Incidence and predictors of high vitamin D in premature infants with very low birth weight

Introduction

VLBW infants are born with significantly decreased reserves of vital minerals and vitamins necessary for proper bone mineralization leading to metabolic bone disease (MBD) and risk development of rickets of prematurity. Clinicians have understood the connection between vitamin D and prevention of MBD in the newborn population for over 80 years, with formal guidelines first published in the early 20th century by the Institute of Medicine followed by the American Academy of Pediatrics (AAP) [1], [2], [3], [4]. Implementation of these guidelines, in addition to fortification of human milk and the use of preterm high mineral containing formulas, has led to the decreased incidence of radiographically defined rickets of prematurity from 50 % to 10 %–20 % in the extremely low birth weight population (birth weight < 1,000 g) [5].

The third trimester of pregnancy is vital for optimal mineralization of the fetal skeleton with as much as 80 % of the calcium and phosphorus transfer across the placenta occurring during this period [6], 7]. Preterm infants are at a further disadvantage as immature gut development further hastens absorption of calcium and phosphorus [8]. Vitamin D enhances the absorption of intestinal calcium by both dependent and independent mechanisms [9].

The AAP recommends 200–400 IU/day vitamin D for preterm infants [10]. International guidelines also recommend supplementation, however, at higher doses, ranging from 400 IU/day to 1000 IU/day, the latter recommended by ESPGHAN (European Society for Pediatric Gastroenterology Hepatology and Nutrition) [11].

These doses have been shown to prevent rickets of prematurity with additional reported benefits in reducing the risk of respiratory and autoimmune conditions later in childhood [12]. Wicklow et al. reports poor gross motor development at 6 months of age in infants that received higher doses vitamin D supplementation, however, due to low power, is unable to comment on an association with 25[OH]D levels. Literature is robust in regard to the clinical impact of vitamin D deficiency; however, it is limited with respect to vitamin D toxicity. Higher doses have been shown to increase serum concentrations of 25-hydroxyvitamin D (25[OH]D), however, there is an unknown benefit for skeletal or extra-skeletal health [13]. Hypervitaminosis D may lead to hypercalcemia, which can present with a range of symptoms, such as anorexia, abdominal pain, vomiting, constipation, polyuria, polydipsia, nephrocalcinosis, renal failure, and ataxia [14], 15]. There is a paucity of data regarding adverse outcomes of hypervitaminosis D in the VLBW population and the safety and advantages of increased doses and serum concentrations continue to be a subject of debate [16].

Few studies have reported the incidence of hypervitaminosis D. Even fewer, mostly small single-center studies, have evaluated protocols designed to maintain optimal 25[OH]D concentrations using different supplementation doses. Mathilde et al. reported a 33 % incidence of hypervitaminosis D when implementing a targeted 25[OH]D supplementation protocol. Scarcity of studies and limitations in power prohibit understanding of the relationship between 25[OH]D and extra-skeletal effects [17].

Our center launched a vitamin D deficiency screening algorithm for VLBWs in 2021 and have anecdotally noted hypervitaminosis D, leading us to study and better understand this phenomenon. We hypothesized that hypervitaminosis D among the VLBW population is more prevalent in lower gestational age (GA) and birth weight (BW).

Materials and methods

Subjects and eligibility criteria

This was a retrospective cohort study conducted among VLBW infants delivered between March 2020 and October 2023. We included newborns with a GA ≤ 32 weeks and/or BW ≤ 1,500 g. Patients meeting the above criteria but unable to provide specimens were excluded. An institutional algorithm for screening for vitamin D deficiency and supplementation and management of 25[OH]D was strictly adhered to (Figure 1). Prenatal vitamins were recommended for all mothers, but compliance was not assessed in this study. All patients receive 400 IU/day cholecalciferol. Human milk is used exclusively for this population, with donor human milk provided if mother’s own milk is unavailable. The protocol was approved by our institutional Feinstein Institutes for Medical Research Institutional Review Board.

Figure 1:

Institutional algorithm for vitamin D supplementation and management. MVI (multivitamin), TPN (total parenteral nutrition).

Results

There were 130 infants with at least one 25[OH]D lab values, who were included in this analysis, among the 148 observations. Sixty-three (48.46 %) were male, 97 (74.6 %) were born via Cesarean section and 33 (25.4 %) via vaginal delivery. Median (IQR) maternal age was 31 (27–36) years, while 43 (33.1 %) of mothers had diagnoses of hypertensive disorders and 19 (14.6 %) were diabetic. 79 (60.8 %) received a full course of antenatal betamethasone, while 51 (39.2 %) received a partial course or no steroid therapy. The median (IQR) for GA was 30.4 (28.2–31.6) weeks, while 14 (10.8 %) were SGA, 22 (16.9 %) were intrauterine growth restriction (IUGR). 96 (73.9 %) were diagnosed with respiratory distress syndrome (RDS) of which 45 (34.6 %) required surfactant, and the median (IQR) LOS was 40.5 (28–59) days.

