A preliminary assessment of the use of household budget survey data for the prediction of individual food consumption.

OBJECTIVE To compare individualised household budget survey (HBS) data with food consumption values derived from individual nutrition surveys (INSs). SETTING Four European countries: Belgium, 1980-85 and 1987-88; Greece, 1993-94; Norway, 1992-94; and the United Kingdom, 1985-88. DESIGN Household budget survey data were individualised with non-parametric models. Individual nutrition survey data were converted into 'HBS-like' estimates, with the application of recipe-based calculations and yield factors for weight changes during food preparation. To correct for over- and underreporting or recording in both surveys, quantities (in g day(-1)) of 14 principal food groups were expressed as fractions of total food quantity (in g day(-1)). For each food group, INS and HBS-derived mean values were calculated for 24 research units, jointly defined by country (four countries involved), gender (male, female) and age (younger, middle-aged and older). Pearson correlation coefficients were calculated and correlation diagrams were drawn. CONCLUSION The results of this preliminary analysis show that there is value in the nutritional information derived from HBSs. However, additional and more sophisticated work is required, in order to derive reliable point and interval estimates of individual food consumption based on HBS data.