Comparison of nutrient intake determined by four dietary intake instruments

We compared nutrient intake estimates among four commonly used food intake instruments: 24-hour recall, a 2-day diet diary, a modified diet history, and a quantitative food-frequency questionnaire. The four instruments were administered to each of fifty people selected by systematic sampling from Washington County, Maryland. Using an analysis of variance for a complete block design, we analyzed data by mean differences across all instruments, and by simple correlations and Kappa statistics between instrument pairs. The food-frequency questionnaire gave the highest mean estimates, and the modified diet history estimate means fell between the food-frequency questionnaire and the other two instruments. However, when adjusted for calories (nutrients/1000 kcal), the food-frequency questionnaire estimates for nutrients were no longer consistently higher than those of the other instruments. Total fat, oleic acid, cholesterol, protein, sodium, iron, thiamin, and niacin were most consistently estimated by two or more instruments as measured by correlation coefficients. Crude fiber, calcium, and vitamins A and C gave the least consistent estimates among all four instruments. The correlation coefficients and Kappa scores were modest for all instrument pair comparisons, but the majority were statistically significant at p ≤ 0.005. All data showed greater consistency between instruments that measured either long-term intake (modified diet history and food-frequency questionnaire) or short-term intake (diet diary and twenty-four hour recall). In addition, demographic variables (age, sex, and marital status), as well as differences in caloric intake, affected instrument estimates of some nutrients.