Discovery and characterisation of dietary patterns in two Nordic countries: Using non-supervised and supervised multivariate statistical techniques to analyse dietary survey data

This Nordic study encompasses multivariate data analysis (MDA) of preschool Danish as well as pre- and elementary school Swedish consumers. Contrary to other counterparts the study incorporates two separate MDA varieties - Pattern discovery (PD) and predictive modelling (PM). PD, i.e. hierarchical cluster analysis (HCA) and factor analysis (using PCA), helped identifying distinct consumer aggregations and relationships across food groups, respectively, whereas PM enabled the disclosure of deeply entrenched associations. 17 clusters - here defined as dietary prototypes - were identified by means of HCA in the entire bi-national data set.

These prototypes underwent further processing, which disclosed several intriguing consumption data relationships: Striking disparity between consumption patterns of Danish and Swedish preschool children was unveiled and further dissected by PM. Two prudent and mutually similar dietary prototypes appeared among each of two Swedish elementary school children data subsets. Dietary prototypes rich in sweetened soft beverages appeared among Danish and Swedish children alike. The results suggest prototype-specific risk assessment and study design.