Background/Aims: Timely identification and treatment of children with vitamin D deficiency is important. However, vitamin D status determined through serum 25(OH)D sampling can be painful for young children and is not routinely funded in New Zealand. Predictive questionnaire-based tools offer a non-invasive alternative. Our objective was to develop a questionnaire-based tool to predict vitamin D deficiency in NZ preschool-aged children.
Methods: Dried capillary blood spots were collected from 1329 children during late winter to early spring for 25(OH)D measurement by LC-MS/MS. Caregivers completed a questionnaire describing their child’s demographics and factors known to affect vitamin D status. Predictors of 25(OH)D<25nmol/L and 25(OH)D<50nmol/L were identified using multivariable logistic regression in a randomly-selected sub-sample (n=929) for development of two predictive tools, which then were validated by receiver operating characteristics (ROC) analysis (n=400).
Results: Serum 25(OH)D levels were <25nmol/L in 7% of children and <50nmol/L in 49%. The tool predicting 25(OH)D <25nmol/L had a sensitivity of 42%, specificity 97% and AUC 0.76 (95%CI 0.67-0.86,P<0.001). The tool predicting 25(OH)D <50nmol/L had a sensitivity of 52%, specificity 66% and AUC 0.63 (95%CI 0.57-0.68,P<0.001).
Conclusions: Despite the large sample size, the questionnaire-based tools were unable to accurately predict vitamin D deficiency in these preschool-aged children, suggesting that blood 25(OH)D testing be used to determine vitamin D status. Further research into safe and inexpensive methods of assessing vitamin D status in vulnerable populations is indicated.
Funding source: Health Research Council of New Zealand