Dietary patterns in children: Associations with ethnicity, socioeconomic position, BMI and body composition

In the Netherlands and many other high-income countries, the prevalence of childhood overweight and obesity is unequally distributed among ethnicity and socioeconomic position (SEP). This thesis focuses on the question which dietary patterns can be derived in 5-year old children and how these dietary patterns are associated with ethnicity, socio-economic determinants, longitudinal BMI and measures of body composition. In Chapter 2 we describe the four dietary patterns that were derived using the data-driven Principal Component Analyses (PCA) method: a snacking (high intakes of savoury and sweet snacks, fruit drinks, refined breakfast products and low intakes of whole-grain breakfast products), full-fat (high intakes of full-fat spreads, full-fat cheese, pasta dishes and low intakes of low-fat spreads and low-fat cheese), meat (high intakes of low- and high-fat meat, sauces, potatoes and refined grain products for warm meals) and healthy (high intakes on the food groups water and tea, vegetables, fish, fruits and whole grain product for warm meals) dietary pattern. Children of non-Dutch origin scored high on the snacking and healthy pattern, children of Turkish origin scored high on the full-fat pattern and children of African Surinamese origin high on the meat pattern. Children of lower educated mothers scored above all high on the snacking pattern. In Chapter 3, we observed that also other indicators of SEP, that is a lower level of paternal education, lower level of household finance and lower neighbourhood socioeconomic status (SES) were associated with higher snacking pattern scores. Yet, a lower maternal education level conferred the highest risk. A lower level of household finance was an additional risk factor for higher snacking pattern scores within the group of middle-high educated mothers. We observed that Children of Dutch origin and higher educated mothers were more often normal weight at age 5, and they developed more often normal weight at age 10 if they were underweight or overweight at age 5, compared to children of non-Dutch origin and lower and middle high educated mothers (Chapter 4). Overall, higher scores on the PCA-derived healthy dietary pattern were unexpectedly positively associated with weight development in most groups, regardless of ethnic origin and maternal education level, whereas higher scores on the PCA-derived full-fat pattern were negatively associated with weight development. In Chapter 5, we applied an a priori diet quality score, which reflects the extent to which the intake within 10 food groups complies with the age-specific Dutch food-based dietary guidelines for children aged 4 to 8 years. We observed an average score of 4.8 (± 1.1) out of a possible 10 points, indicating that in our cohort, about half of the age-specific dietary guidelines were met. Only the intake of whole-grains met the recommendation. We observed that children with higher diet quality scores had more often a higher educated mother, a lower weight, lower BMI and a lower Fat Mass Index (FMI) at age 12 years. In Chapter 6 we choose to further study which dietary patterns were relevant for BMI and measures of body composition at age 5 stratified by maternal educational level. We applied the hybrid Reduced Rank Regression (RRR) method and derived dietary patterns that explained the maximum of variation in BMI, FMI and body fat-free mass (FFMI) at age 5. Pattern 1 was characterised by high intakes of low-fat and healthy food items in all maternal education level groups, and explained the maximum of variation in BMI and FMI, in our cohort at age 5. Higher scores on pattern 1, were associated with higher BMI and a higher odds of being overweight/obese at age 10, and higher FMI at age 12 in all maternal education level groups.