Abstract
This thesis is about the spatial analysis of public health issues in sub-Saharan Africa. In a first study, we analysed the spatial variations of HIV prevalence and the socio-behavioral determinants among people aged 15 to 49 in Burundi using 2010 Demographic and Health Survey (DHS) data. In a second study, we analysed the evolution of the spatial distribution of stunting prevalence among under-three children in Côte d’Ivoire using 1994, 1998-1999 and 2011-2012 DHS data. For both studies, we first predicted the prevalence levels of the public health issue under study in unsampled points by the kernel method or ordinary kriging. We then identified clusters using the Kulldorff’s method and zones of different risks using the oblique partitioning method of the study area. Finally, we used spatial modeling to identify associated risk factors. These spatial analysis techniques have not only many advantages but also many limitations that we emphasize.