Small area estimation techniques: focus on under-five mortality data in uganda

Abstract

Abstract

In Uganda, using survey data, estimates of under-five mortality have only been available at national and regional levels. This study utilized small area estimation techniques in a Hierarchical Bayes framework to derive estimates of relative risk of under-five mortality up to District level. The study utilized the Uganda Demographic and Health Survey data of 1995, 2001 and 2006 in the investigations. Results show that the Poisson-gamma model could provide reliable estimates for relative risk of under-five mortality. Results reveal that compared to the modeling approach, utilization of the traditional Standardized Mortality Ratio (SMR) could potentially be associated with very high undesirable coefficient of variations (>100%). The modeling approach has added advantage over the commonly used SMR by estimating under-five disease risk for a particular district and smoothening using adjacent district estimates. The study further reveals that it is possible to utilize small area estimation techniques together with national survey data to generate relative risk of under-five mortality for districts in Uganda. These results are potentially useful for targeting District decentralized system level of governance with high relative risk of under-5 mortality. Key words: under-five mortality, district, small area estimation, disease mapping, Poisson-gamma, log-normal, Uganda.