Cross-context equivalence and agreement of healthy diet metrics for national and global monitoring: a multicountry analysis of cross-sectional quantitative 24-hour dietary intake studies

Background: Despite national and global commitments to improve nutrition, there are no universally accepted indicators for at-scale monitoring of diets. Several metrics have been proposed and used, but they vary in their comprehensiveness and validity in capturing the properties of healthy diets and, potentially, their interpretability across contexts.

Objective: The objective of this study was to assess the cross-context equivalence and agreement of healthy diet metrics.

Design: Quantitative 24-h dietary intake data from 57,456 nonpregnant females (15-49 y) in 21 countries were used to construct the food group diversity score (FGDS) and Minimum Dietary Diversity for Women indicator; the Global Diet Quality Score (GDQS) and its submetrics: GDQS+ and GDQS-; the Global Dietary Recommendations (GDR) score and its submetrics: noncommunicable disease (NCD)-Protect and NCD-Risk scores; and the All-5 indicator. Associations between (standardized) measures and indicators were quantified by fitting linear and logistic mixed-effect models across World Bank country income classifications, respectively. The levels of agreement between measures were assessed using Bland-Altman plots of z-scores, whereas rank correlations were assessed using Spearman's ρ.

Results: The consumption of healthy food groups was positively associated with concurrent intakes of unhealthy food groups, in particular in high-income countries. Hence, metrics constructed using both healthy and unhealthy food groups (i.e., GDR score and GDQS) showed relatively less discriminatory capacity across country income classifications than their respective submetrics or FGDS. Standardized metrics of healthy (e.g., FGDS and GDQS+) and unhealthy food group consumption (i.e., NCD-Risk and GDQS-) exhibited strong agreement and correlations.

Conclusion: Composite metrics weighting both healthy and unhealthy food groups have limited cross-context equivalence, because a wide range of diets can theoretically return similar scores. Healthy food group (sub)metrics performed comparably, likely indicating strong underlying construct validity (i.e., dietary diversity and nutrient adequacy). For national and global monitoring, refinement and validation of unhealthy food group metrics (i.e., moderation) is recommended to complement healthy food group metrics.