Can a food frequency questionnaire be used to capture dietary intake data in a 4 week clinical intervention trial?

Collecting dietary data in the clinical research setting is labour intensive and can be burdensome for study participants. The aim of this study was to assess the agreement between data obtained from 2 different dietary assessment methods, a 74-item semi-quantitative food frequency questionnaire (FFQ) and 3-day weighed food records (WFR) used to estimate dietary intake over the preceding month. One hundred and fifty nine subjects, aged between 31 and 74 years (53 males, 65 females), enrolled in a clinical trial at the Commonwealth Scientific and Industrial Research Organisation, Division of Health Sciences and Nutrition, (CSIRO HSN) Adelaide, Australia. Group mean intakes and individual mean intakes estimated by the two measures were compared. One hundred and eighteen (91%) three-day WFR and their corresponding FFQ were analysed. Pearson correlation coefficients ranged from 0.22 for cholesterol to 0.78 for alcohol (median 0.41). Mean energy and nutrient intakes were within ± 20% difference. The FFQ gave lower carbohydrate intake estimates, percentage energy from carbohydrate (P <0.001) and dietary fibre (P <0.05) and gave higher percentage energy from saturated fat estimates, poly-unsaturated fatty acids (P <0.001) and mono-unsaturated fatty acids (P <0.05). Subjects were also ranked into quintiles and the quintiles cross-tabulated. The FFQ classified more than two thirds of the subjects within ±1 quintile difference for all nutrients. We conclude that this FFQ can capture similar information as WFR and may be used for estimation of dietary intakes over a relatively short time in clinical intervention trials.