Abstract
OBJECTIVE: To assess the accuracy of three self-administered food frequency questionnaires (FFQs) to measure dietary calcium intake in healthy adults. DESIGN: Estimates of dietary calcium intake from one previously validated and two recently developed FFQs were compared with those from 7-day food records. SUBJECTS/SETTING: Healthy adults enrolled in an outpatient study of calcium supplementation completed the 36-page Dietary History Questionnaire (DHQ), a 3-page Calcium Questionnaire, and a 1-page Short Calcium Questionnaire. Subjects then completed a 7-day food record. MAIN OUTCOME MEASURES: Differences between calcium intake reported on FFQs and calcium intake from food records were compared. STATISTICAL ANALYSES: Spearman correlations were used to measure associations among variables; Bland-Altman pairwise comparisons were conducted to assess systematic and magnitude biases. RESULTS: We studied 341 subjects, 74.5% female, mean (+/-standard deviation) age of 38+/-11 years and body mass index (calculated as kg/m(2)) of 31.8+/-7.1. Mean (+/-standard deviation) food record calcium intake was 896+/-380 mg/day; data from all three FFQs were positively related to food record calcium intake, but accounted for <40% of the variance in food record dietary calcium intake (DHQ: r(2)=0.21; Calcium Questionnaire: r(2)=0.33; Short Calcium Questionnaire: r(2)=0.37; all P<0.001). The DHQ underestimated daily calcium intake (systematic bias: -94 mg/day, P<0.001; magnitude bias r=-0.40; P<0.001), whereas the Calcium Questionnaire overestimated calcium intake (systematic bias +177 mg/day, P<0.001), but had no significant magnitude bias (r=-0.09; P=0.11). The Short Calcium Questionnaire showed minimal systematic bias (+34 mg/day, P=0.09), but had magnitude bias (r=-0.33; P<0.001). CONCLUSIONS: All three FFQs performed reasonably well at estimating dietary calcium intake compared to food records; each may be appropriate for use in select clinical and research settings.