Variability in nutrient and food intakes among older middle-aged men. Implications for design of epidemiologic and validation studies using food recording.

The authors conducted a dietary methodology study in 1984 in Finnish men aged 55-69 years in order to validate two dietary assessment instruments being used in the US-Finland Alpha-Tocopherol, Beta-Carotene Lung Cancer Prevention Trial. Twelve 2-day food records collected from 162 men over a 6-month period, including every day of the week, served as the reference measure. This report focuses on three important questions for investigating diet and disease relations: 1) How many days are necessary to classify "usual" intake? 2) Is there loss as a result of using consecutive days? 3) Which days are necessary for assessment and classification of "usual" diet? A repeated-measures regression model was used to estimate the variance components and the effects of consecutive days, weekday (weekday vs. weekend), and season. Correlations between the averages of different numbers of days of food records and "true" usual intake were examined along with the resulting attenuations in relative risk. Results suggest that 7-14 days are required to adequately classify most individuals into categories of intake for most nutrients and some foods. There appears to be some loss of information from using consecutive days rather than days further apart. Weekday/weekend differences in mean intakes are slight, and the rank ordering of individuals appears to be preserved. A moderate seasonal effect is shown for classification of fruits, but only a slight one is seen for micronutrients and berries. Implications for the design of epidemiologic and validation studies are discussed.