In nutritional epidemiology, the traditional approach consists of estimating the association between single food/food group or nutrient and the risk of a health event, accounting for confounding factors. New methods have been developed to look at food as a whole through holistic approaches. Initially, two types of methods have been described: a priori methods based on dietary scores and a posteriori methods based on statistical dimension reduction techniques based on the structural properties of consumption data. Various scores have been described in the literature reflecting typical diets, adequacy to national and global dietary recommendations or prevention of pathologies. These scores allow the characterisation of individual diets in a descriptive or etiological approach and comparisons between studies. The a posteriori methods are sample-dependent and are therefore less straightforward to compare between studies, although certain patterns emerge in a coherent manner in several studies. More recently, hybrid methods have emerged that combine exploratory methods with existing scientific knowledge. Each of these methods exhibits advantages and limitations, including more or less subjective methodological decisions that need to be validated.
展开▼