OBJECTIVE: This study (1) examines spatio-temporal variation in diarrhoea prevalence and (2) for 2013 identifies and maps the factors associated with diarrhoea prevalence at district level.
METHODS: Data were drawn from Demographic Health Surveys (1990, 1999, 2003, 2008 and 2013). Moran index was used to analyse spatial dependence and clustering of diarrhoea prevalence in 2008 and 2013. Geographical Weighted Poisson Regression analysis was used to identify factors associated with diarrhoea prevalence for 2013.
RESULTS: Diarrhoea prevalence was higher in rural than in urban areas. Prevalence exhibited statistically significant spatial variation, but temporal variation and spatial dependence were not significant. Locally, diarrhoea prevalence hot spots clustered among five states in the North East zone. Non-improved sanitation, children 6-23 months not breastfed, dung floor, relative poverty, unemployed mothers and Gini coefficient were main predictors of diarrhoea prevalence.
CONCLUSIONS: Results of spatial analysis improved understanding of local spatio-temporal variation in diarrhoea prevalence and underlying factors. Intervention strategies should emphasize behaviour change regarding washing of hands and feeding utensils before and after feeding children, exclusive breastfeeding, safe water, improved sanitation and hygiene, particularly in hot spot states.