Evaluation of a Web‐Based Dietary Questionnaire Used to Assess Soy Isoflavones Intake in Young People

BACKGROUND Soy foods are good sources of high‐quality protein, essential fatty acids, and isoflavones and, thus, can potentially improve the overall diet quality. However, increased soy foods consumption have raised some concerns because of the known estrogenic properties of isoflavones. Isoflavones intake data in adolescent populations is scant. Existing dietary assessment questionnaires focused on soy isoflavones are for adult populations and none of these had been validated in adolescents. OBJECTIVE To assess the isoflavones intake of an adolescent population highly exposed to soy using innovative dietary assessment methods that tap the technological skills of adolescents. METHODS We developed a 151‐item web‐based semi‐quantitative food frequency questionnaire (FFQ) which included 37 soy foods. Multiple 1‐day food records with digital photographs was used as the criterion method. A total of 55 12–18 year‐old adolescents who met the inclusion criteria and passed the training requirements for the study completed the web‐based FFQ. Over a 2–3 month period, they provided 6 non‐consecutive 1‐day food records which covered 5 weekdays and 1 weekend day. They took pictures of foods to be eaten at each eating occasion and simultaneously sent them with their food record for one day via text messaging. Follow‐up calls were made the following day to verify the accuracy of the food records which were checked against the photographs. Total soy isoflavones, daizein, genistein, and glycitein intakes on both the food records and web‐based FFQ were calculated using the Nutrition Data System for Research nutrient database and existing meat analogs database in our institution. Wilcoxon Signed Ranks tests, correlations with energy adjustment and correction for multiple measurements, cross‐classification analyses, and Bland‐Altman plots using SAS version 9.4 were conducted to evaluate the performance of the web‐based FFQ against the food records with digital photographs. RESULTS The web‐based FFQ significantly overestimates the intake of all the isoflavones by about 200–300%. Deattenuated correlation values after energy‐adjustment are 0.67 for total isoflavones, 0.63 for daidzein, 0.64 for genistein, and 0.56 for glycitein. Degree of agreement within one quartile are 82%, 84%, 82% and 80% for total isoflavones, daidzein, genistein, and glycitein, respectively; gross mismatching is between 0% (total isoflavones) and 6% (glycitein). Bland‐Altman plots consistently show overestimation of the web‐based FFQ but a relatively even distribution of points inside the limits of agreement. CONCLUSION The web‐based FFQ is relatively accurate in ranking adolescent intake of soy isoflavones. However, precaution must be taken when using absolute intake as the exposure variable. Further improvement of this web‐based FFQ would be needed to reduce the overestimation bias. Support or Funding Information Soy Nutrition Institute