Geospatial analysis of food composition data to estimate population micronutrient intake

Micronutrient deficiencies are widespread globally with highest prevalence in low- and middle-income countries. Spatial mapping of dietary micronutrient supplies and deficiency risks is important for identifying subnational variations that are often masked by national averages. This is particularly relevant for mineral micronutrients where geospatial variability of grain composition could influence the micronutrient intakes and status of the population. However, at present, food composition data are typically reported in national or regional food composition tables (FCTs), a structure that precludes the reporting and use of finer scaled, georeferenced food composition data. Hence, the aim of this thesis was to assess the implications of using georeferenced food composition data to estimate risk of inadequate intake in context with localised food systems.

First, a scoping review (Chapter 3) was conducted to assess the availability and geographic location of food composition data and its metadata for use in sub-Saharan Africa. In Chapter 4, a framework to compile reproducible, reusable, efficient and transparent food composition datasets, was developed. Using selenium in Malawi as case study, in Chapter 5, the different levels of spatial aggregation of high resolution (~250 m) maize selenium concentration data to estimate dietary selenium intakes was evaluated. A geospatial model was developed and applied comparing 10 levels of spatial aggregation to georeferenced (i.e. point data) plasma selenium concentration data among women (15-49 years old). Finally, in Chapter 6, the smallest level of spatial aggregation for maize selenium concentration (small area) and single maize values from national and regional food composition data were matched to apparent food intakes for dietary selenium intake estimations. The implications of using small area selenium content of maize and maize products vs single (national level) values from Malawi and from publicly available FCTs for estimating the risk of apparent inadequate selenium intakes was evaluated for Malawi.

This study’s results highlighted the need for more geospatial information in food composition data, and the value of using high spatial resolution food composition data (including improved data systems), particularly when estimating sub-national apparent selenium intakes and inadequacy risks. The use of small area food composition data may increase the accuracy when identifying hotspots of micronutrient inadequacies which could help make informed decisions with regards to investments and resource allocation for nutrition surveillance, and the development and evaluation of interventions, particularly in contexts where dietary diversity is low and food systems are highly localised.