AUTHOR=Mitry Patricia , Wawro Nina , Six-Merker Julia , Zoller Dorothee , Jourdan Carolin , Meisinger Christa , Thierry Sigrid , Nöthlings Ute , Knüppel Sven , Boeing Heiner , Linseisen Jakob 

TITLE=Usual Dietary Intake Estimation Based on a Combination of Repeated 24-H Food Lists and a Food Frequency Questionnaire in the KORA FF4 Cross-Sectional Study

JOURNAL=Frontiers in Nutrition

VOLUME=6

YEAR=2019

URL=https://www.frontiersin.org/journals/nutrition/articles/10.3389/fnut.2019.00145

DOI=10.3389/fnut.2019.00145

ISSN=2296-861X

ABSTRACT=<p><bold>Background:</bold> Estimation of usual dietary intake poses a challenge in epidemiological studies. We applied a blended approach that combines the strengths provided by repeated 24-h food lists (24HFLs) and a food frequency questionnaire (FFQ).</p><p><bold>Methods:</bold> At least two web-based 24HFLs and one FFQ were completed by 821 participants in the KORA FF4 study. Consumption probabilities were estimated using logistic mixed models, adjusting for covariates and the FFQ data on consumption frequency. Intake amount of a consumed food item was predicted for each participant based on the results of the second Bavarian Food Consumption Survey (BVS II). By combining consumption probability and estimated consumption amount, the usual food intake for each participant was estimated. These results were compared to results obtained without considering FFQ information for consumption probability estimation, as well as to conventional FFQ data.</p><p><bold>Results:</bold> The results of the blended approach for food group intake were often higher than the FFQ-based results. Intraclass correlation coefficients between both methods ranged between 0.21 and 0.86. Comparison of both methods resulted in weighted kappa values based on quintiles ranging from fair (0.34) to excellent agreement (0.84). Omission of FFQ information in the consumption probability models distinctly affected the results at the group level, though individual intake data were slightly affected, for the most part.</p><p><bold>Conclusions:</bold> Usual dietary intake data based on the blended approach differs from the FFQ-based results both in absolute terms and in classification according to quintiles. The application of the blended approach has been demonstrated as a possible tool in nutritional epidemiology, as a comparison with published studies showed that the blended approach yields reasonable estimates. The inclusion of the FFQ information is valuable especially with regard to irregularly consumed foods. A validation study including biomarkers of dietary intake is warranted.</p>