Irregular dietary intakes impairs estimations from food records. Biomarkers and method combinations can be used to improve estimates. Our aim was to examine reproducibility from two assessment methods, compare them, and validate intakes against objective biomarkers. We used the Malmö Offspring Study (55% women, 18-71 y) with data from a 4-day food record (4DFR) and a short food frequency questionnaire (SFFQ) to compare (1) repeated intakes (n = 180), (2) intakes from 4DFR and SFFQ (n = 1601), and (3) intakes of fatty fish, fruits and vegetables, and citrus with plasma biomarkers (n = 1433) (3-carboxy-4-methyl-5-propyl-2-furanpropanoic acid [CMPF], β-carotene and proline betaine). We also combined 4DFR and SFFQ estimates using principal component analysis (PCA). Moderate correlations were seen between repeated intakes (4DFR median ρ = 0.41, SFFQ median ρ = 0.59) although lower for specific 4DFR-items, especially fatty/lean fish (ρ ≤ 0.08). Between-method correlations (median ρ = 0.33) were higher for intakes of overall food groups compared to specific foods. PCA scores for citrus (proline betaine ρ = 0.53) and fruits and vegetables (β-carotene: ρ = 0.39) showed the highest biomarker correlations, whereas fatty fish intake from the SFFQ per se showed the highest correlation with CMPF (ρ = 0.46). To conclude, the reproducibility of SFFQ data was superior to 4DFR data regarding irregularly consumed foods. Method combination could slightly improve fruit and vegetable estimates, whereas SFFQ data gave most valid fatty fish intake.
Keywords: biomarker; citrus; dietary assessment methods; fish; food intake; fruits; reproducibility; validation; vegetables.