Empirically Derived Eating Patterns Using Factor or Cluster Analysis: A Review

Abstract

This paper reviews studies performed to date that have employed cluster or factor analysis to empirically derive eating patterns. Since 1980, at least 93 studies were published that used cluster or factor analysis to define dietary exposures, of which 65 were used to test hypotheses or examine associations between patterns and disease outcomes or biomarkers. Studies were conducted in diverse populations across many countries and continents and suggest that patterns are associated with many different biomarkers and disease outcomes, whether measured by cluster or factor analysis. Despite clear differences in approaches and interpretations, there is some evidence that underlying eating patterns are revealed by either method. Although the research considered herein has created a meaningful body of literature, refining both the factor and cluster analysis methods will help to further establish eating patterns as a sound dietary assessment method.

Dietary patterns and significance of nutrition for women with low-risk pregnancy

“…The energy-rich pattern reported in the study by Bouwland-Both et al [19] was characterized by the presence of food such as bread, breakfast cereals, margarine, nuts, sweets, and non-alcoholic drinks, a little different composition from that of the pattern of the present study although having the same denomination. This can be justified by the different methodologies used in each one of those studies due to the subjectivity involved in some steps of the factor analysis to reduce the factors, including the criteria used to determine the number of factors to extract and the labeling of the dietary patterns, which is also chosen by the researchers [10,16,27].…”

“…The assessment of eating habits using patterns emerged as an alternative method in nutritional epidemiology and is justified by the fact that people do not eat certain foods or nutrients in isolation but rather in combination during meals [10,11]. The analysis of patterns of food intake during pregnancy has been widely used, as shown in the studies carried out by Brantsaeter et al [1], Uusitalo et al [2], Gaillard et al [3], among others.…”

“…Dietary patterns are commonly named by the researchers [10,16]; however, we searched the literature for denominations of eating patterns with composition similar to that of the patterns in this study. The healthy pattern was found with the same name in other studies involving pregnant women with similar food group distribution [2,18].…”

“…The dietary patterns were identified using factor analysis and Principal Component Analysis (PCA) following these steps: use of Kaiser-MeyerOlkin (KMO) test to measure suitability of data for factor analysis with a cut-off value ≥0.5 and the Bartlett's Sphericity Test (BST), with p≤0.05; extraction of factors by defining the number of factors to retain using the eigenvalues and the Cattell's scree test (scree plot), in which the points at which the curve changes drastically (and the "scree on the hill slope" begins) indicate the maximum number of factors to retain and the subjective interpretation of the dietary patterns identified [10]; rotation of the factors retained using varimax, an orthogonal rotation method; each major component was then interpreted based on the foods with factor loadings (correlation coefficients between dietary variables and factors) ≥0.3 or ≤-0.3; and lastly, each major component was named according to the food present in each factor [10,16]. It is worth mentioning that negative charges indicate an inverse correlation between the food item and the dietary pattern and positive charges indicate a direct correlation between them [17].…”

Diversity of eating patterns and obesity in older adults—A new challenge

“…In terms of epidemiology, the use of empirical analyses of derivative standards may reflect the true eating behaviors of a population and allow for improved planning and effective proposals to promote healthy eating habits [10]. Therefore, from the hypothesis that the eating habits of older adults are diversifying and exerting a strong influence on the increased prevalence of obesity, this study aimed at identifying eating patterns of the older…”