Secondary Logo

Journal Logo


Identifying deaths from AIDS in South Africa: an update

Groenewald, Pama; Bradshaw, Debbiea; Dorrington, Robc; Bourne, Davidd; Laubscher, Riab; Nannan, Nadinea

Author Information
doi: 10.1097/01.aids.0000166105.74756.62
  • Free

On the basis of a 15% sample of death notification forms for the period 1997–2001 processed by the national statistical office, which gave the underlying cause of death as HIV for approximately 8.7% of all deaths [1], Groenewald et al. [2] suggested that the official mortality statistics for South Africa underreport HIV/AIDS deaths, with only 39% of HIV/AIDS deaths being reported as such. The national statistical office recently published a new mortality report based on all registered deaths between 1997 and 2003 [3]. In this report, deaths with the underlying cause recorded or coded as HIV account for a much smaller proportion (2.1%) of the total deaths than was the case with the 15% sample.

This discrepancy is probably mainly attributable to a change in the coding system, from one which allowed for some interpretation of the details provided on the death notification forms to a strict automated mechanistic coding practice using Automated Classification of Medical Entities software (ACME 2004.02) developed by the United States National Center for Health Statistics [4]. The strict coding practice results in AIDS cases certified with synonyms or euphemisms such as ‘retroviral disease’ or ‘acquired immunosupression’, being classified as ‘other viral diseases’ or ‘certain disorders of the immune system’, instead of AIDS if there is no other immediate cause recorded. If the immediate cause of death (e.g. tuberculosis) is specified with these synonyms or euphemisms, this will be identified as the underlying cause, in favour of a less well specified condition.

The impact of the change in coding practice can be seen by comparing the leading causes of death from the sample data for 1997–2001 with those from the complete data for 1997–2003 (Fig. 1). The trend in HIV disease for the period 1997–2001 in the sample data is upward, increasing from approximately 4.2 to 8.5% over the period. However, according to the new coding of all death notification forms, the proportion of deaths caused by HIV remains virtually constant at 2% over the same period. In contrast, the proportion of deaths caused by tuberculosis, influenza and pneumonia and intestinal infectious diseases increase over this period, and account for a larger proportion of the deaths in the full data than in the sample. In addition, deaths caused by ‘certain disorders of the immune mechanism’ rise from 1% to approximately 2.5% over the years according to the new report, whereas there were very few (< 0.02%) in the sample data. Clearly, with the new coding system, a larger proportion of what would previously have been coded as HIV deaths have been coded to the immediate cause of death (tuberculosis, pneumonia and intestinal infectious diseases) or to an ill-defined immune disorder than was the case with the sample data. A rough estimate, without access to the detailed data, suggests that only approximately 8% of all AIDS deaths in 2002 were coded as such in the new report.

Fig. 1:
Proportion of deaths caused by AIDS-related conditions, events of undetermined intent (injuries) and ill-defined causes of mortality for Statistics South Africa sample data 1997–2001 (top) and Statistics South Africa total registered deaths 1997–2003 (bottom).
Fig. 1
Events of undetermined intent;
Fig. 1
ill-defined causes of mortality;
Fig. 1
Fig. 1
HIV disease;
Fig. 1
influenza and pneumonia;
Fig. 1
intestinal infectious disease;
Fig. 1
certain disorders involving the immune mechanism. *2003 incomplete data. Source: Adapted from Statistics South Africa 2002 [1] and Statistics South Africa 2005 [3].

Statistics South Africa are to be congratulated on producing a very useful report on the causes of death between 1997 and 2003, which clearly states the causes of death that are recorded on death notifications in South Africa and describes how the underlying cause of death has been coded. South Africa is among a few countries in sub-Saharan Africa that produce cause-of-death statistics. This report has fuelled the debate about the extent of mortality as a result of AIDS in South Africa, a debate that at times has overshadowed the most important finding in the report, namely, the huge increase in mortality of 57% over the period 1997–2002, concentrated in young adults, especially women, between the ages of 20 and 44 years and children less than 5 years. Only a small proportion of this increase (approximately 10%) can be explained by population growth and increased completeness of death registration (approximately 5%). Given that the number of deaths as a result of unnatural causes decreased slightly between 1997 and 2003, the increase in deaths is attributable to natural causes. The cause-of-death pattern between 1997 and 2003 demonstrates marked increases in AIDS indicator conditions, which together with evidence from the antenatal seroprevalence surveys that HIV seroprevalence has been rising steadily during the 1990s [5] strongly suggests that AIDS is responsible for this increase.


1. Statistics South Africa. Causes of death in South Africa, 1999–2001: Advance release of recorded causes of death. Statistical release P0309.2. Pretoria: Statistics South Africa; 2002.
2. Groenewald P, Nannan N, Bourne D, Laubscher R, Bradshaw D. Identifying deaths from AIDS in South Africa. AIDS 2005; 19:193–201.
3. Statistics South Africa. Mortality and causes of death in South Africa, 1997–2003: findings from death notification. Statistical release P0309.3. Pretoria: Statistics South Africa; 2005.
4. United States National Center for Health Statistics. Automated Classification of Medical Entities (ACME) software version 2004.02. Available at: Accessed: 23 February 2005.
5. Department of Health. Summary report: National HIV and Syphilis Antenatal Seroprevalence Survey in South Africa 2003. Pretoria, South Africa: Directorate Health Systems Research, Research Coordination and Epidemiology, Department of Health; 2004.
© 2005 Lippincott Williams & Wilkins, Inc.