Comparison of direct and indirect childhood mortality trends in high HIV settings

Chifundo Kanjala, University of Cape Town

This study has the goal of assessing the consistency of the childhood mortality trends constructed from the direct and the indirect methods in high HIV settings. The data used are for Kenya and Malawi collected during the period 1990 – 2004. Trends of childhood mortality are constructed by applying the Brass Children Ever Born Children Surviving method to census and Demographic and Health Survey (DHS) data. Trends of childhood mortality are also constructed using direct estimates obtained from applying the synthetic cohort life table analysis. The results reveal that the trends of childhood mortality from the two methods are generally different. While the direct estimates generally give well defined trends, the trends from the indirect estimates are more erratic. However, being less erratic does not confirm the correctness of the trends. It is possible for the trends to be biased to the same extent and in the same direction.