(P29-039-23) DietR - A dietary analysis tool for ASA24 and NHANES dietary data in R

Objectives: Our objective was to develop an R package, DietR, with functions to clean and analyze dietary data obtained through the Automated Self-Administered 24-hour Dietary Assessment Tool (ASA24) or National Health And Nutrition Examination Survey (NHANES) dietary data. 

Methods: We developed the DietR package to enable the analysis of dietary data with R. Package scripts, and associated tutorials guide users through analyzing food and nutrient intake (available at https://computational-nutrition-lab.github.io/DietR/). Diet R wraps functions from existing R packages such as vegan for application of ecological metrics (i.e. alpha-diversity) to dietary data and ggtree for visualization of tree-based metrics. Summary statistics of variables can be obtained, and the percentage of kilocalories from macronutrients can be visualized. Dietary pattern clustering analyses can be performed by principal component analysis (PCA), k-means, and principal coordinate analysis (PCoA) using a tree-based dietary pattern metric. Permutation-based statistical tests (i.e., PERMANOVA) can also be performed to test the separation of dietary patterns.

Results: As an illustration of the DietR package, we applied package functions to analyze the relationship between legume intake diversity and body measurements in NHANES 2015-16. We defined four dietary intake groups based on legume alpha-diversity (Shannon diversity, intake of multiple different legumes, nuts, and seeds consumed over two days) for n = 4,207. Waist circumference and BMI data were available for n = 4,038. Despite higher energy intake (mean difference 251 kcal, p-value < 0.0001), the highest legume alpha-diversity group had lower BMI (mean difference −2.3, p < 0.0001) and lower waist circumference (mean difference −6.2 cm, p < 0.0001), after adjustment for age, gender, kcal, and fiber as appropriate.

Conclusions: DietR offers a streamlined set of functions to clean and analyze dietary data obtained through ASA24 or NHANES. The application of these functions offers easy methods to derive a posteriori dietary patterns and to calculate dietary diversity metrics, which allows for the exploration of different dimensions of diet.

Funding Sources: Internal institutional start-up funds from the University of Minnesota supported this project