(P24-023-24) Sociodemographic Factors Associated With Nutrient Diversity in Ghana

Objectives: In the nutrition transition, predominately staple-based diets tend to diversify to include highly processed foods with low variation in nutritional quality. Food trees, or dendrograms, can be used to assess food diversity based on differences in the foods’ nutrient content. The study objectives were to create a dendrogram of foods purchased in Ghana, use the dendrogram to assess the nutrient diversity of household food purchases, and to examine relationships between nutrient diversity and sociodemographic factors.

Methods: This study analyzes nationally representative cross-sectional food expenditure data collected from 13,679 households from the Ghana Living Standards Survey 7 (GLSS 7) (2016-17). We created a dendrogram of the 100 nutritionally distinct foods in the survey list of foods purchased for home consumption. To equally weight nutrients in the dendrogram, nutrients were scaled by the adult male RDA/AI and standardized (µ = 0, σ = 1). We scored nutrient diversity by summing the cophenetic distances between all foods the household purchased, and expressing that sum as a percentage of the total summed distances in the dendrogram of 100 foods. To assess relationships between household nutrient diversity scores and sociodemographic factors, we used multivariable linear regression with survey sampling weights.

Results: The food dendrogram indicated nutritionally distinct foods (e.g., oils, nuts & seeds, or pulses) amidst broad clusters of nutritional similar foods (e.g., refined grain products, fruits, vegetables, or fish and meat). The mean household nutrient diversity score was 28.1 (σ = 12.2) with a range of 0.5 to 73.8. Female-headed (β = 3.8 , p </em> < 0.001) and maternal households (β = 4.4, p < 0.001) were positively associated with the nutrient diversity score, as well as the household’s log-transformed monthly income (β = 4.5, p < 0.001). Residence in rural areas (β = -2.8, p < 0.001) and in northern regions of Ghana (β = -7.4 to -10.2, p < 0.001) exhibited negative associations.

Conclusions: Female-headed and maternal households were positively associated with nutrient diversity in Ghana, independent of household income and residence context. Future research should use food dendrograms to explore relationships between nutrient diversity and individual nutrition outcomes across food systems types.

Funding Sources: U.S. Department of Agriculture