Narrative Review of the Validity and Cross-Context Equivalence of Healthy Diet Metrics and Their Data Collection...

Valid, sensitive healthy diet metrics that are comparable across contexts are needed for global monitoring. The healthy diets monitoring initiative identified 4 field metrics as potentially fit for purpose: global diet quality score (GDQS), global dietary recommendations score, minimum dietary diversity for women (MDD-W), and Nova ultra-processed food score.

To review whether these 4 healthy diet metrics 1) accurately predict food and nutrient intakes; 2) accurately differentiate the average or prevalence of food and nutrient intakes; 3) respond to changes over time; 4) are comparable across contexts; and 5) can be collected using their proposed brief assessment methods while preserving predictive accuracy.

Peer-reviewed literature was searched and extracted from PubMed, Web of Science, and Google Scholar, including preprints and grey literature from the latter. Evidence on the accuracy of the field metrics and methods was qualitatively assessed against the aforementioned objectives, considering the underlying theory of change and study design, as well as the direction and magnitudes of the observed associations or effects.

Increments in GDQS+ and MDD-W predicted higher composite metrics of nutrient adequacy. MDD-W was sensitive to changes in nutrient intakes over time. MDD-W cutoffs showed limited variability across contexts and population groups. Higher GDQS and global dietary recommendation scores and lower Nova ultra-processed food scores were associated with lower intakes of food and nutrients to moderate. The predictive accuracy of field methods for nutrient adequacy was maintained for GDQS and MDD-W. No study explicitly investigated how field metrics differentiate averages or prevalence of reference metrics across countries.

MDD-W demonstrated comparatively stronger predictive accuracy for nutrient adequacy, with a lower burden method, than GDQS+. Further research is required to determine the predictive accuracy of field-friendly metrics measuring moderation across contexts and time. Complementary metrics that can be collected simultaneously on a large scale are needed for global monitoring.