Am. Fehily et T. Hopkinson, DO WE NEED INFORMATION ABOUT PORTION SIZES TO RANK INDIVIDUALS BY THEIR NUTRIENT INTAKES, Journal of human nutrition and dietetics, 6(2), 1993, pp. 101-111
Nutrient intakes calculated from 7-day weighed intake records for a co
mmunity sample of 99 men and women were compared with those estimated
from the same records, but with mean portion weights substituted for t
he actual weights. For many nutrients, approximately half of the mean
portion records yielded intakes within 10% of the weighed record value
s. Correlation coefficients between the two estimates of intakes were
high (0.64-0.97). Between 58 and 74 subjects were classified in the sa
me third of the nutrient distributions and only between zero and four
classified in opposite thirds. Using estimated portion weights is like
ly to reduce the degree of misclassification only slightly. It is ther
efore concluded that for most nutrients a diary with standard portion
weights may be as likely to detect diet-disease relationships in popul
ation studies as a diary with estimated weights and only slightly less
likely to detect such associations than a weighed intake survey.