[Reduction modality of a dietary questionnaire]

Abstract

The data relative to the average daily intake of 7 nutrients and total energy obtained through the answers to a dietary questionnaire, consisting of 152 items and administered to 200 patients (admitted for acute pathologies) has been analysed. The aim has been to elaborate a questionnaire consisting of a more limited number of items but maintaining a great part of the informative contents from the original questionnaire. For this aim, a multiple regression model with a selection procedure for independent variables, of the stepwise type, has been used, where the dependent variable is the daily average intake of the nutrient and the independent variables are the consumption of each food included in the complete questionnaire (QC). By combining the results of the model with the 7 nutrients and total energy, 3 reduced questionnaires have been produced (QR): the first two are based on the sole criteria of the inter-individual variance of the intake of the nutrient, explained by the consumption of the foods included in the QR, the third by adding the foods that offer an adequate contribution in order to explain the supply of the nutrient under examination. The application potentials of the 3 QRs have been analysed both by evaluating the inter-class correlation coefficient, deduced by an concordance/discordance matrix relative to the quintilis values of the nutrient deduced by the QC and by each QR, and by simulating their use in a case-control study; for this aim, the attenuation of the odds ratio trend and the increase of the sample dimension has been calculated following the use of a QR instead of a complete one.(ABSTRACT TRUNCATED AT 250 WORDS)