Subjects were stratified into two cohorts (A) 25[OH]D < 60 ng/mL and (B) 25[OH]D ≥ 60 ng/mL. 60 (46.2 %) had hypervitaminosis D at least once during the hospital course. There were 13 subjects with insufficient 25[OH]D levels, defined as 25[OH]D levels < 30 ng/mL. There was a significant difference between GA and BW between the two cohorts (30.4 weeks vs. 29.3 weeks, p=0.002 and 1471.1 g vs. 1233.4 g, p = 0.001, respectively). Length of stay was significantly longer in patients with 25[OH]D ≥ 60 ng/mL (p < 0.0001). There was a significantly greater proportion of maternal HTN in cohort A compared to cohort B (44.3 % vs. 20.0 %, p = 0.003). Hypervitaminosis D was more common among patients who received mechanical ventilation (p = 0.011) and approached statistical significance in patients that required surfactant replacement therapy (p = 0.053) (Table 1).

IUGR was included in the initial screen for potential predictors of hypervitaminosis D. The initial screen was necessary due to the larger number of predictors, relative to the number of patients with hypervitaminosis D. The univariate logistic regression for IUGR resulted in p = 0.3151, and based on our predetermined rule (p < 0.20), it was not included in the multivariable logistic regression.

Additionally, we assessed this association using a two-way table of IUGR by hypervitaminosis D. Eight patients (36.4 %) had hypervitaminosis D among the 22 patients with IUGR, while 52 patients (48.2 %) among the 108 patients without IUGR, which corresponds to p = 0.3552, based on the Fishers’ exact test, confirming the earlier conclusion. Variables studied were considered for their plausibility to influence this outcome, and variables with p < 0.20, based on a univariate logistic regression, were included in a multivariable logistic model to assess their associations with the primary outcome. The variables selected were GA, maternal age > 35 years, maternal hypertensive disorders, RDS and need for surfactant replacement therapy. We subsequently investigated the association of the selected predictors with the primary outcome using multiple logistic regression. We excluded surfactant replacement therapy (p = 0.87) and RDS (p = 0.34) from the final model. Our finding indicates that increasing GA (OR=0.77, 95 % CI 0.64–0.92) and maternal HTN (OR = 0.32, 95 % CI 0.14–0.73) were associated with decreased incidence of excess 25[OH]D during the hospital stay (Table 2). We kept in the model the indicator for maternal age > 35 years (OR = 2.2, 95 % CI 0.96–5.06), due to its clinical importance. The association between 25[OH]D level and the calcium level was assessed using a linear mixed model for the 255 observations with both measurements available at the same time. We used a fixed effect for the time of the measurement and a random effect for subject to account for the correlated observation from the same subject. We did not find a significant association between 25[OH]D level and the calcium level (beta = 0.001, 95 % CI −0.001 to 0.003, p = 0.2534). There was not a statistically significant association between time and the calcium level (beta = −0.012, 95 % CI −0.042 to 0.018, p = 0.4325). Figure 2 presents a scatter plot of 25[OH]D level vs. calcium level by time.

Figure 2:

Scatter plot of vit D level vs. calcium level by time.

Discussion

In our study, we categorized 25[OH]D levels <30 ng/mL as insufficient and levels ≥30 ng/mL–60 ng/mL as sufficient. Hypervitaminosis D was defined as ≥60 ng/mL at any point during the hospital stay. In this study, hypervitaminosis D was a common finding in the VLBW population. While there has been significant focus on an optimal range for 25[OH]D for the prevention of rickets of prematurity, no consensus exists on optimal upper limit of serum 25[OH]D levels. 25[OH]D ≥ 60 ng/mL have been discussed as a risk factor for vitamin D toxicity [18], [19], [20], [21]. While significant data exists on vitamin D deficiency and prevention of rickets of prematurity in this population, a paucity of literature exists on the incidence of hypervitaminosis D and whether this phenomenon is associated with any clinical sequelae [17], 22]. This observation is similar to previous studies in VLBW infants which saw an incidence ranging from 33 % to 75 % [17], 23]. However, these studies used cut off below 50 ng/mL.

The current study was performed over several years which allowed for inclusion of seasonal influence on 25[OH]D, knowing there is a seasonal impact on 25[OH]D concentration [24]. 25[OH]D is transferred from mother to fetus early in pregnancy. Therefore, a mother with vitamin D deficiency will likely have a newborn with insufficient levels as well. It is well known that seasonal variations in 25[OH]D concentrations exist, therefore, mothers affected by seasonal variation will lead to neonatal seasonal changes in 25[OH]D levels. Our NICU environment includes minimal natural light exposure [25].

