Data-Driven Dietary Patterns and Diet Quality Scores: Reproducibility and Consistency in Sex and Age Subgroups of Poles Aged 15–65 Years

• Main Authors: Joanna Kowalkowska, Lidia Wadolowska, Jolanta Czarnocinska, Grzegorz Galinski, Anna Dlugosz, Dorota Loboda, Magdalena Czlapka-Matyasik
• Format: Article
• Language: English
• Published: MDPI AG 2020-11-01
• Series: Nutrients
• Subjects: FFQ; dietary assessment; diet quality; dietary pattern; PCA; reproducibility
• Online Access: https://www.mdpi.com/2072-6643/12/12/3598
• ISSN: 2072-6643

This study aimed to assess: (i) the test–retest reproducibility of identification of data-driven dietary patterns (DPs) derived using a Principal Component Analysis (PCA) and hypothesis-driven DPs (diet quality scores); (ii) the consistency of data-driven DPs with diet quality scores in sex and age subgroups of Poles aged 15–65 years. The study involved 504 subjects (55.6% of females). Data on food consumption frequency (33 food items) were collected twice with a two-week interval using the Dietary Habits and Nutrition Beliefs Questionnaire (KomPAN^®) in a self-administered version (test and retest). Two major data-driven DPs (‘Prudent’ and ‘Western’) were identified in the total sample, sex groups and four age groups separately from test and retest data. Two diet quality scores were analysed: Pro-Healthy-Diet-Index-10 (pHDI-10) and Non-Healthy-Diet-Index-14 (nHDI-14). Tucker’s congruence coefficient indicated fair-to-good similarity of data-driven DPs between test and retest for all study subgroups, except for males. Across study subgroups, the intraclass correlation coefficient (ICC) between the test and retest ranged from 0.56 to 0.86 for ‘Prudent’ DP and 0.57 to 0.82 for ‘Western’ DP, with the lowest values in males. The ICC (test vs. retest) ranged from 0.84 to 0.88 for pHDI-10 and 0.75 to 0.88 for nHDI-14. Comparing the data-driven DPs and diet quality scores, the Spearman’s correlations ranged from 0.63 to 0.93 between ‘Prudent’ DP and pHDI-10, and from 0.60 to 0.81 between ‘Western’ DP and nHDI-14. The test–retest reproducibility of data-driven DPs and diet quality scores and their consistency were acceptable in most of the study subgroups, with a tendency to be higher for pro-health than unhealthy DPs. Data-driven DPs were more reproducible in females than males. The reproducibility of diet quality scores tended to be better in males than females and was the highest in 25–44-year-olds. The KomPAN^® questionnaire can be recommended to use data-driven DPs and diet quality scores to describe the habitual diet in people aged 15–65 years.