Objective: To describe methods and dietary habits of a large population cohort.
Design: Prospective assessment of diet using diet diaries and food-frequency questionnaires, and biomarkers of diet in 24-h urine collections and blood samples.
Setting: Free living individuals aged 45 to 75 years living in Norfolk, UK.
Subjects: Food and nutrient intake from a food-frequency questionnaire on 23 003 men and women, and from a 7-day diet diary from 2117 men and women. Nitrogen, sodium and potassium excretion was obtained from single 24-h urine samples from 300 individuals in the EPIC cohort. Plasma vitamin C was measured for 20 846 men and women.
Results: The food-frequency questionnaire (FFQ) and the food diary were able to determine differences in foods and nutrients between the sexes and were reliable as judged by repeated administrations of each method. Plasma vitamin C was significantly higher in women than men. There were significant differences in mean intake of all nutrients measured by the two different methods in women but less so in men. The questionnaire overestimated dairy products and vegetables in both men and women when compared with intakes derived from the diary, but underestimated cereal and meat intake in men. There were some consistent trends with age in food and nutrient intakes assessed by both methods, particularly in men. Correlation coefficients between dietary intake assessed from the diary and excretion of nitrogen and potassium in a single 24-h urine sample ranged from 0.36 to 0.47. Those comparing urine excretion and intake assessed from the FFQ were 0.09 to 0.26. The correlations between plasma vitamin C and dietary intake from the first FFQ, 24-h recall or diary were 0.28, 0.35 and 0.40.
Conclusions: EPIC Norfolk is one of the largest epidemiological studies of nutrition in the UK and the largest on which plasma vitamin C has been obtained. Methods for obtaining food and nutrient intake are described in detail. The results shown here for food and nutrient intakes can be compared with results from other population studies utilising different methods of assessing dietary intake. The utility of different methods used in different settings within the main EPIC cohort is described. The FFQ is to be used particularly in pooled analyses of risk from diet in relation to cancer incidence within the larger European EPIC study, where measurement error is more likely to be overcome by large dietary heterogeneity on an international basis. Findings in the UK, where dietary variation between individuals is smaller and hence the need to use a more accurate individual method greater, will be derived from the 7-day diary information on a nested case-control basis. 24-h recalls can be used in the event that diary information should not be forthcoming from some eventual cases. Combinations of results utilising all dietary methods and biomarkers may also be possible.