S. Voss et al., OBESITY AS A MAJOR DETERMINANT OF UNDERREPORTING IN A SELF-ADMINISTERED FOOD FREQUENCY QUESTIONNAIRE - RESULTS FROM THE EPIC-POTSDAM STUDY, Zeitschrift fur Ernahrungswissenschaft, 36(3), 1997, pp. 229-236
The phenomenon of underreporting of dietary intake has been observed p
reviously in many epidemiologic studies. In this study it was investig
ated whether dependencies exist between energy intake obtained by a se
miquantitative, self-administered food frequency questionnaire and lif
estyle or anthropometric factors, particularly obesity. The study popu
lation consisted of 2 531 subjects, men aged 40 to 64 years and women
aged 35 to 64 years from the general population of Potsdam and the sur
rounding areas. First, subjects were allocated into quintiles of the r
atio 'reported energy intake (EI)' to 'calculated basal metabolic rate
(BMR)' as a measure of age and weight adjusted energy intake. No appa
rent dependencies between socio-economic variables and the ratio EI/BM
R were observed. Among anthropometric variables, BMI and related measu
res of obesity were inversely related to the ratio EI/BMR in men and w
omen. While dietary intake was directly related to the ratio EI/BMR in
absolute quantities, energy adjusted intake of fat, protein, carbohyd
rate, and alcohol was found to be independent of this ratio. Energy ad
justed food group consumption was also found to be independent of the
ratio EI/BMR, showing only slightly increasing trends across quintiles
of EI/BMR for cereals and fats, and a slightly decreasing trend for s
weet foods in women. When subjects were classified into three categori
es of BMI, reported energy intake decreased across categories. Estimat
ed energy expenditure based on BMR was increasing with BMI categories.
A close direct relationship was observed between BMI categories and t
he difference between reported energy intake and estimated energy expe
nditure. It is concluded that obesity is a major determinant of underr
eporting. Energy adjusted dietary variables were found to be largely i
ndependent of such methodological influences.