A Systematic Assessment of National, and Regional Under-Five Mortality Rate by Wealth Quintiles and Identification of Countries with Outlying Levels Using a Bayesian Hierarchical Time Series Model

Abstract

National-level estimates of the under-five mortality rate (U5MR) may mask disparities at the subnational level. We developed a Bayesian hierarchical time series model to assess disparities in the U5MR by household-level socioeconomic status (measured through wealth quintiles). In the model, country-specific ratios of wealth-quintile-specific U5MRs are modelled as the product of an expected ratio and a country-specific deviation from that expected ratio. We obtained quintile-specific estimates for 96 countries from 1990 (or the earliest year of data collection) to 2015 and pinpointed countries with unusually large or small disparities. We find that important differences exist between children from the poorest and richest households with respect to survival up to age five.