HIV & AIDS - Appendix 3 - STATS SA response to the Medical Research Council's interpretation of deaths

VIRUSMYTH HOMEPAGE

PRESIDENTIAL AIDS ADVISORY PANEL REPORT
APPENDIX 3
STATS SA response to the Medical Research Council's interpretation of deaths

(Officially Approved)
Stats SA
Pretoria, Tuesday 11 July 2000

The analysis of deaths statistics is very important in shedding light on the changing profile of diseases and causes of death. The statistics on overall deaths show the total number, and its age and sex distribution. It is only through the analysis of causes of death that one can attribute the observed level of mortality to certain causes of death. However, even with the availability of causes of death statistics, the problem of not stating the cause of death or mis-stating it, leads to some uncertainty in determining the exact contribution of the different causes of death.

The Medical Research Council (MRC) is reporting to have observed an increase in adult mortality in the younger age groups and is attributing this to HIV/AIDS. It supports this view by comparing 1999 statistics with that of 1990. The 1990 deaths statistics is shown to have the distribution of deaths towards the older ages (pre-AIDS) while the 1999 deaths is shown to have the distribution of deaths towards the younger adult ages. Stats SA has several problems with this interpretation and would like to suggest alternative approaches to the problem. The main points are as follows:

1. The 1990 deaths statistics are not geographically comparable with those of 1999.

The breakup of South Africa into the RSA (former RSA) and the former TBVC states took place over the period from 1976 to 1990. As each TBVC state ‘gained independence’ it was removed from the South African statistical system. The 1990 statistics were therefore referring to death statistics in which the former TBVC states were excluded, while the 1999 statistics included all these former homeland ‘states’.

2.The distribution of deaths among Africans and those of non-Africans is different

In South Africa, Africans are a demographically younger population than whites or Indians or other Asians. As such the distribution of deaths among Africans is different from that found among the other population groups. The distribution of deaths among Africans tends to be more concentrated in the younger ages, and in the young adult ages while the distribution among whites and Indians or other Asians is more concentrated in the older ages. The Figures 1 and 2 attached show the 1990 distribution of deaths among Africans in the former RSA and Figure 3 shows the distribution of deaths among whites, Indians or other Asians and coloureds. Figure 1 shows that among African males in RSA in 1990 there was an appreciable number of deaths in the younger age groups. Among African females however, Figure 2 shows that this pattern is less marked. Among Indians and other Asians and whites, the pattern is reversed with an appreciable number of deaths in the older age groups. The 1999 deaths profile show similar profile as the 1990 one for Africans, albeit toned down a little because of the contribution of the other population groups. Because of the larger number of Africans in the population, they tend to swamp the graph. But large numbers of Africans were excluded from the 1990 illustration of the Medical Research Council. As such, the 1999 profile is not a drastically new profile as portrayed.

3.Causes of death statistics show that the largest proportion (27%) of deaths among males were attributed to unnatural causes.

The causes of death statistics for 1995 for all population groups and all of South Africa show that the largest proportion of causes of death among males was unnatural causes as shown in Figure 5. The breakdown shows that 27% of South African males die of accidents and violent deaths. This profile is therefore not necessarily fully attributable to HIV/AIDS.

4. A skewed distribution of deaths does not directly translate into a skewed distributed of death rates.

The profile of deaths reflects the effect of the age composition. A high number of deaths do not mean a high death rate. Only when the age structure is taken into account can we begin to say that deaths is high in certain age groups. The deaths rates computed for 1996 do not show a remarkably skewed distribution.

[Graphs attached]

See the .pdf file for Graphs (134 pages 1 Mb).

Contents | Continue

VIRUSMYTH HOMEPAGE