Associations Among Diet Quality-Related Biomarkers, Block Food Frequency Questionnaire, and 24-hour Dietary Recall

Abstract

Abstract Objectives Self-reported dietary data suffer from high measurement error and findings that rely on them are of limited value. We determined the association of diet quality related biomarkers with associated measures from Block Food Frequency Questionnaire (FFQ) and Automated Self-Administered 24-hour dietary Assessment Tool (ASA24); and the association of FFQ and ASA24. Methods This cross-sectional single-visit study included 127 subjects (18–60 years, BMI ≥18.5 kg/m2). Diet quality related biomarkers were collected from pre-planned subsample (N = 33) to measure - plasma alkylresorcinol for whole grain, omega-3 index, serum fatty acid composition for dairy, serum carotenoids for fruit and vegetable, and serum selenium for seafood intake. Self-reported dietary intake data for whole grain, Omega-3 index, fatty acid, carotenoids, and selenium were calculated per instrument guidelines using FFQ and ASA24. Biomarkers were analyzed by Mass Spectrometry Facility, TTU. Outcomes with missing data were handled via multiple imputation with predictive mean matching. Spearman's correlation coefficient (using R statistical software) were used to assess the association of biomarkers with self-reported measures (N = 33), and the association between FFQ and ASA24 (N = 33, and N = 127). Results Diet quality related biomarkers were not associated with self-reported intake (all P > 0.07), except, omega-3 index was significantly correlated with reported intake in FFQ (P = 0.01). Significant associations were reported for whole grain, omega-3 index, and dairy intake between FFQ and ASA24 (P = 0.02, P = 0.01, P = 0.05 respectively; N = 33). After further analysis with N = 127 between FFQ and ASA24 significant association were reported in whole grain (P = 0.02), Omega-3 index (P = 4.90e-3), dairy (P = 1.79e-8), and seafood intake (P = 6.06e-4), but not carotenoid intake (P = 0.96). Conclusions A significant association between measures via FFQ and ASA24 suggests consistency in self-reporting and also the likelihood these measures do not capture the time-frames they purport to – but rather a self-belief/representation of habitual intake patterns. Interestingly, the association of one of the biomarkers with self-reported intake suggests the necessity of including larger sample to better determine validity of self-reported data. Funding Sources Texas Tech University.