The HELENA online food frequency questionnaire: reproducibility and comparison with four 24-h recalls in Belgian–Flemish adolescents

Abstract

To investigate the reproducibility of the HELENA Food Frequency Questionnaire (FFQ) developed as a basis for an online tailored intervention and to compare the FFQ with four computerized 24-h recalls. A convenience sample of 48 Belgian-Flemish adolescents completed four computerized 24-h recalls, the FFQ and a retest 1-2 weeks later. Correlations, the Wilcoxon's signed rank test and Bland and Altman's plots were used to compare the intakes of 22 food groups, energy (kcal), fibre (g, g/1000 kcal), vitamin C (mg, mg/1000 kcal), calcium (mg, mg/1000 kcal), iron (mg, mg/1000 kcal) and fat (g, % from total energy). Reproducibility correlations were good (0.46-0.90). De-attenuated correlations between both methods were high (> or =0.51) for all nutrients and nutrient densities, except for fibre (g), vitamin C (mg) and percent energy from fat (respectively 0.23, 0.40 and 0.30). On a food group level correlations were significant (r(unadjusted)>0.3) for most (17/22) food groups. No systematic differences were found between test and retest; comparison with the recalls resulted in a significantly higher intake of bread, breakfast cereals, other snacks, vegetables, potatoes, coffee and tea, alcoholic beverages, energy, and most nutrients. No significant difference was found for calcium, whereas an underestimation was found for nutrient density of calcium and fat. The overestimation in seven food groups, resulting in an overestimation of energy, fibre, iron and vitamin C, and underestimation of percentage energy from fat, needs to be considered when the estimated intakes are used for tailored feedback.