Simulated validation of a brief food frequency questionnaire

Abstract

Nutrient assessment by the brief 45-item food frequency questionnaire used in the New York State Cohort was simulated by reducing the food list and frequency-of-use categories of the extensive 131-food item semiquantitative questionnaire used in the Health Professionals Follow-up Study. As a validation of the brief questionnaire, daily nutrient intakes were calculated following simulation procedures for 127 men who completed the extensive questionnaire, and the values were compared with those from two 1-week diet records in the same population. Deattenuated correlation coefficients for 25 energy-adjusted nutrient values ranged from 0.12 for iron to 0.76 for vitamin C (mean, 0.26). The brief questionnaire appears reasonably effective in assessing relative dietary intakes of total fat, saturated fat, monounsaturated fat, carbohydrate, dietary fiber, carotene, vitamin C, calcium, potassium, and magnesium. Modest improvements in the correlations for most nutrients were seen when the number of frequency-of-use categories was increased. Correlations decreased when nutrient intakes were calculated using regression coefficients instead of a conventional nutrient database.