RELATIVE VALIDITY AND REPRODUCIBILITY OF A CALCIUM AND VITAMIN D FOOD FREQUENCY QUESTIONNAIRE (FFQ) AMONG MALAYSIAN PRE-ADOLESCENT CHILDREN

Abstract

Background:

Calcium and vitamin D are essential nutrients for peak bone mass development and nutritional needs during early puberty in children. Hence the assessment of calcium and vitamin D intake and their food sources are important. There is a lack of valid dietary assessment tool available in Malaysia to measure these nutrients. Compared to 24-hour dietary recalls and food records, the food frequency questionnaire (FFQ) to assess calcium and vitamin D intake is considered more suitable for large-scale studies as it is simple to understand, easy to administer and impose less respondent burden. Thus, the objective of this study was to develop and assess the validity and reproducibility of a calcium and vitamin D FFQ in preadolescent Malaysian children. Method: The study was conducted in 2 phases: (1) development of FFQ and (2) relative validity and reproducibility of the FFQ. In Phase 1, the food intake data previously collected from 236 pre-adolescent children, aged 8 to 12 years old from national primary schools in Kuala Lumpur was used to develop the FFQ for this study. In total, 464 food codes were grouped into 71 food items based on conceptual similarities. The percentage contribution of calcium provided by a particular food item to the population’s total consumption of calcium was calculated using the equation reported by Block et al. The final calcium food list, which captured the top 90% of the total population’s calcium intake, was incorporated with a vitamin D food list and then developed into a 30-item semi-quantitative FFQ for calcium and vitamin D for this study. In phase 2, the FFQ was administered to 243 pre-adolescent children aged 8 to 12 years old from three national primary schools in Kuala Lumpur. Sociodemographic data were collected using a questionnaire. Total body bone mineral density (BMD) was measured using dual-energy X-ray absorptiometry (iDXA) (Lunar iDXA, GE Healthcare, USA). A total of 2 ml of blood was collected to measure serum 25-OH vitamin D by LC-MS method. The participants were interviewed using the FFQ for their calcium and vitamin D intake in the past month and a 1-week diet history (DH) was collected as the reference method. The participants repeated the administration of the FFQ at 6 and 12 months later to test reproducibility. Energy, macro- and micronutrients intake from the DH were analysed based on the Nutrient Composition of Malaysian Foods, Energy and Nutrient Composition of Food, Singapore and nutrient labelling by manufacturers. The validity of the FFQ was assessed by comparing it with the DH using paired-sample t-test, Bland-Altman analysis, cross classification analysis, weighted Kappa agreement and Pearson correlation coefficient. The reproducibility of the FFQ was tested by comparing the repeated administrations of the same FFQ using Pearson/Spearman correlation, intra-class correlation and Bland-Altman analysis. Pearson correlation was performed between the BMD and 25- OH vitamin D values against the FFQ at baseline. Results: After removing four energy intake outliers, a total sample size of 239 was analysed. The mean age of the participants was 10.1 ± 1.0 years and they were predominantly Malays (90.4%) and prepubertal at Tanner Stage 1 (94.6%). There was no significant difference in the intake of calcium and vitamin D from the FFQ vs DH (calcium FFQ: 330 ± 203 mg/day vs calcium DH: 347 ± 180 mg/day, p > 0.05; vitamin D FFQ: 1.6 ± 1.6 µg/day vs vitamin D DH: 1.4 ± 1.6 µg/day, p > 0.05) and the intakes from FFQ were significantly correlated with DH (calcium: r = 0.329; vitamin D: r = 0.223, p < 0.001). Weighted Kappa showed significant acceptable agreement between the two methods as well (calcium: 0.238, p < 0.001; vitamin D: 0.209, p < 0.001). Cross classification showed that on average 45% of the participants were correctly classified into the same tertiles by FFQ and DH (calcium: 45.6%, vitamin D: 44.4%), exceeding 33.3 % by chance. Bland-Altman analysis however had varying results of acceptable to no agreement between FFQ and DH (calcium: mean difference: -17 mg; LOA: -454mg, 419mg, p =0.047; vitamin D: mean difference = 0.1 µg; LOA: -3.7 µg, 4.0 µg; p = 0.804). Intra-class correlation coefficients (ICC) indicated that the FFQ was not reproducible (calcium: r = 0.258, p < 0.01; vitamin D: r = 0.358, p < 0.001) between baseline to 12 months. None of the pairwise comparisons of the FFQ reported an acceptable agreement for both calcium and vitamin D intake in Bland-Altman analysis, except for vitamin D intake between baseline and 6 months (vitamin D: mean difference = 0.1 µg; LOA: -4.3 µg, 4.0 µg; p = 0.648). Total body BMD was not correlated (r = 0.036, p = 0.577) with the mean intake of calcium from FFQ while serum 25-OH vitamin D was significantly correlated (r = 0.203, p < 0.01) with vitamin D intake from the FFQ. Conclusions: These findings indicate that the newly developed FFQ is valid for measuring the intake of calcium and vitamin D in this population of predominantly Malay pre-adolescent children aged 8 to 12 years old but not reproducible for long term intakes.