Reproducibility of an Italian food frequency questionnaire for cancer studies: Results for specific food items

To evaluate the reproducibility of a quantitative food frequency questionnaire (FFQ) used in a case-control study on cancer of the breast, ovary and digestive tract, we compared the result of a 98-item questionnaire administered twice at an interval of 3-10 months (median = 5.4 months) to 452 volunteers (144 males and 308 females, median age = 50 years) from three Italian provinces (Pordenone, Genoa and Forli). Spearman correlation coefficients (r) for intake frequency of 87 dietary items ranged from 0.35 (''chicken or turkey, boiled'') to 0.84 (''wine''). Most coefficients were between 0.60 and 0.80, only two being below 0.40 and five equal or above 0.80 (mean r = 0.59). The concordance of the two measurements tended to be somewhat better for alcoholic and non-alcoholic beverages, bread, cereals and first courses, fruits and summary questions at the end of each section of the questionnaire than for side dishes, sweets and desserts. Also, the reproducibility of 11 subjective questions, such as those concerning the amount of fat in seasoning and the intake of garlic or salt, seemed to be high. Age, sex, educational level of the volunteers and interval between the two FFQ did not have a large or systematic impact on the concordance of the two measurements. In conclusion, the present study has shown a good level of reproducibility of our questionnaire and has provided a few important hints on ways of improving the description of various food items.