Estimating food intakes in Australia: validation of the Commonwealth Scientific and Industrial Research Organisation (CSIRO) food frequency questionnaire against weighed dietary intakes

Background:  There is a dearth of knowledge about the foods that Australian adults eat and a need for a flexible, easy-to-use tool that can estimate usual dietary intakes. The present study was to validate a commonly used Australian Commonwealth Scientific and Industrial Research Organisation (CSIRO) food-frequency questionnaire (C-FFQ) against two 4-day weighed food records (WFR), as the reference method.
Methods:  The C-FFQ, as the test item, was administrated before the WFR. Two 4-day WFR were administrated 4 weeks apart. Under-reporting was established using specific cut-off limits and estimated basal metabolic rate. Seventy-four women, aged 31–60 years, were enrolled from a free-living community setting.
Results:  After exclusion for under-reporting, the final sample comprised 62 individuals. Correlations between protein intake from the WFR and urinary urea were significant. Overall agreement between FFQ and WFR was shown by 'levels of agreement' (LOA) and least products regressions. There was presence of fixed and proportional bias for almost half the nutrients, including energy, protein, fat and carbohydrates. For most of the nutrients that did not present bias, the LOA were 50–200%. Agreement was demonstrated for percentage dietary energy protein and fat; carbohydrate; and absolute amounts of thiamine, riboflavin, magnesium and iron. However, relative intake agreement was fair to moderate, with approximately 70% of (selected) nutrients exact or within ±1 quintile difference.
Conclusion:  The C-FFQ is reasonable at measuring percentage energy from macronutrients and some micronutrients, and comprises a valuable tool for ranking intakes by quintiles; however, it is poor at measuring many absolute nutrient intakes relative to WFR.