[Development and validation of a semi-quantitative food questionnaire based on a population survey].

An auto-administered, semi-quantitative food frequency questionnaire was developed to assess evolution of dietary habits in the Geneva population. For this purpose 626 adults, aged 35 years or more and residents of Canton Geneva, were interviewed using a 24 hour recall technique. The questionnaire comprises the 80 food items or homogeneous food groups contributing to more than 90% of the intake in calories, proteins, lipids, carbohydrates, alcohol, cholesterol, calcium, vitamin D and retinol, and to more than 85% of dietary fibers, carotene and iron. To complete the food frequency questionnaire, participants indicate for each of the 80 food items or groups their frequency of consumption and whether their usual portion is equivalent to a reference portion listed in the questionnaire (median portion in the sample), smaller (first quartile of the portion distribution in the sample) or larger (third quartile of the portion distribution in the sample). The questionnaire can be filled in about 20 minutes. To test the new questionnaire we submitted it to a sample of 27 men and 29 women randomly selected from the 626 subjects who participated to the original survey. Mean intakes obtained using the food frequency questionnaire agree well with those obtained using the 24 hour recall for all subgroups defined by age, sex or nationality. The food frequency questionnaire tends to slightly underestimate the absolute intakes of alcohol, fibers and calcium since it is based on a subsample of all food items consumed according to the 24 hour recall interviews. These results suggest that the semi-quantitative food frequency questionnaire is a good alternative to 24 hour recalls to assess mean nutrient intakes in the adult population of Canton Geneva.