(P12-07-26) A Flexible Cloud-Based System for Food-Frequency Questionnaire Design, Administration, Nutritional and Statistical Analysis, and Reporting Results in Population Surveys

Objectives: To assemble and test a package of Google Suite components for automating and enhancing various aspects of food-frequency questionnaire (FFQ) design and use in diverse populations.

Methods: To develop the system, we adapted a pen-and-paper semiquantitative FFQ validated for use in the Thailand National Health Examination Survey. The system consists of (1) a Google Sheet for each study participant that contains tabs for (a) the FFQ in different languages (depending on user preference), (b) linked food composition data, (c) harmonized nutrient reference values, (d) interim calculations, and (e) a personalized report that incorporates lifestage, bodyweight, and iron/zinc bioavailability data and outputs tables and figures summarizing nutrient intake, adequacy probability, food sources of nutrients, diet quality scores, and equivalent statistics for the rest of the population; (2) a Drive folder for storing all Sheets; (3) a Colab notebook for statistical analyses across Sheets; and (4) an administrator Sheet and embedded Apps Scripts for organizing metadata and results from all participants and tabulating statistical results.

Results: We tested the system among 150 university students in Bangkok (21.6+3.5 years, 76% female) who received brief (5 mins.) instructions before self-administering the FFQ. Energy intake fell within 800–4000 and 500–3500 kcal/day for 96% of men and women, respectively. Median time to completion for the 162-item FFQ was 40 mins. User experience was improved by real-time validation of responses and generation of results, and the facility for participants to contribute interactively to population parameters and statistics (e.g., factor loadings for dietary patterns derived from the population feed back into each participant’s personalized report to compute pattern adherence and present distributions for different subgroups of the population for comparison).

Conclusions: A system of integrated Google Suite components facilitates efficient collection and analysis of FFQ data from populations and flexibly accommodates diverse instrument designs and reference data. Ongoing research involves characterizing relative validity of different modes of administration.

Funding Sources: Mahidol University Strategic Research Fund, The Rockefeller Foundation, National Institute of Diabetes and Digestive and Kidney Diseases