Non random distribution of child undernutrition in Ethiopia: Spatial analysis from the 2011 Ethiopia demographic and health survey

Abstract

Background: Child undernutrition showed geographical inequalities due to variations in contextual determinants from area to area which indicates that location is an important factor in child undernutrition. However, there are limited studies on spatial epidemiology of child undernutrition in Ethiopia. This study was aimed to identify the SaTScan spatial clusters of child undernutrition in Ethiopia. Methods: Nutritional indices of children (0-59 months) with Global Positioning System (GPS) location data were accessed from the 2011 Ethiopia Demographic and Health Survey (EDHS) after getting permission from the MEASURES Demographic and Health Survey (DHS) Program. The Bernoulli Model was fitted using SaTScan™ software, version 9.4. for SaTScan cluster analysis. Log Likelihood Ratio (LLR) test was used for each SaTScan cluster and size of the scanning SaTScan cluster to test the alternative hypothesis that there is an elevated risk within the SaTScan cluster compared to outside the SaTScan cluster. Less than 0.05 for LLR was considered as statistically significant level. Results: The SaTScan spatial analysis result detected Liben, Afder and Borena administrative zones around the South East Ethiopia as the most likely primary spatial SaTScan clusters (LLR = 28.98, p < 0.001) for wasting. In the Northern, Middle, North East and North West areas of Ethiopia particularly from all administrative zones of Amhara, Tigray, Afar, Ben. Gumz regional states and East Welega and North Showa zones from Oromiya Regional State (LLR = 60.27, p < 0.0001) were detected as the most likely primary SaTScan clusters for child underweight. Also in the Northern, Middle, North East and North West areas of all administrative zones of Tigray, Amhara, Ben. Gumz and Afar regional states and West and North Showa and East Welega from Oromiya Regional States (LLR = 97.28, P < 0.0001) were primary SaTScan clusters for child stunting. Conclusion: The study showed geographical variability of child stunting, underweight and wasting in the Country which demands risk based local nutritional interventions. Further study will be important to assess the temporal nature of the problem and to identify community level factors that create this spatial variation.