Literature is limited on the effects of various dose ranges. In this study, subjects received 400 IU/day of cholecalciferol, the recommendation by the AAP [10]. International guidelines also recommend supplementation, however, at higher doses, ranging from 400 IU/day to 1,000 IU/day, the latter recommended by ESPGHAN [11]. Fort et al. investigated the effect of different doses of vitamin D supplementation for the first month of life in ELBWs, concluding that higher doses (800 IU/day vs. standard dose 200 IU/day) resulted in the majority of patients developing hypervitaminosis D at 28 days of life.

Our study did not reveal any notable abnormalities in serum calcium, phosphate or alkaline phosphatase which happened to be a similar finding in a report by Niko et al. Although hypervitaminosis D often leads to hypercalcemia, this was not observed in our study, likely due to secondary mechanisms that tightly regulate calcium homeostasis. One other theory that may contribute to this finding is that vitamin D metabolism differs among preterm newborns, leading to different clinical and biochemical presentation during hypervitaminosis D [26].

Prior studies have demonstrated that vitamin D enhances the expression of alveolar surfactant protein B and vascular endothelial growth factor, suggesting a role in lung maturation and potential reduction in RDS severity. Reduced serum 25[OH]D levels have been observed in newborns with RDS compared to term controls, supporting an association between vitamin D deficiency and impaired respiratory outcomes [27]. However, in our study, higher 25[OH]D levels were paradoxically associated with the need for mechanical ventilation and surfactant therapy, contrary to existing literature, indicating that further research is needed to clarify this relationship. Moreover, clinical reports of vitamin D toxicity describe symptoms consistent with hypercalcemia, including irritability, poor feeding, growth failure, gastrointestinal disturbance, and vomiting [28], 29]. When applying our cut-off for elevated levels and evaluating multiple neonatal morbidities—including RDS, sepsis, urinary tract infection, necrotizing enterocolitis, and intraventricular hemorrhage—we observed that hypervitaminosis D was more common in patients requiring mechanical ventilation and surfactant replacement therapy. Notably, higher vitamin D levels were associated with lower gestational age and birthweight, with both differences reaching statistical significance (Table 1). Taken together, although evidence in neonates remains limited, existing studies in older pediatric populations suggest an inverse relationship between vitamin D deficiency and pulmonary disease, emphasizing the complexity of this association and the need for further research in neonatal cohorts [12].

Another important consideration is the absence of a standardized algorithm across NICUs for measuring 25[OH]D. Current practice is variable, often relying on individual provider preference rather than consistent criteria. There are no formal society guidelines recommending routine serum 25[OH]D monitoring in preterm infants, including from the AAP. This lack of standardization complicates the identification of infants at risk for deficiency or hypervitaminosis and underscores the need for evidence-based protocols to guide monitoring and supplementation in preterm infants.

Previous case reports have commented on prolonged use of vitamin D supplementation leading to hypervitaminosis D [30]. VLBWs often require prolonged lengths of stay, and therefore commonly receive prolonged periods of supplementation. Upon discovery of hypervitaminosis D, any additional vitamin D supplementation was discontinued. There were no episodes of hypovitaminosis D following discontinuation of vitamin D supplementation. 25[OH]D levels were not immediately repeated and were drawn again per the protocol described in the manuscript (Figure 1). Our study also noted that the longer the hospital course the more likely the risk of hypervitaminosis D. Maternal hypertension was noted as a protective factor against the risk of hypervitaminosis D. It is understood that during preeclampsia there is decreased perfusion to the placenta leading to intrauterine growth restriction [31]. During prolonged periods of uteroplacental insufficiency, it is possible that vital nutrients and minerals are suboptimal transferred to the developing fetus. Of concern to our population, the transfer of vitamin D may be greatly limited leading to a state of vitamin D deficiency at birth, reducing the risk of developing hypervitaminosis D later during hospital stay.

Conclusions

Hypervitaminosis D is common among VLBWs, and routine monitoring and adjustment of vitamin D supplementation is warranted. Maternal HTN was associated with reduced likelihood of excess vitamin D. The data suggest a possible association between advanced maternal age and hypervitaminosis D; however, this association was not statistically significant (p = 0.06). Hypervitaminosis D was more common among ELGANs (≤28 weeks), and patients who received surfactant replacement therapy and mechanical ventilation. This retrospective, observational, single center study had limitations. These included missing data, for example, when infants were discharged early when meeting discharge criteria ahead of routine milestones. Future studies should include partnerships with obstetricians to gain further understanding of the maternal laboratory profile, including maternal and cord serum 25[OH]D at time of birth. Further, multi-center studies are needed to determine optimal nutritional supplementation with longitudinal follow to understand the clinical impact of excess 25[OH]D on comorbidities in preterm infants. In addition to future multicenter studies, the establishment of standardized guidelines for 25[OH]D screening and supplementation in the NICU is warranted. Such protocols could reduce practice variability, prevent both deficiency and excess, and ultimately improve clinical outcomes in this vulnerable population.