Validity evidence for two brief fruit and vegetable assessment instruments among 3rd grade students

Abstract

Brief dietary instruments with validity evidence are used to assess fruit and vegetable intake in adults because they are inexpensive and easy to administer compared to 24-hour dietary recalls or assessment with biomarkers. However, use of these tools with young children without prior examination of evidence for validity may compromise data accuracy. The purpose of this study was to evaluate the validity evidence for two brief methods/instruments to estimate fruit and vegetable intake among 3rd-grade children. One instrument used a single retrospective question each about usual fruit and usual vegetable intake illustrated with measuring cup pictures (FVQ) to estimate amounts. The other instrument was a food record based on A Day in the Life questionnaire (DILQ) that measures intake frequency. Third grade students from one school and one community center (n = 107) in the metropolitan Minneapolis/St. Paul area participated in the study in 2012 and 2013. Three 24-hour dietary recalls were conducted using Nutrition Data System for Research (NDSR) to validate the FVQ and one recall to validate the DILQ. The first recall was conducted on the same day the two questionnaires were completed and the other two recalls were conducted within a 2-3 week period of time. Recalls were collected for 2 weekdays and 1 weekend day. An algorithm derived from National Health and Nutrition Examination Survey (NHANES) data was used to convert frequencies measured by the DILQ into amounts. Bland-Altman and deattenuated Pearson correlation coefficients were used to compare agreement between intakes assessed with the 24-hour dietary recalls and the FVQ and DILQ. Deattenuated Pearson correlations were moderate between 24-hour dietary recall and DILQ results (0.38 for fruit and 0.38 for vegetables). Deattenuated Pearson correlations between FVQ and 24-hour dietary recalls results were strong (0.54 for fruit and 0.58 for vegetables) based on the low reliability of multiple 24-hour dietary recall data. Therefore, the strength of these correlations is misleading. Participants overestimated fruit intake using both tools and vegetable intake using the FVQ. Participants underestimated vegetable intake by 0.37 cup using the DILQ. The limit of agreement for all comparisons was high. Therefore, evidence of validity for the FVQ and DILQ was lacking for evaluation of fruit and vegetable consumption in the current sample of 3rd grade students. Other assessment methods/instruments should be considered for use with younger children.