A 51-item calcium-focused food frequency questionnaire is a reliable tool to assess dietary calcium intake in postmenopausal women

Given the lack in a valid biomarker to assess dietary calcium intake (dCai), reproducible estimation of usual dCai is crucial for better understanding of its interaction with health outcomes in specific populations. This study tested the hypothesis that a calcium-focused food frequency questionnaire (FFQ) may be used to estimate dCai of women ≥50 years residing in a multicultural environment (Montreal, Canada). One hundred and eight women (age, 63.1 ± 7.7 years; 98% postmenopausal) completed the FFQ twice and 4 nonconsecutive 24-hour recalls (24HRs) over 1 month. Medians of dCai were compared by Wilcoxon signed rank test. Reproducibility and relative validity of the FFQ were assessed by Spearman correlation (rs) and Cohen's weighted kappa (κw). Agreement was further assessed by cross-classification by quartiles, Bland-Altman plot, and sensitivity and specificity analyses. The median (interquartile range) dCai estimated by the FFQ and 24HRs were 723 (524-1033) mg/d and 854 (666-1068) mg/d, respectively (P < .001). The FFQs had a strong correlation (rs = 0.72, P < .001) and moderate agreement (κw = 0.55). The FFQ and 24HRs were moderately correlated (rs = 0.65, P < .001). Cross-classification showed moderate agreement (κw = 0.42), with 85% of the participants classified into identical or contiguous quartiles and 2.8% into extreme opposite quartiles. According to the Bland-Altman plot, the FFQ underestimated dCai with a bias of 99 mg/d (95% limits of agreement, −677 to +480 mg/d). Sensitivity and specificity of identifying intakes <1000 mg/d were 90% and 57%, respectively. This FFQ is a useful tool to discriminate dCai <600 and ≥1000 mg/d in postmenopausal women and to rank dCai in epidemiological studies.