Visualization of Dietary Patterns

Abstract

Diet quality indices, such as the Healthy Eating Index 2010 (HEI‐2010), include numerous components to capture the multidimensionality of diets, but typically only the total index score indicative of overall diet quality is presented. This project examined the merit of radar graphs for depicting multidimensional dietary patterns and how those patterns vary in the US population over time and by different characteristics.National Health and Nutrition Examination Survey (NHANES) dietary intake data were used to estimate HEI‐2010 component scores for the US population, over time (from 1999/2000 to 2009/2010), by socioeconomic status (i.e. poverty income ratio), and by age and sex categories. Each component score was scaled from 0‐100% and plotted. A perfect HEI‐2010 score, representing a diet that conforms to the 2010 Dietary Guidelines for Americans, would be depicted as a line around the border of the radar plot.Points on all axes of the radar plots show the relative scores for the components, and the overall shape provides an indication of the dietary pattern, with larger shapes generally indicating higher dietary quality. The spacing between different groups' points on each axis shows the difference in quality for that component between those groups. The plots show that some components—such as whole fruit and dairy—vary greatly, while others—such as sodium and total protein foods—vary less. The components that vary greatly drive the differences in overall diet quality between groups.Radar plots are a useful tool to conceptualize dietary patterns. They allow for visualization of differences in patterns between groups, and trends in dietary patterns over time. However, as the shape of the patterns depends on the order of the components presented on the plot, it is important for the order of components to be standardized across analyses using this tool.