is expected to produce a population prevalence that is around 90% of the estimates of the current 2000 model.

"Part of the problem with the Malan articles are his frequent insinuations that there is something sinister about such improvements to our projection models," says the University of Cape Town's Professor Rob Dorrington, one of the country's foremost actuaries dealing with AIDS.

"Actuaries do the best they can with the information available at the time," says Dorrington. "This means that models are changing constantly as we learn more about the epidemic. The new ASSA models are based on the latest research and ASSA is certainly not part of any conspiracy to exaggerate numbers or to 'quietly retire' our models."

Actuaries use mathematical modelling which is based on making various assumptions about available data. The main sources of their AIDS predictions have been the annual HIV prevalence survey of pregnant women collected by the Department of Health and the estimates of deaths produced by the Medical research Council. They use these figures then make a number of assumptions about relationships between the women and the general population in order to come up with an overall HIV prevalence rate for the country.

"But," says ASSA's AIDS convenor David Schneider, "modelling sexual behaviour is complex". There are a number of variables, such as the rate of HIV transmission, as well as a number of completely unknown parameters.

"A 10% decrease is significant and obviously we are delighted to see the numbers coming down. But there has never been any deliberate distortion," he adds.

Instead, says Schneider, the new model has developed because actuaries have had access to fresh research. The antenatal survey results have been complimented by the country's first representative HIV prevalence survey conducted by the HSRC in 2002. This was based on a saliva sample of 8 840 all South Africans over the age of two, and is thus more reliable than figures drawn only from pregnant women.

In addition, the new model includes different calculations about fertility (the initial model was based on countries that had a higher fertility rate than South Africa), and new calculations about the survival rate of people from HIV.

South Africans are also beginning to change their sexual behaviour, and this is having a positive impact of HIV prevalence particularly amongst younger people.

But this is the realm of mathematical assumptions and calculations that is far removed from human beings living with HIV and dying from AIDS.

Professor Carel van Aardt from Unisa's Bureau of Market Research bases his AIDS predictions on "hard data" such as available death-registration data.

"A few years ago, industry requested that we move away from modelling to more realistic data sets," says Van Aardt. "We have a few years of death-registration data available. This we break down according to age and sex and compare with what would be considered a normal mortality pattern for a country such as ours," he says.

The pattern that emerges from the death registration data tells a very clear story. Before 1996, the mortality pattern was as expected. But in 2000, for example, more than two and a half times the number of men aged 35 to 39 died and three times the number of women aged 25 to 29 died than were supposed according to a normal pattern of death.

"We can't assume that all these deaths are AIDS-related, as South Africa has a high rate of injury-related mortality. But even taking this into account, this cannot explain the excess deaths. In addition, there is an underregistration of death registration," says Van Aardt.

Based on 12 sets of "data", including sets with “hard data” such as death registration and antenatal clinic data, Van Aardt says his demographic-research division estimated an overall 2003 HIV-prevalence rate of 14.87 percent although he says that the HIV epidemic is at a different stage in every province so it is more useful to look at provincial figures than one national figure.

The HSRC figure for South Africans over the age of two was 11,4% and the new ASSA figure is expected to be closer to that.

Dorrington points out, there is not even agreement over how many South Africans there are let alone how many have HIV, while Van Aardt says weather predictions used to be nothing more than educated guesses.

While South Africans should be concerned that we get accurate AIDS figures, we have no reason to believe that there is a conspiracy aimed at denying these to us.

This article was commissioned by This Day and appeared on January 22. Please contact Health-e News Service for permission to reprint.