Anthropometric changes and dietary patterns: a comparison between cohort studies in Brazil and the Netherlands

Abstract

To control worldwide obesity prevalence and to promote healthy eating habits are major challenges in public health. Even with economic and cultural differences, low-middle income and high-income countries face the same challenge of reducing the prevalence of overweight and obesity. Since foods are not consumed in isolation, studies in the field of nutritional epidemiology are pointing to the necessity of analyzing diet holistically. In this context, the dietary pattern (DP) analysis emerged as a complementary approach to the traditional methods of diet analysis. Several approaches to DP analysis exist, with a main distinction in whether they are predefined or data-driven. A downside of data-driven patterns is that they are usually difficult to compare between populations. Reduced rank regression (RRR) analysis is a hybrid method, which aims to more directly relate the data-driven steps of DP identification to an outcome of interest. This characteristic of RRR analysis gives the possibility to derive DP from different populations using the same intermediate variables and a comparison could be made considering that RRR identify outcome-related DP. This study aims to compare DP derived by RRR in Brazilian and Dutch cohorts and to estimate how these DP are associated with anthropometric changes. Data will be used from the Longitudinal Study of Adult Health and the Rotterdam Study. Dietary data were obtained using food frequency questionnaires. Anthropometric data will be analyzed throughout the follow-up period. To verify the association between DP and anthropometric changes linear mixed-effects models will be applied, adjusted for possible confounding factors previously described in the literature. (AU)