Abstract
In South Africa, deaths from HIV/AIDS are often misclassified as being caused by another condition, according to a study by IHME researchers. The study,
Exposing misclassified HIV/AIDS deaths in South Africa, found that more than 90% of HIV/AIDS deaths from 1996 to 2006 were incorrectly attributed to other causes.
Research objective
HIV/AIDS is highly prevalent in South Africa, with HIV/AIDS cases in the country accounting for 17% of the global burden of HIV/AIDS in 2007. Studying HIV mortality using data from South Africa’s vital registration system is difficult because the data are often based on inaccurate cause of death coding. Issuers of death certificates may be unaware of an individual’s HIV status, in part because fear of stigma prevents people from getting tested. These factors result in an underreporting of deaths from HIV/AIDS.
This study was conducted to quantify deaths from HIV/AIDS that are wrongly attributed to other causes in South Africa’s death registration data and to adjust for this bias. This research is part of ongoing work by IHME to provide rigorous, comparable, and current scientific measurement of causes of death.
Research findings
In 1996 to 2006, 2.0% to 2.5% of all registered deaths in South Africa were attributed to HIV/AIDS. However, this reflects only about 10% of all HIV/AIDS mortality. The study found that the true HIV/AIDS cause-specific mortality fraction rose from 19% to 48% over that period, indicating that as many as 94% of HIV/AIDS deaths were misattributed to other causes. The misclassification was especially evident in young to middle-aged women and middle-aged and older men, although the uncertainty ranges of the various sensitivity analyses are wide.
Of the 48 causes of death examined, 14 were identified as “source” causes, or causes to which HIV deaths were likely misclassified according to the assumption that true relative rates are largely driven by biology. These causes were: tuberculosis; sexually transmitted diseases excluding HIV infection; intestinal infectious diseases; selected vaccine-preventable diseases; parasitic and vector-borne diseases; meningitis and encephalitis; respiratory infections; other infectious diseases; maternal conditions; nutritional deficiencies; endocrine, nutritional, blood, and immune disorders; noncommunicable respiratory diseases; other digestive diseases; and “garbage” codes (including ill-defined and unspecified causes of death that should not be underlying causes). Relative rates of death from these 14 causes were higher in South Africa than in the global standard for people aged 20 to 45 years, and most of these relative rates also displayed a clear time trend paralleling the rise of the HIV infection epidemic. In people aged 5 and older, garbage codes, tuberculosis, respiratory infections, and respiratory diseases (only in people aged 60 or older) contributed most to death misclassification.
Analytical approach
Data on deaths were compiled from the World Health Organization’s (WHO) mortality database according to a list of 48 causes of death, and population data were obtained from the United Nations World Population Prospects 2008. Data were grouped as South African or global (all other countries), and South African vital registration data were corrected for underrepresentation of deaths based on a model from the Actuarial Society of South Africa. Age-, sex-, and cause-specific death rates were computed for both groups, and reference death rates were identified for the global and South African datasets by averaging the death rates in several older age groups for each sex and cause. Relative death rates were computed by age, sex, and cause as the ratio of the observed death rate to the reference rate.
For the 48 causes of death examined in this study, the researchers used the relative death rates to identify source causes, meaning they were causes to which HIV deaths were likely misattributed. Causes without misclassification showed consistent relative rates between South African and the global data, and causes with misclassification showed marked deviation from the global relative rates in people aged 20 to 45 years. For the source causes, the global relative death rates were applied to each year of South African reference rates, and the excess deaths in the source causes were assigned to HIV/AIDS.
Policy implications
High-quality health statistics are instrumental in health planning, decision-making, program evaluation, and monitoring progress. The researchers suggest that this empirical method could be useful for improving data quality and accurately estimating HIV/AIDS mortality in South Africa, although rigorously training physicians to properly prepare death certificates should be the ultimate goal for countries looking to improve the quality of their death certification data. This approach could be transferable to other settings with moderate-to-large epidemics of HIV infection where death registration data may not accurately reflect HIV/AIDS mortality.