A Validation Study of the Automated Self-Administered 24-Hour Dietary Recall for Children, 2014 Version, at School Lunch.

BACKGROUND
Obtaining valid and reliable estimates of usual dietary intake at a reasonable cost is a challenge in school-based nutrition research. The Automated Self-Administered 24-Hour Dietary Recall for Children, 2014 version (ASA24 Kids-2014), a self-administered, computerized 24-hour dietary recall, offers improved feasibility over traditional interviewer-administered 24-hour recalls.

OBJECTIVE
This mixed-methods study examined ASA24 Kids-2014's validity for measuring dietary intake from National School Lunch Program lunches.

PARTICIPANTS/SETTING
After 24% attrition, 96 middle-school students from three urban schools in eastern Pennsylvania participated in the study. A subsample of 27 participants completed qualitative interviews. Data were collected in the spring of 2014.

MAIN OUTCOME MEASURES
Self-reported ASA24 Kids-2014 data were compared to direct observations of school lunch, which served as the criterion measure. Dependent variables included eight meal components selected from the National School Lunch Program guidelines (fruit, vegetables, grains, protein-rich foods, dairy, oils, solid fats, and added sugars). A supplemental interview collected qualitative data regarding students' perceptions of content and substantive validity.

STATISTICAL ANALYSES
The Wilcoxon signed rank test and Spearman's ρ examined criterion-related validity; qualitative content analysis examined content and substantive validity.

RESULTS
Participants inaccurately recalled food items eaten at lunch, as 58% of foods were reported in error. However, among foods recalled correctly, no statistically significant differences emerged for estimates of portions consumed for six meal components (fruit, vegetables, grains, protein-rich foods, oils, and added sugars). In addition, statistically significant positive correlations emerged between ASA24 Kids-2014 and direct observation for all estimates. Qualitative data identified students' interest and motivation, comprehension, memory, and English-language fluency as relevant sources of error.

CONCLUSIONS
Middle school students have difficulty recalling food items eaten at school lunch; however, they are somewhat successful at estimating intake of accurately recalled foods using ASA24 Kids-2014. Like many self-administered computerized recalls, it remains limited by substantial error. Findings have implications for the development of ASA24 Kids-2014 among diverse youth in urban school settings.