Testing the accuracy of demographic estimates in countries with generalized epidemics

Objective: To assess the accuracy of demographic estimates that include the effects of HIV/AIDS on adult mortality. Design: To compare estimates of demographic indicators based on UNAIDS/WHO estimates and projections with newly available estimates based on cohort studies, hospital records, national surveys and other sources of data. Methods: New information has become available recently from a number of sites in Africa on the ratio of mortality among the HIV-positive and HIV-negative population, the proportion of all adult deaths attributable to AIDS, and the number of orphans. These data are compared with the same indicators calculated from UNAIDS/WHO estimates to assess the accuracy of those estimates. Results: Differences between demographic indicators based on UNAIDS/WHO estimates and study-based estimates are generally within the uncertainty range of the UNAIDS/WHO figures. Conclusion: Demographic estimates based on surveillance data and demographic models are close enough to study-based estimates to be useful for advocacy and medium-term planning. However, significant differences do exist that should be taken into account for short-term planning.