Objective: Compare the validity and reliability of a one-week fruit, juice and vegetable (FJV) food frequency questionnaire (FFQ) that does not require averaging in response categories and one-week food recognition form (FRF) against three to four 24-hour dietary recalls (24hdr) among mostly African-American boys and 17 to 20 year-old adults for possible use in subsequent intervention studies.
Design: In Study One, the FFQ was administered to 40 boys in four Boy Scout (BS) troops on two occasions separated by two weeks. The FRF was administered to 36 boys in four other BS troops after school on six school days, covering seven consecutive days. All these boys completed four 24hdr, including one Sunday. In Study Two, 56 17 to 20 year-old adults completed the FFQ during telephone interviews on two occasions two weeks apart. In between, they completed three 24hdr by telephone interview on one weekend and two weekdays.
Subjects/setting: Participants in Study One were members of eight urban BS troops, and in Study Two were high school (HS) seniors and recent HS graduates. BS data were collected at times scheduled for troop meetings or immediately after school. The young adult group completed telephone interviews in the evenings and on weekends.
Main measures: This research compared a FJV FFQ and a FJV FRF (incorporating the same 24 foods) against three to four 24hdr conducted using NDS software from the University of Minnesota.
Statistical analyses performed: Descriptive statistics, intraclass correlations, Spearman correlations, corrections for attenuation.
Results: The first FFQ provided substantially higher mean consumption estimates for both groups. The FFQ had moderate reliability across assessments and moderate validity coefficients for fruit and total FJV combined for the BS sample. The FRF had higher validity coefficients for juice. The FFQ had moderate reliability and poor validity among the young adult group. Correction for measurement error in the 24hdr substantially increased validity coefficients to high levels for BS fruit and total FJV and to moderate levels for the HS groups, with some exceptions.
Applications/conclusions: FFQ more validly assessed fruit and combined FJV consumption for the nine to 14 year-old BS but overestimated mean values. FRF means were closer to those of the 24hdr and were more valid for assessing juice consumption. The FFQ had moderate validity among the young adult group but overestimated consumption. Further research is needed on simpler methods of reporting diet in these age groups.