Validity of self-reported dietary intake at school meals by American Indian children: the Pathways Study

To examine the validity of a modified diet record-assisted 24-hour recall in third-grade (8 to 10 years old) American Indian children.

The children were trained to record their food intake using diet records, and then they recalled their 24-hour food intakes, using the diet records as memory prompts, during interviews by trained staff using the Minnesota Nutrition Data System (NDS; version 2.6, 1993, Food database version 8A, Nutrient database version 23; Nutrition Coordinating Center, University of Minnesota, Minneapolis). The modified method added training in portion size estimation. Direct observation of the children's intakes during school meals was used to validate the accuracy of their self-reported recalls.

Eighty third-grade children recruited from schools from four of the American Indian Nations participating in the Pathways Study. Statistical analyses performed Pearson correlations were used for nutrient level data. A mixed regression model (PROC MIXED), with no other fixed effects and site as a random effect, was used to test the null hypothesis that the difference between recalled and observed intakes was zero (H(o): beta(o)=0). Food intake data were obtained from the Nutrition Data System Record Reports.

There were no significant differences between recalled and observed energy intakes for the school meals combined or for either meal individually. Percentages of energy intake from fat, protein, and carbohydrate from recalls were not significantly different from those observed for the combined school meals. Pearson correlations for energy and energy-providing nutrients ranged from 0.52 to 0.86 for both meals, from 0.55 to 0.86 for school lunch, and from 0.61 to 0.86 for school breakfast. Agreement between recalled and observed food items was 75%. Children recalled 57% of food quantities within +/-10% of observed quantities.

At the group level, American Indian children were able to accurately report the macronutrient proportions of their total energy intake, and their reporting of total energy intake (+13% of criterion) compares favorably with that of other ethnic groups of children of similar age. They were able to accurately recall the majority of foods that they were independently observed consuming during school meals.