Comparison of a food frequency questionnaire using reported vs standard portion sizes for classifying individuals...

Individual intakes of retinol, carotene, vitamin C, and folacin calculated from a food frequency questionnaire using reported portion size were compared with intakes calculated using standard portion size information. Data from a case-control study to determine the association of nutrient intake and risk of cervical dysplasia were supplemented by standard portion size information from the US Department of Agriculture and reanalyzed. Significant mean differences were found between intake calculated from reported portion size data and that calculated from standard portion size data for all nutrients. Correlation of nutrient intakes obtained by the two methods of data collection ranged from .73 to .92. Calculation of the rho statistic, measuring the consistency of classification of participants into groups of high, medium, and low nutrient intake, led to values ranging from .55 to .71, indicating some misclassification of study participants. To determine the effect misclassification had on the study outcome, odds ratios were calculated using nutrient amounts obtained from both methods of collecting portion size data. Results indicate that replacing reported portion size data with standard portion size data may lead to conflicting outcomes for specific nutrients in research concerning the relationship between diet and disease.