Trend estimation of child undernutrition indicator at micro-level administrative units in Bangladesh using remote-sensed data
Conference
64th ISI World Statistics Congress - Ottawa, Canada
Format: IPS Abstract
Keywords: sattelite-data
Tuesday 18 July 10 a.m. - noon (Canada/Eastern)
Abstract
This study aims to estimate trends of chronic undernutrition (stunting) in under-five year old children in Bangladesh for micro-level administrative domains (64 districts and 544 sub-districts) using remote-sensed data (such as, night-time light intensity data as a proxy measure of urbanization; precipitation, land surface temperature and normalized difference vegetation index (NDVI) as environmental factors), with data from six rounds of the Bangladesh Demographic and Health Survey (BDHS) over the period of 2000-2018, in which surveys the geographic coordinates of the sampled clusters are available.
Bayesian multilevel time-series models are developed assuming the outcome variable follows a binomial distribution with the mean as the estimated stunting prevalence extracted from the micro-data.
These models borrow cross-sectional, temporal, and spatial strength in such a way that they can interpolate stunting levels in non-survey years for all small domains conditional on the relationship of outcome variables with the remote-sensed data. The remote-sensed variables make a significant contribution to borrowing strength over space and time. Estimates for higher aggregation levels are obtained by aggregation of the small domain predictions for examining numerical consistency in stunting prevalence from micro to macro levels.
Results show that the national-level trend is in steep decline over the period, from about 50% in 2000 to about 30% in 2018. The trends at district level shows some districts with higher stunting levels over the last two decades have consistently higher vulnerability, while others vary more. At sub-district, the direct estimates, which were too volatile and 30--50% domains were uncovered in all surveys, are considerably improved through using multilevel time series modelling.
Findings of the study provide data-driven evidence for monitoring the progress in meeting nutrition goals to date at the detailed administrative domains in Bangladesh, using the accessible remote-sensed data to improve the precision and reliability.