Validation of a Photo Atlas for Food Portion Estimation for a National 24-hr Recall Dietary Survey in Kenya

Abstract

Objectives: Reliance on respondents’ individual recall ability for accurate dietary data collection is always a challenge. We developed a photo atlas to aid respondents in recall of foods eaten during the interactive 24 hr recall multiple pass approach for data collection during the Kenya National Micronutrient Survey 2011. Photographs of foods are preferred for estimating portion sizes because the amount eaten is easier for the respondent to visualize. However, it was necessary to validate the photo atlas before using it. Methods: Seven common Kenyan food items were chosen for validation of the atlas: cassava root, tomatoes, eggplant, raw chicken wings and legs, watermelon and fried dough. Each color photo displayed three differently sizes of each food item to capture “large,” “medium,” and “small”. Actual food portions were used for comparison with the photos. Volunteers were asked to rate the level of similarity between the actual-life food item and the item in the photograph on a scale of 1 to 5. Five indicated “Perfect Match” and one indicated “Not a Match.” 84 volunteers from KEMRI and Kibera were conveniently selected for the validation exercise. Results: Exact matches were highly observed in all sizes of whole foods (onions (52.4; 61.9%; 59%), tomatoes (59.5%; 58.3%; 63.1%) and eggplants 35.7%; 27.4% and 63.1%) compared to sliced food items. No significant differences observed between female and male matches of food items. Conclusions: Identification of difficult to recall foods can be presented as food albums and used as dietary data collection tools.