Validation of an App-based Food Frequency Questionnaire: Comparison With 24-hour Diet Recalls Among Women in two Rural Areas of Kenya

A new mobile application with a new semi-quantitative Food Frequency Questionnaire (FFQ) was developed as a quick dietary assessment tool. This study aimed to test the validation of food and nutrient intake measurements from an app-based FFQ, which included 49 dishes and food items, using non-consecutive two-day 24-hour diet recalls as the standard. A total of 179 women aged 19–70 years (mean (SD) = 33.7 (9.9)) from two rural areas of Kenya, Vihiga and Kitui counties, were included. Dietary assessment by the two dietary survey methods was conducted from November to December 2021. The intake of macro- and micronutrients, and food groups between the two methods were compared by Spearman's rank correlation coefficients, cross-classification agreement, and Bland-Altman plots for assessing the validation of the app-based FFQ. The median correlation coefficients between energy-adjusted intakes estimated by the app-based FFQ and the 24-h diet recall were 0.19 (0.10 (protein)–0.26 (fat)) for macronutrients, 0.24 (0.11 (sodium)–0.46 (vitamin B12)) for micronutrients, and 0.26 (0.07 (seasonings)–0.45 (pulses)) for 17-food groups. The median percentage of classification into the same quartiles between the two methods was 31% for macronutrient, 32% for micronutrient and 39% for food group. While 9% for macronutrient, 8% for micronutrient and 10% for the food group were misclassified into the opposite categories. Bland-Altman Plots showed acceptable agreement for macronutrients intake between the two methods. The app-based FFQ was shown to be able to rank or categorize individuals according to their intake of most of the nutrients and food groups we examined. This dietary assessment tool can be used for quantitative dietary data collection. Further research is needed to improve the validation of the app-based FFQ, as it has been shown that intakes of staple foods and sugar are underestimated and others are overestimated, and that estimating intakes of some specific nutrients, such as protein, is challenging. Key words: Diet assessment, food frequency questionnaire, 24-hour diet recall, validation, Kenya