Prompting methods affect the accuracy of children's school lunch recalls

Objectives To evaluate the feasibility of implementing 3 specific prompting methods among students in the first and fourth grades (mean age=7.2 and 10.1 years, respectively), to validate recall accuracy of first- and fourth-grade students against observation, to develop a single measure of inaccuracy that cumulated errors in reporting food items and amounts without allowing underreporting and overreporting to cancel each other, and to establish information regarding the measure of inaccuracy for use in future studies. Design Children were interviewed the morning after they were observed eating lunch provided by the school. Interviews included free recall, nonsuggestive prompted recall, and specific prompted recall (either preference, food category, or visual). Subjects/setting Ninety-six children (48 students per grade) were studied-32 per the specific prompting method-stratified by grade, sex, and ethnicity (African-American or white). Specific prompting method was randomly assigned to children within strata. Statistical analysis Food items were categorized and weighted as: combination entree=2, condiments= 1/3, and other=1, To calculate inaccuracy of recall, absolute differences between amounts reported and observed eaten were calculated for each item and then multiplied by each item's weight; these values were then summed across all items for each child. Inaccuracy of recall was determined before specific prompting and after specific prompting and for the difference (inaccuracy after specific prompting minus inaccuracy before specific prompting). Results Before specific prompting, median inaccuracy was 2.7 servings for the 48 first-grade students and 1.7 servings for the 48 fourth-grade students. The median difference in inaccuracy from before to after specific prompting was 0 for both grades. Specific prompting increased recall accuracy for 9 first-graders and 12 fourth-graders, but decreased recall accuracy for 21 first-graders and 7 fourth-graders, and did not change recall accuracy for 18 first-graders and 29 fourth-graders. Among children whose recall accuracy increased after specific prompting, 5 first-graders and 8 fourth-graders received prompting for food category, Applications/conclusions An-tong first-grade students, specific prompting (either preference, food category, or visual) may hurt more than help recall accuracy. Among fourth-grade students, prompting for food category yields small gains in recall accuracy with minimal losses. Validation studies are needed with larger sample sizes to determine prompting methods that produce more accurate dietary recalls from children of various socioeconomic status groups, other racial/ethnic groups, and at other meals.