Cluster analysis of individuals with similar trends of fat intake during childhood and adolescence: a new approach to analyzing dietary data

Abstract 

To determine the age at which dietary habits emerge, it is necessary to document the changes in dietary intake, beginning early in life and continuing through late adolescence, of an individual. We examined fat intake as a percentage of energy intake in 228 participants of the Dortmund Nutritional and Anthropometric Longitudinally Designed Study with at least ten 3-day dietary records collected yearly between 1985 and 2002 between the ages of 2 and 18 years. After defining a distance matrix with 3 criteria to measure similarity between individual fat intake trends during childhood and adolescence (=fat intake pattern), cluster analysis was used to identify groups with similar individual fat intake pattern. Cluster analysis resulted in 4 clusters consisting of 35 to 81 subjects with differences in (1) mean fat intake, (2) standard deviation of intraindividual fat intake, and (3) fat intake trends over time. No differences of sex, age, or study participation characteristics were found between clusters. Our analysis shows that cluster analysis is an appropriate tool to identify different fat intake patterns during childhood and adolescence, starting early in life.

Keywords: Fat intake, Children, Adolescents, Distance matrix, Cluster analysis