Associations between self-reported vegetable and fruit intake assessed with a new web-based 24-h dietary recall and serum carotenoids in free-living adults: a relative validation study

The aim of the present study was to assess the relative validity of a new web-based 24-h dietary recall (R24W) in terms of vegetable and fruit (VF) intake assessment using serum carotenoid concentrations as reference biomarkers. A total of seventy-four women and seventy-three men (mean age 47·5 (sd 13·3) years; mean BMI 25·5 (sd 4·4) kg/m(2)) completed the R24W four times to assess their VF intake. Serum carotenoids were obtained from 12-h fasted blood samples and measured by HPLC. Raw and de-attenuated partial Spearman's correlations were performed to determine how usual vegetable and/or fruit intake was associated with serum carotenoids. Relevant confounders were selected using a stepwise regression analysis. Finally, cross-classification was used to determine agreement between intake of VF and serum carotenoids. Intake of total dietary carotenoids was significantly associated (r 0·40; P < 0·01) with total serum carotenoids (without lycopene). Total VF intake was also associated with total serum carotenoid concentrations without lycopene (r 0·44; P < 0·01). HDL-cholesterol, waist circumference and age were identified as confounders in the association between total VF intake and total serum carotenoids (without lycopene). De-attenuated partial correlation adjusted for these confounders increased the associations between dietary carotenoids and total serum carotenoids without lycopene (r 0·49; P < 0·01) and between total VF intake and total serum carotenoids without lycopene (r 0·48; P < 0·01). Almost 80 % of respondents were classified in the same or the adjacent quartile for total VF intake and total serum carotenoids without lycopene, while less than 6 % were classified in the opposite quartile. Overall, these observations support the appropriateness of the R24W to assess the dietary intake of VF.