The use of cluster analysis to derive dietary patterns: methodological considerations, reproducibility, validity and the effect of energy mis-reporting

Over the last three decades, dietary pattern analysis has come to the forefront of nutritional epidemiology, where the combined effects of total diet on health can be examined. Two analytical approaches are commonly used: a priori and a posteriori. Cluster analysis is a commonly used a posteriori approach, where dietary patterns are derived based on differences in mean dietary intake separating individuals into mutually exclusive, non-overlapping groups. This review examines the literature on dietary patterns derived by cluster analysis in adult population groups, focusing, in particular, on methodological considerations, reproducibility, validity and the effect of energy mis-reporting. There is a wealth of research suggesting that the human diet can be described in terms of a limited number of eating patterns in healthy population groups using cluster analysis, where studies have accounted for differences in sex, age, socio-economic status, geographical area and weight status. Furthermore, patterns have been used to explore relationships with health and chronic diseases and more recently with nutritional biomarkers, suggesting that these patterns are biologically meaningful. Overall, it is apparent that consistent trends emerge when using cluster analysis to derive dietary patterns; however, future studies should focus on the inconsistencies in methodology and the effect of energy mis-reporting.