Bone and Muscle Health in Kids (BONUSKids)

Vitamin D is a modifiable regulator of bone mineral metabolism and muscle function; yet little is known about the role of in utero vitamin D exposure on bone mineral accrual and muscle development in early childhood. Observational findings suggest associations between prenatal vitamin D status and infant bone, lean and fat mass and muscle function, but data from intervention trials is either unavailable or limited to early infancy. The investigators of the proposed study have previously shown that prenatal vitamin D supplementation safely improves maternal-fetal vitamin D status and perinatal calcium metabolism. Given the current interest in the potential role of vitamin D in perinatal health, adequately powered and well-designed randomized trials are required to investigate the effect of maternal vitamin D supplementation on offspring musculoskeletal health and body composition.

The investigators recently conducted the Maternal Vitamin D for Infant Growth (MDIG) trial (NCT01924013), a randomized placebo-controlled dose-ranging trial of vitamin D supplementation in pregnancy and lactation in Bangladesh, where severe vitamin D deficiency is common. Women at 17-24 weeks' gestation were randomized to 1 of 5 dose groups comprising a prenatal:postpartum regimen of placebo:placebo, 4200:0, 16800:0, 28000:0 or 28000:28000 IU vitamin D3/week until 26 weeks postpartum. Enrolment (n=1300) was completed in September 2015, and all infants were delivered by February 2016. The MDIG trial was primarily designed to determine the effect of maternal vitamin D on infant length at 12 months of age, with follow-up continuing until infants reached 24 months of age (completed in March 2018).

In the proposed observational follow-up study, the investigators will leverage the design and infrastructure of the MDIG trial to investigate effects of vitamin D on early childhood bone and muscle outcomes in a sample of MDIG participants at 4 years of age (target n=600 with 120 participants per maternal vitamin D treatment group; to allow for an expected attrition rate of 15%, the required sample size will be inflated to 140 children per group, giving an overall target sample size of 700 children).

The primary objective is to quantify the effects of prenatal vitamin D supplementation versus placebo on offspring total-body-less head (TBLH) bone mineral content and density. Secondary objectives are to examine the effect of vitamin D on TBLH lean and fat mass, and muscle function. Bone, lean and fat mass will be measured using dual-energy x-ray absorptiometry (DXA). Muscle strength (handgrip test), anthropometry, dietary intake and vitamin D status will also be assessed, in addition to biomarkers of bone turnover that may affect musculoskeletal health in young children.

The proposed follow-up study of a large randomized trial will contribute new evidence regarding the effects of vitamin D supplementation during pregnancy on musculoskeletal health in childhood.

Data analysis

Owing to the large contribution of head BMC values to whole-body BMC, in addition to difficulties in obtaining accurate head positioning without motion artifact in young children, TBLH BMC will be considered as the primary outcome in the present study. While TBLH BMC and TBLH areal BMD are closely related and complementary measures of bone mass, and both will be examined in the main analyses, TBLH BMC will be considered as the primary outcome measure. Whole-body measurements will be examined as secondary outcomes, in addition to BMC and areal BMD measurements of the head alone. DXA-derived estimates of lean and fat mass will also first be explored as TBLH measures, and later explored as whole-body measurements. Primary analyses will be conducted without adjustment for covariates; however, adjustment for concurrent body size, age or child sex will be explored in additional analyses.

Placebo versus high-dose prenatal vitamin D supplementation: To address the objectives with maximum statistical efficiency, the 2 high-dose prenatal vitamin D groups will be combined for a comparison of children of mothers who received 28000 IU/week of vitamin D prenatally (with or without 28000 IU/week of vitamin D postpartum) versus those children whose mothers received placebo prenatally. Differences will be expressed as mean differences with 95% CIs, and tested for statistical significance using t-tests. In sub-group analyses, we will stratify by baseline maternal vitamin D status (25(OH)D ≥ 30 nmol/L vs < 30 nmol/L) or infant sex (boys vs girls). A restricted analysis will also be completed, limited to term-born infants only (≥37 weeks' gestation).

Dose-response effect of prenatal and postpartum vitamin D supplementation: For the dose-response analysis including data from all treatment groups, we will regress each outcome (as a continuous variable) on the assigned dose of vitamin D (as a categorical variable) during the prenatal period (with the 2 high-dose prenatal vitamin D treatment groups combined). In this way, mean differences of each prenatal supplementation group will be compared to the placebo group, to test the effect of prental supplementation only. We will also regress each outcome on the average weekly dose of vitamin D assigned as a continuous variable during the prenatal period. Similar regression models will be conducted using more precise estimates of the vitamin D dose received, based on manufacturer analysis of the vitamin D composition of each batch of tablets provided and individual compliance with study tablets. We will use linear regression splines to accommodate non-linear outcome-dose relationships if warranted based on non-parametric visualization (e.g. LOWESS). To ensure precision of the 95% CIs, and therefore ensure robust inferences, 95% CIs from the linear regression models will be obtained using bootstrapping (with 1000 replications).

To explore potential effects of prenatal only versus prenatal plus postpartum vitamin D supplementation, a similar dose-response analysis will be conducted across all 5 treatment groups (i.e., disaggregation of the 2 high-dose prenatal vitamin D groups), for which we will regress each outcome (as a continuous variable) on the vitamin D treatment group assigned during the prenatal and postpartum periods (as a categorical variable). Mean differences of each supplementation group will therefore be compared to the placebo group. Estimates of the 95% CIs from all regression models will be obtained using bootstrapping (with 1000 replications).

Further exploratory analyses of primary and secondary outcomes: Multivariable regression models will enable adjustment for other potential determinants of bone mass, muscle function and body composition including: baseline covariates that differed by treatment group (if present) and thus may confound treatment effects; child characteristics at the time of assessment (e.g. vitamin D status); maternal characteristics (e.g. education level) and socioeconomic status of the household. We will include measures of height or weight (e.g., height-for-age Z-scores) as covariates to test their roles in mediating the effect of the vitamin D intervention on bone mass or other outcomes (muscle strength, body composition). Linear regression models will also be used to assess between-group differences in calcium metabolism and bone turnover markers.