Under-5 mortality is often used—perhaps implicitly—as a measure of “population health”. But what is happening to adult mortality in Africa?
In a recent working paperi , we combine data from 84 Demographic and Health Surveys from 46 countries, and calculate mortality based on the sibling mortality reports collected from female respondents aged 15-49. The working paper is available here  and the database we used for the analysis can be found here .
We find that adult mortality is quite different from child mortality (under-5 mortality)1. This is perhaps obvious to most readers, but is clearly illustrated in figure 1 . While in general both under-5 and adult mortality decline with per-capita income, and over time, the latter effect is much smaller for adult mortality, which has barely shifted in countries outside Africa between 1975-79 and 2000-04.
But in sub-Saharan Africa, contrary to under-5 mortality everywhere and to adult mortality outside of Africa, adult mortality increased between 1975-79 and 2000-04 and the relationship between adult mortality and income became positive in Africa as indicated by the upward sloping line in 2000-04.
This diverging and dramatic trend for sub-Saharan Africa is mainly driven by the HIV/AIDS epidemic.
Figure 3  separates out the African countries in which the prevalence of HIV was above 5 percent in 2001 and those where it was below 5 percent (according to UNAIDS 2010ii) and compares them with countries outside of Africa. The effects of high levels of HIV/AIDS on adult mortality are readily apparent, with the average adult mortality rate almost tripling between 1985-89 and 2000-04 (from a probability of around 1.7 percent to almost 5 percent) in the high prevalence sub-Saharan African countries. Importantly, however, mortality increased somewhat in the lower prevalence sub-Saharan countries as well, from around 1.6 percent in 1985-89 to over 2 percent in 2000-04.
Let’s pause and try to understand the human tragedies behind these statistics. The death toll from HIV/AIDS is larger than the casualties caused by conflict.
We chose the 5 countries for which we could compute mortality estimates for 2000-04 and which had the highest HIV prevalence in 2001: Lesotho (24.5%), Namibia (16.1%), Swaziland (23.6%), Zambia (14.3%) and Zimbabwe (23.7%). We compare them with African countries (Uganda, Côte d’Ivoire, DR Congo, Liberia and Sierra Leone) for which we could also compute mortality estimates for 2000-04 and which experienced conflicts (the right panel of figure 4 ). Notice that some of these countries have also fairly high levels of HIV prevalence. In Uganda mortality continued to increase after the end of the conflict in 1986—a result likely related to the increase in HIV prevalence (which was 7.0 percent in 2001).
In the countries in which HIV is widespread, namely in the southern African countries, mortality rates are higher than those in countries that experience conflict. The exception is Rwanda (figure 5 , with a different scale for the vertical axis) where conflict escalated into genocide—and where this appears as a dramatic peak in mortality which subsequently reverts to non-conflict levels. By contrast, elevated levels of mortality in the high-HIV prevalence countries are sustained and, to-date, increasing.
The news cycle alerts us regularly about the dramatic developments of conflicts in Africa. But the war that is really ravaging the continent is a silent one, killing slowly.
1 Child mortality is defined as the estimated number of children who will die before their 5th birthday out of 1000 live births in any given year. Adult mortality, as we define it here, is the probability that an adult ages 15-49 who is alive at the beginning of any given 5-year period dies during that period.
i de Walque, Damien and Deon Filmer. "Trends and Socioeconomic Gradients in Adult Mortality Around the Developing World", World Bank Policy Research Working Paper 5716, June 2011.
ii UNAIDS. 2010. Global report: UNAIDS report on the global AIDS epidemic 2010. UNAIDS. http://www.unaids.org/globalreport/Global_report.htm .
FIGURES (click on each figure to see a larger image)