Comparison of Children's 24-Hour Recalls and Food Records Using Two Methods of Analysis
AuthorNielsen, Katherine M.
StatisticsView Usage Statistics
The purpose of the study was to determine which dietary assessment method, the 24-hour recall or the food record, would yield data that equates the greatest to the Spears Point-of-Sale-Dietary Assessment-Tool (Spears POS-DAT). The second purpose of the study was to determine which analysis method, the conventional approach or the reporting-error-sensitive approach, accurately reflected a child's reported dietary intake determined by 24-hour recall or food records in comparison to the Spears POS-DAT.A descriptive, quantitative study was conducted. Subjects were fourth and fifth grade children enrolled in the 2006-2007 school year at a local charter elementary school (n=40). For two weeks Spears POS-DAT values were obtained at lunch to determine the actual amount of foods eaten. During this time, students kept a 3-day food record and were interviewed to obtain a single day 24-hour recall. Data was analyzed for 24-hour recalls and food records using two different analysis methods. The conventional method compared mean reported intake to mean reference intake (from the Spears POS-DAT) by converting food into energy and macronutrients. The reporting-error-sensitive method classifies items as matches, omissions, or intrusions and amounts as corresponding, over or unreported. Paired t-tests, correlation coefficients, report rates (reported amounts/reference amounts x 100) and correspondence rates (corresponding amounts from matches/reference amounts x 100) were calculated.Conventional analysis for 24-hour recalls found significant differences for mean protein (p=0.04) and carbohydrates (p=0.05), but not for energy (p=0.09) or fat (p=0.33). Report rates ranged from124%-131%, indicating over-reporting. Food records had no significant differences between mean reported and mean reference amounts for energy and macronutrients, with food record report rates ranging from 95% to 112%.Reporting-error-sensitive analysis found significant differences between energy and macronutrients for 24-hour recalls and food records when comparing mean reference amounts to mean corresponding amounts from matches (all p-values < 0.001). Correspondence rates for 24-hour recalls were 61% to 67% and 60% to 67% for food records. Based on conventional analysis, food records were more accurate than 24-hour recalls; this was not substantiated by the reporting-error-sensitive analysis. Conventional methods overestimated reporting accuracy. The reporting-error-sensitive analysis method provided additional information regarding reporting accuracy. Further studiesare needed to confirm these findings.