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

Abstract

Given the lack in a valid biomarker to assess dietary calcium intake (dCa_i), reproducible estimation of usual dCa_i 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 dCa_i 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 dCa_i were compared by Wilcoxon signed rank test. Reproducibility and relative validity of the FFQ were assessed by Spearman correlation (r_s) 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) dCa_i 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 (r_s=0.72, P<.001) and moderate agreement (κw=0.55). The FFQ and 24HRs were moderately correlated (r_s=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 dCa_i 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 dCa_i <600 and ≥1000 mg/d in postmenopausal women and to rank dCa_i in epidemiological studies.

Keywords: Dietary calcium intake; Food frequency questionnaire; Middle-aged women; Nutrition assessment; Relative validity.