Validation of a food frequency questionnaire for measuring dietary intake of isoflavones and lignans using the method of triads.

Background: Emerging epidemiologic evidence suggested a protective association between dietary phytoestrogens and multiple chronic diseases, although validation studies on the 2 phytoestrogen subclasses, isoflavones and lignans, estimated from food frequency questionnaires (FFQs) are lacking.

Objectives: We evaluated the performance of the FFQ for measuring isoflavone and lignan intake among 892 free-living adults in the Anhui Lifestyle Validation Study (ALVS).

Methods: From July 2021 to July 2022, participants from the ALVS completed a 141-item FFQ and donated fasting morning urine specimens twice, 1-year apart, and completed twelve 24-h dietary recalls (24HDRs) that consisted of 4 quarterly 3 consecutive 24-h dietary recalls (i.e. 2 weekdays and 1 weekend day in each season) between the 2 FFQ assessments. The reproducibility coefficients were assessed by the intraclass correlation coefficient between estimates from the 2 FFQs. The validity coefficients (VCs) were evaluated by the method of triads to calculate correlations between estimates from dietary assessment tools and true dietary intake using biomarkers (the mean of 2 measurements of urinary concentrations of 4 isoflavone metabolites and 4 lignan metabolites), 24HDRs, and FFQs.

Results: The FFQ-derived isoflavones (median: 23.6 mg/d; IQR: 12.1-41.5 mg/d) and lignans (median: 19.7 mg/d; IQR: 11.9-30.7 mg/d) showed good reproducibility with the same intraclass correlation coefficient of 0.55 (95% confidence interval [CI]: 0.46, 0.62). Estimated VCs between the FFQs and theoretical true intake were 0.65 (95% CI: 0.31, 1.00) for isoflavones and 0.48 (95% CI: 0.24, 0.86) for lignans. Compared with FFQ-based assessments, urinary biomarkers showed the same validity in assessing dietary lignans (VC: 0.48; 95% CI: 0.21, 0.88), but dietary isoflavones showed a lower VC of 0.22 (95% CI: 0.03, 0.47).

Conclusions: FFQ shows reasonable performance in measuring habitual intake of isoflavones and lignans, although urinary isoflavones are unlikely to perform as well as FFQ-based assessment of dietary isoflavones.