Relative validity of a food group–based dietary screener for cardiometabolic risk-related nutrients in Korean adults
Background Dietary guidelines express chronic disease risk as numerical intake thresholds, yet existing tools are poorly suited to checking individual adherence. Comprehensive food frequency questionnaires are burdensome, 24-h dietary recalls (24HRs) require repeated administration, and single-nutrient screeners miss multi-nutrient imbalance. Objective To evaluate, against 24HRs, the relative validity of NutriView (NV), a 103-item food group–based screener that simultaneously assesses seven cardiometabolic risk-related nutrients, with agreement in classifying intake against guideline thresholds as the primary outcome. Methods In this cross-sectional study, 185 Korean adults aged 19–75 years (77 controls, 108 with self-reported cardiometabolic disease) completed NV once and two nonconsecutive 24HRs on a weekday and a weekend day, in randomized order. Habitual intake was estimated as a 5:2 weekday-to-weekend weighted average; analyses covered the seven target nutrients, energy, and three macronutrients. Intakes were energy-adjusted (residual method). Agreement was assessed by Spearman correlation, Bland–Altman analysis, and quartile cross-classification. Classification against the 2025 Dietary Reference Intakes for Koreans was assessed by positive and negative percent agreement (PPA, NPA), Cohen’s κ, and prevalence-adjusted bias-adjusted κ (PABAK). Results NV items covered 74.1% of distinct reported foods and captured 77.5–96.4% of reported nutrient intake, largely through group-level matching; for sodium, only 20.2% of intake matched an individual item, yet coverage reached 83.2%. Energy-adjusted correlations were 0.171–0.422, and 73.5% of participants fell in the same or an adjacent quartile (chance expectation 62.5%). PPA was highest for calcium (94.2%) and dietary fiber (87.6%), though with correspondingly low NPA (28.6 and 40.0%), indicating overclassification. Cohen’s κ was low throughout (0.044–0.231), largely reflecting skewed risk prevalence; PABAK was higher for calcium (0.784) and dietary fiber (0.676). Agreement for sodium did not exceed chance (κ 0.044; PABAK 0.038), with underdetection predominating. Detection could not be evaluated for trans fat or added sugar, each with fewer than 15 screen-positive participants. Conclusion In this study, NV captured most reported intake and classified intakes of calcium, dietary fiber, saturated fat, and cholesterol consistently with its screening purpose. Performance varied by nutrient, indicating that NV is suited to initial screening — identifying individuals needing fuller dietary assessment — rather than to quantitative estimation of intake.
Authors
- Haeun Park
- Woo Jeong Kim (ORCID: https://orcid.org/0000-0002-9678-7973)
- Donghoon Lee
- Jaewan Park
- Kyeongjin Seo
- Kyuhoon Seong
- Seoung-Uk Ha
- Jung-Min Kim
Institutions
- Seoul National University (KR)
- Severance Hospital (KR)
- Seoul National University Bundang Hospital (KR)
- Yonsei University Health System (KR)
Publication Details
- Journal
- Frontiers in Nutrition
- Published
- 2026-11-19
- DOI
- https://doi.org/10.3389/fnut.2026.1984011
- Primary Topic
- Nutritional Studies and Diet
- Type
- article
- Field-Weighted Citation Impact
- 0.00