Do Global Dietary Nutrient Datasets Associate with Human Biomarker Assessments? A Regression Analysis

Background: To assess the state and trends of global nutrition, our best tools are nationally representative assessments using human biomarker assays, yet these are expensive and logistically challenging. We instead often rely on more easily produced global nutrient datasets-measures of nutrients provided by the diet-as a proxy for nutritional intake and deficiency, due to their greater geographic and temporal coverage. However, the accuracy of global nutrient datasets is questionable. Objective: We aimed to test whether estimates of inadequate dietary intake derived from existing global nutrient datasets reliably associate with biophysical deficiency. Design: We performed linear regressions of estimates of inadequate dietary nutrient intake derived from three global nutrient datasets-Global Dietary Database, Global Nutrient Database, and Global Expanded Nutrient Supply (GENuS) model-against the existing suite of nationally representative biomarker survey data for three key nutrients of global concern in two vulnerable demographic groups: zinc, folate, and vitamin A in females of childbearing age; and zinc and vitamin A in children younger than 5 y. Results: We found significant associations (P < 0.1) for only 3 of 22 regressions between global nutrition datasets and biophysical deficiency: zinc for females of childbearing age from GENuS and Global Dietary Database, and zinc for children under 5 y from GENuS. Folate and vitamin A show no reliable relationship between nutrient datasets and independent biomarker surveys. Applying the successful models for zinc to the accompanying full datasets yield estimates of global zinc deficiency of 31%-37% for these demographic groups. Conclusions: We found that few estimates of nutritional inadequacy from global dietary datasets are associated with more direct measures of biophysical deficiency from biomarker studies. Researchers and policymakers must be cautious when applying global nutrient datasets to questions of global health and use them for limited applications.