Reproducibility and validity of 7-day food records

The reproducibility of the 7-day food record method, as used in The Glostrup Population Studies, has been tested by 40 adult volunteers, mainly hospital staff, recording their dietary intakes in two 7-day periods. To validate the calculated protein intake a 24-h urine specimen was collected from all participants in period II, but for various reasons urine specimens were only obtained from 31 in period I. There were significant differences in the intakes of energy, protein and some vitamins and minerals between the two periods, but the nutrient densities, except for vitamin A, were equal. The mean energy distributions in the diet showed no significant differences between the two periods. The weekly variance ratio (S2within/S2between) was between 0.25 and 0.80 for most nutrients. The only ratios above 1.0 were found for vitamin A, vitamin C, vitamin D, vitamin B12 and chromium, selenium and sodium. The weekly coefficients of variance (CV) were for most nutrients less than 30%. In period II the mean protein intake estimated from the urine samples was 74.7 (+/- 15.9 SD) g/day, whereas the protein intake calculated from the diet records was 70.6 (+/- 18.4) g/day. The average estimated protein intake for the 31 people who collected urine in period I was 72.1 (+/- 18.3) g/day, whereas the calculated intake was 78.0 (+/- 22.7) g/day. In neither period were the mean differences significant.