A Systematic Review of Principal Component Analysis-Derived Dietary Patterns in Japanese Adults: Are Major Dietary Patterns Reproducible Within a Country?

Kentaro Murakami, Nana Shinozaki, Aya Fujiwara, Xiaoyi Yuan, Ayumi Hashimoto, Hitomi Fujihashi, Han Chieh Wang, M. Barbara E. Livingstone, Satoshi Sasaki

Research output: Contribution to journalArticle

7 Citations (Scopus)


Principal component analysis (PCA) has been widely used in nutritional epidemiology to derive dietary patterns. However, although PCA-derived dietary patterns are population-dependent, their reproducibility in different populations is largely unexplored. We aimed to investigate whether major dietary patterns are consistently identified among different populations within a country and, if so, how similar these dietary patterns are. We conducted a systematic review of PCA-derived dietary patterns in Japanese adults using PubMed and Web of Science for English articles and Ichushi-Web and CiNii databases for Japanese articles. We assessed the reproducibility of major dietary patterns using congruence coefficients (CCs), with values ≥0.80 considered to represent fair similarity. From 65 articles (80 studies) included in this review, 285 different dietary patterns were identified. Based on the names of these patterns, major dietary patterns were Western (n = 34), Japanese (n = 12), traditional (n = 10), traditional Japanese (n = 9), healthy (n = 18), and prudent (n = 9) patterns. When assessment was limited to high-quality data (i.e., studies based on a sample size ≥200 and use of a validated dietary assessment questionnaire or multiple-day dietary record), the median CC was low for Western (0.44), traditional (0.59), and traditional Japanese (0.31) patterns. Conversely, the median CC was 0.89 for healthy, 0.86 for prudent, and 0.80 for Japanese patterns; and the proportion of pairs with a CC ≥0.80 was 87.3%, 64.3%, and 50.0%, respectively. Characteristics shared among these 3 dietary patterns included higher intakes of mushrooms, seaweeds, vegetables, potatoes, fruits, pulses, and pickles. In conclusion, this systematic review showed that some of the major dietary patterns are relatively reproducible in different populations within a country, whereas others are not. This highlights the importance of careful interpretation of PCA-derived dietary patterns. Our findings in Japan should be confirmed in different countries and globally. This study was registered at https://www.crd.york.ac.uk/prospero/ as CRD42018087669.

Original languageEnglish
Pages (from-to)237-249
Number of pages13
JournalAdvances in nutrition (Bethesda, Md.)
Issue number2
Publication statusPublished - Mar 1 2019



  • congruence coefficient
  • dietary pattern
  • factor analysis
  • food consumption pattern
  • principal component analysis
  • reproducibility
  • similarity

ASJC Scopus subject areas

  • Food Science
  • Medicine (miscellaneous)
  • Nutrition and Dietetics

Cite this