By Statistics South Africa

10 Oct. 2001


Statistics South Africa has difficulties with the Medical Research Council's findings that 40% of adult deaths are the result of HIV/AIDS.

There are many questions about HIV/AIDS that we in South Africa cannot adequately answer at present. For example, we do not have sufficient information about the probability of transmission of the disease for different groups of people in the country.

There is a wide range of reasons, which explains our present lack of knowledge. For example, HIV/AIDS is not a notifiable disease. Therefore, it is difficult to trace the disease and its manifestations from its sources of origin. We have incomplete registration of births and deaths in the country, particularly in rural areas. Death data are essential for monitoring the extent of the epidemic. We also need information on the sero-prevalence rate in the country as a whole. At present, this rate for the country is calculated from women attending public antenatal clinics, which is not necessarily a representative sample.

The only recourse we have at present regarding estimating the extent of the epidemic is through demographic modelling. This is why we welcome the Medical Research Council's report. It contributes to the debate on both the changing profile of the disease and causes of deaths in South Africa.

But this type of model has its own built-in risks.

* It may be based on incomplete and possibly inaccurate or unrepresentative data.

* Each model is based on certain underlying assumptions, which are sometimes difficult to test.

* Demographic modelling could over- or under-estimate the extent of a phenomenon such as an epidemic, depending on these assumptions.

* Assumptions about the probability of transmission of a disease may make the model unstable.

* Projections of mortality from a possibly inaccurate base (for example the 1985 base used in the MRC model) against which to compare changes in mortality over time is also open to question.

Stats SA therefore believe that at present it is difficult to model the HIV/AIDS epidemic adequately or accurately, since the country requires much better empirical data.

Stats SA would like to initiate such empirical studies be undertaken in the country which links life circumstances and living conditions of households, and attitudes and behavioural practices within them, to the prevalence of the disease.

The only way in which such a study can be undertaken is through bringing the household survey expertise of Stats SA together with the expertise of other bodies. For example, nurses in the Department of Health could draw blood or saliva samples from selected households for testing, the MRC could give conceptual and methodological inputs, and the expertise of the SAIMR or other bodies could be used for blood or saliva testing.

After wide consultation with relevant stakeholders, a questionnaire could be developed, which covers living conditions and life circumstances of South Africans, with particular emphasis on those circumstances that may place certain people at higher risk than others regarding contracting the disease. The survey could also ask questions on knowledge, attitudes and behavioural practices in relation to the transmission of the disease.

This questionnaire could be administered to a representative selection of households throughout the country. The size of the sample would depend on funds available, but it should be a large one (at least 60 000 households), since larger samples are required to measure the extent of less common phenomena.

Following on questionnaire administration, a sample of blood could be drawn from members of the household, or a sample of saliva extracted. This sample could be sent for HIV/AIDS testing. The data from the results of the tests could be linked to the questionnaire information. Analysis of the importance of each risk factor could then be undertaken, in relation to the presence or absence of HIV/AIDS.

There are indeed certain issues that would need to be addressed before such a study could be undertaken, for example, having the issue of the ethics of this type of research debated in public and through an ethics committee. Another issue could be how to test, for example, whether to embark on confidential, or anonymous testing, as against giving people information and counselling about their HIV status. The blood testing among pregnant women is done anonymously. Ongoing confidentially, for example throughout the linking process of questionnaire completion and the results of blood or saliva testing, would need to be guaranteed. Respondents would need to be assured on the issue very safe testing procedures.

In spite of these issues, such studies have been successfully undertaken in other developing countries such as Zambia. It is the only scientific way to determine the size of the epidemic.

Dr Sulaiman Bah at (012) 310 8678 or 083 305 5781
Dr Ros Hirschowitz at (012) 310 8955 or 082 454 6209