Objectives : To compare self-reported dietary intakes using food frequency questionnaires (FFQs) adjusted for 7-d food checklists against recovery biomarkers (doubly labeled water (DLW) for energy; 24-h urine for protein, sodium, potassium).
Methods : Over 12-months, participants aged 50-74 years, from the Interactive Diet and Activity Tracking (IDATA) Study, were asked to complete 2 FFQs, 2 7-d, 32-item food checklists, 2 24-h urine collections, and 1 DLW administration. Participants who completed at least 1 FFQ, 1 food checklist with ≥5 days, and had DLW and urine results were included (N=742). FFQ energy, protein, sodium, and potassium intakes and densities were calculated both directly (unadjusted) and adjusted by the ratio of reported intakes, calculated as the average percentage of days with reported intake from the 7-d food checklist divided by the summed FFQ frequency for each food group. The discrepancy between FFQ values and biomarkers were determined by comparing mean unadjusted and checklist-adjusted values to biomarkers. Attenuation factors and validity coefficients were determined for all unadjusted and adjusted FFQ values.
Results : Overall, adjusting FFQ values by estimates from the 7-d food checklist resulted in higher values for protein, sodium, potassium, and energy. For both men and women, FFQ energy, protein, sodium and potassium values were lower than biomarker estimates, but all differences were reduced when adjusted for checklist values. The greatest reduction in differences between self-reported values (unadjusted vs. adjusted) and biomarkers were seen for sodium and energy intake in females (-36% to -21% and -29% to -14%, respectively), and total energy intake in males (-29% vs. -14%). , The checklist-adjusted potassium values were higher than the biomarker values, but for men, they were closer to the biomarker. Finally, the adjustment process improved the level of attenuation and validity coefficients for each nutrient and energy.
Conclusions : Though past studies have shown 24-h recalls and food records have higher correlations with recovery biomarkers than FFQs, our results suggest that, simple adjustment of FFQs using food checklists reduces reporting error and may improve the power to detect relationships between dietary exposures from FFQs and health outcomes.
Funding Sources : N/A