Development and reproducibility of a brief food frequency questionnaire for assessing the fat, fiber, and fruit and vegetable intakes of rural adolescents

Objective: To describe the systematic development and reproducibility of a food frequency questionnaire (FFQ) designed to meet the specific research requirements of the Goals for Health cancer prevention intervention program for rural middle school children. Design: A 4-step process was used to develop a brief FFQ for scoring intakes of total fat, fiber, and fruits and vegetables. The resulting questionnaire consisted of 25 food frequency items and 10 supplemental questions. Reproducibility of the questionnaire was determined by comparing responses at the beginning and end of a 4-month interval. Subjects: Study subjects were sixth- and seventh-grade students attending middle schools in rural areas of Virginia and upstate New York. Seventh-grade students participated in the pilot study, and sixth-grade students participated in the reproducibility study. The final version of the FFQ was completed twice by 539 sixth graders. After exclusions for missing and unreliable data, the usable sample size was 415. Boys were somewhat more likely than girls to be excluded for missing data. African-American students comprised 32% of the population. Statistical analyses performed: Each food frequency item was associated with 3 scores--a fat score, a fiber score, and a combined score for the number of servings of fruits and vegetables. Means and standard deviations were determined for nutrient variables, differences between repeat administrations were tested for significance by paired t test, and Pearson correlation coefficients were calculated for nutrients and for individual food items. Results: Correlation coefficients for nutrient scores were 0.58 for fat, 0.49 for fiber, and 0.51 for fruits and vegetables. For individual food items, correlations ranged from 0.24 to 0.59 (mean=0.41). Applications/conclusions: Using a systematic approach to developing a study-specific FFQ for rural adolescents is feasible. Further, the reproducibility of the Goals for Health questionnaire was demonstrated for the 3 nutrient scores it was designed to measure. This developmental approach may be readily adapted to other populations, study designs, and nutrients of interest. The validity of the questionnaire remains to be tested.