Development and Evaluation of a Novel Diet Quality Screener for Global Use

Abstract

Objectives: Appropriate tools for measuring and tracking diet quality on a global scale are lacking. Such instruments need to be easy-to-use across various settings, and based on a metric associated with non-communicable diseases (NCDs) and key nutrient intakes. We developed and evaluated a diet quality screener among women in a high-income country setting. Methods: The Prime Diet Quality Score (PDQS), a new food-based global diet quality index, shown to predict multiple NCDs among large U.S. cohorts, and associated with nutrient intake adequacy in low and middle-income countries (LMICs), was used to develop a short-form diet screener. Participants were recruited via Amazon MTurk to complete two 24-hour diet recalls (ASA24) and two screeners (PDQS-24HR and PDQS-30D) 7-30 days apart. Spearman rank correlation and linear regression were used to evaluate correlations and associations between usual nutrient intakes assessed by ASA24 and the PDQS values. We also compared the associations of the two score values with nutrient intakes by including them simultaneously in a regression model, and using a Wald test. Results: The final sample included 290 U.S. women of reproductive age (41 ± 11). The median time required for completing the PDQS-24HR was 4.9 minutes, compared to 21 minutes for the ASA24. The rank correlations of the PDQS-24HR with energy-adjusted intakes of some key nutrients were 0.53 (fiber), 0.50 (β-carotene), 0.37 (folate), 0.31 (vitamin B6), 0.36 (vitamin C), 0.40 (vitamin E), 0.48 (potassium), 0.51 (magnesium), 0.25 (zinc), 0.21 (iron), 0.17 (PUFAs), −0.34 (added sugar), and −0.19 (SFA) (all P < 0.05). Correlations of the PDQS for vitamin B1 and calcium were null. Associations of nutrient intakes with the PDQS-30D were not significantly different from those with the PDQS-24HR. Conclusions: The results of an initial evaluation of a stand-alone, PDQS-based diet quality screener among U.S. women are promising. Associations between the PDQS and nutrient intakes were of variable strength but in the expected directions. The lack of correlation for calcium intake warrants further examination: it may be partially attributed to the neutral scoring of key sources of this nutrient in U.S. diet. Future work should include piloting the screener across LMICs, and among children and adolescents.