Estimating adult mortality rates in the context of the AIDS epidemic in sub-Saharan Africa: analysis of DHS sibling histories

Recent efforts to model the demographic effect of the AIDS pandemic in sub-Saharan Africa have far outnumbered empirical studies of adult mortality levels and patterns in AIDS-affected countries of the region. There is still a paucity of population-ba sed data on adult mortality for nearly all countries in the region. Using data from recent Demographic and Health Surveys (DHS) of six countries and one in-depth DHS, this paper examines the use of sibling histories to directly estimate rates of adult mortality. The countries studied include Uganda, Zambia, Central African Republic, Côte d'Ivoire, Zimbabwe, Malawi, and Tanzania. Rates of adult male and female mortality are presented at the national level in comparison to estimates obtained from other published sources, where available, and for subnational areas where cohort and other mortality studies have been recently conducted. The results indicate surprising consistency with external data and, on the whole, underscore the expected but hitherto only sparsely documented association between residence in high HIV-prevalence areas and sharply elevated mortality risk during the relevant adult ages. The cases of Zambia and Uganda in particular provide clear evidence of very high adult mortality levels among both men and women. In general, the findings of the study demonstrate that DHS-type sibling histories represent a promising, relatively untapped source of data that will add to our understanding of adult mortality dynamics in Africa. The paper discusses some of the advantages and potential limitations of the data and derived mortality estimates.