We evaluated the performance of two reduced dietary questionnaires (RDQ). They were derived from a more complex questionnaire used to investigate the relation between diet and gastric cancer in a case-control study recently concluded in Italy. The full dietary questionnaire (FDQ) used in that study included 181 questions on the frequency/portion size of consumption of 146 different foods and drinks. Two different RDQs were defined, including subgroups of the original food items able to reproduce at least 90% (RDQ90) or 75% (RDQ75), respectively, of the total variability captured by the FDQ on the nutrient intake of controls. Each RDQ provided estimates of nutrients highly correlated with the estimates produced by the FDQ. For different quintiles of nutrient intake, we compared the estimated relative risk obtained by means of the RDQs and the FDQ. For various nutrients, the differences in relative risks were relatively minor, below 20%, but were larger for protein and beta-carotene. These analyses indicate that a RDQ is able to quantify adequately various aspects of the Italian diet for the purpose of epidemiologic inference.
(C) Lippincott-Raven Publishers.